The goal in information retrieval is to locate relevant documents in response to a user’s query at the same time retrieving as few as possible of the irrelevant documents. One possible approach to this problem is to use the Singular Value Decomposition (SVD) which models documents and queries as vectors in reduced space. The components of the vector are determined by the term weighting scheme, a function of the frequencies of the terms in the document or query. In this paper, we discuss term weighting schemes and the results from experiment on Malay text retrieval using a set of Malay document collection.

Singular Value Decomposition, term weighting, Malay document