Wise search engine based on LSI

Abstract

The objective of this work is to provide, as a search engine, latent semantic indexing (LSI), which is a classical method to produce optimal approximations of a term-document matrix and has been used for textual information mining. The use of this technique is examining mine content which based web document, using keyword features of documents. Experimental results show that together with both textual and latent features LSI can extract the underlying semantic structure of web documents, thus improve the search engine performance significantly. © 2010 Springer-Verlag Berlin Heidelberg.