Applications of Machine Learning in Detecting Unethical Sources of Raw Materials in Supply Chains in the Cosmetic Industry

The cosmetic industry maintains a global and complex supply chain network, where products composed of multiple raw ingredients are transported to numerous countries. Due to consumer pressure and regulatory requirements, ethical sourcing issues have attracted the attention of global consumers and businesses. This study proposes a machine-learning (ML)-oriented approach for identifying unethical sources in the cosmetic supply chain. We develop an ML-based framework by applying supervised learning, unsupervised learning and natural language processing (NLP), which includes text classification algorithms using a recommendation system to create training modules tailored to specific respondents. The proposed framework identifies unethical sourcing at any stage within the supply chain, thereby enhancing transparency, traceability, sustainability and business ethics as well as gaining the trust of consumers. While existing models have been successful in identifying certain unethical practices, they often face challenges such as limited scalability, high false positive rates, and the need for extensive labeled data. This study aims to address these limitations by developing a more robust and scalable machine learning framework. The results show that our ML models effectively detect potential unethical practices at early stages before harming reputations and adopt outlier detection in real-time monitoring or early warning systems with minimum configuration effort due to frequent changes in violation patterns or control mechanisms faced by practitioners. This study demonstrates that ML can aid governments and industries in implementing sustainable practices through stronger governance of business activities along their supply chains.