Classification for Relevance of Scientific Articles : Examples for Food and Feed Risk Assessment

This thesis explores an innovative classification approach to determine the relevance of scientific articles in the domain of food and feed risk assessment. During my internship at Innovamol Consulting Srl, I undertook an exciting project aimed at transforming the categorization process. We developed an automated system that efficiently classifies articles as relevant or irrelevant, accompanied by probabilities, using advanced machine learning algorithms. By pursuing various fine-tuned classification models with different feature extraction techniques, the fine-tuned logistic regression classifier model emerged as the top performer, significantly reducing the time previously invested in manual analysis. This transformative approach enhances efficiency, empowers data-driven decision-making, and opens up new possibilities for advancements in the food and feed risk domain. Our system promises to drive productivity and make accurate assessments, ultimately optimizing operations and fostering further progress in the industry.