| With the rapid development of network, social networking sites(SNS) gradually penetrated into people’s lives.The mobile internet in recent years has became increasingly popular, and SNS applications connects the traditional networks and mobile platforms,making more and more resources being shared on the net and resulting in the sharp increase in the amount of information.The huge amount of information in the social network provides massive resources to the user,but also makes the user feel difficult to get what they really want.It is hard for common people to get what they are interested. At the same time, vertical search technology has been catching a growing number of professional users’ attention, and its application in various fields has been widely studied.Vertical search technology is used in this article to help the SNS users get interested information. The article researches on key techniques of vertical search system, focused crawler,search results clustering and semi-structured information extraction and then propose performance improvement program for the SNS page.The traditional feature extraction methods don’t fit SNS pages.This paper puts forward an improved feature extraction method which integrates MI algorithm and x2statistics algorithm’s different performance on terms with high and low frequency,and at the same time introduces within-class word frequency factors and location factors to enhance the feature extraction effect.Use of HITS algorithm in the analysis of the SNS pages’ link is sometimes unreasonable.Based on the in-depth study of the algorithm of calculation of mutual enhancement relationship and topic keeping,here the paper raise an improved algorithm,using which to guide crawler’s fetch strategy we can get a significant improvement in the performance of fetching SNS pages.In addition, this paper raises an improved proposal on searched results clustering and semi-structured information extraction based on the traditional algorithm.The average similarity of text-based clustering algorithm based on K-means algorithm makes the clustering process enhanced on noise resistance,and also makes the pre-selected class center be more representative.Single-page multi-record information extraction algorithm,which supports the SNS pages using of AJAX technology,is raised based on the characteristics of the SNS pages.Significants improvements on application of vertical search in SNS can be seen through the experiments data after the improvements in these two areas.Finally, we developes The Party history education platform-Paving Stones and integrates the vertical search system designed into it to push information from outside to users.Integration tests show that the use of this vertical search system designed for SNS obtains good results and can meet the individual needs of different users. |
