Mitigating Bias in Recruitment: A Practical Approach to CV De-identification Considering Privacy Sensitive Information
摘要
Curriculum Vitae (CVs) often contain sensitive personal information, which poses risks of discriminatory use by employers during the hiring process. To address this, there have been various research approaches to de-bias the various phases of the hiring process. However, these initiatives often fall short of resolving the core issues effectively. In this paper, we propose a process and proof of concept aimed at de-identifying Privacy Sensitive Information (PSIs) in CVs. We run rigorous evaluations to de-identify relevant PSI outlined in the EU GDPR (e.g., art. 9) and synthesize findings from state-of-the-art research on text and CV de-identification. Our designed prototype is able to firstly identify PSI specific to CVs and then de-identifies these PSI within the Europass CV structure. This novel approach demonstrates a promising path towards promoting fairness and preserving privacy in the recruitment process.