Search-In-a-Box SIB

The first generation of corporate intranets, labeled by some a s, attempt to implement some of the vision of the Semantic Web coupled with personalization and groupware suites. These systems are quite primitive in their text processing and thus perform patchily. Moreover they have huge installation costs and restrictions on interoperability with other systems. Cost effective information management software for organizations, libraries, and educational and research institutions are thus non-existent. With the advent of distributed search just over the horizon (for instance, allowing the database-backe d to be exposed to search), the generally recognized perception that keyword search is a wasteful paradigm, and the recognition that search is a productivity tool that needs personalization, we propose to research and develop software with the following features:

• it can operate as an appliance (i.e., plug into network and play with minimal maintenance);
• it is open source (encourages longevity and non-corporate usability);
• it employs language models hidden from the user so that named entities, synonyms, topics and genre implicitly dealt with to ease the user frustration associated with keyword search;
• it employs probabilistic query evaluation models to overcome the lack of URL (link based) information in some intranets and to carefully integrate relevant synonym information, etc.;
• it integrates personalization tightly with search to further improve the user experience, but moreover so that all subsequent groupware tasks such as message routing, information alerts, filtering, etc. can be easily integrated at a later date.

The research project will build on existing theoretical advances funded by the Academy of Finland, and couples with industry partners in key application areas so that prototypes can demonstrate and test the open source systems developed. This multi-disciplinary research will be performed by the Complex Systems Computation Group at Helsinki Institute for Information Technology, the Tampere Unit for Computer Human Interaction at the University of Tampere, and the Technology Programme at the Helsinki Institute for Physics. A significant part of the research is in cooperation with the University of California at Berkeley, Tsinghua University in Beijing and supported by the FP6 projects ALVIS. In addition to the strategic research activities, the project will develop a series of useable prototypes starting from the first year. This is necessary to fully understand the interactions and requirements of individual modules.

SIB related publications can be found on the CoSCo research group publications page.