Sem-Rank: a Page Rank Algorithm Based on Semantic Relevancy for Efficient Web Search

Abstract

The search engine produces numerous result for a single query, but the question is how relevant the result towards the query. The results provided by the search engines are not that much relevant to the query submitted, due to the ranking measures adopted by the search engines. The search engines used now a day return results based on the contextual information not by the content of the document. The ranking measure adapted by the familiar search engine also based on number of visits and time spent on the web page. For a personalized search the search engine has to produce some personalized results according to the user interest. We propose a new ranking algorithm using the semantic measures, which shows the semantic relevancy of the web document towards the query submitted. The query categorization is performed by computing same semantic link measure for the query towards set of domain ontology O. Given a web document set Ds of n documents, the category of each document Di is identified by the semantic link measure between each document term set Ts and set of concept or classes from domain ontology O. The document which comes under the category of query are identified and ranked based on the semantic link measure. Based on semantic link measure and depthness we compute a cumulative semantic measure which represents the document closure about a domain. We used domain ontology and word net to compute these measures for efficient ranking.