"A Fresh Perspective on Job Analysis"
A Fresh Perspective on Job Analysis: Extracting Job Information from Job Vacancy Announcements
Posted: 6 Dec 2022
Vladimer Kobayashi*
University of Amsterdam
Stefan T. Mol
University of Amsterdam Business School
Gábor Kismihók
affiliation not provided to SSRN
Evangelos Kanoulas
University of Amsterdam
Abstract
Online vacancies provide a novel and rich source of job information that may complement traditional job and labor market analyses. Common information that are used as input to job analysis are worker attributes and work activities. Previous studies aimed at extracting these pieces of information from vacancy data have by and large leveraged methods that rely on prespecified keywords during extraction which can be laborious, require subject matter expertise to screen the keywords and may introduce and/or replicate historical biases. To address these issues, this paper proposes the use of state-of-art text classification techniques coupled with a rich set of inductively derived features that incorporate word-based, syntactic, and grammatical features of text found in vacancies. The results showed our approach is effective in sorting vacancy content into worker attributes and worker activities. Furthermore, we demonstrate how the extracted job information can be used to extract task groups and cluster jobs, and how to validate the extracted information by comparing it to an existing job taxonomy that was independently constructed by experts.
Visit the web post here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4281602
