| Difficulties with web-based full text information retrieval (IR) systems include spurious matches, manually intensive document sifting, and the absence of communication or coordination between users. Furthermore, it is not clear how to best design a Computer-Supported Collaborative Work (CSCW) system which possesses coordination mechanisms in the inherently stateless Web environment. Document collections in an Intranet are a particularly interesting problem; electronic documents contain the potential for improving information throughput and knowledge transfer in organizations. The emergence of Intranet infrastructure creates new IR demands as all users can easily publish documents online. The associated problems include inefficient discovery and proliferation of documents regardless of quality, which in turn implies the mismanagement of the authors' expertise. This thesis investigates how to design an IR system to support KM in organizations. I propose an architecture and implement a novel knowledge management support system, Annotate, to address these issues.; As a technical background, I first describe heuristics for designing software applications on the Web which focus on capturing usage history and altering the hypertext interfaces the user sees during the application session.; Next I apply these heuristics to introduce Annotate, a system which enables document annotations, and captures global usage history. Annotate provides improved data and metadata clues to guide the user in a search session and leverages the intelligence of the collective user community to enhance the power of traditional full text search. These features are enabled by novel coordination mechanisms between users, and between users and system administrators, made possible by Annotate's architecture.; The system is tested in a field setting and propositions are developed for its effective use.; The contributions are: (1) An architecture for enhancing IR and KM in an organizational setting, (2) Improving the semantics of the organization's document collections, (3) Analysis of preconditions for successful adoption of a system which leverages user intelligence via annotations, and (4) Proposal for follow-on experiments and technical enhancements.; The thesis concludes with how Annotate can be used to explore different KM policies in organizations. |
