Information extraction method from a resume (CV)

Abstract

An approach to information extraction from a short and poorly structured text document such as a resume (CV) is suggested. The computer-based resume processing is an actual interesting application problem. There are a number of web-sites for centralized CVs allocation oriented at different employers. An employer is often more interested in some peculiar features connected to professional achievements and knowledge of the applicant, not in a resume as a whole. Extraction of such peculiar information from a CV is a problem itself, especially if the CV is organized in an arbitrary form, poorly structured and contains grammatical mistakes. The suggested paper is devoted to the processing of CVs of this type. There is a short review of the existing approaches to information extraction from a CV, a keyword-based approach is selected and founded from the viewpoint of efficient information extraction the employer is interested in. The specificity of the approach is emphasized for the case when keywords define text blocks with a constant conceptual content. In this case, another problem arises, which is connected with the definition of such blocks. An approach based on a clustering technique is suggested, so each cluster is associated with a corresponding text block. At the same time, the technical realization of the approach suggested remains open for future investigations. The paper provides examples illustrating the described text extraction technique from a CV used in order to get a relevant answer to an arbitrary query.