A scalable framework for learning from implicit user feedback to improve natural language understanding in large-scale conversational AI systems

TL;DR

This paper introduces a scalable, domain-agnostic framework that leverages implicit user feedback from live interactions to enhance natural language understanding in large-scale conversational AI systems, demonstrating significant improvements across multiple domains.

Contribution

It presents a novel automatic method for improving NLU by utilizing implicit feedback and dialog context, applicable across various domains in large-scale systems.

Findings

Improved NLU accuracy across 10 domains.

Effective use of implicit user feedback for supervision.

Enhanced user satisfaction inference from interaction data.

Abstract

Natural Language Understanding (NLU) is an established component within a conversational AI or digital assistant system, and it is responsible for producing semantic understanding of a user request. We propose a scalable and automatic approach for improving NLU in a large-scale conversational AI system by leveraging implicit user feedback, with an insight that user interaction data and dialog context have rich information embedded from which user satisfaction and intention can be inferred. In particular, we propose a general domain-agnostic framework for curating new supervision data for improving NLU from live production traffic. With an extensive set of experiments, we show the results of applying the framework and improving NLU for a large-scale production system and show its impact across 10 domains.