Efficient And Secure Collaborative Filtering Through Intelligent Neighbour Selection
ECAI'04: Proceedings of the 16th European Conference on Artificial Intelligence(2004)
摘要
In this paper, we introduce novel neighbourbood formation and similarity weight transformation schemes for automated collaborative filtering systems. We define profile utility, which models the usefulness of user profiles for collaborative filtering as a function of the items they contain. We demonstrate that our approach leads to more efficient and scalable collaborative filtering when compared to a benchmark k-Nearest Neighbour approach, while providing system accuracy and coverage to the same standard. In particular, we show that our approach is completely secure against the malicious attacks outlined in the paper, whereas k-NN proves very vulnerable.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络