A pragmatic analysis of query expansion based on unsupervised learning

In this paper, we examine the significance of expansion of the user query by two techniques namely Efficient Clustering-By-Direction and Theme Clustering. These two techniques produce the clusters of keywords extracted from the set of retrieved documents for the user query. The former clustering is based on statistical approach whereas the latter clustering is based on semantic approach. Empirical analysis of set of keywords that provides the importance of user´s need produce the narrow search results. The clusters are further analyzed to provide the most appropriate keywords from the group of documents.