New Similarity Measure for Web Documents Retrieval

Web information retrieval (WIR) faces the challenges in finding more relevant information due to unstructured characteristic of the data and the tremendous growth in the size of the information. As a result, the users in search of information face difficulties in obtaining specific information on the web. The similarity measures play significant role in serving the machines to deal with the natural language. Various similarity functions used in the context of WIR to match between query and the documents. In this study, a new version of similarity measure for documents retrieval (SMDR) has been proposed. The experiments are conducted on various queries over CACM and RCV1 document collections to evaluate and analyse the performance of SMDR. The standard particle swarm optimization algorithm is used to optimize the entire WIR process. The comparative study is carried out considering the three important similarity measures in the context of WIR such as Cosine, Dice and Jaccord. The results show that significant improvement in the performance of the proposed similarity measure in retrieving the set of relevant documents.