Clive Best(JRC-CEC), Joint research centre of the European Commission, CCRISPRA, Italy, Clive.Best@cec.eu.int -- Clive_Best_CV -- Clive_Best_Research

Jens Christensen, Project Officer, European Commission DG XIII/F and DG XIII/C, Brussels, Jens.CHRISTENSEN@cec.eu.int --Jens_Christensen_CV

Terence Mayes (UK), Glasgow Caledonian University Department of Learning and Educational Development, J.t.mayes@gcal.ac.uk --Terry_Mayes_CV

Jacobijn Sandberg(NL), University of Amsterdam Dpt. of Social Science Informatics , Sandberg@swi.psy.uva.nl -- Jocobijn_Sandberg_CV --Jocobijn_Sandberg_Research

Pier-LuigivEmiliani(I), Institute of Research on Electromagnetic Waves "Nello Carrara" (IROE) National Research Council, Ple@iroe.fi.cnr.it

Herman Maurer(A), (GRAZ University of Technology) IICM, HMC (JR) and AWAC (ARCS, Hmaurer@iicm.edu, Hmaurer@iicm.tu-graz.ac.at --Herman_Mauer_CV, Herman_Mauer_Research

Nicolas Balacheff(F), CNRS-Laboratoire Leibniz-IMAG, Nicolas.balacheff@imag.fr -- Nicolas_Balacheff_CV-- Nicholas_Balacheff_Research

Bjoern Pehrson(S), Royal Institute of Technology, bjorn@it.kht.se

Rebecca Alden(B) ECOTEC Research and Consulting Ltd. Senior Consultant, Great Britain, Becky_Alden@ecotec.com -- Becky_Alden_CV -- Becky_Alden_Research

Matthias Rohrbach(D), Fraunhofer IAO MT Information Engineering, Mathias.Rohrbach@iao.fhg.de -- Matthias_Rohrbach_CV -- Matthias_Rohrbach_Research

Roberto Carneiro(P), Grupo FORUM, Roberto.Carneiro@forum.pt -- Roberto_Carneiro_CV -- Roberto_Carneiro_Research

Wilhelm Bruns(D), Bremen university – Artec Research centre Work & Technology, Bruns@artec.uni-bremer.de --Wilhelm_Bruns_CV -- Wilhelm_Bruns_Research

Mariono Sanz (E), Consorcio de los Quesos Tradicionales de España, Mariano.sanz@eun.org -- Mariano_Sanz_CV

Born in 1941 in Vienna, Austria. Study of Mathematics at the Universities of Vienna (Austria) and Calgary (Canada) starting in 1959. System Analyst with the Government of Saskatchewan (Canada) in 1963. Mathematician-programmer with IBM Research in Vienna 1964--1966. Ph.D. in Mathematics from the University of Vienna 1965. Assistant and Associate Professor for Computer Science at the University of Calgary, Canada, 1966--1971. Full Professor for Applied Computer Science at the University of Karlsruhe, West Germany, 1971-1977, and Visiting Professor at SMU, Dallas, and University of Brasilia (Brazil) for three months, each, and at the University of Waterloo, during the same period. Full Professor at the Graz University of Technology since 1978. In addition, director of the Research Institute for Applied Information Processing of the Austrian Computer Society 19983-1998, chairman of Institute for Information Processing and Computer Supported New Media since 1988, and director of the Institute for Hypermedia Systems of Joanneum Research since April 1990. Adjunct Professor at Denver University 1984--1988. Professor for Computer Science at the University of Auckland, New Zealand, in 1993 (on leave from Graz). Honorary Adjunct Professor since October 93. Honorary Doctorate Polytechnical University of St. Petersburg (1992), Foreign Member of the Finnish Academy of Sciences (1996).

Author of fourteen books, about 500 scientific contributions, and dozens of multimedia products. holder of patent for optical storage device. Editor-in-Chief of the Journals J.UCS and J.NCA. Chairperson of steering committee of WebNet and ED-MEDIA Conference series. Member of program committee of numerous international conferences. European Representative on ICCE.. Project manager of a number of multimillion-dollar undertakings including the development of a colour-graphic microcomputer, a distributed CAI-system, multi-media projects such as "Images of Austria" (Expo'92 and Expo'93), responsible for the development of the first second generation Web system Hyper-G, now Hyperwave (http://www.hyperwave.com) and various electronic publishing projects such as the "PC Library", "Geothek" and "J.UCS" and participation in a number of EU projects (e.g., LIBERATION, HYMN, EUROPE-MMM, HYPDOC, EONT, and others).

Main research and project areas: networked multimedia/hypermedia systems (Hyperwave); electronic publishing and applications to university life, exhibitons and museums, Web based learning environments; languages and their applications, data structures and their efficient use, telematic services, computer networks, computer assisted instruction, computer supported new media, and social implications of computers.

Graz, Fall 1999

Thomas Dietinger,
Institute for Information processing and Computer supported new Media (IICM),
Graz University of Technology, Austria, (tdieting@iicm.edu),
Austrian Web Application Center (AWAC), Austrian Research Centers Seibersdorf (tdieting@awac.at)

Christian Eller,
Graz University of Technology, Austria, (eller@zid.tu-graz.ac.at)

Christian Gütl,
Institute for Information processing and Computer supported new Media (IICM),
Graz University of Technology, Austria, (cguetl@iicm.edu)

Hermann Maurer,
Institute for Information processing and Computer supported new Media (IICM),
Graz University of Technology, Austria, (hmaurer@iicm.edu)

Maja Pivec,
Faculty of Mechanical Engineering, University of Maribor, Slovenia (maja.pivec@uni-mb.si)

Institute for Information processing and Computer supported new Media (IICM),
Graz University of Technology, Austria (mpivec@iicm.edu)

1. Introduction

 

A system called GENTLE (GEneral Networked Training and Learning Environment) [Dietinger et al. 1998a] is currently being developed at the Institute for Information Processing and Computer supported new Media which deals with many aspects of web based training. Its main purpose is to provide an integrated platform for students and teachers that offers all functionality that is required for web based training, like communication and collaboration tools, course and user administration, etc. Most of this functionality has already been implemented or is nearing completion. However, we decided not to develop a new authoring application for course modules because there already exist numerous good ones on the market. Our intention is to improve integration of those tools and even more important to provide methods for easy administration and reuse of already existing materials: The creation of high quality courseware is still quite cumbersome and expensive and superfluous work like recreating an already available module should be avoided.

 

The demand on such a module repository is flexible categorization and thus ease of location of required information and modules, in combination with simple usage. Especially efforts that can only be accomplished by IT and domain experts have to be reduced to an absolutely required minimum and substituted by automatic methods like computerized generation of content description, or incidental tasks that do not bother users. It also has to be considered that not all data may reside within the system, but can also be found externally at a different place or just somewhere on the Internet. This means that the system does not have control of outside modules and thus might not modify or add e.g. meta data. Additionally, external information may be dynamic and change without notice. Nevertheless, it might be valuable to include this kind of information because of topicality and quality.

 

Another important feature is personalization: by this we mean that a group of users (or a single user) may have their own view about the meaning of some modules, because the same data can be used differently in various situations: E.g. a chapter about multi media concepts might be introductory or advanced material, depending on the course topic. This can only be accomplished if situation dependent meaning and object oriented properties like title, creation time, author etc. are separated, because the first one can change with the point of view. Adaptive views can also be used to automatically suggest certain modules to authors if they specify a course profile that describes the meaning and aim of the course by e.g. a table of contents etc. The system can achieve this by trying to match the profile. This information about the course can also be used for suggesting suitable courses to students, if students also own a profile about their preferences.

 

The whole system could not only be used to find appropriate course components but also for retrieving relevant information in more general terms, like a background library for students or within a knowledge management or intranet system to be used by employees. In the following chapters we will examine already existing strategies and present a concept for an adaptive, associative, joint, dynamic and location independent module and information repository.

 

 

2. Meta Data Alone are Not Sufficient

 

The most obvious storage of categorized data is to save it in a data-warehouse with a static hierarchical structure similar to the file system organization of a PC. For example the History of Hypertext and data about HTML are saved in the Hypertext subcategory, which belongs to Hypermedia that is categorized under Computer science. Such inflexible data organization brings many disadvantages. The static keyword categories must be very precisely specified and regularly updated and reorganized, so that categorizing and storing of data can be efficiently supported. Otherwise stored data cannot be found again because they have not been categorized properly. Such a complex structure results in poor performance.

 

A possible solution to avoid the inflexibility of mentioned hierarchically structured data-warehouse is the usage of metadata. Nowadays, there are many standardization initiatives e.g. Dublin Core, Warwick framework, LTSC, IMS and others, trying to define the metadata structure [Weibel et al. 1998][Daniel et al. 1998][Hodgins & Wason & Duval 1998]. The standardization efforts are coming from different backgrounds e.g. technology suppliers, aviation industry, learning technology users, etc.. Because all of the different initiatives are aware of the great importance of the metadata for categorization, structuring of knowledge and finding relevant information, they try to be compliant within reasonable boundaries of other initiatives.

 

One of the definitions still being discussed within IEEE Learning Technology Standardization Committee (LTSC) is that Metadata are data about data, meaning that metadata are providing us with information about certain objects e.g. documents. This additional information about the objects can be information about the author of the object, language used, date of creation, topic, key words, possible use of the object, conditions for use, etc. We can add supplementary attributes that explain e.g. different possibilities to use the object, pedagogical values or other categories like learning style, learning level and prerequisites, which are relevant for re-use of the objects within the learning environment. Thus, metadata can be seen as a feature that facilitates finding of specific requested information (text documents, pictures, video clips, course units, etc.).

 

Metadata compared to the first solution mentioned in this chapter introduces tremendous improvements for the categorization and knowledge management process but there are still some problems left regarding the gathering of relevant, reliable and up-to-date information. In the categorization process different key words are assigned to the object. Exactly which key words provide the most appropriate description of the content or meaning of the object is the decision of the person, creating the metadata. This leaves the difficulty of finding the required object to the user, because it may be described in a different way than expected. The fact that equal key words have different meaning within different contexts is also causing some problems for users, who want to find a specific object. A possible way for useful and sufficient metadata can be probably achieved by combining human knowledge and technology. The disadvantage of metadata being rigidly coupled with the object can be avoided. Grouping of metadata offers a solution to that problem. Grouping of metadata means that metadata which describe physical appearance of the object stay coupled with it. Other metadata which describe some more general concepts e.g. topic, category, key words can be transformed into separate objects. Those meta-objects exist independently and can be linked to many different objects. Meta-objects can be shared within many objects and relations can be defined between meta-objects themselves. One possible implementation of this concept is described in the next chapter.

 

 

3. The Concept of Knowledge Clusters

 

A cluster usually describes a collection of objects that are relatively close to each other according to a chosen criterion. Cluster analysis is used for grouping similar, related objects. One possible application of clustering algorithms is document retrieval. In a dynamic environment content changes are made also in the content indicators attached to the documents. In such cases clustered file organization is preferable because similar sets of content identifiers are automatically grouped into common classes or clusters. The search is performed looking only at those clusters that exhibit close similarity with the corresponding query identifiers. A clustered file produces fast search output.

 

In real life, concepts are never isolated nor very simple, so clusters may overlap allowing that data, documents or part of documents (e.g. sections) belong to more than one cluster. To get better results we can use conceptual clustering. The difference between conventional and conceptual clustering is that in conceptual clustering the entities are grouped based on a conceptual cohesiveness (e.g. set of neighboring examples) [Michalski & Bratko & Kubat 1998]. Unlike statistical clustering methods, these algorithms rely on a search for objects within same or similar concepts. As an example of conceptual clustering we can use the term virus. If we search for virus we can get as a result different documents which describe the topic in computer science or documents which deal with viruses in medicine. Using the conceptual clustering, those documents are put into different clusters because they belong to different concepts.

 

What we need to accomplish is a network of meta data, where each meta data object correlates on one hand with the information it describes (meta data and data may also physically be the same object) and on the other hand with other instances of meta data. We suggest to use pure meta data (without a document content), so called base terms, which are related to other base terms to describe main concepts and thus work as a seed for new clusters. To improve effectiveness and versatility relations themselves should also contain some meta data: A type that specifies what kind of relation exists between two nodes (e.g. sub- and super-concept, cause or result, opposite or synonym, prerequisite, introductory, etc.), a weight value that specifies to what degree this type applies to that relation (e.g. fuzzy values like perfect, good, average, bad etc. expressed by a certain percentage) and a quality value that specifies the reliability of the connection (also percentage). In this way users can give feedback about the correctness of the relation during browsing and searching of the cluster and thus influence weight and reliability of the relation. This has the advantage that a new relation need not be completely correct right from the beginning, but can converge to a commonly accepted status by collaborative voting. Thus it is not important any more whether the creator of the relation is completely trustworthy or not, it just influences the starting reliability value. That means that not all connections have to be created by domain experts but can also be created by other e.g. novice users or algorithms. If we combine these relation attributes with access rights like private, group and public access and ownership even different points of view of concepts are manageable because relations can be seen differently by different users.

 

Similar approaches of sharing ideas and opinions within the society and building the contacts between people who have related interests, opinions, is described in Automated Collaborative Filtering (ACF) and Semantic Transports of information [Chislenko 1997a][Chislenko 1997b]. The semantic transport of information can be seen as a social tool, overcoming the side effects of the present-day individual isolation. Implementations of some fragments of these ideas can be also found in Recommender Systems [Resnik & Varian 1997] such as Alexa (http://www.alexa.com) and Phoaks (http://www.phoaks.com/) or in new generations of search engines like Google (http://www.google.com).



Figure 1: Examples for relation meta data

 

To avoid a possible confusion we would like to define what we understand by using the term knowledge cluster: The combination of a knowledge broker point (KBP), expert knowledge and collaborative community system as well as the automatic processing modules can be seen as a knowledge cluster. The task of knowledge broker points is managing static and dynamic knowledge sources within the learning environment. An automatic process module e.g. a knowledge cluster can help to categorize information entities, to position them as new pieces of knowledge in the system, and to store and to find them when necessary.

 

 

1. Creation of Knowledge Clusters and Addition of Documents

 

Typically, the creation of new knowledge clusters will be started by defining a skeleton of related base terms describing a certain knowledge area by domain experts. As all relations created by trusted experts get assigned a maximum reliability value, the quality of the network is more important than its extent and number of nodes. Additionally depending relations like "A is an introduction to B" and "B is an introduction to C" can dynamically be calculated by multiplying its weight and quality values and need not be created manually [see Figure 2].

 



Figure 2: Determining the total weight

 

However, if in special cases dynamic relations do not result in meaningful values from some point of view, they can be shortened by static relations which have higher priority than the calculated ones but e.g. might not be seen by everybody, because access rights for these relations are restricted [see Figure 3].

 



Figure 3: Modified shortest path rule

 

In this context we speak about a modified shortest path rule because here the shortest path is the one with fewest relations and not with the smallest edge length. It should be noted that in some instances the calculation of two paths of the same length (equals the number of relations between nodes) and same starting and ending node can lead to different results. However, this is absolutely correct because different paths also mean different contexts or conclusions [see Figure 4]!

 



Figure 4 : Context dependant paths

 

New base terms, meta data (including information content such as documents and course modules, or pointers to data like references) and also relations between new base terms and meta data cannot only be created by experts but also by ordinary users or even algorithms like user profiling, content analyzer, etc. The quality of these relations then depends on the status of the user. E.g. an anonymous user or a not too trustworthy algorithm only leads to quality values that are starting quite low, whereas identified users who have gained a certain expertise within the specific knowledge area may create more promising relations.

 

Whenever users browse or access a knowledge cluster they should have a possibility to provide feedback whether the visited relation proved helpful and they thus found what they were looking for or whether the relations had been misleading. Such a feedback could be given just by clicking on a plus or minus sign, indicating approval or rejection, or by a more complex form where the user has the possibility to suggest a different type, weight and quality values. The feedback can also be provided by a profiling algorithm that just analyzes behaviors of users. For examples, users starting all over again with a search after following a lot of relations express with this behavior that they did not find what they had been looking for and thus that these relations in their context had not been very helpful for them. In each case, the influence of the feedback on the attributes of the relation again depends on the trustworthiness of the user or algorithm which provides the feedback. Identified users can gain higher trustworthiness in a certain knowledge area if relations and objects they created for this area get higher ranking through feedback. The effect of this concept is that the whole cluster as well as quality values of contributing users and algorithms converge into a commonly accepted status. This means that newly created relations need not be perfect right from the beginning but are 'polished' on usage. In this way, also the effectiveness of an algorithm which automatically creates new structures only has an influence on the duration as long as the resulting relations evolve to a high quality network influenced by users.

 

 

2. Searching and Browsing Within Knowledge Clusters

 

Access to a knowledge cluster starts by defining an entry point to the data structure. This can either be a common root term or more likely a traditional keyword query on the meta data stored within the base terms and relations. The list of search results can be linearly sorted by priority or even better displayed as several clusters or clouds of documents and relations that indicate which terms belong together and what are their neighboring nodes. From now on users may browse through the clusters by putting the focus on a selected term or altering the filter values. The filter specifies how many and which adjacent nodes and documents will be displayed and thus form a new cluster. We have the following possible filter values: type of relations (e.g. introductory), threshold values for weight and quality, number of relations displayed, visibility expressed by maximum length of path that is dynamically calculated and of course also document meta data like mimetype, creation time, author, e.g. Dublin Core attributes etc. The filter can also be used for the very first display of search results. As mentioned before users can give feedback to the displayed relations or add new ones during the browsing process.

 

Another interesting feature is the so-called rucksack which is used to store documents for later reuse or may be used to execute more complex filtering by combining items collected in the rucksack as additional filter criterions such as 'all items displayed in the neighborhood have to be a sub concept of rucksack item X and an introduction to rucksack item Y' etc. An intelligent rucksack could also log or analyze the behavior of users and define a certain task and user specific profile which can be used to foresee the next action and suggest suitable filter settings and items. Certain rucksacks or more generally speaking agents can also be reconfigured for a specific task like chapter creation (e.g. a chapter consists of an introduction, a main part and a summary) or even the creation of new clusters.

4. The Prototype

 

The prototype, like the rest of the WBT system is based on the Hyperwave Information Server (http://www.hyperwave.com). In fact the concept and features of Hyperwave proved to be very helpful for this special demand, since it supports an object-oriented approach for storing documents and hyperlinks (so-called anchors). All objects within the database, including anchors, may store arbitrary attributes like keywords, owner, access rights etc.. This provides the possibility to use anchors for creating our specialized relations, because the required meta data such as type, weight and quality can be stored within the object. Further anchor-object relations are bi-directional, which means an application can find out which links are pointing to an object and which are pointing from this object to a different one which is also an important requirement. Hyperwave objects without content (like a Remote Object) have been used to implement base terms or external meta data that just have pointers to their accompanying documents. Personalization has been easily added by using access rights which are also applicable to anchors and which are used to specify which user(s) may see a relation.

 

Currently our prototype supports the creation of base term skeletons, adding of documents and personalized relations and simple querying and filtering during cluster browsing. Collaborative voting and the rucksack feature will be added soon. Thus the prototype can be used for jointly administrating a module repository or a reference and background library. Another simple application for the prototype would be shared bookmark archives, which might be helpful when doing task oriented teamwork like student exercises etc..

 

 

5. Conclusion and Future Work

 

As we have seen the concept of applying attributes to relations that interlink meta data provides a possibility to add additional meaning to course material which is a basic prerequisite for further automatic processing. Combined with situation and user dependent views and collaborative rating techniques it can be used as a fundamental model not only for managing courseware repositories and background libraries but also for administrating all kinds of related data. One of the next steps will be to examine the usefulness of this concept for adaptive tutoring which could be realized by courses that are generated on the fly and depending on users preferences. Future work will also include the integration of our ideas about dynamic background libraries and intelligent knowledge gathering [Dietinger et al. 1998b][Dietinger et al. 1999] to expand the model to a generalized and distributed concept.

 

 

6. References

[Alexa] http://www.alexa.com

[Chislenko 1997a] Chislenko A.: Semantic Web - a vision of the future of intelligent Web, (June 1997), http://www.lucifer.com/~sasha/articles/SemanticWeb.html

[Chislenko 1997b] Chislenko A.: Automated Collaborative Filtering and Semantic Transports, ideas of using the next generation of information technologies for automated meaningful distribution of knowledge, (October 1997), http://www.lucifer.com/~sasha/articles/ACF.html

[Daniel et al. 1998] Daniel R. Jr., Lagoze C., Payette S.D.: A Metadata Architecture for Digital Libraries. http://www2.cs.cornell.edu/lagoze/papers/ADL98/dar-adl.html

[Dietinger et al. 1998a] Dietinger T., Maurer H. (1998) GENTLE – General Network Training an Learning Environment. Proc. of ED-MEDIA (1998), Freiburg, and http://wbt.iicm.edu/gentle/papers/edmedia98.pdf

[Dietinger et al. 1998b] Dietinger T., Gütl Ch., Maurer H., Pivec M., Schmaranz K. (1998) Intelligent knowledge gathering and management as new ways of an improved learning process; Proceedings of WebNet98, Orlando, Nov. 7-12, 1998, Florida, USA, p244-249 and http://wbt.iicm.edu/gentle/papers/webnet98.pdf

[Dietinger et al. 1999] Dietinger T., Gütl Ch., Knögler B., Neussl D., Schmaranz K. (1999): Dynamic Background Libraries - New Developments in Distance Educations Using Hierarchical Interactive Knowledge System; J.UCS Vol 5/No. 1, p2-10

[Hodgins & Wason & Duval 1998] Hodgins W., Wason T., Duval E.: Learning Object Metadata (LOM) v2.5a, Dec. 23 1998, IEEE Learning Technology Standards Committee (LTSC), http://ltsc.ieee.org/doc/wg12/LOMdoc2_5a.doc

[Michalski & Bratko & Kubat 1998] Machine Learning and Data Mining Methods and Applications (ed.: Michalski R.S., Bratko I., Kubat M.), John Wiley & Sons Ltd., 1998.

[Resnik & Varian 1997] Resnick, P.,Varian, H., "Recommender Systems", Comm. of ACM, Mar.1997, Vol.40 No.3, p.56-58

[Weibel et al. 1998] Weibel S., Kunze J., Lagoze C., Wolf M.: Dublin Core Metadata for Resource Discovery, The Internet Society, September 1998, ftp://ftp.isi.edu/in-notes/rfc2413.txt.

A critical look at Web Based Training efforts

H. Maurer

IICM, Graz University of Technology, Austria

hmaurer@iicm.edu

 

 

Abstract

When looking at the large number of attempts in Web Based Training (WBT), i.e. efforts to use Intranets or the Internet for educational or training purposes, it is noticeable that many do not go beyond providing pretty Web pages and a bit of communication. There are some more general approaches trying to provide integrated learning environments: GENTLE, (a General Networked Training and Learning Environment), developed at the author's institute, is one such approach. Although such systems are a major step forward concerning WBT we claim that WBT must not be seen in isolation but rather as a part of a more general area of activity, knowledge management (KM). In this paper we discuss what we mean by KM, that WBT is really nothing but knowledge transfer (KT) and hence part of KM, and what a typical KT system like GENTLE should offer in terms of functionality.

 

 

1. INTRODUCTION

 

In this paper we show that Web Based Training (WBT) in the sense of "using networks to improve the efficiency and timeliness of knowledge transfer" must be seen as part of what is now often called Knowledge Management (KM).

 

Knowledge in the sense as we commonly use it resides in the brains of people. It is self-evident that within any group of persons the combined knowledge of this group is much larger than the knowledge of any individual. Yet, although we talk about this combined knowledge or ‘group knowledge’it is really fictitious: since human beings (unfortunately ?) have no direct brain-to-brain connection (and even their communication channels are severely handicapped due to a missing organ [1]) group knowledge is fragmented between the individuals, persons often not knowing or understanding what is clear for others. After all, this is why we have teaching and learning processes where in some laborious fashion knowledge is transferred to some extent between individuals, be it teacher to learners or be it in a working group between ‘equals’. This is also why we write memos, reports and books: we try to share knowledge with others. This is also why we have seminars, discussions and why we talk to each other. In the past, it was accepted as a sad fact of life that there is no easy way to increase knowledge of individuals and that no easy route to efficiently share knowledge exists.

 

However, the pressure to more efficiently transfer knowledge to obtain highly qualified people that stay at the edge of the state of the art throughout life, and that can acquire new knowledge (=learn) whenever the need arises has much increased over the years, mainly due to the acceleration of growth of information (knowledge ?) and partly due to high cost of education, training and re-training. Also, the fact that persons within an organization do not know what others are doing or what they know has lead to much waste and duplication of efforts: it is in this connection that the following frustrated sigh of the chairman of the board of a large international group can be understood: ‘If our employees only knew what our employees know, we would be the top leading company’. There is yet another development that is constantly accelerating: in traditional organizations their main assets have been property, buildings, machinery, inventory, etc.; in high-tech (high-brain) organizations the main asset has become more and more the knowledge in the brains of employees. Putting this together, techniques for increasing, transferring and archiving knowledge will be crucial factors for high performance organizations, be it companies, governmental organizations, teaching institutions, or what have you.

 

It is this problem area that Knowledge Management (KM) including Knowledge Transfer (KT) is trying to tackle. Thus, the basic aim of KM is to nurture and to increase the knowledge of individuals and to make sure that knowledge can be easily shared with others and (at least to some extent) remains even if the persons involved become unavailable. Hence, KM has a strong human and organizational component; however, for knowledge sharing, knowledge transfer and knowledge archival it is also necessary to computerize knowledge as much as possible. Thus, KM has also a strong IT component involving the transfer of (parts of) human knowledge (HK) into computer systems, techniques that assure that computerized knowledge (CK) is interlinked, is enriched continuously and that persons can extract the computerized knowledge through queries or more formalized processes to again use it as human knowledge.

 

 

2. IS KNOWLEDGE MANAGEMENT (KM) MORE THAN A BUZZ-WORD?

The most basic issue when talking about KM is surely: what is "knowledge"? After all, we have heard (in historic sequence) about data-processing, information-processing and now knowledge-processing. Some people might claim that a new buzz-word "KM" has been created, without really new concepts or techniques behind it. It is crucial to understand that this is not the case, that KM is indeed a new and remarkable phenomenon. To explain this it is best to start with the definition of knowledge. Outside IT, knowledge is often defined as "digested information" (= human knowledge). To be specific, if you buy a book, you have acquired information; once you have read and understood (= digested) the book, it is knowledge.

This definition has a severe drawback from the point of view of IT: knowledge (by definition) resides only in the brains of people, but can never reside in computer systems. Another definition of "knowledge" is computerized knowledge, more precisely "structured, interlinked and continuously growing information, the growth due to interactions with people". This second definition is much more specific from the IT point of view; it is clear that techniques to improve the management of knowledge defined in this fashion can be developed both by organizational changes and by suitable IT tools and can be potentially measured in whatever organization is at issue. However, critics of KM have tried to argue that KM defined as "computerized knowledge" is just glorified information processing and does not justify a new term, let alone a new concentrated R+D effort. What is overlooked in such arguments is that the two definitions are actually closely related: "knowledge" in the sense of "human knowledge" can be mapped onto "knowledge" in the sense of "computerized knowledge": I.e., the "real" knowledge residing in the heads of people (HK) can be represented by a powerful "shadow" knowledge (CK) residing in computer.

 

The challenge of KM is to make this "shadow" as "close to the real thing" as possible. The main credo of modern KM is that this is possible if sophisticated techniques are used that allow for the creation of CK from HK in a variety of ways, that allow users to re-create HK from CK (even system initiated) and that permit users not just the passive retrieval of information, but the structuring, linking, annotation and "massaging" of such information using a variety of tools that are becoming visible on the horizon. In addition to actively structuring and interlinking information (thereby creating new "views", new "associations", new "knowledge structures") users must be able to increase knowledge by "communicating" with documents: when users find whatever information, they should be able to ask arbitrary questions that are answered directly by the system. This feature might sound like science-fiction on first reading, but it can be accomplished to a large extent by a number of novel techniques; the most powerful one is using the fact that a large information base is used by many persons; hence, if a question asked cannot be answered by the system it will be answered (possibly later) by some human expert. However, and this is the crucial idea, once it has been answered this question/answer dialogue has enriched the system, hence the system is potentially capable of answering the same or a similar question posed by other persons later automatically. We believe that this approach is the first that interweaves the idea of an information/knowledge base that keeps growing because of user observation and of interactions (including questions) of users with the usual more mundane aspects of KM.

 

Lest we overstate the potential of KM: many areas of knowledge are understood so little that there is currently no hope to formalize them, and hence no hope to computerize them. For example, to quote [14] ‘the mental mechanisms involved in planning are not fully understood and cannot be fully automated.’ This is why we cannot map HK as such into CK, but just a mere ‘shadow’. However, as will become quite clear, much can be achieved nevertheless.

 

 

3. A GENERAL MODEL FOR KNOWLEDGE MANAGEMENT (KM)

 

Many sources such as OVUM [8] describe KM as an area with much growth to be expected. Many approaches to KM, where the importance of information technolgoy versuch organizational aspects varies are e.g. described in [9]. Two short introductions on KM are [10] and [11].

 

Concentrating on the information technology aspects our view of KM as allowing the ‘archiving, expanding, sharing and transfering’ of knowledge within groups and how ‘knowledge’ in the heads of people (‘human knowledge’ – HK) can be mapped into ‘shadow knowledge’ (‘computerized knowledge’ – CK) but can thence lead to new human knowledge (and thus expanding the knowledge sharing cycle mentioned previously) is expressed best by Fig. 1. The labels of the arrows represent different functions as will be explained below.

 

---

---

 

---

 

 

 

 

---

 

 

 

 

 

1. Figure 1: Model of a KM System

 

 

Fig. 1 shows that in a KM system humans and computers are involved. Knowledge as such resides as Human Knowledge (HK) in the heads of persons; parts of this knowledge can be made to reside as Computerized Knowledge (CK) in a networked computer system. It is sometimes also called "shadow knowledge" – since it is just a weakish image of the "sums of human knowledge".

The arrow labeled 1 in Fig. 1 corresponds to direct human-human knowledge sharing; the arrows 2-4 correspond to converting HK into CK; arrows 5-6 do the "opposite", i.e. make CK available to humans, while the arrow labeled 7 increases CK "automatically".

 

The arrow labeled 1 subsumes traditional knowledge sharing, be it over a cup of coffee, a seminar, a classical training course, just a phone-call, etc.

 

The arrows labeled 2 correspond to the situation when users explicitly convert some of their HK into CK by explicitly inserting a document, by authoring modules of courseware, by writing an annotation, creating a link or creating a "new view" of existing documents. Modern systems support such functions to various degrees. They play a particularly important role where classical WBT (i.e. knowledge transfer, KT) is involved.

 

Due to the costliness of explicitly converting HK into CT other ways of doing this are also necessary. The currently most realistic and implementable one is implicit knowledge creation, i.e. to assure that activities that persons are carrying out, anyway, (like writing a report, an email, or minutes of a meeting) are automatically stored (with sufficient meta-data) as CK. This approach is represented by the arrows labeled 2 in Fig. 1.

 

The most sophisticated way of obtaining CK form KH is by either "communication monitoring" or by "user observation". In the first case, relevant parts of discussions in emails, chats, news groups or the like are extracted and turned into CK. In the second case, actions of users are observed and recorded as "templates" for later use by other persons. It is clear that the system has to have sufficient "cleverness" to create knowledge this way. Indeed there is still a third way that the system can create knowledge: at a point in time when the user is, hopefully, not distracted by such action, the system asks a question to the user to expand its CK. To give a concrete example, when a user enters "coffee house" as synonym to "coffee shop" in some list the system may ask: "Should I also make "coffee house" and "coffee shop" synonymous in other dictionaries?" All such systemic acts of knowledge creation are symbolized by the arrow labeled 4 in Fig. 1.

 

The arrows labeled 5 in Fig. 1 represent situations when persons are explicitly asking for knowledge. This can be in the form of queries of various types; in the form of "goal oriented learning" (where a user asks for the explanation of a complex issue and - by asking questions - the system determines the necessary sequence of explanatory modules suitable for the knowledge level of that user); or in the form of a "courseware" package (see Chapter 4) possibly customized for the user at issue.

 

The arrows labeled 6 in Fig. 1 represent systemic actions: by observing a user the system decides to offer knowledge from its CK, assuming that this will help the user. This case (like the case of systemic actions discussed earlier) is the most open-ended one, and opens a wide area of research.

 

The arrow labeled 7 in Fig. 1 represents also systemic actions, but his time inside the CK: based on various techniques (semantic dictionaries, meta-information, user observation, usage patterns, etc.) the system autonomously increases its CK over time. In particular, questions asked by some user and (possibly later) answered by others are a valuable source of continuous enrichment of the CK. Thus, although information is the substratum of transferring, archiving and creating knowledge, knowledge is much more than passive information in isolation: KM has to do much with interaction, collaboration and a community that keeps enriching information content in a specific context. This is particularly true of KM in corporate environments, a point detailed a bit more in [15].

 

 

4. THE ROLE OF KT IN KM

 

Observe that the CK can enrich itself without complex systemic actions by creating links (based on keywords and such), tables of contents, etc. and by carrying out question/answer book-keeping. I.e., when one person asks a question which is answered by someone else (possibly much later) the system records this and makes the answer automatically available when a similar question is asked. The research issues here are to solve the "similarity" problem better and better as time goes by and to even deal with situations when users do not ask questions since they do not realize that they have problems.

 

Note, also, that once such features and all kinds of background information are available in a KM system, and this system allows synchronous and asynchronous communication plus private and collaborative annotations many of the features of a GEneral Networked Training and Learning Environment for Knowledge Transfer are already in place particularly if the CK comes with explicitly tailored courseware modules, see [5], [7], [12], and [13].

To implement KM a number of facts have to be observed: first, with the spread of WWW users should have a WWW feel, i.e. must be able to work with their usual browser: accessing the KM must be as easy as accessing an arbitrary WWW server; second, however, the server accessed must have a host of possibilities not found on ordinary WWW server: it must allow user authentication on various levels, must allow (full-text) searchable background archives including non-HTML documents such as PDF, PowerPoint or Winword files, it must supply a host of data-base facilities, version control, and much more.

 

To act as platform for a GEneral Networked Training and Learning Environment for Knowledge Transfer in the sense of serious instruction, training and learning it also particularly requires strong communicational features (like integrated mail, discussion forums, chats); it is of crucial importance that individuals can annotate and link together information for themselves or members of a group, it needs the facility to search in background documents ("digital libraries") as discussed above but including electronic books (e.g. technical dictionaries) and journals supplied by publishers, and it particularly needs the facility that question/answer dialogues between users (in this case mainly between learners and experts) are recorded for later use. However, KT has also the need for very specific components absent in other KM applications: much CK ("courseware") has to be purposefully explicitly created. This requires tools (‘authoring tools’, ‘module-reuse’-techniques, special sets of meta-data, etc.) to support the process; the whole area of student performance analysis up to exams or certifications that could lead to privacy issues in other KM-applications is critical for the success of KT; synchronous communication facilities (e.g. audio- or video-conferences) play a prominent role in KT; the process of author/teacher/tutor/student/course administration is intrinsic to KT; etc.

 

The KT-demand (often mentioned under Web Based Training, Distance Teaching, and many other terms) is huge, not just or not even mainly for routine educational institutions but mainly for companies for on-the-spot, on-demand, in-time, goal-specific training, learning, or just re-learning. Observe also that a certain amount of KT is existing in any KM system! Thus, although KT is a logical extension of KM it does require a substantial amount of further functionality.

 

KT in systems used for serious training or learning purposes requires special courseware modules and some special learning strategies. The most common strategy is "step by step" learning: much as in school, material is presented to learners (and hopefully internalized by them!) with no specific goal in mind except understanding material in a certain area (mathematics, bookkeeping, cooking, or whatever). In the "goal oriented approach" a certain specific aim is formulated and the system computes and offers a "minimal learning path" to achieve this aim: this technique is much more useful for on-demand and on-the-spot learning or re-learning. Both techniques suffer from the fact that working through the material is often boring. Two possibilities to alleviate this problem are currently pursued by the author's group: one is "learning by situation", the other is learning using VR friends. In the first case, a complex situation is shown (using a movie-clip, or pictures, etc.). Then the learner has to "think hard" to choose one of 3-5 alternatives. For each alternative chosen either detailed information why it is correct or wrong is presented, or else a resulting situation is shown, again with a choice of alternatives; we believe that this technique increases the "involvement" of learners very much and hence both fights boredom and increases retention. In the second technique a "VR friend" is used as guide and mentor who helps the learner (the VR friend is shown as real face or as a cartoon-like character) but becomes increasingly annoyed or friendly depending on the learning progress: we believe that his additional motivation [16] is particularly helpful for school children.

 

For our basic KM efforts we have chosen Hyperwave [2], [3] as only system available right now that has most features required for KM: the short whitepaper [4] gives a good indication of what we mean. On top of Hyperwave we have developed a number of KT modules, the whole system is called GENTLE, see [5], [7], [12], [13]. In addition to what one would expect from a WBT system like ‘easily’ customizable material, good communication and administration facilities GENTLE allows heterogeneous background archives (that are all full-text searchable) with well-definable search scopes, a unique annotation facility for private and group use including the important facet that questions asked by one student in one spot and that have been answered by an expert can be seen by other students; this eliminates many unnecessary dialogues, enriches the course material and gives feedback where material should be changed or supplemented. With the planned incorporation of further learning strategies (goal oriented learning, situation-based learning, and VR-friends as mentioned above) the system GENTLE is probably one of the more interesting WBT systems currently around, and in productive use by a number of big organizations. For a glimpse at some of the features access the URL in [6] (Attention: wbt not www), and try out e.g. the course ‘Multimedia Systems’ with its note-, question/answer-, forum- and business-card features.

 

 

5. SUMMARY

 

In this paper we have shown that networked-based training systems (WBT) are natural ‘spin-offs’ of much more powerful concepts that will change how we will work and live: networked knowledge management systems that draw their power not just from pre-defined courseware and digital background libraries but from the large number of users that keep enlarging – almost involuntarily – the knowledge stored, by interacting with the system.

6. REFERENCES

 

[1] Carlson, P., Maurer, H.: Computer Visualization, a Missing Organ and a Cyber-Equivalency; Collegiate Microcomputer X, 2 (1992), 110-116.

[2] Maurer, H., HyperWave: The Next Generation Web Solution; (Ed.), Addison-Wesley Longman, London (1996)

[3] http://www.hyperwave.de/documentation

[4] http://www.hyperwave.com/whitepaper

[5] http://www.iicm.edu/gentle.htm

[6] http://wbt.iicm.edu/

[7] http://www.iicm.edu/gentle

[8] http://www.cnnfn.com/digitaljam/newsbytes/11898.html

[9] Davenport, T., Prusak, L.: Working Knowledge: How Organizations Manage What They Know; Harvard Business School Press, Boston (1998)

[10] Sivan, Y.: The PIE of Knowledge Infrastructure: To Manage Knowledge We Need Key Building Blocks; WebNet Journal 1,1 (1999), 15-17.

[11] Maurer, H..: Web-Based Knowledge Management; Internet Watch, Computer, March 98, IEEE, 122-123.

[12] Maurer, H.: Using the WWW System Hyperwave as the Basis of a General Networked Teaching and Learning Environment; CIT, vol. 6,1 (1998), 63-72 (special issue).

[13] Dietinger, Th., Maurer, H.: GENTLE – (GEneral Networked Training and Learning Environment); Proceedings of ED-MEDIA & ED-TELECOM 98, Freiburg, Germany, AACE, Charlottesville, USA (1998), 925-930.

[14] Robillard, P.N.: The Role of Knowledge in Software Development; Communications of the ACM Vol. 42, No.1 (1999), 87-92.

[15] Maurer, H:: The Heart of the Problem: Knowledge Management and Knowledge Transfer; Proc. ENABLE 99, Espoo, Finland (June 1999).

[16] Holzinger, A., Maurer, H.: Incidental Learning, Motivation and the Tamagotchi-Effect; Proc. CAL, London (1999).

F. Wilhelm Bruns

Professor, Production Informatics/ Shaping Technology

Department of Mathematics & Informatics, Bremen University

Research Center ARTEC (Work, Environment, Technology)

Tel. 0049 218 7307, Fax -- 4449, bruns@artec.uni-bremen.de, www.artec.uin-bremen.de

Born in Hannover, Germany (1945)

 

Grades

Dipl. Ing., Aeronautical & Astronautical Engineering, Berlin University (1971)

Dr.-Ing., Fluid mechanics/Numerical Mathematics, Berlin University (1979)

2. Research Interest

Computer Supported Modeling and Simulation, Tangible Man-Machine Interfaces

Formal and non-formal aspects of complex systems design, qualitative reasoning

Tele-Teaching & -Learning, Learning from experience, work-process oriented learning

Theories of learning, action orientation, forming of mental and physical models of action

3. Awards/Honors

DAAD Fellowship for Research/Studies of Biomechanics at Stanford University (1971/72)

4. Occupation

Lecturer for Numerical Mathematics/Praktische Mathematik, Berlin University (1972-77)

Scientist at German Environmental Agency/Umweltbundesamt (1979-82)

Foundation of Software-Company C+S GmbH for Automation-Technology (1982-87)

Senior Lecturer for Teacher Education in Metal-Engineering, Bremen University (1987-90)

Professor for Shaping Technology/Production Informatics, Bremen University (since 1990)

5. Selected Publications

Transatlantic research and field studies in vocational training bringing together

distributed work and learning places, developers and technology providers, researchers bridging the gap between work-process knowledge and cognitive abstraction, locations of theory and practice

Participation in projects of open distributed learning from working (e.g. Stanford Learning Lab’s initiative: Global Learning Laboratory Network, Carnegie Mellon’s initiative: Intelligent Workplace – A Living Laboratory)

Networks for quality of work-technology-education in mechatronics

Evaluation Methodologies. Exchange of research and course evaluators

PROFESSOR TERRY MAYES is Head of the Department of Learning and Educational Development and Director of the Centre for Learning and Teaching Innovation at Glasgow Caledonian University. He moved to Glasgow Caledonian in 1996. Until then he was Director of Research in the Institute for Computer-Based Learning at Heriot-Watt University from it's formation in 1990. Previously he was Deputy Director of the Scottish Human-Computer Interaction Centre at Strathclyde University from 1986-1990.

QUALIFICATIONS: Bsc (Hons) Biological Chemistry- University of London

Maitrise de Chimie - University of Bordeaux

YEAR OF BIRTH: 1971

NATIONALITY: British

LANGUAGES: English, French

CURRENT POSITION: Senior Consultant, ECOTEC Research and Consulting Ltd

Rebecca Alden is a Senior Consultant in the Innovation and Policy Support Group at ECOTEC’s Brussels Office.

Since joining ECOTEC, Rebecca has been involved in a wide range of projects involving the provision of technical assistance to European Commission Programmes. In particular Rebecca has been responsible for the preparation, writing and quality control of a range of dissemination tools.

Rebecca was the project manager of the TETRISS Project, a support project of the Education and Training Sector of the Telematics Applications Programme. A key role of TETRISS was to disseminate information on the activities and results of the projects of the Education and Training Sector and the Educational Multimedia Task Force, using a variety of media from a multimedia web site to paper based reports and information sheets.

Rebecca is currently project manager of PROACTe (Promoting Awareness and Communicating Technologies in Education), a project set up to support, valorise and facilitate the dissemination and take-up activities of the new educational technology projects funded by the European Commission. PROACTe is a web-based service providing a context for European Research into technologies to support learning.

DESCRIPTION OF PROACTE

PROACTe is the new co-ordinating hub of the education and training projects in the European Commission’s Information Society Technology Programme. Communication is the key to PROACTe, and we will encourage dialogue between the projects themselves and with others in the field. The PROACTe website, to be launched later this year, will act as a focus, with online chat facilities, and a variety of useful multimedia content. Projects will be able to access online resources to develop their promotional activities, including a database of important media contacts, and other players in education technology. All projects will be documented on-site, so interested parties can immediately access up-to-date information about the progress of the IST programme. PROACTe will also maintain a weekly news-service, covering the most important developments in the sector.

Name : Jens Pihlkjær Christensen

Date of birth : 1942.10.23

Nationality : Danish

28, rue de la Gare, L-3377 Leudelange, Luxembourg

- "Akademiingeniør", Maskin, produktion. 1968

Ingeniørakademiet

(B.Sc. Mechanical Engineering, Spec.: Production Techniques and Management)

- "HD", Organisation. 1974

Handelshøjskolen i København

(B.Econ., Spec.: Management and Organisational Theory).

- "Computer Science " at the University of Copenhagen. Part 1. Study suspended.

- Systems Analyst, A/S Bladkompagniet, Copenhagen, 1969 - 1971.

- Head of Systems Group, Municipal Hospital Services of Copenhagen, 1971 - 1980.

- Senior Consultant, Birch & Krogboe Consultant Engineers, Virum, 1980 - 1991,

- Project Officer, European Commission DG XIII/F and DG XIII/C, Brussels, 1991 -

.4. Work in the Commission of the European Communities

- Short term expert assistance to the AIM programme, DGXIII/F, Autumn 1988.

- Proposal evaluator February 1989, AIM Programme.

- DG XIII/C (former XIII/F), AIM office. Expert, May 1989 - July 1991.

(AIM = Advanced Informatics in Medicine)

- DG XIII/C-4 (former XIII/F), AIM Office, Scientific Officer, August 1991 -April 1996.

- DG XIII/C-5, TIDE office, Scientific Officer, May 1996 – April 1998.

(TIDE = Telematics Initiative for Disabled and Elderly)

-DGXIII/C-2, Telem. Appl. Public Administrations, Scientific Officer, April 1998 – February 1999.

-DG INFSO/D-3, Multimedia Education & Training, Scientific Officer, February 1999 –

-Function areas of activities/responsibilities in European Commission: National and European strategies for dissemination and implementation of Health Care Informatics/Telematics. Budgeting for unit, supervision of administration of contracts and cost statements, monitoring of submission of deliverables and reports. Personnel/staff issues. Co-ordination of contract negotiations, project officer for bigger R&D projects, concerted actions and strategic studies. Central role in creation of FP Work Programmes. Organised project evaluations and reviews, concertation meetings, workshops and conferences. Socio-economic studies; exploitation of results; contractual/legal aspects including closing of projects; resource management, including patient classification systems; data protection, confidentiality & information security; medical record; regional and institutional information and communication systems/networks; decision support techniques, including evidence based medicine and clinical guidelines; virtual reality; rehabilitation, including functional electrical stimulation; relation to member states and European health care telematics policies/strategies; human computer interface; vocational training, life long learning, international co-operation.

Co-editor of the "High Integrity Systems" periodical from Oxford University Press, UK, and main editor of the book series "Studies in Health Technology and Informatics" from IOS Press, NL. Has edited several books on medical informatics and written numerous articles and papers.

Member of "The Danish Society of Chemical, Civil, Electrical and Mechanical Engineers".

Member of "Danish Society for Biomedical Engineering".

Member of the board of "Danish Society for Medical Informatics" since 1973.

Associated to the board of EFMI, European Federation of Medical Informatics.

Member of the Local Organizing Committee for the MIE96 Conference in Copenhagen, 1996.

Has been both privately and publicly employed as leader of diverse groups. Several years of experience as senior consultant in medical informatics, decision support and communication systems in the Middle East: Saudi Arabia, Egypt and Kuwait. This included strategies, legal and technical issues. Has successfully been the leader of small and medium seized multi-disciplinary groups. Is an experienced speaker and lecturer. Had a central role in the writing of the Work Programme for the 5FP for Administrations sector in IST.

1. EDUCATION

Telecommunications Engineer Madrid Polytechnic University

 

Final Degree Project: "Communications Environment of a Telematic Training System" which consisted on the design and implementation of a multimedia e-mail system to be used inside a distance training communications

3. PROFESSIONAL EXPERIENCE

1. LANGUAGES

Spanish: Mother tongue

English: Advance

German: Intermediate

French: Basic

2. PUBLICATIONS

• RED (Distance Education and Professional Training Magazine) "Tele-education. Virtual Classroom Environments". Nº 21, October/January 1998. Ministry of Education and Culture.
• Proceedings of the Summer Course of the Escuela Técnica Superior de Ingenieros Industriales de Madrid "Telematic Tools for the Engineering of the Distance Training" (8-12 July 1996).
• The Multimedia YearBook 1996. "Spain". TFPL
• Fundesco Newsletter. "Training and New Technologies". March 1996
• Fundesco Newsletter. "TEN Project: Development of a trans-european training network". February 1996
• The Multimedia YearBook 1995.: "Spain". Interactive Media Publications
• Fundesco Newsletter. "The Multimedia Sector in Spain". September 1995

1. CONFERENCES and MEDIA

• Telework Seminar. Organised by the Confederación de Empresarios de Pontevedra. Vigo 26^th of May, 1999. Tittle: "A practical vision of Teleworking".
• Ciberoamerica 97 (1^st International Teleeducation Seminar) the 22^nd , 23^rd of April, 1997 in Lima (Peru). Title: "Promoción y Gestión de Programas Educativos en Hispanoamérica".
• Eurochinatel (Telematics Multimedia Applications for Education & Training) the 2^nd , 3^rd and 4^th of April, 1997 in Beiging (China). Title: "Satellite Virtual Classroom Environments".
• Interview in "La Aventura del Saber" of TVE2 the 11th of December 1996
• First Annual Concertation Meeting (Telematics Applications Programme) the 2^nd and 3^rd of December, 1996. Title: "Creation of a Real Time Virtual Classroom using satellite".
• Conference on Educative Multimedia organised by CEDETI (Centro para el Desarrollo Tecnológico Industrial) the 30^th October, 1996.
• The XV European Conference on Distance Training organised by AECS (European Association of Distance Training) the 7th-10th May in Madrid. Title: "TEN Project".
• The VII Jornadas sobre Televisión, Formación y Empleo (VII Conference on Television, Training and Employment), University of Alicante (Spain) 20-21 May. Title: "TEN Project: Development of a Pan-European Training Network".

BALACHEFF Nicolas Charles Henri

Born on march 25, 1947 in Mayence (Germany)

French citizen

Directeur de recherche au CNRS

Director of Leibniz Laboratory (UMR 5522)

Laboratoire Leibniz - IMAG (UMR 5522)

Doctorat d'état Didactique des mathématiques (1988), Thèse de troisième cycle informatique (1978), DEA Informatique (1974), Capes mathématiques (1971), Maîtrise de mathématiques (1970)

Professional experience

Research director CNRS (Leibniz-IMAG, Grenoble 1995-… ; LSD2-IMAG, Grenoble 1992-1995 ; IRPEACS, Ecully, 1988-1992), Associate Professor (Université de Grenoble 1, 1988), Assistant Professor (INPG-ENSIMAG 1972-1987).

Since 1975, my main research domain is mathematics education with a special emphasis on the learning of proof, and questions related to the modeling of students conceptions. I then moved to problems related to the design of learning environments, in relation to student knowledge modeling and teacher-machine cooperations. This new direction of research started in 1988 have been developed in cooperation with the microworld project Cabri-géomètre. A first period (project TéléCabri) has been devoted to the implementation of an experimental platform for distant teaching and learning in the Academic Hospital of Grenoble, a second period (project Bagheera) now developing is devoted to the design and implementation of a multi-agents environment for the teaching and learning of mathematics involving learners, teachers and specialized agents (including a automatic theorem prover, ATINF).

Leibniz Laboratory, Grenoble

The main focus of my research activity could be summarised in the following claim: "Supporting isolated learners and their teachers" Let's take an example: How to help students being cured at home or in an hospital not to loose contact with schooling? How to help them to continue their study, whilst it is largely acknowledge that such an activity may even help them to recover. This problem is rather important, since in an average town (half million people) it is a problem for about 600 students who spend at least three weeks in an hospital. A solution is in some cases proposed by means of structures involving teachers who visit students where they are, in their wards, their working place, their home, etc. But such solutions have to face difficulties like the following : each student needs a personalized intervention on several different topic on different level, the students are spread over large region requiring quite a lot of travelling for the teachers, and it may well happen that unexpected reasons prevent the teacher from providing her help when planed (this is especially true in the case of students being cured).

A solution to such problems is offered by current technologies of communication which allow teachers and students to interact in a friendly way through a video chanel and to share a workspace on a screen. A technology which makes the teacher available at any time, in a complementary way to on the site interactions. We have implemented and studied such a solution in the TéléCabri project. We focussed in this project on the possibility offered to the users to share a common world of experimentation : Cabri-geometry, and on teachers' monitoring of students activities. The key difficulty in the use of these technologies is that if on one hand they allow a good quality of communication and provide students with high quality learning material, on an other hand they do not support the interaction between the students and their teachers or trainers at the knowledge level.

Teachers expect tools which could inform them about students activities in such rich environments, they look for machines with which they could cooperate. On the other hand a machine alone can face a very limited range of learning problems. Indeed this is one of the more serious deadlock of the design of ITSs as autonomous systems. Multi-agent technology and distributed architectures allow today to think of teaching-learning communities, including artificial and human agents, co-operating with a common target: to stimulate and enhance human learning. This is the core of our current project: Bagheera.

Some thoughts about the US/EU cooperation on technologies for education and training

We may have some advantage in not considering at the same level primary education, secondary and post-secondary education, and corporate training / vocational education.

1. about the first, I think that it will not be a "real" market but still there will be an agreement that it is a universal service, a duty for our nations. In this case may suggest to associate the US and EU in supporting the acquisition of basic skills and knowledge in less favored countries or using education to help in improving the social status of the poorest in our respective countries.
2. considering the rest, there is here a huge market on which we are more in competition than potential cooperators. On the other hand, the cultural conception of the role of secondary and post-secondary may be different in our different societies. Then I think that the best is to push the idea of a collaboration at a basic research level with a specific emphasis on multidisciplinary approaches (which means that sociologists, psychologists, researchers in ergonomic, and others should be involved in coming workshops)

We could propose the creation of a cyber tool for collaboration which could include : information about the state of the art in the domain of tele-learning, bibliographic resources, freeware repository dedicated to tele-learning, and may be a joint on-line resource which could be use by all those who want to engage in tele-learning actions as providers, designers or researchers (or even as users !). This could be considered as a "virtual technology center" like those recommended to president Clinton by the "President's information technology advisory committee" (august 98).

Some key words to discuss :

• Teacher/trainer and machine partnership

• A new understanding of learning space (web) vs learning environment (software)

• Education as an emerging process, and a continuing process

• Simulation and augmented reality for learning

• Allowing the use of the web by "ignorants" (the non elitist computer)

• Standards, interoperability, reusability of environments

• Ethical issues related to the use of tele-training (assessments, certification, …)

• Cultural issues, learning for all

• Epistemological issues, what is learned meaning and change of meaning

Curriculum Vitae

Dr. Clive Best

Clive Best has a Phd in High Energyy Physics, and has worked at CERN, Rutherford Lab and the JET project before joing the Joint Research Centre. Since 1993 he is head of Multimedia Applications Sector, where he has led the development of a number of new Web techniques and information systems. His group now concentrate on Web based educational systems and resource discovery. He is chairman of the CEOS Web Technology working group and collaborates with USGS, NASA and OGC on standards for Geospatial data. He has experience of metadata standards, Z39.50 and Interoperability issues.

Joint Research Centre

The Joint Research Centre provides scientific and technical to EU policies. It is a service of the

European Commission and acts as a reference centre of science and technology for the Union. The JRC is composed of 8 scientific Institutes distributed over 5 sites. A wide reange of disciplines are

covered from Nuclear Safety to Health and Consumer Protection following the main lines of the

Researcg Framework programme of the EU. Research work concerning the Information Society is covered by the Intsitute for Systems Informatics and Safety (ISIS) based at Ispra Italy. Education and Training is one of the three information socitey areasd covered by ISIS.

Ideas on Cooperation

----------------------

Standardisation - Metadata IMS/Dublin Core

Schools Interoperability Framework ? A European version ?

- currently this is very much US centric and is based around the US schools system, but the

technology and software are general and important. We could begin discussion about adapting it for

Europe.

Open Software Exchange

- develop free software solutions. eg. School Web sites etc.

US/EU cooperation on Resource Repositories.

- work together to produce high quality eductaional resource repositories on the Web

EU-US exchanges of ex

Dr Jacobijn Sandberg studied Psychology at the University of Amsterdam with a special focus on cognitive psychology and education. After her graduation in 1982 she worked at the University of Utrecht were she was involved in a project developing a teaching program for elementary mathematics. From 1986 onwards she has been working at the department of Social Science Informatics.of the University of Amsterdam Her areas of research inlude: intelligent tutoring systems, AI and Education, the development and use of ontologies, human computer interaction, distance learning, and the evaluation of educational applications. She worked on several European projects which were partly funded by the European Commission: Eurohelp (Esprit P280); LEAST (Delta D7042; EAST (Delta D2016); HUMAN (Esprit P8282); GRASP (Telematics); Lilienthal (MM 1016); CREDIT (MM 1032). The latter two projects are still running. Liliental develops a network of pilot schools which provides distance training for students aiming for a PPL (Private Pilot License). CREDIT (Capabilities Registration, Evaluation, Diagnosis and advice through Internet Technology) develops an assessment and accreditation system allowing the assessment and accreditation of previously gained knowledge and skills coupled with advice on further training possibilities. The CREDIT-project started in September 1998 and will finish in February 2001. Jacobijn Sandberg acts as co-ordinator for the project.

1. Sandberg, J.A.C. & Barnard, Y.F. 1997. Deep Learning is difficult. Instructional Science 25:15-36.
2. Wielinga, B., Sandberg, J.A.C., & Schreiber, G. 1997. Methods and Techniques for Knowledge Management: What has Knowledge Engineering to offer? Expert Systems With Apllications, 13, 73-84.
3. Sandberg, J.A.C. & Andriessen, J. 1997. Where is AI and how about education. In du Boulay, B. & Mizoguchi, R. (Eds), Proceedings of the AI and Education Conference, p. 545-525.
4. Anderiessen, J. & Sandberg, J.A.C. 1999. Where is education heading and how about AI? Journal of Artificial Intelligence in Education, p. 130-150.

 

5. Sandberg, J.A.C., Christoph, L.H., & Emans, B.S. (in press).Tutor Training: a systematic investigation of tutor requirements and an evaluation of a training. British Journal for Educational Technology.

 

6. Sandberg, J.A.C., Anjewierden, A.A., Groothuismink, J., de Hoog, R. & Giebels, M. (in press). British Journal for Educational Technology.

Matthias Rohrbach

C.V.:

Dipl.-Ing. Matthias Rohrbach is a senior consultant and scientist at the

Fraunhofer Institute for Industrial Engineering (Fraunhofer IAO) in

Stuttgart, Germany since 1996. As project manager of several national and

international projects he was/is working in the field of human-computer

interaction, online/offline multimedia information/sales/training, internet

communities and electronic technical documentation.

He holds a diploma degree in electrical engineering from the University of

Karlsruhe, Germany, with focus on digital signal processing. After

finishing his studies at the universities of Stuttgart and Karlsruhe he was

CEO of a multimedia service company.

Background

As an employee of the IAO department "Information Engineering" Mr. Rohrbach

was involved as a consultant in a number of projects in the field of

learning environments. Starting with CBT-projects for the industry the

department has now its focus on web based training for schools,

universities and industrial users.

The integration of new internet technologies like audio communication,

multi-user platforms and virtual reality has always been seen as an

essential part in order to improve distance learning platforms. The

research group concentrates on the idea that all the participant's senses

have to be triggered with multimedia content. Furthermore, a strong

interaction component is demanded.

Not only the new technology and the user-friendly application design is

very important for the unit, but also the social and pedagogical aspect of

computer supported learning and training. As an example for a successful

internet distance learning project (EC-funded) one can mention the VIRLAN

environment, where primary school children are able to learn different

languages together with native speakers.

Interest

Since the head of the Fraunhofer IAO, Professor Bullinger, is also the head

of the institute for technology management at the University of Stuttgart

(IAT), he as a strong interest in developing and using new learning

environments for lifelong learning. The aim is to develop and establish

distance learning technologies for multi-user education/learning platforms.

Finally, a second (virtual) world for world wide learning and knowledge

management shall be built.

Several European research projects like Virtual Blackboard or Invite will

start soon and the IAO is very interested in sharing distance learning

know-how with US-partners, who want to join the projects. Furthermore, the

IAO is very interested in establishing a long term co-operation with the

partners. An exchange of scientists and know-how during and after the

project's runtime would be appreciated very much.

Status: Married, 9 children

Born: 10 May 1947

Education: Chemical Engineering, B.Sc., M.Sc., (Lisbon, Portugal)

Economics of Education, Adv. Studies, M.Ed. (U..K.)

Curriculum Development, Adv. Studies (Coleraine, U.K.)

FKC, Presentation Fellow King's College - U.K. (Doctor Ed., H.C.)

Former Cabinet Positions:

- Minister of Education and Sports (1987-91)

- Secretary of State for Regional and Local Government (1981-83)

- Secretary of State of Education (1980-81)

- Senior Advisor to the Minister of Foreign Affairs (1978)

Main International Assignments:

- WORLD BANK (IBRD and IDA): Senior consultant for projects in Latin America, the Caribbean, Africa, Central and Eastern Europe (education, integrated rural and urban development, public management, governance) (since 1975)

- OECD: since 1978, senior consultant on educational policy, manpower planning and public management

- UNESCO: since 1978, senior consultant and expert on education and training policy, as well as on future studies

- CALOUSTE GULBENKIAN FOUNDATION: since 1983, senior consultant on educational development in Africa

- IICA: expert and advisor on rural development in Brazil (1986)

- EU: Occasional advisor on policy planning, new information technologies, education and human resources development (since 1986 to date)

- COUNCIL OF EUROPE: General Rapporteur on the Future of European Education (Structures and Policies, 1994); policy expert and special consultant to the Russian educational reform (1997); consultant on Education Finance, Governance and Administration in Bosnia and Herzegovina (1999).

Academic Responsibilities:

- Professor, Education and Economics of Human Resources, Catholic University- Lisbon (1983/92)

- Professor, Multicultural Education, Catholic University - Lisbon (since 1992)

- Chairman, Study Centre on Portuguese Speaking Peoples and Cultures, Catholic University - Lisbon (since 1985)

- Vice-President, National Institute of Administration - (1983-91)

- Research Director, Portugal Year 2000, Calouste Gulbenkian Foundation (1983-86)

- President, Instituto Fontes Pereira de Melo (public and local management institute) (1983-86)

- Assistant Professor, Lisbon Technical University (1970-74)

• Research Director, Future and Strategic Studies on Education, Ministry of Education (1998)
• Research Director, New Job Skills and Minimum Literacy Platform in Europe, TSER, EC (1997/98)
• Research Director, The Future of Multimedia Content in Portugal, APDC (1998).
• Research Director, An outlook into the future of the Internet, Telecommunications and the Information Society in Portugal, ICP (1998)
• Research Director, Digital Economy in Macau, CPTTM (1999)
• Research Director, Content Industry in Portugal, Ministries of Economy and Culture, (1999-2000)

Other Former Positions:

- President and CEO, TVI-Televisão Independente, SA - a private national television network (1992-96)

- Director-General, Ministry of Education (1973-79)

- Director and main Editor, TEMPO - University Press (1965-70)

Other Relevant Activities:

- Consultant co-ordinator to SIGMA, EU/OECD joint venture designed to implement State reforms in Central and Eastern European Countries (1992/95)

- Co-ordinator, Task-Force for the Planning and Implementation of the new Catholic University of Angola (1992/97)

- OECD Examiner of the Turkish Educational Policy (1985), French Educational Policy (1993) and French Higher Educational Policy (1997)

- Member of the International Commission for Education in the 21st Century (UNESCO, 1993 to date)

- Member and Vice-President of the Information Society Forum set-up in the European Commission under the auspices of Commissioner M. Bangemann (1995 to date)

- Member and Vice-President of the Bureau which co-ordinates the European Commission Study Group on Education and Training, under the auspices of Commissioner E. Cresson (1995 to date).

- Chairman of the 5-year assessment panel of ESPRIT - EU Strategic Research Program on Information Technologies (1996/97.

- Member of the Senior Expert Group INFO 2000 (EU content industry program – 1997-98)

- Chairman of the Interim Evaluation Panel of INFO 2000 (1998)

- Professor of public policies at the Macau Institute of European Studies (1997-2000)

- Professor of education innovation and projects and the Macau InterUniversity Institute (1999-2000)

• Project director for the EU-Macau Public Administration Training Programme (1998-99)
• President of the Foundation for the Macau Portuguese School (1998-99)
• Member of the Calouste Gulbenkian Foundation General Council (1997-99)
• Member of the Telecel (Cellular Phone Operator) General Council (1998-99)
• Member of the Consultative Committee of the Institute of Education of the Catholic University (1998-99)

Publications:

- About 350 articles, papers and books on education, training policies,public/private management, media development and national/international affairs

- Director of a school encyclopaedia edited under the Bertelsmann Group - Activa Multimedia (13 volumes and 7 CD-ROMs) (1996-98)

- Editor-in-Chief of the Journal of Education "Colóquio/Educação e Sociedade", Fundação Calouste Gulbenkian (1997/99)

DATE COUNTRY STATUS MISSION PROJECT

Oct - Dec 77 Brazil Ed.Planner Identification Rural Basic Education

Ap. - May 78 Brazil Gen.Educator Preparation Rural Basic Education

Ap. - Jun 79 Brazil Ed. Planner Appraisal Rural Basic Education

Oct - Dec 79 Brazil Ed. Planner Appraisal Rural Basic Education

Jan - Feb 81 Brazil Ed. Planner Supervision Rural Basic Education

Agriculture and R.

Development

Mar - Ap. 81 Haiti Mis.Leader/ Preparation Basic Education

/Ed. Planner

Jul - Aug 81 Haiti Ed. Planner Appraisal Basic Education

March 83 Brazil Ed. Planner Preparation Urban BasicEducation

June 83 Brazil Ed. Planner Appraisal Urban BasicEducation

Sep - Oct 84 Brazil Gen.Educator Supervision Rural Basic Education

June 85 Brazil Gen.Educator Supervision Rural Basic Education

Sep - Oct 85 Brazil Gen.Educator Supervision Urban BasicEducation

Jun - Aug 86 C.Verde Educ.Administ. Pre Appraisal Basic Education

Aug - Sep 86 Brazil Ed. Planner Preparation Basic Education

Dec. 86 Cape Textbook Pre Appraisal Basic Education

Verde Specialist

Jan - Feb 87 Brazil Ed. Planner Preparation Basic Education

Feb 97 El Salvador Policy Expert Policy Appraisal Basic/Rural Education

Aug 97 Mexico Educ. Policy Preparation Basic Education

Expert

Nov 97 Mexico Educ.Policy Pre Appraisal Basic Education

Expert

Jan-Nov 99 Bosnia and Educ. Policy Sector Study Educ. Finance, Governance

Herzegovina Expert and Administration

DATE COUNTRY STATUS MISSION PROJECT

Sep - Oct 75 Guinea- Ed. Planner Sector Country

-Bissau Study

Oct. 77 Cat. VI Ed. Expert Education

Confer. Reform

Jan 94 - to date Many Ed. Expert Member of the Intern.Commission on Ed. for the 21st century

Apr 98 Australia Ed. Expert Asia-Pacific Regional Conference on Ed. for the 21^st

Century

Sep 98 France Expert 21^st Century Dialogues

Sep 99 Germany Expert 21^st Century Dialogues/EXPO 2000

3. BRAZIL (Gov.of)

DATE COUNTRY STATUS MISSION PROJECT

Sep - Oct 85 Brazil Ed. Planner Strategic long Term Planning in

Basic Education

4. OECD

DATE COUNTRY/CITY STATUS MISSION PROJECT

Mar 86 Japan (Kyoto) Expert Review Japanese Educ. System

Sep - Oct 86 Turkey Educator/ Turkish Educ. Policy Review

/Examiner

Oct. 86 Paris Public Personnel Policies and Productivity

Manag.Expert incentives in Public Management

Jan. 87 Japan/ Education International Conference on the

/Kioto Future of Education

June 87 Turkey Educator/ Turkish Educ. Policy Review

/Examiner

Jul. 87 Paris Chairman/ Personnel Policies and Productivity

/Public Manag. incentives in Public Management

Expert

1993-95 France Examiner French non-Univ. Educ. Policy

1995-97 France Expert Futures Studies at OECD

1997 France Examiner French Higher Educ. Policy

Sep 99 France Expert-Key Government of the Future

Note speaker

5. COUNCIL OF EUROPE

1993-94 France General Study on Ed. Structures and Policies

Rapporteur

Nov 97 Russia Ed. Policy Advise on Educ. Concept and

Expert Legislation

Jan-Jul 99 Bosnia and Educ. Policy Sector Study on Educ. Finance, Governance

Herzegovina Expert and Administration

6. EUROPEAN UNION

May 93 Bologna Expert Carrefour des Sciences et de la Culture

1994-98 Brussels VicePresid Study-Group on Educ. And Training

Oct 94 Leiden Expert Carrefour des Sciences et de la Culture

1995-96 Brussels VicePresid Information Society Forum

Jun 96 - Dec 96 Brussels Chairman Evaluation Panel on ESPRIT

1996 - 99 Brussels Expert Senior Expert Group INFO 2000

Oct 97 - April 97 Warsaw Ed. Policy EU/PHARE educ. Policy paper

Nov 98 Vienna Expert IST 98

Oct 99 Bruges Expert XVII Carrefour des Sciences et de la Culture

Nov 99 Helsinki Expert IST 99

5. CALOUSTE GULBENKIAN FOUNDATION

DATE COUNTRY STATUS MISSION PROJECT

Dec. 83 S.Tomé and Prince Mis.Leader/Planner Sector and Strategy Analysis

Mar - Apr 85 Guinea- Bissau Mis.Leader/Planner Sector and Strategy Analysis

Dec. 85 Cape Verde Mis.Leader/Planner Sector and Strategy Analysis

Roberto Carneiro

Portuguese Catholic University, Lisbon, Portugal

Study Center on Peoples and Cultures

roberto.carneiro@cepcep.ucp.pt

and

Pedro Conceição

Instituto Superior Técnico, Lisbon, Portugal

and

The University of Texas at Austin, Austin, Texas, USA

pedroc@uts.cc.utexas.edu

2. Introduction

Which development patterns can be identified within the Portuguese industry? What kind of qualifications are Portuguese firms requiring in order to face the challenges imposed by the need to absorb new technology in those development processes? How are the Portuguese firms re-adjusting to face increasing competitiveness pressures in the global marketplace? Which broader social issues emerge from these phenomena, particularly that of the exclusion of low-skilled workers, and what policies can contribute to favor their social and economic inclusion? These broad questions motivated the research discussed in this paper.

We go beyond the perspective that analyzes the impact of technological change on employment as the result of a substitution of capital for labor and its host of consequences, such as the different returns that may emerge to different types of skill. We have a broad understanding of technology that encompasses a range of scientific, technical, cultural, social and organizational change. We aim at understanding the processes through which the Portuguese firms nurture "economic learning", where "economic learning" is epitomized by the notion that some firms seem to be able to accommodate institutional, cultural, technological and market changes better than others (Mathews, 1996). We are interested in the dynamics of new knowledge production, diffusion and adoption, which requires firms’ to generate endogenously an ability to learn.

The paper builds upon the analysis of data that resulted from a detailed survey of two industrial sectors, chosen as representative of two main Portuguese industrial categories: a traditional Portuguese sector (shoe and leather industry), and a sector associated with the economies of the "second" industrial revolution (electrical and electronics components and equipment). The survey was performed in conjunction with a series of structured interviews with leaders of firms from the two sectors. Most qualitative description will be based on notes taken during these interviews, and may, therefore, lack a formal reference to a literary source or to quantitative data. Following is a summary of the paper main results.

Earlier TSER research, developed under the Minimum Learning Platform project, has shown that Portugal seems to be partially at odds with the existing trends of the knowledge based economies (Kirsch, 1998). Indeed, the country has shown a remarkable ability to incorporate and sustain in the workforce a large proportion of workers lacking an "appropriate" formal education. This original feature explains in part the comparatively low Portuguese unemployment rates, which contrast with those of the broader European landscape.

We illustrate this feature by analyzing the performance of a traditional sector, leather and footwear, which has been prospering while absorbing sizable contingents of workers with low levels of formal education. We will see how this traditional sector has been competitive in spite of its scarcity in highly qualified people. We offer an interpretation for the performance of this traditional sector stemming from a learning-by-doing dynamics, and contrast it with the performance of a more high technology-intensive sector, the electronics and electric industry, where formal learning processes have been institutionalized.

Since growth in the leather and shoe industry has been accompanied by steady sector productivity gains, our conjecture is that growth in this industry was driven by improved "learning abilities" within the sector, rather than by the mere augmentation of the levels of labor and capital. We will not be able to identify the drivers of this productivity growth (maybe international competition or the contracting of Portuguese firms by multinationals with strict requirements) but we emphasize that our interest lies in the way the learning process itself occurred, which is reflected in increased productivity and output.

Furthermore, explanations of the productivity growth in terms of increases in human capital as measured by standard proxies, such as levels of formal education, appear to be grossly insufficient since, as we noted, the level of formal education in the industry is persistently low. Consequently, we are led to the hypothesis that the Portuguese workers have acquired high levels of on the job performance by learning through experience and practice, rather than through formalized education and training methods. Clearly, this type of learning is as valuable as the one developed formally, despite the difficulties in its acknowledgement and measurement. Therefore, we are led to propose a nation-wide system for the recognition and accreditation of the non-formal competencies acquired through firm-based work experience. However, we must stress that this does not entail a generalizable judgement on the higher merits of learning-by-doing vis-a-vis learning in schools. Rather, it is a policy proposal to acknowledge a specificity of the Portuguese reality, in which human capital seems to be embodied in informal, rather than formal, competencies, at least in some sectors of the economy.

Likewise, this analysis is not normative in terms of recommendations for the future. We are sensitive to the possibility of decreasing marginal returns to the learning-by-doing informal practices. This hypothesis receives support from the recent evolution in the leather and shoe sector, and the prospective analysis expressed by some of its leaders, reflecting a growing demand for managerial skills and for qualified employees, associated with an increased technological sophistication of the value chains, especially within the most innovative firms. A possible saturation of the growth potential yielded through learning-by-doing could require higher levels of formal education, in order for the sector to acquire the "cognitive" resources needed to deal with new knowledge. This new knowledge exists both in terms of production technologies and in the access to markets through ever more complex marketing and distribution practices. Activities such as branding, design, franchising, e-commerce, customer service, de-localization, to mention only a few, may need to be, at least partially, internalized, if the sector is to retain high value-added functions. This internalization may be a challenge that can only be tackled through better-educated people.

By contrast with the leather and shoe industry, the electronics and electric sector (electronics, for brevity) has prospered with comparatively higher levels of formal education, although at much lower growth rates. This industry has been able to be a technological leader in some domains, reaching worldwide competitiveness standards in well-defined segments, a significant accomplishment in a field characterized by cutthroat competition and constant technological change. However, its development pattern was largely based on the acquisition of foreign technology for production (importation of equipment, components, and raw materials) and exportation of manufactured products. Largely, these market shares in exports seem to benefit from what Krugman (1995) has called the slicing-up of the value chain, the feature of producing a final good in a number of stages in a number of locations. Different components for the automobile industry are the prime example.

It is not clear whether this modern industry, despite its high levels of formal education and even formalized research activities, has been able to endogenize a learning ability capable of leading to the internal design and production of capital intensive equipment.

From this brief description of some of the paper main results, we are left with a mixed message. On the one hand, we have a traditional sector that, with very low levels of formal education, has shown to be remarkably resilient and internationally competitive. On the other hand, we have a modern sector, with much higher levels of formal education, which has remained competitive in global markets, but has shown a much slower growth pace.

From this mixed message we retain the following key ideas. Portugal is a country handicapped by decades of low investment in education, particularly during the authoritarian regime that ruled the country between 1926 and 1974. The formidable efforts conducted in recent years to bring the Portuguese formal educational attainment level up to the Western European norm are beginning to bear fruits. The present generation of school age population enjoys roughly the same educational opportunities of its European peers. However, the incoming flow to the labor force has low incremental impact on the overall stock (around 2% replacement rate per year). Thus, the heavy concentration of public resources in catching up the delays in initial education activities has entailed a related toll: the production of a dual labor force – reflected in both the employed and unemployed population– with a clear-cut age divide.

This demographic duality is also reflected in a somewhat atypical industry segmentation: on one hand, a large traditional cluster, which has had to operate and compete, albeit its older and precariously educated workforce, running side by side, on the other hand, with a modern industrial sector that has been able to follow the international path by turning to a younger and better educated labor. Thus, Portugal has shown a very specific development path in the knowledge economy, one in which learning and knowledge accumulation through informal processes are as important, if not more, as the formal learning mechanisms. Consequently, Portugal has been able to integrate in its workforce large amounts of low-skilled workers, who learn and develop professionally on the job.

To acknowledge and recognize this striking feature of the Portuguese economy, we suggest the creation of a mechanism for the accreditation of informally acquired competencies. These informal learning dynamics – at least, in a first transitional stage affecting the more traditional segments of the economy - seem to generate much higher rates of growth and levels of economic performance than the ones that exist in more high technology intensive industries requiring further levels of formal education. In these high technology sectors, the performance of the country seems to follow a well-known pattern: imports of technology and equipment to manufacture final products according to exogenously specified designs. Consequently, these high technology sectors may need to acquire the informal learning ability of the more traditional sectors in order to arrive at the ability to endogenously produce technology and designs.

Ironically, the message seems to be reciprocal. Not only that the more traditional sector will have to enter a new phase of larger investments in the appropriation of formal education, but also that the "formally educated sector" needs to learn with the less literate one as a condition for a more flexible and sustainable modern industrial sector. We are in the realm of the interplay between technology and culture.

The paper is in five parts. Section 2 provides a very brief overview of the relationship between technology and employment. Our main purpose is to acknowledge our conceptual framework of analysis, in which we emphasize "learning" processes to explain the economic performance of the sectors. Section 3 gives the details of the methods and describes the data, and in section 4 we discuss the results. Section 5 summarizes our main conclusions.

Innovation, a consequence of human creativity, is both a blessing and a curse. There is probably no other expression that encompasses this dual nature better than the one proposed by Schumpeter: innovation is a process of creative destruction. As noted by Landes (1969), before the industrial revolution new technology would, literally, destroy and create entire communities, normally localized in a well-defined geographic region. In fact, in pre-industrial times, production techniques were often tied to groups or social organizations that were able to codify and use those technologies. A new and improved technology would not only replace previously existing techniques, but would lead to the dismemberment of the entire community which was based on the older technology. In his novel The Gift of Stones, Jim Crace describes how the emergence of the Bronze Age destroyed stone-age communities. The corporations of the Middle Ages are also a good example of social groups locally tied to a single technique and organized around the transmission and accreditation of tacit knowledge and skills. R. Wagner, in his unique genius, provides us with a colorful narrative of the struggle of power between innovative and conservative forces in professional corporations in his celebrated opera "The Meistersingers of Nuremberg".

Modern theories of long-run economic growth emphasize the role of technology in driving long-term prosperity. Two conceptualizations of the process of technological-driven economic growth have been most influential. The first derives from the marginalist school of economics that, in the late 19^th century, developed the idea of a production function. Capital and labor are the side by side ingredients of production. Labor and capital interact in a process of production of wealth that is limited by the current level of technology. The total maximum level of production, or total output of the economy, can be represented by the production function: Y=F (K,L), in which Y is output, K is capital, L is labor, and F represents the process of transforming the factors of production into outputs. At the aggregate level, Solow (1956, 1957) showed that the pure accumulation of physical capital and labor was not sufficient to account for all the observed growth. He attributed the component of growth that went beyond the accumulation of physical capital and labor to technological change. This is an equilibrium perspective, in which resource allocation is mediated in free markets by pricing in a competitive environment.

More recently, some conceptual efforts have attempted to portray economic growth as being dependent on what we could call with generality knowledge accumulation through endogenous "learning" processes (see Bruton, 1998). One major thrust of recent scholarship has attempted to extend the pure Solownian formulation to encompass microeconomic impacts of technological innovation, including Schumpeterian competition. Very broadly, these extensions can be viewed as attempts to formalize learning processes, which are reflected in improved skills in people and in the generation, diffusion, and usage of new ideas. Included in these learning outcomes is organizational learning, which reflects social processes driven by collective cultures and appropriates management attitudes. In a broad characterization of these perspectives, one could say that the ability to continuously generate skills and ideas (which is to say, to accumulate knowledge through learning) is the ultimate driver of an economy long-run prospects (World Bank, 1997).

As we saw, the work of Solow (1956, 1957) showed that the accumulation of physical assets and labor was insufficient to account for even a small part of the observed growth. The introduction of factors such as human capital (Schultz, 1960; Becker, 1993) and technology (Nelson, 1959) to the equations attempting to account for economic growth was largely motivated by that deficiency. Denison (1967), as we mentioned before, used even more sophisticated techniques to try to circumscribe the "unexplained" component of growth. But the fundamental issue is that the roles of human capital, technology, and of the other factors proposed by Denison in promoting growth were difficult to include in formal models and hard to measure. Therefore, the way in which these factors were introduced into models of growth reflected these deficiencies. In particular, formal models failed to incorporate the dynamics of innovation conceptualized and described by Schumpeter.

Moreover, in the neoclassical literature, the accumulation of physical capital in the form of machinery and "industrial capacity" was regarded as the distinguishing factor in explaining differences in the levels of economic growth across countries. Technology should be, according to the neoclassical formulation, freely available to all, which meant that most of the effort should go to increases in the levels of capital, through savings and, when income was insufficient to promote high enough savings, through foreign investment (the well-known capital gap). This perspective still informs much of the current policies (Easterly, 1997).

More systematically, Romer (1994) describes our inheritance from the initial neoclassical models in three key points. First, knowledge (of which technology and human capital are part) is either a public good (such as in the case of technology or R&D results, largely exaggerating the social returns resulting from spillovers) or a private good (in the case of human capital, largely neglecting the huge externalities associated with education). Secondly, knowledge is exogenous, determined outside the economic context. Finally, growth is a process that exhibits diminishing returns to the traditional factors of production, capital and labor.

The work of a generation of economists and other social scientists has fought the tendency to oversimplify the impact of new skills and ideas on development (the linear model of innovation has been a consistent target), and the conceptual framework proposed by Schumpeter has been a constant guide for theorizing about growth. The body of work of these scholars has provided sophisticated conceptual insights into the way that technology is related with economic growth (good overviews are included in Fuhrer and Little, 1996, and especially in Stoneman, 1995). The "new growth theories" are a prime instance of the effort to introduce some of those insights into the formal economic modeling framework inherited from Solow. Romer (1994) provides a non-technical overview of the main existing variants. Nelson (1997) and Solow (1997) provide critical assessments of new growth theories from opposing perspectives. While Nelson criticizes these theoretical efforts on the basis that they do not add anything significantly new to scholarship in the area, Solow claims that new growth theory provides almost a distraction from the fundamental aspects of economic growth, which should not be concerned with modeling technological change.

Regardless of the validity of the new growth theories, very much under dispute in the specialized literature, we want to stress that there is an increased effort to incorporate the analysis of Schumpeter and many other social scientists concerned with the behavior and incentives associated with technological change in a formalized framework that stresses the ability to learn as the main driver of long-term growth. The origins of these efforts date back to the work of Arrow (1962), which is praised and cited as the origin of formalized efforts to account for the ability to learn in the context of economic development. Other examples include the work of Pasinetti (1993), that uses a modeling framework inherited from Ricardo. However, his main point, as the title of his work conveys, is to investigate the economic consequences of human learning. The concept of economic learning also reflects the idea that some economies are able to prosper in a changing environment (Mathews, 1996), whether the origin of change is in new technology or in shifting preferences. These and other contributions view economic growth and employment generation as resulting from an endogenous ability to learn.

By the same token, Carneiro (1994) has developed a theory on educational cycles explaining how formal systems evolve from traditional production or consumer-driven paradigms, centrally commanded, to advanced innovation-driven learning organizations of a highly decentralized nature. In the latter case, schools and formal institutions emerge as fundamental nodes of networked learning systems in conjunction with firms and other social organizations.

Along with these conceptual efforts, rapid advances in scientific and technical knowledge, and declining costs of producing, diffusing, and processing information (due to advances in information and communication technologies, ICT) are transforming the organization of social and economic activity worldwide, leading to the emergence of the so-called knowledge-based economies (World Bank, 1997). The availability of quantitative data showing the growing importance of knowledge is still scarce. Some people argue that we are not even living in a fundamentally different world. The discussion of this controversy is beyond this paper, but it is clear that the empirical advances have not accompanied the important theoretical breakthroughs in a better understanding of knowledge-based growth (see, for an overview of the difficulties with the measurement of knowledge in growth models, Howitt, 1996).

An important conjecture, supported by many empirical studies, is that the increasing horizontal importance of ICT and its pervasiveness has contributed to a skill-biased unemployment in industrialized countries (CEDEFOP/TSER, forthcoming). Traditional sectors, especially the primary sector, but increasingly in manufacturing, are reducing their overall shares of employment in favor of services, including producer and professional services (OECD, 1994).

Therefore, the safeguard that traditionally existed for low skilled workers in the old manufacturing sectors is shrinking. Even in services, which can absorb large amounts of low skilled workers, the growth of franchising and standardization, coupled and enhanced by better use of ICTs, is demanding people with managerial and technological abilities (Conceição and Heitor, 1998). Recent European research provides evidence on how employment growth and sustainable job creation is much more contingent on information-rich services than on traditional services (Carneiro, 1997).

The Portuguese situation seems ill fitted to the general description of the knowledge-based economy briefly outlined above, as we saw in the introduction. The remaining of the paper deals with the description and analysis of the specificity of the Portuguese situation, whose complexities are illustrated by the study of two representative industrial sectors. However, it is important to stress that, as seems to happen with generality in the modern market-based economies, Portugal is also dependent on the ability to learn of its industries. What seems to be remarkable in Portugal is that the reliance on informal, on-the-job-type methods of learning has allowed for an accumulation of knowledge in the absence of advanced formal education, leading to an unusually high rate of absorption of low skilled.

The research presented in this paper is supported by quantitative and qualitative data elicited from a survey conducted with the support of two industrial associations linked to the two business sectors we researched: shoe and leather industry, and the electric and electronics industries. These surveys were complemented by structured interviews with senior executives of some of the most prominent firms in each sector. The results of these interviews were merged with some aggregated data from the surveys in a forthcoming report, which will be a major source for the characterization of those industries. We thoroughly reviewed the questionnaires and the raw data of the survey, and have chosen a selection of it for our analysis. We ran an in-depth analysis of the survey sections dealing in particular with the notion of knowledge workers. Descriptions of the data are given throughout the paper, but a brief overview of the data is given below.

Two surveys and sets of interviews were made one for each sector. The population (number of firms) size for L&S was 1 600 and for the E&E sector 1 300. Two samples were constructed, with the help of the sectoral industry associations, to be representative of each sector. Sample sizes were 170 for the L&S and 150 for the E&E, representing circa 85% and 80% of total sectoral outputs, respectively. All the firms within the sample were questioned, and the number of replies was 37 in the L&S sector and 42 in the E&E sector. The questionnaire was complex and long, and, consequently, some of the replies were incomplete. Therefore, in the analysis of the results in the next section, we are often limited to using only a fraction of the responses.

The data obtained from the surveys resulted in a rich and large set of information, of which we were forced to choose only a subset for practical reasons. In our selection of data, we were oriented towards variables associated with the firms’ "stock of knowledge" and its utilization, since our ultimate aim is to enhance the understanding of the processes through which firms within each sector are "learning", in the sense defined in section 2. To measure the stock of knowledge we use two types of indicators associated with the firms’ stock of human capital and with the dynamics of ICT adoption. We describe below the specific indicators that were constructed from the data available, beginning with the human capital indicators. For this purpose, we define indirect measures of human capital (HC), and aim at measuring the levels of ICT adoption and utilization by incorporating a breakdown of different levels of knowledge workers (KW) according to a typology of in-firm ICT use.

To measure human capital we use a very rich database that was provided by the survey. In fact, the survey asked each firm to characterize each member of the workforce both by level of formal education, and by professional category. In this manner, the survey addresses the classical problem of translating occupational structures of the workforce into educational requirements. Therefore, we are able to construct indexes of human capital that, beyond the proxy of formal education, also include the occupational profile of the employee. In this case, we assume that even if someone has got a very low level of formal education, his or her position as an executive certainly reveals a level of experience and knowledge accumulation that ought to be accounted for as human capital. Moreover, someone holding a position of high responsibility within a firm, who typically has to make strategic decisions, is dealing, essentially, with knowledge integration and information.

Therefore, we define three levels of human capital: H2, H1, and L. These three levels were inspired by the model proposed by Romer (1990), in which the stock of human capital is divided in H1 and H2. The economy is described in terms of two production functions. Output is a direct function of L and H1, basically traditional labor and traditional human capital. Output is also dependent on capital, and on a knowledge augmentation factor A, which incorporates externalities associated with the innovative content of capital. H2 is the component of human capital that is used in producing A through research and development.

Here we do not assume firms engage necessarily in formal research but, with Rosenberg (1990), argue that even when innovative activities are not institutionalized, firms develop search and exploration activities that yield the very same type of externalities that accrue from formalized research and development. We assumed that these activities, which are "externality generating" in the sense proposed by Romer, depend primarily on the professional categories which include high levels of decision making. Therefore, H2 for us is associated with the number of executives within the firm, following the rationale given above for considering these occupations "knowledge intensive".

We should stress that our assumption about H2 attempts to portray, more than human capital in the traditional sense, the level of people engaged in "knowledge intensive" decision-making. Our assumption is that the search and exploration activities associated with, for example, introduction of new technologies and innovation, depend on people with leadership roles in the firm. Well beyond the provision of a measure for the number of executives by their educational levels, it is clear that H2 is closely correlated with the structure and dimension of the firm’s leadership. A high level of H2 may, therefore, also indicate a bureaucratic structure and redundancy, but still, we have decided to experiment with this indicator. Again, we should stress that differences in the educational level of managers should be reflected in H2, to the extent that we weigh the number of managers according to their educational level, as we describe in detail below.

Firms were asked to fill in a table as shown in Figure 1. The table shown below indicates also the areas corresponding to the different levels of human capital, making clear which were the differentiating criteria.

While the discriminating factor for H2 is the level of responsibility within the firm, the difference between H1 and L is the formal level of education, following the proxies regularly used to account for levels of human capital.

Finally, to compute the human capital in each category we use weights associated with the level of formal education attained by the workers. We consider five educational levels: university education (weight=17); polytechnic (short-cycle) higher education (15) - in Portugal corresponding to 3 years of higher education, while university education requires on average 5 years; technical education (12); secondary education (12); primary education (5); and incomplete primary education (3). The value of 5 for primary education comes from the evolution of the definition of primary education from 4 (prior to 1968), to 6 (1968), to the current 9 years (1986). We assume the number 5, midway between the more ancient requirements of 4 and 6 years, but this is a purely arbitrary assumption. Since our aim is to compare the levels of human capital across sectors, the specific number chosen is largely irrelevant.

To characterize the knowledge workers we measure the level of adoption and utilization of ICT that is given in rich detail in the survey data. Arguably, ICT have been the enabling technologies that mostly contribute to the emergence of the digital economy. The adoption of these technologies, in and of itself, will not tell us whether firms are better equipped to learn. As extensively discussed in the literature, these technologies are helpful when dealing with codified knowledge (Conceição and Shariq, 1998). Nonetheless, we use ICT adoption here as a proxy for the firms’ fluency and sophistication in dealing with information.

In the survey, firms were asked to indicate the number of people that use a computer generically at least one hour per day, and subsequently their time-loads in using specific computer applications.

We define KW3 as the index of people that use sophisticated computer applications, such as CAD/CAM, with respect to the number of people that use the computer one hour per day.

KW2 corresponds to the index of workers that use connecting and advanced communications applications. In the survey, firms were asked to state how many people use e-mail and how many use the Internet (WWW). Since the same person can use both e-mail and the Internet, we constructed this index with reference to the total number of workers those uses a computer multiplied by two. In this manner, the index remains always below one; an index of one indicates that every person uses both e-mail and the Internet.

Finally, KW1 includes those that use the computers more generically. Four specific types of cumulative applications were surveyed: word processing, worksheets, presentation tools, and use of CD-ROMs. Therefore, to construct this index, the denominator is defined as four times the number of workers that uses the computer at least one hour per day, again to ensure that the index remains between zero and one.

It is important to stress out the different nature of the HC and the KW indexes. In fact, H2, H1, and L correspond to levels of human capital. Their upper bounds are limited by the weights we attributed to the maximum "allowed" qualification in each category. Consequently, L cannot be weighted beyond 5, and H1 and H2 are limited by 17. In turn, the KW indexes measure ICT intensity utilization indexes and are bounded between 0 and 1.

Knowledge workers are not necessarily those with the highest level of educational achievements. Table 1 shows that the correlation values between the levels of human capital and the indices of ICT utilization are low both in the leather and shoe (L&S) sector and in the electronics and electrical industry (E&E). No discernible patterns or systematic relationships emerge.

In fact, our interest on knowledge workers derives partially from the need to understand the way in which low-skilled workers contribute to productivity enhancements through the usage of ICT.

The measures of human capital and of knowledge workers developed above give us a static characterization of the stock of knowledge existing in each sector. Therefore, we need to have information about the dynamics of the firm. The survey also inquired firms about processes of change associated with innovation. Firms were asked about their past innovative performance, differentiating between generic organizational and strategic innovations, on the one hand, and technological and production innovations, on the other.

Associated with the questions of whether innovations were, or were not, part of the firm, several different items were investigated. Firms were asked to specify which category of innovation had occurred, what had been its impact, and which sector of the firm it had affected. Questions were also put concerning the sources of innovation.

The picture that emerged from this rich questionnaire is, naturally, complex. Rather than trying to characterize the innovation dynamics of each sector, we use the innovation indicators drawn from the surveys throughout the paper to illustrate and strengthen our argument. However, a particularly strong emphasis on the innovative performance of the firms will be given towards the end of section 4, when we attempt to summarize and integrate the main empirical results obtained. Once again, though, we emphasize that our aim in this paper is not the study of innovation, in and of itself, in the two sectors.

Beyond innovation indicators, firms were also inquired about investments in human capital (through training), in ICT, and in upgrading knowledge workers. This information also contributed to the dynamic perspective we are able to picture in the paper, despite the fact that we do not present in this paper all the data available.

Finally, we should also note that throughout the paper many observations on the performance, the history, and the prospects of the industries studied will not be accompanied by formal references to the literature. These observations resulted from the background research on the industrial sectors, but more fundamentally from the structured interviews with leaders within each industry.

Source: OECD STAN.

Source: OECD STAN.

Part of the explanation for this dynamic behavior is certainly associated with the lower barriers to entry in this sector, in comparison with the ones existing in the E&E industries. Lower levels of capital investment are required in the L&S sector, where a small firm can exist with virtually no major production equipment. In contrast, investments in physical capital are omnipresent in the E&E sector, particularly the importation of advanced machinery for production. The level of technological sophistication is also lower in the L&S sector than in the E&E, facilitating entry. Nonetheless, the highly entrepreneurial behavior of the L&S sector must not be downplayed. Other "low technology", labor intensive, sectors have not shown an entrepreneurial dynamism comparable with the one that is evident in the L&S industry in Portugal.

In fact, industry development is facilitated within the context of a network, especially if the network is circumscribed within a geographic cluster. This has been a topic of major interest in recent scholarship (Porter, 1990; Nelson, 1991; Lundvall, 1992; Feldman, 1994; Jacobs, 1996; Karlsson, 1997). The effectiveness of an industry-wide network is based on its "knowledge base", composed of the existing technological and market competencies, physical infrastructure, people’s skills and know-how. A robust result of scholarship is that, beyond an intensification of the use of the existing "knowledge-base", communication and interaction between firms is key to establish an endogenous "learning infrastructure", which will ensure the continuous revitalization of the knowledge-base. High rates of firm creation, as the L&S sector exhibits, are an empirical translation of this dynamic "learning infrastructure". The build-up and improvement of this dynamic "learning infrastructure" is largely based on personal relationships, supplier-customer connections, unacknowledged sharing of tacit knowledge, and other seamless interactions between people and firms, mediated primarily through informal contacts which are facilitated by geographic concentration.

Another important aspect of the L&S sector is associated with the important role that its industrial association, APICCAPS (from a Portuguese acronym) has played in the development of the sector. Once again, geographic proximity may have played a role in the effectiveness of APICCAPS. This industrial association has provided opportunities to firms for upgrading their workforce, by running an industry-specific training center, and to scale-up research and development projects relevant for a large set of firms. This has contributed to increase the levels of skills and also the access to new technology. Small firms tend to gain in a disproportional way in comparison with their investments, due to the strong economies of scale that result from the synergistic effects associated with externalities from R&D and training. Furthermore, APICCAPS engages their member firms in active benchmarking, namely through the support of visits to international fairs and exhibitions.

In this context, it is clear that the geographic proximity of firms in the L&S industries compounded with the role of APICCAPS explain part of the learning dynamics that the sector exhibits. However, geographic proximity alone cannot fully account for the extraordinary performance of the sector, nor, on the other hand, answer the question of why the E&E sector has not clustered in the same manner, creating what could be called, if such a concentration existed, a Portuguese "Silicon Valley". On the other hand, and industry association, effective as it may be, is not able, in and of itself, to generate a successful cluster of firms. The establishment of industry clusters depends on the emergence of an intricate web of relationships between organizations and individuals, which is associated with specific complex cultural environments and with contingent historical paths, of which the emergence of an effective industry association may be more the result than the cause. Recent literature has interpreted the emergence of clustering processes as resulting, once again, from an imbedded industry-wide ability to learn (Glaeser, 1998; Cook and Morgan, 1998). Learning is a cumulative process. Over and above the existence of institutions for formal learning, such as schools and universities, successful people and firms provide more opportunities for learning-by-doing and learning-by-interaction (Glaeser, 1998). Cooperation among firms, but also copying, and even grossly imitating successful experiences, are important drivers of these industry-wide learning processes.

In face of this discussion, we interpret the growth dynamics of the L&S sector in terms of the sector ability to develop an endogenous capability to learn largely based on informal processes. In contrast, development of the E&E sector was implicitly attributed to major investments in physical capital and machinery, a "traditional" way to absorb technology. These hypotheses were validated by the interviews reports, and are strengthened by the data presented in Table 3, which characterizes the two sectors in terms of the organizational and strategic innovations introduced by the firms that answered the questionnaires.

Table 5 compares the average values of ICT utilization intensities in the two sectors. Of the 30 L&S firms of the reduced sample, 5 indicated no workers that had used a computer at least one hour per day, while in the E&E sector only 2 firms indicated a total absence of computer usage. Whether the firms that reported zero KW did so because of difficulties in answering the questions, or because they did not, indeed, have any knowledge workers, we cannot determine. In at least one instance we can be certain that there were reporting problems, since one of the two companies in the E&E sector that reported zero knowledge workers is the Portuguese branch of a large IT multinational. Nonetheless, the difference in the level of reporting of zero KW is a clear indication that the L&S sector exhibits a lower level of ICT usage than the E&E industries

In general, the "cloud" of E&E points drifts upward and to the right, denoting high intensities of utilization of KW1 and KW2, typically above 0.25 for both. On the other hand, a majority of firms in the L&S sector is constrained in the (0, 0.25; 0, 0.25) area, with 4 firms barely escaping this region, and the remaining firms scattered throughout the graph, mostly with low levels of at least one type of KW.

In the L&S sector there seems to be a "great divide" between the firms that have achieved some proficiency in the use of KW1 alone, and those that utilize both KW1 and KW2. Of the 25 firms that reported the existence of knowledge workers in the L&S sector, ten, or 40%, said they used KW1, but utilized no connectivity technologies associated with KW2. Therefore, one could say the remaining 15 firms have "jumped" into a higher stage in which both KW1 and KW2 are present.

It is worthwhile to investigate the behavior of these 15 L&S firms that use both KW1 and KW2. If we take out of the analysis three "outliers" (a firm with KW2=1; one with KW2=0.5 but very low level of KW1; and, finally, one other firm with KW1=0.5 but with KW2 almost equal to zero), the remaining 12 firms are distributed as shown in Figure 4. The correlation coefficient is remarkably high, indicating an almost perfect linear relationship between KW1 and KW2 for the L&S firms that have jumped the barrier of Internet and e-mail adoption.

The behavior exhibit by the L&S firms, we argue, further reinforces our hypothesis that informal and tacit knowledge exchanges prevail in the L&S sector. Face-to-face contacts, trust, empathy, maybe spying, and other subtle human relationships cannot be achieved only through electronic interaction. Therefore, 40% of the firms that use ICT in the L&S sector have not tried to use the electronic connectivity tools permitted by e-mail and the Internet. Moreover, those that utilize KW2, the remaining 60%, seem to be doing it in parallel with the adoption of KW1. This last behavior can be the result of a "technology push" by software and hardware vendors (after all, the Internet icon emerges prominently in any PC desktop nowadays), or the mere exercise of a "fashion pull". There is little evidence that it is a manifestation of an explicit necessity felt by the industry toward the establishment of electronic connections. However, we must be cautious in putting forward this interpretation. We cannot establish whether the almost perfect correlation between KW1 and KW2 in the L&S sector corresponds to the effects of late "computerization" in the sector, or to a trend of the sector’s future. Nonetheless, the contrast with the E&E sector, which we will explore below, unveils some dramatic difference in ICT usage across the two sectors.

In fact, electronic interaction technologies, important facilitators for the communication of codified knowledge across remotely located sites, seem to be key for the E&E firms, or at least as important as horizontal and generic use of computer applications. A more refined analysis of the E&E firms gives us even deeper insights. Let us consider a 45 degree diagonal in Figure 3, separating the firms that exhibit a comparatively higher rate of Internet and e-mail usage from those relying on more prosaic office productivity tools. Which differences exist between the firms that are located above and below this diagonal?

Of our reduced sample of 38 firms, 12, or about one third, are located above the diagonal, and the remaining two-thirds below the 45 degree imaginary line. The most important difference between these two clusters is that the majority of the ones located above the diagonal are multinational or transnational firms, while those below the diagonal are primarily national firms. In fact, as Table 6 shows, almost two-thirds of the equity ownership of the firms above the diagonal is foreign, while in firms below the diagonal only one third is controlled by foreigners.

Therefore, our argument is that firms in the E&E sector displaying stronger and tighter links with interests outside of Portugal need to adopt comparatively more electronic connectivity tools than those firms that operate primarily in the Portuguese domestic market. Or else, they simply reflect multinational standard behaviors concerning the utilization of electronic communication. It should also be noted that those few firms above the diagonal that have not foreign owners are very much engaged in exports and even, in at least three cases, internationalization. A further reinforcement of this view is provided by the results reported in Table 7. We already showed in Table 3 the share of innovative firms that relied on different types of strategic alliances. Table 7 provides a more detailed information, showing for the firms in the E&E sector the nature of the alliances with international partners. It is clear that, regardless of the specific type of alliance to be considered, firms above the diagonal rely comparatively much more on international partners than do the other firms located below the diagonal.

In the E&E sector we found a more profuse usage of KW1 and KW2, but were able, still, to identify two clusters, according to the relative level of KW2 usage vis-a-vis KW1. In the group of firms with a higher intensity of KW2 relative to KW1 we discovered that foreigners owned the majority of the firms’ equity, while the remaining Portuguese controlled firms were very active in exports and internationalization. On the other hand, firms in which usage of KW1 is relatively higher than KW2 tend to have a more domestic focus. Consequently, firms in the first cluster need sophisticated electronic communications tools such as e-mail and the Internet.

In Table 5 we saw that the average utilization of KW3 was higher in the L&S sector, but the information provided by Figure 6 makes the contrast much more impressive, and rules out the possibility that the average is driven by one or two outliers. There is, in fact, one such outlier, which we will discuss below, but its position does not affect the entire distribution. Note that the distribution of KW3 in the L&S sector is to the right of the distribution in the E&E sector for high values of KW3. In fact, the first interval of KW3 intensity considered in the graph includes almost 80% of the firms in the E&E sector, while only slightly more than half of the L&S firms are included in this lowest of the intervals. Even if the "outlier" were to come from the highest to the lowest bin, the percentage of firms here would still be only 60%, and the remaining of the distribution would, obviously, remain the same.

Empirical evidence is provided to sustain our beginning hypothesis: that the traditional L&S sector has relied mostly on informal learning processes while the modern sector has based its progress on formal education and training coupled with heavy capital investment. The two different knowledge accumulation paths have proven equally successful in their respective fields although the L&S companies were able to display a relatively impressive record both in terms of output and productivity as well as in terms of employment generation.

Amsden, A. (1989). Asia’s Next Giant: Latecomer Industrialization in Korea, Oxford University Press: New York.

Arrow, K. (1962). "The Economic Implications of Learning by Doing", Review of Economic Studies, 28, 155-73.

Barro, R.J. (1996). "Determinants of economic growth: a cross-country empirical study", NBER wp 5698.

1. Barro, R.J., and Sala-i-Martin (1995). "Two-sector models of endogenous growth with special attention to the role of human capital", Chapter 5 in: Economic Growth, by Barro and Sala-i-Martin.

Becker, G. S. (1993). Human Capital- A Theoretical and Empirical Analysis with Special Reference to Education, 3^rd edition, Chicago: The University of Chicago Press.

Benhabib, J. and M.M. Spiegel (1994). "The role of human capital in economic development. Evidence from aggregate cross-country data", Journal of Monetary Economics, vol.34 no.2, pp. 143-173.

Benhabib, J., and M.M.Spiegel (1991). "Growth accounting with physical and human capital accumulation", Economic Research Report nr. 91-66, C.V. Starr Center for Applied Economics, New York University.

Bruton, H. J. (1998). "A Reconsideration of Import Substitution", The Journal of Economic Literature, XXXVI (June): 903-936.

Carneiro, R. (1994). "A dynamic de evolução dos sistemas educativos" [ "The evolutionary dynamics of educational systems"] , Colóquio Educação e Sociedade, Vol. 6/94 (July): 13-60, Fundação Calouste Gulbenkian, Lisbon, Portugal.

Carneiro, R. (1997). "Emprego: Sim ou Não na Sociedade da Informação" [ "Employment: Yes or No in the Information Society"] , Cybernet, Grupo Forum, Lisbon, Portugal.

Carneiro, R. et al. (1998). Future scenarios of the Internet and e-commerce in Portugal, Grupo Forum: Lisbon, Portugal (forthcoming).

Conceição, P., Heitor, M. V. (1998). "Perspectivas sobre o Papel da Universidade nas Economias Baseadas no Conhecimento"["Perspectives on the Role of the University in the Knowledge Based Economies", in Portuguese], Colóquio Educação e Sociedade, Nova Série, 2 (March): 70-98, Fundação Calouste Gulbenkian, Lisbon, Portugal.

Conceição, P., Shariq, S. (1998). "The Emerging Role of Universities in the Digital Economy- Preliminary Observations on the Patterns of Demand for Knowledge and Challenges and Opportunities Facing Universities in the 21^st Century", Colóquio Educação e Sociedade, 2 (March), 99-109, Fundação Calouste Gulbenkian: Lisbon, Portugal.

Cook, P., Morgan, K. (1998). The Associational Economy: Firms, Regions, and Innovation, Oxford University Press: Oxford, UK.

Crace, J. (1988). The Gift of Stones, The Ecco Press: London.

Denison, E. (1967). The Residual Factor, OECD, Paris, France.

Denison, E.F. (1962). "The sources of economic growth in the United States and the alternatives before us", Committee for Economic Development, Washington.

Easterly, W. (1997). "The Ghost of Financing Gap: How the Harrod-Domar Growth Model Still Haunts Development Economics", Policy Research Working Paper 1807, The World Bank: Washington, D.C.

Fagerberg, J. (1994). "Technology and international differences in growth rates", Journal of Economic Literature, Sept. 1994, pp. 1147-1175.

Feldman, M. P. (1994). The Geography of Innovation. Kluwer Academic Publishers, Dodrecht, The Netherlands.

Fuhrer, J. C., Little, J. S., (eds.) (1996). Technology and Growth: Conference Proceedings, Federal Reserve Bank of Boston: Boston, MA.

Gavron, R., Cowling, M., Holtham, G., Westall, A. (1998). Creating the Entrepreneurial Society, IPPR- Institute for Public Policy Research: London, UK.

Glaeser, E. L. (1998). "Are Cities Dying?", Journal of Economic Perspectives, 12(2), Spring: 139-160.

Howitt, P. (1996). "On Some Problems in Measuring Knowledge-Based Growth", in Howitt, P. (ed), The Implications of Knowledge-Based Growth for Micro-Economic Policies, University of Calgary Press: Calgary, Canada.

Jacobs, D. and De Man, A. P. (1996). "Clusters, Industrial Policy and Firm Strategy: a Menu Approach". Technological Analysis and Strategic Management, Vol. 4, 1996, pp. 425-437.

Jones, P. and Barr, A. (1996). "Learning-by-doing in Sub-Saharan Africa: Evidence from Ghana", Journal of International Development, Vol. 8, No. 3, pp. 445-466.

Karlsson, C. (1997). "Product Development, Innovation Networks, Infrastructure and Agglomeration Economies", The Annals of Regional Science, 31: 235-258.

Kirsch, J. L. (1998). "Devenir des bas niveaux de qualification: comparaison des situations nationales". Mimeo. CEREQ: Marseille.

Kyriacou, G. (1991). "Level and growth effects of human capital: A cross-country study of the convergence hypothesis", New York University Economic Research Report: 91-26, May.

Landes, D. (1969). Prometheus Unbound: Technological Change and Industrial Development in Western Europe from 1750 to the Present. Cambridge University Press.

Landes, D. (1998). The Wealth and Poverty of Nations: Why Some are so Rich and Some so Poor. New York: W. W. Norton & Company.

Landes, D. S. (1992). "Homo Faber, Homo Sapiens: Knowledge, Technology, Growth, and Development", Contention, 1(3).

Lau, L., D. Jamison and F. Louat (1991). "Education and productivity in developing countries: An aggregate production function approach", World Bank PRE Working Paper Series No. 612.

Lucas, R.E. (1988). "On the mechanisms of economic development", Journal of Monetary Economics, vol.22, pp.3-40.

Lundvall, B.A. (1992). National System of Innovation – Towards a Theory of Innovation and Interactive Learning, London: Printer Publishers.

Maddison, A. (1987). "Growth and slowdown in advanced capitalist economies: techniques of quantitative assessment", Journal of Economic Literature, vol.25, no.2, pp.649-698.

Mankiw, N., D. Romer and D.N. Weil (1992). "A contribution to the empirics of economic growth", Quarterly Journal of Economics, 107:2, pp.407-438.

Mathews, J. (1996). "Organizational Foundations of the Knowledge-Based Economy", Employment and Growth in the Knowledge-based Economy, Paris: OECD.

Nelson, R. (1991). National Innovation Systems, Oxford, UK: Oxford University Press.

Nelson, R. R. (1959). "The Simple Economics of Basic Scientific Research", Journal of Political Economy, 67, 297-306.

Nelson, R. R. (1997). "How New is Growth Theory?". Challenge, 40(5), 29-58.

Nordhaus, W.D. (1969). "An economic theory of technological change", American Economic Review, 59, pp.18-28.

OECD (1998). STAN Data Set, Directorate for Science, Technology, and Industry, Paris: OECD.

Pack, H. (1994). "Endogenous growth theory: intellectual appeal and empirical shortcomings", Journal of Economic Perspectives, vol.8, no.1, pp.55-72.

Pasinetti, L. L. (1993). Structural Economic Dynamics: A Theory of the Economic Consequences of Human Learning. Cambridge University Press: Cambridge, UK, New York, USA.

Pencavel, J. (1990). "The contribution of higher education to economic growth and productivity: a review", CEPR publication nr.191, Stanford University.

Petit, P. (1995). "Technology and Employement", in Stoneman, P. (ed.). Handbook of the Economics of Innovation and Technological Change. Blackwell: Oxford, UK, and Cambridge, MA.

Porter, M. (1990). Competitive Advantage of Nations, Free Press, New York.

Reynolds, P., Storey, D. J., Westhead, P. (1994). "Cross-National Comparisons on the Variation in New Firm Foundation Rates", Regional Studies, 28(4): 443-456.

Romer, P. (1990); "Endogenous Technological Growth". Journal of Political Economy, 98(5), s71-s102.

Romer, P. (1994). "The Origins of Endogenous Growth", Journal of Economic Perspectives; 8(1), 3-22.

Romer, P.M. (1990). "Human Capital and Growth: Theory and evidence", Carnegie-Rochester Conference Series on Public Policy, no.32, pp. 251-286.

Schultz, T. (1960). "Capital Formation by Education", Journal of Political Economy, 68(6), 571-583.

Schumpeter, J. (1911). The Theory of Economic Development.

Schumpeter, J. (1943). Capitalism, Socialism and Democracy.

Solow, R. M. (1956). "A Contribution to the Theory of Economic Growth", Quarterly Journal of Economics, 70, 1, 65-94.

Solow, R. M. (1957). "Technical Change and the Aggregate Production Function", Review of Economics and Statistics, 39 (August), 312-320.

Solow, R.M. (1997). Learning from Learning by Doing. Stanford University Press: Stanford, CA.

Steedman, H. (1998). "Low Skills – How the Supply is Changing Across Europe". European trends in occupations and qualifications, CEDEFOP (forthcoming).

World Bank (1997). World Development Report 1998: Knowledge for Development, (Annotated Outline). Mimeo.