Using Information Extraction and Evolutionary Algorithms to Improve Matchmaking on the Labor Market

Abstract

This thesis identifies a significant lack of attention to using unprocessed documents as input for automated matchmakers on the labor market. To address this lack of attention, a job matching application is proposed that uses unprocessed Curriculum Vitae (CVs) and job descriptions as input, avoiding the great restrictions that many other approaches described in the literature pose on the data. This is achieved by parsing the documents using Information Extraction techniques. Afterwards, an Evolutionary Algorithm is used to optimize the weight (relative importance) of finding a match on each extracted field.

The obtained results show that there is a significant difference in performance be- tween the application described in this thesis and statistical full-content matching. This indicates that the identified lack of attention to unprocessed CVs and job descriptions is unjust, as this approach still has unexplored potential. An interesting suggestion for future work is to combine the Information Extraction component from this thesis with the rule and ontology based approaches currently popular in the literature. This way, the latter approaches can elevate the previously imposed restrictions on the input data, and accept unprocessed CVs and job descriptions