Cross Domain Collaborative Filtering Recommender System for Academic Venue Personalization based on References

Finding the most appropriate venue to submit a scientific discovery plays a crucial rule in the acceptance results. However, an author is overwhelmed with the huge number of academic venues and often limits himself to known conferences where he already has publications. The use of recommender systems seems to be a good solution to filter out irrelevant items. Most recommender systems work on a single domain. Cross domain recommender systems allow to leverage ratings from multiple domains to improve recommendation results. In this paper, we propose a cross domain collaborative filtering based approach to suggest suitable academic venues. Our recommender system is able to deal with authors changing interests. It is also useful for young researchers who have not publications yet. We formalize the cross-domain problem for academic venues by investigating not only authors past publications, but also authors, from the reference list, past publications. Experimental results demonstrate the efficiency of our method.