In this paper, we present a preliminary study of the problem of characterizing and categorizing the interests (e.g., reading, jogging, java) that the individuals disclose on social networks. As these interests are expressed in natural language, we are confronted with a disambiguation problem in a limited context, as social network profiles have usually limited textual content. The approach we present here disambiguates an interest of an individual by determining a Wikipedia article that describes it; the other interests disclosed by the individual form the context. The results that we obtained on 392 interests of 50 user profiles of the social network LiveJournal are encouraging.