A cohesive page ranking and depth-first crawling scheme for improved search results

Abstract

The quality of the results collections, displayed to users of web search engines today still remains a mirage with regard to the factors used in their ranking process. In this work we combined page rank crawling method and depth first crawling method to create a hybridized method. Our major objective is to unify into one search system, the page rank method and the depth first crawling method to get into crawling the web at an efficient rate for a better recall and precision. The page ranking and depth first systems respectively are aimed at producing increase recall and ranking backlinks of a query to a matched document. The unification now is to help produce additionally, the most authoritative pages in a search result, thereby increasing the precision thereof. The methodologies adopted are the document and query likelihood models. Documents or corpora of known measures in query types, recalls and precision from the Text Retrieval Conference (TREC), the Initiative for Evaluation of XML retrieval (INEX) and REUTERs collection, were used as work bench for evaluation of the system. The results obtained showed significant improvement from results if implemented using the depth-first method and page ranking method as a single entity.

Keywords: Page Rank, Authoritativeness, Outbound link, Inbound link, Depth-First, crawling