Method

k-nearest neighbors

Prediction or retrieval from nearest examples.

17 cases from 12 companies apply k-nearest neighbors. 7 were checked in an online A/B test; 2 report results in production.

Typical tasks: Learn user and item embeddings, Retrieve and rank from millions of items, Recommend to new users and new items, Rank search results and 8 more. Companies: Zillow, DoorDash, Mercari, eBay, LinkedIn, Pinterest and 6 more. Often combined with: Embedding fine-tuning, Approximate nearest neighbors, Contrastive learning, LLM labeling, Dense retrieval and 53 more.