Characterizing a social bookmarking and tagging network

Abstract

Social networks and collaborative tagging systems are rapidly\ngaining popularity as a primary means for storing and sharing data among \nfriends, family, colleagues, or perfect strangers as long as they have common \ninterests. del.icio.us is a social network where people store and share their \npersonal bookmarks. Most importantly, users tag their bookmarks for ease of \ninformation dissemination and later look up. However, it is the friendship \nlinks, that make delicious a social network. They exist independently of the \nset of bookmarks that belong to the users and have no relation to the tags \ntypically assigned to the bookmarks. To study the interaction among users, the \nstrength of the existing links and their hidden meaning, we introduce\nimplicit links in the network. These links connect only highly “similar” users. \nHere, similarity can reflect different aspects of the user’s profile that makes \nher similar to any other user, such as number of shared bookmarks, or \nsimilarity of their tags clouds. We investigate the question whether friends \nhave common interests, we gain additional insights on the strategies that users \nuse to assign tags to their bookmarks, and we demonstrate that the graphs \nformed by implicit links have unique properties differing from binomial random \ngraphs or random graphs with an expected power-law degree distribution.