Folksonomy-Based Ontological User Interest Profile Modeling and Its Application in Personalized Search Xiaogang Han 1 , Zhiqi Shen 2 , Chunyan Miao 1 , and Xudong Luo 1 1 School of Computer Engineering, Nanyang Technological University, Nanyang Ave, Singapore 639798 {hanx0009,ascymiao,xdluo}@ntu.edu.sg 2 School of Electrical and Electronic Engineering, Nanyang Technological University, Nanyang Ave, Singapore 639798 zqshen@ntu.edu.sg Abstract. Information overload on the Internet is becoming more and more in- sufferable. The accurate representation of user interests is crucial to a successful information filtering system that are used to solve the issue of information over- load. To model the users’ interests more effectively, this paper investigate how to collect user tags from folksonomy and map them onto an existing domain on- tology. The experiment that integrates our user interest profile model to a Web Search Engine shows that our approach can accurately capture user’s multiple interests at the semantic level, and thus the personalized search performance is significantly improved compared with the state-of-the-art approaches. Key words: User Modeling, User Interest Profile, Folksonomy, Ontology 1 Introduction The amount of information resources on the Web nowadays is so enormous that the retrieval of relevant information is getting more and more difficult. To take this chal- lenge, various personalization, recommendation, and information filtering techniques are developed. The main idea behind these techniques is to adapt relevant information to users according to their short and long-term interests [7]. As the popularity of Web 2.0, users can not only consume contents from the Inter- net but also contribute contents to the Internet. Thus, social tagging systems, such as Delicious, 3 Last.fm, 4 and Flickr, 5 are developed to enable users to tag online resources (bookmarks, music, images, and so on) with freely chosen non-hierarchical annotations (e.g., tags). Tags provide a textual vector representation of the resource’s features, re- gardless of the type of the resource. Different from automatically generated metadata and the metadata annotated by the authors of the resources, user-created tags are a reflection of the user’s own interest space for online resources. As the users share their tagging with the public, the collaboratively created tagging data in a social tagging system constitute a folksonomy, which explicitly provides abundant information about their interests. For example, [21] analyzed personal data in folksonomies and investi- gated how to generate and represent multiple user interests; and [17] harvested user interests across multiple social networking sites. 3 http://delicious.com/ 4 http://www.last.fm/ 5 http://www.flickr.com/