Item-based collaborative filtering recommendation algorithms
Proceedings of the 10th international conference on World Wide Web
Mining newsgroups using networks arising from social behavior
WWW '03 Proceedings of the 12th international conference on World Wide Web
Proceedings of the 2006 international workshop on Mining software repositories
Modeling relationships at multiple scales to improve accuracy of large recommender systems
Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining
Measurement and analysis of online social networks
Proceedings of the 7th ACM SIGCOMM conference on Internet measurement
Recommendations in taste related domains: collaborative filtering vs. social filtering
Proceedings of the 2007 international ACM conference on Supporting group work
Recommending topics for self-descriptions in online user profiles
Proceedings of the 2008 ACM conference on Recommender systems
TANGENT: a novel, 'Surprise me', recommendation algorithm
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
Collaborative filtering recommender systems
The adaptive web
Content-based recommendation systems
The adaptive web
Hi-index | 0.00 |
Social networking service has become very popular recently. Many recommendation systems have been proposed to integrate with social networking websites. Traditional recommendation systems focus on providing popular items or items posted by close friends. This strategy causes some problems. Popular items always occupy the recommendation list and they are usually already known by the user. In addition, items recommended by familiar users, who frequently communicate with the target user, may not be interesting. Moreover, interesting items from similar users with lower popularity are ignored. In this paper, we propose an algorithm, UBI, to discover unknown but interesting items. We propose three scores, i.e., Quartile-aided Popularity Score, Social Behavior Score, and User Similarity Score, to model the popularity of items, the familiarity of friends, and the similarity of users respectively in the target user's personal social network. Combining these three scores, the recommendation list containing unknown but interesting items can be generated. Experimental results show that UBI outperforms traditional methods in terms of the percentages of unknown and interesting items in the recommendation list.