Lazy learned screening for efficient recruitment

Abstract

The transition from traditional paper based systems for recruitment to the internet has resulted in companies in getting a lot more applications. A majority of these applications are often unstructured documents sent over email. This results in a lot of work sorting through the applicants. Due to this, a number of systems have been implemented in an effort to make the screening phase more efficient. The main problems consist of extracting information from resumes and ranking the candidates for positions based on their relevance.

We develop a system that can learn how to rank candidates for a position based on knowledge obtained from earlier screening phases. This Candidate Ranking System (CRS) is based on Case-based Reasoning, combined with semantic data models. The systems performance is evaluated in conjunction with a large international Job company and a software company in an actual recruitment process.