The Development and Validation of a Campus Recruiting Expert System Using Expert Opinion and Historical Data

This article describes the development and validation of an expert system used for screening entry level candidates for employment in an accounting firm. The system is designed to assist a recruiter conducting campus interviews in deciding which candidates to call for a second interview at company premises. Such expert systems have the potentail for improving consistency and efficiency of the many human recruiters' decisions, and for minimizing personal biases or the use of unlawful criteria (e.g., race, gender, or age) in these decisions. Whereas a hypothetical expert system for this recruiting problem has been previously described in the literature, we actually built one. We describe how we built the necessary knowledge base and how and why our inference engine calculates the probabilities of acceptance, hold, and rejection based on a combination of expert opinions and historical data base. The system is validated by comparing it against recent case-by-case decisions of human experts. We recognize the limitations of our study and discuss the lessons learned from this development effort