Towards An Ontology-Based Knowledge Base for Job Postings

This paper presents an approach that identifies and qualifies job demands by analyzing job postings on recruitment websites as unstructured sources of knowledge using an ontology-based knowledge base. This ontology provides an integrated view for exploring and querying data at a real time. It captures terms and relationships to facilitate the sharing and re-use by applications. For data extraction, a rule-based technique is used to extract concepts instances to populate the ontology. Several techniques are proposed to enhance the performance and accuracy such as text processing and named entity recognition. To validate the approach, an application in the IT domain is built and experimented. The performance of the approach is evaluated based on the quality of the instance extraction step using evaluation metric F1-score, which is commonly used for information extraction problems.