A Web Page Scoring Method for Local Web Search Engines

摘要

Web-page scoring is a method to improve Web search-engines by assigning a score to each page according to its importance. The PageRank algorithm implemented for Google is a well known efficient scoring method for WWW search-engines, whereas it is not efficient for searching a local Web. For the latter case, text matching is usually used for computing scores and the hyperlink structure of Web-pages is wasted. Although a method for scoring local Web-pages called the HotLink method has been proposed, it is not well established because the scores depend on how to extract a tree structure, which is unknown, from the Web-graph. In this paper, we solve the problem of the HotLink method by considering all shortest-path trees and taking the average score. As a result, the scores are independent of the selection of a tree, which makes the scores robust. We also propose an efficient algorithm to compute this average score in O(|V||E|) time where V and E is the set of pages and hyperlinks of a local Web-graph, respectively. Experimental results show that our new scoring method captures important pages in a local Web.