The Study Of Detection Model For Special Group In Microblog

With the rapid development of network and information technology, there are moreways for people to interact with each other. Especially on the process of generation andpropagation of a topic in society, the microblog platform, as a means of spending information,plays a very important role in the social interaction and connects people invisibly. This paperaims to look for a specific group when given a specific topic related to it. Specific groupmeans that, in the microblog platform as a social media, although there is less or no directrelationship among users in this group on the microblog platform, the users hold the sameviewpoint or focus on the same problem. This group is difficult to found with traditionalgroup detection methods. They usually have following characteristics. Firstly, they are faraway from large communities with few followees and followers secondly, they rarely publishmicroblogs or publish microblogs with long intervals of time; thirdly, they only communicatewith members in specific groups, etc. In this paper, specific group is the object to study. Thereis also a comparison between the concept of specific group and community, and some relatedresearch about the key technologies of semantic analysis and relationship analysis. In orderto effectively extract the features of a special group, TF-IDF algorithm based featureextraction technology is put forward in this paper; semantic similarity algorithm is improvedand the indicator user-topic relevance i.e. UTR is proposed; in order to effectively measurethe potential semantic links between users the indicator of common attention level is alsoproposed. Then this paper mainly studies micro-blog special group detection model, andproposes algorithm to make user semantic analysis, user relationship analysis, and candidateselection. Three parts of the model, i.e. the model framework, the model definition, and themodel iteration process are formed. Finally experiments are set up to find the special groupwhich is related to the topic "only child losing" using micro-blog special group detectionmodel. Results are analyzed by user semantic analysis and user relationship analysis. It isfound that the model can effectively find the "only child losing" specific group, and theiteration algorithm saves the network resource and reduces a lot of manual work. In ourExperiments, we compares our micro-blog special group detection model with search rankingalgorithm from microblog platform and improved search ranking algorithm. Micro-blogspecial group detection model performs better than the other two algorithms.