News Retrieval Based On Latent Semantic Index And Clustering

Abstract- Web is a collection of heterogeneous as well as unstructured set of news articles. This paper presents a novel approach to retrieve relevant news articles from heterogeneous and unstructured collection of articles. Efficient retrieval requires analysis of news articles based on keyword. Two problems that occur in the analysis of news articles are synonymy and polysemy. In this paper we present a News Retrieval approach based on Latent Semantic Index LSI and Clustering. It includes projection of keyword-news article matrix into small spaces called clusters. After that clustering approach is used to group relevant articles into clusters.