Research on social network based on meta-search engine
Yang Shen,Liu Zi-tao,Cheng Luo,Ye Li
DOI: https://doi.org/10.1109/WISA.2009.21
2009-01-01
Abstract:In order to solve the problem that we can only collect data from one single data source at some fixed time after mining the keywords in a rather superficial level, and to take full use of the information returned by search engines to construct the social relationship network based on the semantic link of the searched subject, we do the regular research by using the ROST Content Mining System which helps to undergo the process of page monitoring, content analysis and social network mining based on the pages returned from the four search engines (Google, Baidu, Sougou and Youdao). In the mining process, we adopt the cross-page framework adaptive algorithm which helps to solve the instability problem of the HTML framework codes, to extract information from the acquired web pages. Then we extract the cooccurrence set of high-frequency characteristic words to create the tridimensional social network graph by adopting the progressive search algorithm in the meta-search engine to extend the attribute set of the keywords. Finally, we conducted three typical case studies. They are the comparison of the coverage rate between Google and the meta-search engine, the dynamic changes in real-time network based on the meta-search engine and the progressive mining of effective content in meta-search engine, which all showed the advantages of the method in which we proposed the meta-search engine, as we could have more data sources, stronger real-time dynamic monitoring capacity, and deeper progressive searching ability. So we propose this meta-search engine method which can be used in social network study, aiming to develop the quality of the social network based on content mining, observe the hiding relationships in deeper levels and widen the research scope of content mining. © 2009 IEEE.