Method

Text clustering

Grouping texts or dialogues by embeddings or keywords.

8 cases from 8 companies apply Text clustering. 1 was checked in an online A/B test; 3 report results in production.

Typical tasks: Build an assistant over internal docs, Learn user and item embeddings, Evaluate models, Moderate content at scale and 2 more. Companies: DoorDash, Raiffeisenbank, Yandex, Alfa-Bank, MegaFon and 3 more. Often combined with: UMAP, HDBSCAN, LLM-as-a-judge, Supervised fine-tuning, Overlap annotation and 24 more.