Skills Named-Entity Recognition for Creating a Skill Inventory of Today’s Workplace
Gjorgjina Cenikj, Bozhanka Vitanova, Tome Eftimov · 2021 IEEE International Conference on Big Data (Big Data) · 2021
To trace the skills of the employees and to link them to their responsibilities, organizations need support from methods that automatically support such kind of activities. Even more, the COVID-19 pandemic completely transformed organizations’ workplace and culture, so project managers should take care of the skills of their team members in order to successfully realize a project. For this purpose, we have introduced rule- based named-entity recognition methods that are able to extract soft and technical skills using textual data such as job postings, curricula vitae (CVs), shout-outs, performance reviews, etc. The best model using a fusion of several skills-related dictionaries provides an F1 score of 59%, which can be further used as a base for creating an annotated silver corpus consisting of textual data and all skills found there. Further, we demonstrated the application of the developed model to link the required skills for a certain job title. Being able to trace the skills in an automatic way, we can further link them to tasks and actions to fully understand their economic value.