ONLINE SUPERVISED LEARNING OF NON-UNDERSTANDING RECOVERY POLICIES

Citations Over TimeTop 10% of 2006 papers

Abstract

Spoken dialog systems typically use a limited number of non- understanding recovery strategies and simple heuristic policies to engage them (e.g. first ask user to repeat, then give help, then transfer to an operator). We propose a supervised, online method for learning a non-understanding recovery policy over a large set of recovery strategies. The approach consists of two steps: first, we construct runtime estimates for the likelihood of success of each recovery strategy, and then we use these estimates to construct a policy. An experiment with a publicly available spoken dialog system shows that the learned policy produced a 12.5% relative improvement in the non-understanding recovery rate.

Related Papers

• → Learning mixed initiative dialog strategies by using reinforcement learning on both conversants(2005)37 cited
• → Why Guided Dialog Policy Learning performs well? Understanding the role of adversarial learning and its alternative(2023)
• → A Task-oriented Dialog Model with Task-progressive and Policy-aware Pre-training(2023)