TalentCLEF 2025 Overview Talk
Abstract
Presentation of the TalentCLEF 2025 results. TalentCLEF is an initiative to advance Natural Language Processing (NLP) in Human Capital Management (HCM). In 2025, the lab focused on skill and job title intelligence, addressing the growing need for reliable and fair language models in the labor market. The campaign included two tasks: (A) Multilingual Job Title Matching (English, Spanish, German, Chinese) and (B) Job Title-Based Skill Prediction (English). The evaluation covered monolingual, cross-lingual, and gender bias analyses, attracting 76 teams with over 280 submissions. The results highlight that training strategies impact performance more than model size, establishing the first public benchmark for robust, fair, and transferable language technologies in HCM.
