ADEPT: An Entropy-Driven Dual-Strategy Agent for Interactive Video Retrieval
This work addresses the problem of ambiguous user queries in video retrieval for practitioners, offering an efficient and interpretable interactive solution.
ADEPT introduces a training-free, entropy-driven agent for interactive video retrieval that dynamically selects between ASK and REFINE strategies to bridge the 'Intent-Query Gap'. It significantly outperforms all baselines on two challenging datasets, setting a new performance benchmark.
This research aims to solve the challenge of video retrieval from massive datasets, caused by ambiguous user queries. Prevailing single-round retrieval paradigms face a performance bottleneck, as they lack effective feedback mechanisms to handle complex search intentions. The root cause is the "Intent-Query Gap", where users' intent cannot be captured by a simple text query. To solve this, we propose the ADEPT framework: a training-free agent that pioneers an entropy-driven decision engine to efficiently guide dialogue by dynamically selecting between ASK and REFINE strategies. Experiments on two challenging datasets demonstrate that ADEPT significantly outperforms all non-interactive, heuristic, and Video-LLM baselines. The core contribution of this work is an efficient and interpretable entropy-driven interactive strategy that sets a new performance benchmark for the field of interactive video retrieval.