Offline and Online Satisfaction Prediction in Open-Domain Conversational Systems

TL;DR

This paper introduces ConvSAT, a novel model for predicting user satisfaction in open-domain conversational systems, enhancing responsiveness and adaptability by integrating multiple conversation signals and demonstrating superior performance in real-world and benchmark datasets.

Contribution

The paper presents ConvSAT, a new satisfaction prediction model that combines diverse conversation features, achieving improved accuracy in both offline and online evaluations over existing methods.

Findings

ConvSAT outperforms state-of-the-art methods in satisfaction prediction.

The model effectively integrates multiple conversation representations.

Results show significant improvements in real-time user satisfaction estimation.

Abstract

Predicting user satisfaction in conversational systems has become critical, as spoken conversational assistants operate in increasingly complex domains. Online satisfaction prediction (i.e., predicting satisfaction of the user with the system after each turn) could be used as a new proxy for implicit user feedback, and offers promising opportunities to create more responsive and effective conversational agents, which adapt to the user's engagement with the agent. To accomplish this goal, we propose a conversational satisfaction prediction model specifically designed for open-domain spoken conversational agents, called ConvSAT. To operate robustly across domains, ConvSAT aggregates multiple representations of the conversation, namely the conversation history, utterance and response content, and system- and user-oriented behavioral signals. We first calibrate ConvSAT performance against state of the art methods on a standard dataset (Dialogue Breakdown Detection Challenge) in an online regime, and then evaluate ConvSAT on a large dataset of conversations with real users, collected as part of the Alexa Prize competition. Our experimental results show that ConvSAT significantly improves satisfaction prediction for both offline and online setting on both datasets, compared to the previously reported state-of-the-art approaches. The insights from our study can enable more intelligent conversational systems, which could adapt in real-time to the inferred user satisfaction and engagement.