Mining New Trend on the Web to Improve Search Engine Results

摘要

The "rich-get-richer" phenomenon caused new and high quality pages had no chance to get users' attention. To solve the problem, this paper analyzes new trend on the Web-the relationship between the frequency of pages' updating and the attention of topics, then proposes topic-attentive ranking algorithm in which the attention of relational topics is applied to capture the importance of Web pages. Furthermore, we introduce time series analysis and incremental updating method to the original topic-attentive algorithm to improve the performance of search. Theory analysis and experimental results both show our proposed algorithm could provide more accurate results on the precision, and effectively alleviate the biased ranking problem for new important pages.