Method
Near-duplicate detection
Perceptual hashing, MinHash/LSH and semantic deduplication.
13 cases from 11 companies apply Near-duplicate detection. 2 were checked in an online A/B test; 5 report results in production.
Typical tasks: Moderate content at scale, Catch fraud and abuse, Rank search results, Match entities and 6 more. Companies: Celsus, Sber, LY Corporation, DoorDash, Wildberries, Yelp and 5 more. Often combined with: Embedding similarity matching, Threshold selection, Human in the loop, Word2Vec, Weak supervision and 48 more.
Multimodal Soup: Turning a Heap of Ad Banners into Meaningful Clusters![]() | ![]() | adtech | 2026-05 |
| From Linear Text to a Semantic Graph: Building a Knowledge-Extraction Pipeline for Studying | edtech | 2026-05 | |
Extracting trending keywords from OpenChat messages![]() | ![]() | social | 2025-09 |
Challenges of Labeling Doctor-Patient Dialogues for ASR![]() | ![]() | healthtech | 2025-05 |
| Part 1: DoorDash 2024 summer intern projects | food-delivery | 2024-10 | |
How We Fought Spam in Reviews![]() | ![]() | marketplace | 2024-05 |
| Moderating Inappropriate Video Content at Yelp | marketplace | 2024-03 | |
NER for Marketplace Search Queries![]() | ![]() | marketplace | 2023-05 |
Data-Centric AI: A Review of Methods![]() | ![]() | healthtech | 2023-05 |
| Text Deduplication: A Review | — | banking | 2023-05 |
| Simple, Fast, and Scalable Reverse Image Search Using Perceptual Hashes and DynamoDB | general | 2022-10 | |
| How Instacart Uses Machine Learning-Driven Autocomplete to Help People Fill Their Carts | e-commerce | 2022-08 | |
Workshop: Introduction to the Similarity Search Task![]() | ![]() | telecom | 2022-06 |





