In this paper, we propose a tag recommendation algorithm for profiling the users in Sina Weibo. Sina Weibo has become the largest and most popular Chinese microblogging system upon which many real applications are deployed such as personalized recommendation, precise marketing, customer relationship management and etc. Although closely related, tagging users bears subtle difference from traditional tagging Web objects due to the complexity and diversity of human characteristics. To this end, we design an integrated recommendation algorithm whose unique feature lies in its comprehensiveness by collectively exploring the social relationships among users, the co-occurrence relationships and semantic relationships between tags. Thanks to deep comprehensiveness, our algorithm works particularly well against the two challenging problems of traditional recommender systems, i.e., data sparsity and semantic redundancy. The extensive evaluation experiments validate our algorithm’s superiority over the state-of-the-art methods in terms of matching performance of the recommended tags. Moreover, our algorithm brings a broader perspective for accurately inferring missing characteristics of user profiles in social networks.