Method

Multi-label classification

Assigning several labels per item.

12 cases from 10 companies apply Multi-label classification. 3 were checked in an online A/B test; 2 report results in production.

Typical tasks: Moderate content at scale, Rank search results, Catch fraud and abuse, Detect anomalies and 3 more. Companies: Alfa-Bank, LY Corporation, Walmart, Yelp, Canva, Netflix and 4 more. Often combined with: Text classification, Contrastive learning, Regular expressions, Named entity recognition, Model export (ONNX) and 43 more.