Software Redesign of the Haystack Project Software Redesign of the Haystack Project Acknowledgements

Most current large-scale information retrieval systems are impersonal, allowing uniform access to a non-modifiable corpus of information. The Haystack project is a personal information repository, which employs artificial intelligence techniques to adapt to the needs of its user. This thesis outlines a new overall design for Haystack. The design includes an RDFbased data model, a transaction-based storage system, services which are triggered by patterns in the data, a kernel with the machinery to invoke and execute these services, and a query processing system. In addition, the thesis serves as a valuable resource to future Haystack researchers by explaining and documenting the fundamental ideas of Haystack, as well as the reasoning for all design decisions. Thesis Supervisor: David R. Karger Title: Associate Professor