Exploring Dense Retrieval for Dialogue Response Selection
This work addresses the efficiency problem in dialogue systems for practical applications, enabling direct selection from large or nonparallel corpora, though it is incremental as it builds on existing dense retrieval methods.
The paper tackled the computational cost of neural dialogue response selection models by proposing a dense retrieval model with an interaction layer and tailored learning strategies, achieving superior performance over strong baselines in both re-rank and full-rank evaluations, including human-annotated scenarios with millions of candidates.
Recent progress in deep learning has continuously improved the accuracy of dialogue response selection. In particular, sophisticated neural network architectures are leveraged to capture the rich interactions between dialogue context and response candidates. While remarkably effective, these models also bring in a steep increase in computational cost. Consequently, such models can only be used as a re-rank module in practice. In this study, we present a solution to directly select proper responses from a large corpus or even a nonparallel corpus that only consists of unpaired sentences, using a dense retrieval model. To push the limits of dense retrieval, we design an interaction layer upon the dense retrieval models and apply a set of tailor-designed learning strategies. Our model shows superiority over strong baselines on the conventional re-rank evaluation setting, which is remarkable given its efficiency. To verify the effectiveness of our approach in realistic scenarios, we also conduct full-rank evaluation, where the target is to select proper responses from a full candidate pool that may contain millions of candidates and evaluate them fairly through human annotations. Our proposed model notably outperforms pipeline baselines that integrate fast recall and expressive re-rank modules. Human evaluation results show that enlarging the candidate pool with nonparallel corpora improves response quality further.