Large‐Scale Occupational Skills Normalization for Online Recruitment

Abstract

Job openings often go unfulfilled despite a surfeit of unemployed or underemployed workers. One of the main reasons for this is a mismatch between the skills required by employers and the skills that workers possess. This mismatch, also known as the skills gap, can pose socioeconomic challenges for an economy. A first step in alleviating the skills gap is to accurately detect skills in human capital data such as resumes and job ads. Comprehensive and accurate detection of skills facilitates analysis of labor market dynamics. It also helps bridge the divide between supply and demand of labor by facilitating reskilling and workforce training programs. In this paper, we describe SKILL, a Named Entity Normalization (NEN) system for occupational skills. SKILL is composed of 1) A skills tagger which uses properties of semantic word vectors to recognize and normalize relevant skills, and 2) A skill entity sense disambiguation component which infers the correct meaning of an identified skill. We discuss the technical design and the synergy between data science and engineering that was required to transform the system from a research prototype to a production service that serves customers from across the organization. We also discuss establishing customer feedback loops, and it led to improvements to the system over time. SKILL is currently used by various internal teams at CareerBuilder for big data workforce analytics, semantic search, job matching, and recommendations.

Soft and hard skills identification: insights from IT job advertisements in the CIS region

“…The others apply descriptions of skills from national occupational classifiers ( Cao et al, 2021 ; Colombo, Mercorio & Mezzanzanica, 2019 ; Karakatsanis et al, 2017 ; Lovaglio et al, 2018 ; Botov et al, 2019 ; Amato et al, 2015 ). The rest—use manual mark-up for extracted knowledge domains ( Calanca et al, 2019 ; Sayfullina, Malmi & Kannala, 2018 ; Hoang et al, 2018 ).…”

“…Another group of papers provides separation in two ways. Firstly, authors use manual mark-up and ready skill bases before the stage of skill sets detection ( Cao et al, 2021 ; Hoang et al, 2018 ; Skhvediani et al, 2021 ; Papoutsoglou, Mittas & Angelis, 2017 ; Hiranrat & Harncharnchai, 2018 ; Cegielski & Jones-Farmer, 2016 ). Secondly, researchers obtain “soft” skills mapping using domain experts correction after classifying or clustering job advertisements ( Colombo, Mercorio & Mezzanzanica, 2019 ; Litecky, Igou & Aken, 2012 ; Gardiner et al, 2018 ; Brooks, Greer & Morris, 2018 ; Hussain, Clear & MacDonell, 2017 ; Sodhi & Son, 2010 ; Gurcan & Cagiltay, 2019 ; Radovilsky et al, 2018 ; Giabelli et al, 2020 ; Wu, Shi & Yang, 2017 ).…”

“…Some authors use job titles and their descriptions from vacancies in order to match them with ESCO (European Skills/Competences, qualifications and Occupations), ISCO (International Standard Classification of Occupations), or O*NET (Occupational Information Network) classifiers ( Boselli et al, 2018 ; Cao et al, 2021 ; Colombo, Mercorio & Mezzanzanica, 2019 ; Lovaglio et al, 2018 ; Colace et al, 2019 ; Papoutsoglou, Mittas & Angelis, 2017 ; Giabelli et al, 2020 ). Others use the structure of job advertisements provided on hiring platforms and validate it with domain experts mark-up and professional standards ( Chang, Wang & Hawamdeh, 2019 ; Ternikov & Aleksandrova, 2020 ; Hiranrat & Harncharnchai, 2018 ; Amato et al, 2015 ; Sayfullina, Malmi & Kannala, 2018 ; Litecky, Igou & Aken, 2012 ; Hoang et al, 2018 ; Botov et al, 2019 ; Brooks, Greer & Morris, 2018 ). This research employs text mining techniques such as TF-iDF and n -grams that are used for specific words and phrases extraction.…”

Automated Discovery and Classification of Training Videos for Career Progression

“…This model consists of a job-job and a job-skill network that provide the next job along with associated skills given a current job title. In total, we have 5,425 unique normalized job titles [3] and an average of 15 normalized skill terms [2] per title.…”

“…For each job title-skill pair, the query gets formed in three different ways to try to capture all related videos. Specifically, the queries are generated by (1) skill term (2) skill term + job title (3) "skill term" + job title 2 . The API calls are configured to fetch only video contents in English.…”