Cross-domain Recommender Systems
Recommender systems typically recommend items of the same kind to users (e.g. a new book based on the books you have read before).
In this project we want to investigate whether it is possible to recommend items of one type (e.g. books) based on information on user's prefernces in another domain (e.g. wine) in an attempt to solve or improve the cold start problem.
- Obtain or construct a data set with cross-domain users-preferred items
- Apply data mining techniques to discover patterns
- Implement a cross-domain recommender system
Depending on your interests we can further elaborate our goals, we could look at more complex algorithms or analyze the impact of differences between different datasets.