Applied research of query-biased summary in the relevance judgment of chinese web pages

In order to overcome the shortcomings of insufficient information return by traditional search engine, improve the speed and accuracy of relevance judgment, a new sentence scoring method is proposed, which incorporate weighted query phrases and the structure information into sentence scoring based on the phrase recognition of the query and the analysis of structure of Web pages. Mutual information (MI) between keywords is used to recognize query phrases; Useful information for the summary is mined from the structure of Web pages. Experimental results show that the improved algorithm performs better in the speed and accuracy of relevance judgment than the existing methods.