A Frequent Pattern Mining Technique for Ranking Webpages Based on Topics

In this paper, we propose a frequent pattern mining technique for ranking webpages based on topics. This technique shows search results according to selected topics in order to give users exact and meaningful information, where we use an indexer with the frequent pattern mining technique to comprehend webpages’ topics. After mining frequent patterns related to topics (i.e. frequent topics) in collected webpages, the indexer compares new webpages with the generated patterns and calculates degree of topic proximity to rank the new ones, where we also propose a special tree structure, named RP-tree, to compare the new webpages to the frequent patterns. Since our technique reflects topic proximity scores to ranking scores, it can preferentially show webpages which users want.