Coordinate search method used for reordering search results

Abstract

The invention provides a coordinate search method used for reordering search results. The method mainly aims at improving the ordering quality of search results by combining the search experience of users who have the same search intention. The method comprises four stages of pretreatment stage, user interest learning stage, community calculating stage and ordering score calculating stage. Beginning from query logs of a Web search engine, the method analyzes the search activity of users and establishes a relational matrix taken as a potential knowledge base for inquiring a characteristic word in the pretreatment stage, then learns the interest model of each user from the segment of inquiry session of users, integrates the interest models and potential knowledge bases of users, dynamically calculates the composition range of communities, and finally endows the search experience of community members with different weights according to the similarity between all members and activity users so as to regulate the sequence of search results and improve ordering quality.

Description

technical field [0001] The invention relates to a collaborative retrieval method for reordering retrieval results, which is suitable for collaborative retrieval of Web information. Background technique [0002] The ranking of retrieval results has a direct impact on the quality of user queries. When each document in the retrieval results is given a reasonable score as a sorting basis according to its own relevance and importance, the query process is satisfactory; on the contrary, if the scoring results of the documents in the retrieval results are not reasonable, it will Produces a poor user experience. [0003] To rank documents by evaluating the importance and authority of a specific document, a lot of research has been carried out and many methods have been proposed, among which the two most famous methods are HITS and PageRank. However, the HITS algorithm requires more query processing time and is more susceptible to spam links; while the PageRank algorithm calculates...

AI Technical Summary

Problems solved by technology

However, the implementation process of the existing collaborative retrieval technology ignores an important step of Web collaborative retrieval, that is, the establishment of community

In the existing methods, the composition of the community is pre-defined by the user or the system. This mechanism often ignores the similarity between two users, and assigns the same weight to each member of the community in the ranking process, reducing the the effectiveness of the system

In addition, there are: (1) the type of query must be defined in advance, and the quality of the pre-definition directly affects the quality of sorting; (2) the establishment of the user description file uses the method of support vector machine, if the user's interest changes, the difficulty of updating larger technical issues

Images