Hybrid recommender system with multi-source data and social knowledge graph integration

Accurately recommending items of interests is essential for users to improve their experience. To acquire better performance, a recommender system can utilize multiple sources of data and the social knowledge graph. This can lead to efficient use of information to improve the recommender system. By exploring the data and extract crucial features to feed to designed models, the recommending engine can increase its performance dramatically. Furthermore, a social knowledge graph contains the description and relation of users, which can act as a knowledge base for inference. By fusion of those techniques, the resulting recommender system can be optimized.

Faculty Supervisor:

Peter Marbach

Student:

Yiwen Feng

Partner:

AppDirect Canada Inc

Discipline:

Other

Sector:

Professional, scientific and technical services

University:

University of Toronto