Index

1. List and breif explanation of papers with my opinion on how they can be utilized towards my thesis.

2. My Questions and Plan.

1. HCI papers and other papers regarding to event detection were from following sources.

HCI journals: Interacting with Computers 1997-1998, Human-Computer Interaction (Lawrence Erlbaum Associates) 1999 issues.

(Event) detection: MIT media Lab technical reports, W3C journal.

Let me state in the order that my thoughts have evolved by reading above papers.

First of all, I wanted to get sample data from other people who researched "event detection" At first, motion detection people seemed to be ones who do the event detection. So, I looked at technical papers of the media lab. 3D motion detection by using clusters of 2D points, or computer responses to certain user stimuli such as user frustration - related papers had possibly similar approach with mine. Their common approaches were to develop a mathmetical model of their application. Then description of application or algorithm followed, but usually no experimental data has any focus. i.e., mathmetical modeling is sole approach. 

It was about same with papers from the journal of Human-Computer Interaction. Let me skip this approach because I believe that quantitative analysis should be more emphasized in my thesis.(right, Geoffrey? ) 

Then I encountered this journal, Interacting with computers. This journal introduces itself as interdisciplinary journal , but most of authors have background as computer scientists or human computer interaction researchers ( mostly european countries ). They performed interesting experiments on human computer interface issues , and (therefore) could provide many valuable hypotheses which can be used in my thesis. Agan, however, most of their approaches were to explain their methods in models within human computer interaction such as task model, system model, etc. Experiment data itself is mostly by user study. Following is the more detailed explanation about papers that interested me regarding to my thesis.

1. Multimodal messages: the pen and voice opportunity. : goal of this paper was to show in asyncronous communication if we user both pen and voice, it has considerable potential advantage over single mode communication such as fax. They experimented a text-only communication and text and voice communication over the same group of subjects. This approach will be similar to mine in a sense that my experiment was comparing original SD and persomalized SD.  

2. Supporting user-adopted interface designLThe USE-IT system. : This paper described system features ,then used a metric to evaluate it. (IBM Usablity Satisfaction Questionnares) It interested me because I thought that I might compare my own "content-based" analysis with other usablity analysis metrics.

3. Synergistic modelling of tasks, users and systems using formal specification techniques.: Rather than forming a model to describe an application or features, it is about to "clarify the articulation between the task models and systme models" in CHI design practises. So they created examples to show their new apprach of modeling.

4. Using the LUCID method to optimize the acceptability of shared interfaces. : There were two things that interested me. One was the LUCID method. Similar methods could be used in my thesis too. The other was the shared interfaces by remote users, because WWW is an instance of such environments.

5. The use of the Internet as a research tool: It discusses advantage and disadvantage of using internet at research. As an example, they ask people through user-groups about Seasonal Affective Disorder. :)

6. A psychological investigation of long retrieval times on the WWW : Their experiment was to ask users how well they understood contents of web files which had different loading times at users' sides. When we know loading time of files are important in learning (it could be one of hypotheses in my thesis), it gets more important to measure events and time information at user session!

7. A flexible hypertext courseware on the web based on a dynamic link structure: The idea of adaptable links while a user was learning was interesting but it must be a preliminary experiment because they didn't have any conclusion, modeling, or data.

2. My Questions and Plan

After I read above papers, I strongly feel that they lack in one thing - quantitative data analysis. I was about to change my minds to do mathematical modeling like they did , but more I read, I do think to be able to present useful data will be more helpful than explaining the significance of my system in mathematical domain, frankly. I don't mean that mathematical approach is any less important though. Both should be done, I guess. But how would I legitimate my data? It was so hard for me to find peopole who 1. developed a adaptive learning system (there are quite a few people) and 2. have non-user study data like mine.

So. I think I will rather look at further the DOM and finish my tracking tools. Even though I end up not being able to find other sample data to compare with mine, to have such a quantitative data itself is new one and there are reasons to support why it should be done in that way too. Also, I will focus more on various experiment within Smart Desk environment, rather than employing any more evaluation methods or adding features.

Would you tell me what you think?

Thanks a lot, in advance.

August 28th, 1999

Jen (Myeongjin) Lee