System-Level Natural Language Feedback

TL;DR

This paper introduces a novel framework for utilizing natural language feedback at the system level to improve models, focusing on metric and prompt design, with case studies in search and dialogue systems demonstrating its effectiveness.

Contribution

It presents a new approach for system-level use of NL feedback, combining it with instance-level feedback for enhanced model refinement.

Findings

System-level feedback improves model performance in search and dialogue tasks.

Combining system-level and instance-level feedback yields further gains.

Human-written feedback results in more grounded refinements than GPT-3.5 generated feedback.

Abstract

Natural language (NL) feedback offers rich insights into user experience. While existing studies focus on an instance-level approach, where feedback is used to refine specific examples, we introduce a framework for system-level use of NL feedback. We show how to use feedback to formalize system-level design decisions in a human-in-the-loop-process -- in order to produce better models. In particular this is done through: (i) metric design for tasks; and (ii) language model prompt design for refining model responses. We conduct two case studies of this approach for improving search query and dialog response generation, demonstrating the effectiveness of system-level feedback. We show the combination of system-level and instance-level feedback brings further gains, and that human written instance-level feedback results in more grounded refinements than GPT-3.5 written ones, underlying the importance of human feedback for building systems. We release our code and data at https://github.com/yyy-Apple/Sys-NL-Feedback.