Natural Language Processing on the Balance of theSwedish Software Industry and Higher VocationalEducation

Abstract: The Swedish software industry is fast-growing and in needof competent personnel, the education system is on the frontline of producing qualified graduates to meet the job marketdemand. Reports and studies show there exists a gapbetween industry needs and what is taught in highereducation, and that there is an undefined skills shortageleading to recruitment failures. This study explored theindustry-education gap with a focus on higher vocationaleducation (HVE) through the use of natural languageprocessing (NLP) to ascertain the demands of the industryand what is taught in HVE. Using the authors' custom-madetool Vocational Education and Labour Market Analyser(VELMA), job ads and HVE curricula were collected fromthe Internet. Then analysed through the topic modellingprocess latent Dirichlet allocation (LDA) to classify lowerlevel keywords into cohesive categories for documentfrequency analysis. Findings show that a large number ofHVE programmes collaborate with the industry via indirectfinancing and that job ads written in Swedish consist, inlarger part, of inconsequential words compared to adswritten in English. Moreover, An industry demand withincloud and embedded technologies, security engineers andsoftware architects can be observed. Whereas, the findingsfrom HVE curricula point to a focus on educating webdevelopers and general object-oriented programminglanguages. While there are limitations in the topic modellingprocess, the authors conclude that there is a mismatchbetween what is taught in HVE programmes and industrydemand. The skills identified to be lacking in HVE wereassociated with cloud-, embedded-, and security-relatedtechnologies together with architectural disciplines. Theauthors recommend future work with a focus on improvingthe topic modelling process and including curricula fromgeneral higher education.