Learning to Select External Knowledge with Multi-Scale Negative Sampling

TL;DR

This paper introduces a novel approach for selecting and utilizing external knowledge in task-oriented dialogue systems, significantly improving response quality for out-of-API-coverage queries through advanced negative sampling techniques.

Contribution

It proposes a multi-scale negative sampling method combined with schema-guided knowledge decision to enhance external knowledge selection and response generation in dialogue systems.

Findings

Achieved top ranking in human evaluation of DSTC9 Track-1.

Demonstrated improved response relevance and informativeness.

Validated effectiveness through comprehensive experiments on a public dataset.

Abstract

The Track-1 of DSTC9 aims to effectively answer user requests or questions during task-oriented dialogues, which are out of the scope of APIs/DB. By leveraging external knowledge resources, relevant information can be retrieved and encoded into the response generation for these out-of-API-coverage queries. In this work, we have explored several advanced techniques to enhance the utilization of external knowledge and boost the quality of response generation, including schema guided knowledge decision, negatives enhanced knowledge selection, and knowledge grounded response generation. To evaluate the performance of our proposed method, comprehensive experiments have been carried out on the publicly available dataset. Our approach was ranked as the best in human evaluation of DSTC9 Track-1.