A resume recommender system based on link prediction and greedy reranking.

Description:
As companies all over the world are facing the challenge of talent acquisition and re- tention, recruiters need advanced tools to source candidates quickly. Modern computer algorithms can facilitate the candidate search and improve the hiring outcomes. Rec- ommender systems are used effectively to filter large data sets based on user preferences and past transactions. Link prediction models are used in social graph analysis for pre- dicting potential relationships or transactions between members of a social network. We propose a novel r ́esum ́e recommender system based on link prediction models and item clustering. In building such as system, we must overcome two data-related hurdles: data scarcity and data sparsity. We resolve the first issue by sourcing a large set of empirical data collected from a leading executive search firm. We overcome the second hurdle by using clustering techniques whereby we synthesize links from the historical data based on the consistent representatives for jobs and r ́esum ́es. We compare the proposed system with a baseline content-based r ́esum ́e recommender system.