A search engine usually returns a long list of web search results corresponding to a query from the user. Users must spend a lot of time for browsing and navigating the search results for the relevant results. Many research works applied the text clustering techniques, called web search results clustering, to handle the problem. Unfortunately, search result document returned from search engine is a very short text. It is difficult to cluster related documents into the same group because a short document has low informative content. In this paper, we proposed a method to cluster the web search results with high clustering quality using graph-based clustering with concept which extract from the external knowledge source. The main idea is to expand the original search results with some related concept terms. We applied the Wikipedia as the external knowledge source for concept extraction. We compared the clustering results of our proposed method with two well-known search results clustering techniques, Suffix Tree Clustering and Lingo. The experimental results showed that our proposed method significantly outperforms over the well-known clustering techniques.
PageRank-based Word Sense Induction within Web Search Results Clustering
