Improving Quality of Job Application Pre-processing with Knowledge Graphs

Joseph Porter, Pace University

Abstract

Human Resources (HR) personnel face challenges when preprocessing or evaluating job applications. The main goal of job application preprocessing is to filter through many resumes in order to produce a short list of the most qualified candidates to hire. The lack of domain expertise in high-tech recruiting has led to poorly qualified prospective employees advancing to the short list of candidates considered for hire. The challenge in evaluating prospective employees is made worst given the variety of professional and laymen terms used by job-providers/companies and job-seekers/prospective employees. Job description created by companies and resumes created by prospective employees both use a combination of professional and laymen terms to describes job requirements and job qualification. It is acceptable for one company to use different terms than another to describe a Cloud Software Developer’s position. Similarly, a prospective employee can use different terms than another perspective employee in describing qualifications for a Cloud Software Developer’s position. Ontology is often used to define vocabulary and terms in an application domain. However, the most popular ontology tools in used today are limited in supporting only the single “is-a” relationship which prevent them from describing rich relationship needed to capture terms used during job application evaluation or preprocessing. To fill this gap, Pace Universities Knowledge Graph (KGs) and manual methods of producing KGs extend ontology to support customized relationship like “part of” to better support relationships among various concepts that may be used by professional and laymen in job descriptions or resumes. This research proposes to use KGs to identify custom relationship so that important keywords could be used in a Python Analyzer (PA) Application Tracking System (ATS) that was developed for matching keywords in job descriptions to keywords in resumes. The PA ATS use KG keywords from a text file to compare against job description keywords to produce a subset of keywords for further matching between the job description and a batch of resumes for the job. For each resume, HR personnel are able to see match percentage before and after filters for keyword and keywords with synonyms were applied. This research experiments showed a 10% average improvement in job match accuracy using the PA ATS. Improved matches enhance job application preprocessing. Contribution in this research can also be applied to similar problems that need a PA ATS tool to bridge terminologies in different communities.

Subject Area

Computer science

Recommended Citation

Porter, Joseph, "Improving Quality of Job Application Pre-processing with Knowledge Graphs" (2020). ETD Collection for Pace University. AAI28411859.
https://digitalcommons.pace.edu/dissertations/AAI28411859

COinS