Expert Recommendation for Knowledge Management in Academia

Recommendation systems are not only important in e-commerce, but in academia as well: They support scientists in finding relevant literature and also potential collaboration partners. It is essential that such a recommendation system proposes the most relevant people. Scientometric similarity measurements like co-citation and bibliographic coupling analysis have proved to give a good representation of research activities and hence it can be said that they put authors with similar research together and detect possible collaborations. Our aim is to implement a recommendation system for a target author who searches for collaboration colleagues. The research question is: 1) Can we propose a relevant author cluster for a target scientist? Furthermore we try to apply user data from the social bookmarking system CiteULike. The second research question is: 2) Is this user-based data also relevant for our target scientist and does it recommend different results? Our first outcomes of this work in progress are evaluated by our target authors.

PDF