Controlling Dialogue Generation with Semantic Exemplars
This work addresses the issue of incoherent replies in dialogue generation for users needing more controlled and goal-oriented conversational AI, representing an incremental improvement over existing exemplar-based methods.
The paper tackles the problem of dialogue systems lacking fine-grained control over responses by introducing EDGE, an exemplar-based model that uses semantic frames from exemplars to guide generation, resulting in improved coherence while preserving semantic meaning and conversation goals.
Dialogue systems pretrained with large language models generate locally coherent responses, but lack the fine-grained control over responses necessary to achieve specific goals. A promising method to control response generation is exemplar-based generation, in which models edit exemplar responses that are retrieved from training data, or hand-written to strategically address discourse-level goals, to fit new dialogue contexts. But, current exemplar-based approaches often excessively copy words from the exemplar responses, leading to incoherent replies. We present an Exemplar-based Dialogue Generation model, EDGE, that uses the semantic frames present in exemplar responses to guide generation. We show that controlling dialogue generation based on the semantic frames of exemplars, rather than words in the exemplar itself, improves the coherence of generated responses, while preserving semantic meaning and conversation goals present in exemplar responses.