WIDIT in TREC 2004 Genomics, Hard, Robust and Web Tracks

To facilitate understanding of information as well as its discovery, we need to combine the capabilities of the human and the machine as well as multiple methods and sources of evidence. Web Information Discovery Tool (WIDIT) Laboratory at the Indiana University School of Library and Information Science houses several projects that aim to apply this idea of multi-level fusion in the areas of information retrieval and knowledge organization. The TREC research group of WIDIT, who engages in examination of information retrieval strategies that can accommodate a variety of data environments and search tasks, participated in the Genomics, HARD, Robust, and Web tracks in TREC-2004. The basic approach of WIDIT was to leverage multiple sources of evidence, combine multiple methods, and integrate the strengths of man and the machine. Our main strategies for the tracks were: the use of gene name thesaurus in the Genomics track; query expansion and relevance feedback in the HARD track; query expansion with keywords from Web search in the Robust track, and the interactive system tuning process called “Dynamic Tuning” in the Web track.