Social intercourse based on interest is becoming more and more popular, design of the existing architecture and the business logic of the light blog by taking analysis is implemented. Through recording and analyzing users’ behavior and data profiles, this system can automatically push the new useful information to the specific users who may have an interest on that by data mining. In this paper, the application of the data mining technique, the implementation of TF-IDF algorithm and the solution of the model similarity are mainly explained.