Learning to Select External Knowledge with Multi-Scale Negative Sampling
This work provides an incremental improvement for dialogue systems needing to answer out-of-scope queries by integrating external knowledge.
This paper addresses the problem of answering user requests in task-oriented dialogues that fall outside the scope of available APIs/DB by leveraging external knowledge resources. The proposed method, which includes schema-guided knowledge decision, negatives-enhanced knowledge selection, and knowledge-grounded response generation, was ranked as the best in human evaluation for DSTC9 Track-1.
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.