A Model-Based Approach to Recommending Partners

Searching for and selecting qualified partners is a core task in many business contexts. Empirical research among Germany’s top 1,000 firms discloses that internet-based platforms are effectively used as a personnel marketing channel but cannot increase the matching quality between jobs and candidates. Using erecruitment as an example, we show how the matching quality can be substantially improved by means of a probabilistic latent aspect model developed in this paper. The underlying method incorporates findings from collaborative filtering and hybrid approaches to automated recommendation and is based on a model of personal attributes derived from research on team building and work psychology.