The Study And Implementation Of Tie Strength Prediction Module Based On Microblog

With the rapid development of IT industry, speed broadband network and high-speed mobile networks increase dramatically, and various types of terminal devices that can access network become increasingly popular, which greatly improves the speed and frequency of communication over the Internet. Currently, Microblog has become an important platform for people to exchange ideas online, a place of gathering of network public opinion, and an important window for people to communicate with others and for government to insight into public opinion. Behind the high user activity and a huge amount of microblog posting, there is a great hidden data value. And an accurate understanding of the interaction of microblog users, the relationship between users and their behavior data which contained information and impact through proxy data is significant. Therefore, the paper analyzed and studied the relationship between users based on microblogging platform, the main research content and results are as follows:1. Through the collation and analysis of current study status of relationship between microblog users, tie strength was selected as the study object, and tie strength prediction module was designed based on user attentions and interactions of microblog. Virtual social networks are mapping the real social networks, and the factors which affected the tie strength of characters were also similar. Therefore, this module considered factors affected the degree of tie strength as many as possible in real world, and migrated it to sina microblog platform to extract available information which was reflect the tie strength between the users. Then using similarity algorithm, standard deviation and other mathematical algorithm to transform the available information into specific numerical information, and composite multiple factors through the linear model and analysis quantitatively ultimately.2. The thesis implemented the tie strength prediction system which was based on sina platform. Data acquisition, which played an important part in the system, used two methods to collect data, the one was a method based on simulated login, and the other one was a method based on API provided by sina. And the system implemented the automatic extraction of relationship information of users, personal information, and microblog information. Meanwhile, the combined use of two microblog data collectionmethod not only avoided the limitation of direct use of the API and the incomplete data caused by the unsuccessful login of the user, but also reduced the workload of analysis and extraction of massive data.3. Through using the part of data acquisition of the tie strength prediction system which was based on the tie strength module, the related data was collected, analyzed, and calculated to predict the tie strength of characters. After achieving the final results, a graphic interface was showed to express the results which also proved the correct and effectiveness of the proposed tie strength prediction module designed and the system implemented.Tie strength prediction studied in the thesis could help to build the more accurate division of relationships between the microblog users, and it could support the higher accuracy of recommendation of friends, help to find public opinion and improve the accuracy of it, and identify user’s friends to protect users’ privacy.