Structured and natural responses co-generation for conversational search
Abstract
Duplicate record, see https://ink.library.smu.edu.sg/sis_research/7223/. Generating fluent and informative natural responses while maintaining representative internal states for search optimization is critical for conversational search systems. Existing approaches either 1) predict structured dialog acts first and then generate natural response; or 2) map conversation context to natural responses directly in an end-to-end manner. Both kinds of approaches have shortcomings. The former suffers from error accumulation while the semantic associations between structured acts and natural responses are confined in single direction. The latter emphasizes generating natural responses but fails to predict structured acts. Therefore, we propose a neural co-generation model that generates the two concurrently. The key lies in a shared latent space shaped by two informed priors. Specifically, we design structured dialog acts and natural response auto-encoding as two auxiliary tasks in an interconnected network architecture. It allows for the concurrent generation and bidirectional semantic associations. The shared latent space also enables asynchronous reinforcement learning for further joint optimization. Experiments show that our model achieves significant performance improvements.