Privacy-preserving social DNA-based recommender

This research studies current privacy-preserving and familiarity-based recommender systems and introduces a new recommender system based on these existing techniques that uses DNA-similarity between users of the system as a basis for generating recommendations. Techniques for privacy-preserving DNA-matching are researched and incorporated in the setting of a privacy-preserving recommender system. Somewhat homomorphic encryption is used as the standard encryption for both the privacy-preserving DNA-matching and the recommendation generation. The system is first devised in a semi-honest security model and is then extended to the malicious user model. The recommender system is implemented in both security models and its performance is compared to existing privacy-preserving recommender systems.