SNS is a platform to build social networks or social relations among people based on their social graph It is not satisfied to user�s preference on friend selection in real life. Mean while in prosposed system,we recommend friend by used semantic-based (or)user based on their lifestyle. By taking merits of sensor-rich smartphones, Friend matching graph find out something life styles of people from user-centric sensor data, action to achieve something the similarity of life styles between users, and to advise someone to users if their life styles have most similarity. An extra ordinary quality by data mining, A user�s daily life documents are extracted by using the Hierarchical dirichlet algorithm . Past a certain point ,a similar metric to measure the similarity of life styles between users and enumerate the recommend user�s the action of one object coming forcibly into contact with another. When receiving a request ,It returns a list of social network user with highest recommendation scores to the query user. At last, Friend matching graph combine with another to form a whole the feedback mechanism to further improve the recommendation user accurate. We implemented on the Android-based smartphones, and its performance on both small-scale experiments and large-scale simulations. The results show that the recommendations defect reflect the preferences of users in choosing friends in social network.