A Conditional Variational Framework for Dialog Generation
This work addresses the need for more controllable and personalized dialog generation, though it appears incremental as it builds on existing latent variable models.
The paper tackles the problem of uncontrollable response generation in open-domain dialog systems by proposing a conditional variational framework that generates responses based on specific attributes like genericness or sentiment, with experiments validating its potential for meaningful attribute-aligned responses.
Deep latent variable models have been shown to facilitate the response generation for open-domain dialog systems. However, these latent variables are highly randomized, leading to uncontrollable generated responses. In this paper, we propose a framework allowing conditional response generation based on specific attributes. These attributes can be either manually assigned or automatically detected. Moreover, the dialog states for both speakers are modeled separately in order to reflect personal features. We validate this framework on two different scenarios, where the attribute refers to genericness and sentiment states respectively. The experiment result testified the potential of our model, where meaningful responses can be generated in accordance with the specified attributes.