Tag-based user modeling using formal concept analysis

Social tagging systems have been largely adopted by users as useful and powerful tools to organize, browse and share interested resources on the web. All the tags used and resources collected by a user constitute the userpsilas personal tag space, which contains valuable information that can be used for building and enhancing the user model. In this paper, we propose an approach to mine user profile from onepsilas personal tag space. The proposed approach is based on the theory of concept lattices, which provides a powerful, well-founded, and computationally-tractable framework to model onepsilas personal tag space in which tags and collocated resources are represented and to compute such a transformation. We define several properties for each concept generated to evaluate the degree of such concept to depict the userpsilas interest, and then a novel algorithm is introduced to construct a hierarchical user profile from the built lattice. Experiments are carried out on the dataset collected from delicious and a detailed real-life case study is presented. Results show that the hierarchical user profile can describe the target userpsilas interests well enough, as well as organizing and browsing the tags and collocated resources effectively. Several research issues and technical details are discussed to further improve tag-based user profile for personalized services in the end.