Latent Semantic Indexing and Information Retrieval-A quest with BosSE
Institution: | Universität Heidelberg |
---|---|
Department: | |
Year: | 0 |
Record ID: | 1114975 |
Full text PDF: | http://www.ub.uni-heidelberg.de/archiv/6753 |
This master thesis deals with the implementation of a search engine using Latent Semantic Indexing (LSI) called BoSSE. Four different search types were implemented which allow a search for documents or terms similar to a given term, query or document. These search types are evaluated and the importance of term weighting, exclusion of non content words and the right selection of k for the reduction of dimension are discussed. Furthermore, an introduction to Latent Semantic Indexing (LSI) and an explanation of the Singular Value Decomposition (SVD) is given.