Lessons learnt from using vacancy mining for validating and supplementing labour market taxonomies


Keeping taxonomies up to date is a labour-intensive and hence expensive endeavour. An obvious approach to save costs is to automate at least some of the manual tasks involved in maintenance. Vacancy text provides good insight into the language employers and HR professions currently use when describing job openings. Thus we considered it to be a promising approach to automatically collect and process this data and then compare it to the current vocabulary of labour market taxonomies, thus identifying amendment needs.

Jobfeed (https://www.jobfeed.com/), a big-data platform automatically collecting and processing online vacancies provided the examination material. The two labour market taxonomies to be amended – ‘AMS-Kompetenzenklassifikation’ (on occupational requirements, currently containing almost 29.000 terms) and ‘AMS-Berufssystematik’ (an occupational classification of approx. 84.000 terms) are both used by the Austrian Public Employment Service to structure labour market matching processes and online information systems.


Interested in this talk?

Register for SEMANTiCS conference