Social recommendation is a method which requires the participants of both user's historical behavior data and social network, which generally belong to different parties, such as recommendation system service provider and social network service provider. Considering the fact that in order to maintain the value of their own data interests and user's privacy, none of them will provide data information to the other, two privacy preserving protocols are proposed for efficient computation of social recommendation which needs the cooperation of two parties (recommendation system service provider and social network service provider). Both protocols enable two parties to compute the social recommendation without revealing their private data to each other. The protocol based on the well-known oblivious transfer multiplication has a low cost, and is suitable for the application of high efficiency requirements. And the one based on homomorphic cryptosystem has a better privacy preserving, and is more suitable for the application of higher data privacy requirements. Experimental results on the four real datasets show those two protocols are efficient and practical. Users are suggested to choose the appropriate protocol according to their own need.
Bibliographical noteTitle in Pinyin: shù jù yǐn sī bǎo hù dí shè huì huà tuī jiàn xié yì
- Homomorphic encryption
- Oblivious transfer
- Recommendation system
- Yao's protocol