Trading Vector Data in Vector Databases
This addresses the largely unexplored problem of vector data trading for cross-domain learning, which is incremental as it builds on existing bandit methods to handle specific challenges in this domain.
The paper tackles the problem of vector data trading in vector databases under online learning, where sellers face uncertain retrieval costs and buyers provide stochastic feedback to prices, by proposing a hierarchical bandit framework that jointly optimizes retrieval configurations and pricing. The result shows consistent improvements in cumulative reward and regret reduction on four real-world datasets, with theoretical guarantees of logarithmic and sublinear regret.
Vector data trading is essential for cross-domain learning with vector databases, yet it remains largely unexplored. We study this problem under online learning, where sellers face uncertain retrieval costs and buyers provide stochastic feedback to posted prices. Three main challenges arise: (1) heterogeneous and partial feedback in configuration learning, (2) variable and complex feedback in pricing learning, and (3) inherent coupling between configuration and pricing decisions. We propose a hierarchical bandit framework that jointly optimizes retrieval configurations and pricing. Stage I employs contextual clustering with confidence-based exploration to learn effective configurations with logarithmic regret. Stage II adopts interval-based price selection with local Taylor approximation to estimate buyer responses and achieve sublinear regret. We establish theoretical guarantees with polynomial time complexity and validate the framework on four real-world datasets, demonstrating consistent improvements in cumulative reward and regret reduction compared with existing methods.