Name of company or organization: Community School Networks, Inc. Contact Person: Name Gary Markovits Title President Address 3 City/State/ZIP code Syracuse NY 13244 Telephone # (315) 443-4456 Duration of support by this sponsor for this project (start/end dates) 9/94 - 8/96 Sponsor expectations Assistance in developing optimal computational strategies for implementing their text indexing and query algorithms. Access to NPAC high performance computers. Support investigation of intelligent, concept-based text retrieval from very large text databases Specific objectives Use of large disk arrays (30-70 GB) and multiple nodes of the IBM SP2 for indexing large text databases. Consultation and advice on strategies for parallelizing both the index and query processing. Consultation and joint project supervision on exploring the potential advantages of storing the index in a relational or object oriented database. Relevance to the Center's target industry Study of the use of parallel computers and databases in text retrieval specifically and in information systems more generally is at the cutting edge for this growing industry segment. Plan of approach NPAC and TextWise technical staff jointly analyse indexing and retrieval algorithms and evaluate possible strategies using NPAC expertise in parallel computing and relational database. NPAC and TextWise may jointly supervise student projects that prototype selected strategies. TextWise will be provided occasional periods of exclusive of NPAC high performance computers for indexing large text databases and for experimentation. Purpose of any planned travel None planned. TextWise is local. Industry involvement TextWise is developing potential partnerships with other Infomall members. They are also being approached by large publishing houses that want to make their primary information collections available (for a fee) online with sophisticated search capability. Technical innovation TextWise has unique concept-based retrieval capability that is computationally intensive and requires high bandwidth dataflow. NPAC has an unusual array of expertise that can evaluate the range of possible strategies including both parallel and distributed computing, different input/output configurations of the possible platforms, and use of relational database for the index. Steps to commercialization TextWise is a start-up company that builds on software developed in the SU IST department, mainly under ARPA funding. It has attracted venture capital, and maintains a research and development branch in offices provided by CASE at SU, and a commercial branch in Rochester NY. Progress to date The orginal software has been ported to several platforms and the original interface replaced with a WWW-based query interface. Several meetings have taken place on HPCC strategy, with preliminary conclusions. Several text databases have been indexed using the IBM SP1 and SP2 computers at NPAC. A student summer project supervised at NPAC began looking at the issues involved in using a database for the index and translating queries into SQL. Impact to date The task of productizing TextWiseUs software requires extensive utilization of advanced computational resources and facilities. Without access to the NPAC computing facilities, this process would be very cost prohibitive to a small, startup company. In addition, InfoMall staff personnel have worked closely with TextWise. Once again, highly trained technical specialists may be too expensive to be utilized by most small businesses.