Full metadata record
|dc.description.abstract||A tag is the simplest and effective tool for a user to classify contents actively and share with other users. As social tagging system becomes active, a tag is considered to be a tool to present a user's preference, rather than a simple classification tool. The recent noticeable studies are on network based user profiling comprised of User-Tag-Item and personalized recommendation techniques based on the profiling. Unlike previous works that consider a user's tag independent, the studies use the relation between tags for profiling a user's preference. However, the network based user profiling techniques proposed so far fail to take into account the semantic correlation between tags. In addition, a network based personalized recommendation technique has low precision because of the simple scoring of a tag weight for an item. To solve the problem, this study proposes an item recommendation technique applying a user's preference, which uses foxonomy to cluster tags semantically and then applies PageRank algorithm for profiling users' preference in clustered tags. In addition, by analyzing the suitability of the clustering technique for the tag-based semantic user profiling and comparing precision between the proposed algorithm and a conventional algorithm, this study shows the excellency of the proposed technique.||-|
|dc.title||Tag Network Diffusion-Based Recommendation Technique Exploiting Semantic User Profiling||-|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.