A recommendation system is a software program which endeavors to limit down choices for user in view of their communicated inclinations, past conduct, or other information which can be mined about the user or different users with comparative interests. Existing social networking services recommend friends to users based on their social graphs, which may not be the most appropriate to reflect a user’s preferences on friend selection in real life. We introduce Friendbook, a novel semantic-based friend recommendation system for social networks, which recommends friends to users in view of their ways of life and interest rather than social charts. Inspired by text mining, we model a user’s daily life as life documents, from which his/her life styles are extracted by using the Latent Dirichlet Allocation algorithm. Friendbook discovers life styles and interest of users, measures the similarity of life styles and interest between users, and recommends friends to users if their life styles and interest have high similarity. We propose a unique similarity metric to characterize the similarity of users in terms of life styles and interest and then we construct a friend-matching graph to recommend friends to users based on their life styles and interest. We integrate a linear feedback mechanism that exploits the user’s feedback to improve our system accuracy