Social View Based User Modeling for Recommendation in Tagging Systems by Association Rules

Social tagging systems such as Facebook, YouTube, del.icio.us, Flickr become popular recent years and have achieved widespread success. State-of-art user modeling approaches in tagging systems usually use a vector of weighted tags. Unfortunately, typical user modeling methods using a vector of weighted tags which are based on personal view only and ignore the social view, have some inherent drawbacks. As in a social network like collaborative tagging system, it is subjective and incomplete to profile using only personal view. In this paper, a novel approach applying association rules is proposed to extend user profiles from the social view. The enriched user profile is a harvest from both personal view and social view. Algorithms of personalized recommendations for tags and items are presented. Also experimental results of using the profile we proposed are discussed.