Social networks are becoming increasingly popular and also increasingly valuable platforms. However, since their value is strongly influenced by the information quality of user profiles, assuring the correctness, currency and completeness (3C) of profile information is important for their success. In this paper we present an integrated approach for improving the 3C in social networks. Our approach is based on the idea that, if a user is active in a community which differs from the information community determined by her/his profile information, then there is a quality deficiency which needs to be improved. To identify the exact information quality problem, past community structure is considered. Thus, our approach is applicable to many different social networks and also considers the temporal dynamics of the network. Social network companies can apply it to achieve high-quality profile information, which is essential for many applications (e.g. head hunting on LinkedIn). We demonstrate this contribution, by applying it to the research social network ScholarMate. The initial results show that the activity and information communities of the user do not always overlap and that our approach effectively addresses information quality problems in real-world social networks.