Design And Implementation Of A Social Network User Account Correlation System

With the development of the function of social networking,The number of social network users showing an increasing trend in recent years, the users of social network can use different social networking sites to achieve different needs. Different social network have different functions, due to security considerations and interests, social networking service providers will not allow their site user account associated with other sites’ user accounts, thus creating a resource can not be fully utilized.This thesis mainly studies using the user account association algorithm to make multiple user accounts belong to the same entity users linked based on features of user behavior that extract from content data generated social network user account. Firstly,designing the structutre of the system and divided the system into two parts: data acquisition subsystem and the user account associated subsystem. Secondly, devising and implementing the data acquisition subsystem and the user account associated subsystem in turn, focusing on the study of design of network crawler, the extract of user’s behavior characteristics and improvementing and innovationing of the user account correlation algorithm. Finally, making experiments to test the system and analyzing the output results of the system. The work of this thesis mainly includes the following three aspects:(1) Design a web crawler that can fast and convenient crawled the generate content of the user account on social networking sites and with dynamic operation function, enhanced page parsing function and efficient database access function.(2) Put forward the new writing style features based on N-Gram and screened writing style features to filter out redundant features, improving the system’s processing speed and robustness.(3) Proposed a kind of social network user account association algorithm based on one class classifier and further improvements on the algorithm, improving the practicability and accuracy of the system.Finally, by crawling the user account data of social networking site Google+, Twitter and Facebook to make different system test cases based on the parameters of system and then test system. Through calculate the precision, recall and F-measure of system outputs to analyze and evaluate the performance of the system. In the best condition,the precision of the system can exceed 70%, the recall can reach above 75% and the F-measure reached more than 70%. The results show that the more the amount of data as the user account contained, the better the performance of the system.