A proposed Emergent Skill Extraction Methodology from Unstructured Text

In this paper a system for emergent skill extraction from massive job postings is proposed. The proposed system relies on semantic skill representation in spatial skill space. Based on this semantic skill space, suitable statistics are adopted over the temporal dimension of the job posts to decide the emergent skills. Skills are very diverse and changing over time, not only individuals are affected by these changes but also policy-makers, businesses and educational institutions. So, in such a very dynamical domain we are interested to detect emergent skills and future demands on different skills. Skills are to be first extracted from the unstructured text of job posts. Skills may be phrased in different wordings and there meaning may depend on the context of the job post. Such challenges are to be resolved adopting some sort of reliable skill extraction methodology, suitable skill representation space as well as smart statistical analysis of such representation space. Results based on the proposed methodology on different job posts from well-known job posting portals show very promising results that encourage us to extend this system for more advanced analysis such as skill gap analysis and job post format standardization.