Multi‑criteria Optimization of Search Results at Avito
summary
Avito handles hundreds of millions of search queries daily and faces the problem of satisfying multiple user groups whose ranking preferences conflict. The talk describes a multi-criteria optimization approach for search ranking that balances these competing objectives to produce a single search result ordering.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 3:33Different stakeholder interests
- 7:27Ranking approach evolution
- 12:01User behavior modeling
- 17:17Blender operation
- 23:12Algorithm issues and fixes
- 28:52Metrics and system testing
- 34:32Results and new features
- 38:49Conclusions
- 41:04Q&A
related talks
37mSearch and Recommendations: 3 Sides of the Same Coin #bbuzz
41mDynamic Ranking of Search and Recommendation Results in High-Traffic Systems
26mStreamlining Property Search Across Multiple Platforms
31mEverything in its Right Place: Optimising Ranking in Online Grocery
13mThe State of AI Powered Search and Retrieval
32mMulti-Armed Bandits in Recommendations
35mLearning to Judge
51mHow to Effectively Rank All of Runet's Products