PK-ICR: Persona-Knowledge Interactive Context Retrieval for Grounded Dialogue
This work addresses the need for improved multi-context dialogue grounding, which is incremental as it builds on prior isolated research in persona and knowledge retrieval.
The paper tackles the problem of jointly identifying relevant persona and knowledge for grounded dialogue systems, introducing a novel retrieval method that uses all dialogue contexts simultaneously and achieves computational efficiency via neural QA retrieval models.
Identifying relevant persona or knowledge for conversational systems is critical to grounded dialogue response generation. However, each grounding has been mostly researched in isolation with more practical multi-context dialogue tasks introduced in recent works. We define Persona and Knowledge Dual Context Identification as the task to identify persona and knowledge jointly for a given dialogue, which could be of elevated importance in complex multi-context dialogue settings. We develop a novel grounding retrieval method that utilizes all contexts of dialogue simultaneously. Our method requires less computational power via utilizing neural QA retrieval models. We further introduce our novel null-positive rank test which measures ranking performance on semantically dissimilar samples (i.e. hard negatives) in relation to data augmentation.