Data-driven Skills Taxonomy
A skills taxonomy could show workers other jobs that they can transition to with the skills they have or with minimum retraining.
For this project I explored how an established labour market framework can be used to start developing a map of skills. I used the European Skills, Competences, Qualifications and Occupations (ESCO) framework. ESCO has been developed by the European Commission and includes information on relationships between 2,942 occupations and 13,485 skills/competences.
The link between required skills and occupations could be a useful basis for building the taxonomy, because it provides contexts in which certain skills occur frequently together. That is, certain skills co-occur more often in a certain context, here across occupations listed in the ESCO framework. Additionally I use the similarity between more elaborate skill descriptions as calculated using Natural Language Processing tools. The skills can then be represented as a network with weighted edges according to these attributes, they can be hierarchically grouped into clusters using community detection algorithms.
