Research on Social Relation Extraction of Web Persons

The Web contains a lot of information about personal social relations, how to efficiently and accurately extract social relations of Web persons is the focus of this study. According to the Web information, this paper classifies and defines the social relations, and selects eight kinds of personal relations to extract their feature words based on a Tongyici Cilin, and then uses the limited nature of the feature words to obtain the related personal names sets. At last, this paper calculates the co-occurrence rate of the Web personal names, the related names and the feature words to extract the personal social relations. The experimental results show that this method's average accuracy rate is over 85%.