Thesis Overview

  This thesis proposes that:

1. Our hybrid-search system should provide more efficient query mechanisms than hybrid-search emulation on any one of the other tools independently.
2. Our system should demonstrate an efficient mechanism for storing the semi-structured information used in our system.
3. The system should demonstrate an efficient mechanism for extracting this information.

The next chapter (Chapter ) reviews previous efforts and gives a general overview the technologies used in this thesis. Chapter  describes a model for documents and why we believe hybrid-search will work well in this domain. Chapter  covers the implementation of the Haystack data structures. The Haystack data structures will provide us with the means of modeling the document structure. The service model provides the tools for converting documents into the Haystack data model structures. Chapter  details the Haystack to Lore interface, an essential component of the hybrid-search implementation. Finally, Chapter  discusses some conclusions and describes some future directions for Haystack. Appendix  discusses in great detail the design of the revised service model included in the new release of Haystack.