Publicação

Job Candidate Screening Problem: Data-driven approach to evaluate profile quality on Crafthunt

Ver documento

Detalhes bibliográficos
Resumo:This project addresses the job candidate screening process of the German company Crafthunt, a job platform specialized in the construction industry. With an average of 500 candidate profiles created monthly, manually screening them has become a tedious process that consumes significant resources of the company's small Talent Management Team. Among researchers, this organizational challenge is referred to as the job candidate screening problem. To solve this problem, multiple clustering algorithms and analytical models were employed to create a candidate segmentation based on key profile quality indicators. While the DBSCAN algorithm did not deliver satisfactory results, the application of the k-means algorithm created 4 distinct clusters with a strong silhouette score of 0.69, separating candidates based on their language skills, profile completion, geographical location, and verification status. Through the application of the RFM model, three quality tiers of candidates were built based on their platform engagement. The project successfully met all set objectives and provided Crafthunt with two segmentation approaches along with recommended actions for implementation.
Autores principais:Kohl, Amelie
Assunto:Recruitment Artificial Intelligence Clustering Construction Job Platforms SDG 8 - Decent work and economic growth
Ano:2025
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso embargado
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:This project addresses the job candidate screening process of the German company Crafthunt, a job platform specialized in the construction industry. With an average of 500 candidate profiles created monthly, manually screening them has become a tedious process that consumes significant resources of the company's small Talent Management Team. Among researchers, this organizational challenge is referred to as the job candidate screening problem. To solve this problem, multiple clustering algorithms and analytical models were employed to create a candidate segmentation based on key profile quality indicators. While the DBSCAN algorithm did not deliver satisfactory results, the application of the k-means algorithm created 4 distinct clusters with a strong silhouette score of 0.69, separating candidates based on their language skills, profile completion, geographical location, and verification status. Through the application of the RFM model, three quality tiers of candidates were built based on their platform engagement. The project successfully met all set objectives and provided Crafthunt with two segmentation approaches along with recommended actions for implementation.