Investigating the effective use of LSA matrix comparison for internet search answers

The principal objective of this research project is to analyse and apply the use of Latent Semantic Analysis (LSA) as a support mechanism for Internet searching. The research aim is to improve the standard of search engine results where accommodation in South Africa is the search key. LSA is a theory and a method for extracting and representing the contextual meaning of words by statistical computations applied to a large text section. It analyses word-word, word-passage, and passage-passage relationships. According to Deerwester et al (1990: 391-407), the mathematical component used by LSA uses the singular value decomposition (SVD) technique to decompose the existing text-document matrix. This is done to three other matrices of a very special form, the resulting matrices containing “singular vectors” and “singular values”. These special matrices show a breakdown of the original relationships into linearly independent components or factors (Deerwester et al,). This makes it feasible to compare words by paragraphs, paragraphs by paragraphs, and paragraphs by documents for the relevancy of data. Most of the existing search engines base their information retrieval purely on the keyword search mechanism. This implies that results are retrieved based on the matching of these keywords, ignoring the meaning and the sense they make towards documents to be retrieved. The strength of the proposed design is the ability to use the keyword technique, concentrate on the meaning of words and the sense they make in the webpage document. In order to test, analyse, and apply this technique and its ability, an implementation of a search tool based on LSA technology will be developed.