Imprecise categorization of workstations for the risk assessment related to chemical mixtures

The assessment of occupational exposure to polycyclic aromatic hydrocarbons (PAHs) is a major issue for preventing health risks in the workplace. This study aims to analyze the similarity between various job positions using clustering methods. In the first part of the work, several preprocessing and clustering techniques are tested and evaluated using an entropy measure specifically designed for this problem. The second part introduces a pipeline based on belief functions to account for the possibility that a single job position may belong simultaneously with different degrees to multiple exposure profiles. The results obtained are consistent with expert domain knowledge.