Supervised Key Terms Clustering for Regulatory Monitoring

Abstract

Multivariate analysis, clustering, and classification are common machine learning challenges in food, beverage and consumer product regulation documentations. Using selective key terms from a food, beverage chemical and consumer product lexicon to perform supervised key term clustering helps on both predicting the relevance of regulation documents to their subjects and also explaining the classification and document ranking decisions. The results can also be used as feedback to improve the lexicon.