Semantic similarity-based PageRank using wordnet

Abstract

With the huge volume of web pages that exist today, search engines play an important role in finding the required information. It orders search results by performing link analysis. However, existing link analysis techniques have not considered the semantic similarity among the linked documents for rank calculation. Since links from semantically similar documents are more important than the links from other dissimilar documents, this work introduces a new method for ranking web pages based on the semantic similarity among the web pages and the link structure. Wu and Palmer (1994) measure of wordnet is used to find the semantic relationship between the terms in different documents. Cosine similarity measure is used to find the similarity among the documents. Proposed technique is compared with existing ranking algorithms using the measures precision, recall and F-measure. From the results, it is observed that the proposed method brings more relevant documents to the beginning of the list of search results than the existing methods.