To whom and why should I connect? Co-author Recommendation based on powerful and similar Peers

The present article offers preliminary outcomes of a user study that investigated the acceptance of a recommender system that suggests future co- authors for scientific article writing. The recommendation approach is twofold: network information (betweenness centrality) and author (keyword) similarity are used to compute the utility of peers in a network of co-authors. Two sets of recommendations were provided to the participants: Set one focused on all candidate authors, including co-authors of a target user to strengthen current bonds and strive for acceptance of a certain research topic. Set two focused on solely new co-authors of a target user to foster creativity, excluding current co- authors. A small-scale evaluation suggests that the utility-based recommendation approach is promising, but to maximize outcome, we need to 1) compensate for researchers’ interests that change over time, and 2) account for multi-person co-authored papers.