Latent Semantic Indexing with selective Query Expansion.
Andy Garron, April Kontostathis · 2011
This article describes our experiments during participation in the Legal Track of the 2011 Text REtrieval Conference. We incorporated machine learning, via selective query expansion, into our existing EDLSI system. We also were able to expand the number of dimensions used within our EDLSI system. Our results show that EDLSI is an effective technique for E-Discovery. We also have shown that selective query expansion can be a useful mechanism for improving retrieval results when a specific initial query is provided. We believe that queries that are stated in general terms, however, may suffer from “expansion in the wrong direction ” when certain iterative approaches to incorporating relevance feedback information are incorporated into the search process. I.