Relevance-Based Embeddings: Lightweight Candidate Retrieval via Heavy-Ranker Calls
2026-07-03 • Information Retrieval
Information RetrievalMachine Learning
AI summaryⓘ
The authors address the problem of finding relevant items quickly when a detailed similarity check is too slow. Usually, this is done by turning queries and items into simple vectors to quickly find close matches, but this can lose some accuracy. Their new idea is to use how the slow, detailed model scores items relative to a set of support examples to create better vector representations. They prove these new vectors can closely mimic complex similarity checks and show their method works well on real datasets.
machine learningsimilarity modelapproximate nearest neighbor searchembeddingre-rankingquery representationsupport itemsvector spacedot productcosine similarity
Authors
Kirill Shevkunov, Andrey Ploskonosov, Liudmila Prokhorenkova
Abstract
In many machine learning applications, the most relevant items for a query should be efficiently retrieved. The relevance function is usually an expensive similarity model, making the exhaustive search infeasible. A typical solution is to train another model that separately embeds queries and items to a vector space, where similarity is defined via the dot product or cosine similarity. This allows one to search the relevant items through fast approximate nearest neighbor search at the cost of some reduction in quality. To compensate for this reduction, the found items (candidates) are re-ranked by the expensive ranking model. In this paper, we investigate an alternative approach to candidate selection that utilizes the scores of the expensive model to improve the representations of queries and items. The idea is to describe each query (item) by its relevance to a set of support items (queries) and use these new representations to obtain query (item) embeddings. We theoretically prove that such embeddings are powerful enough to approximate any complex similarity model (under mild conditions). We also investigate the choice of support items, which is a crucial ingredient of the proposed approach. The experiments on diverse academic and production datasets illustrate the power of our method.