Evaluating LLM-based Agents for Multi-Turn Conversations: A Survey

2025-03-28

AI Summary

This paper addresses the lack of a systematic evaluation framework for large language model (LLM)-driven multi-turn dialogue agents. Following the PRISMA methodology, we systematically review 250 studies to propose a dual-dimensional taxonomy—“what to evaluate” and “how to evaluate.” Methodologically, we introduce a novel five-dimensional evaluation framework encompassing task completion, response quality, user experience, memory retention, planning capability, and tool utilization. We further categorize evaluation approaches into four types: human annotation, automated metrics (e.g., BLEU, ROUGE), human-AI collaboration, and LLM-based self-evaluation—the first such classification in the literature. The resulting structured knowledge system constitutes the first comprehensive, principled evaluation framework specifically designed for multi-turn dialogue agents. It establishes a unified benchmark for empirical assessment and supports scalable, multi-paradigm research in conversational AI.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Conversational AI/Dialog Systems
• Multiagent Systems: Agent Communication

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

This survey examines evaluation methods for large language model (LLM)-based agents in multi-turn conversational settings. Using a PRISMA-inspired framework, we systematically reviewed nearly 250 scholarly sources, capturing the state of the art from various venues of publication, and establishing a solid foundation for our analysis. Our study offers a structured approach by developing two interrelated taxonomy systems: one that defines emph{what to evaluate} and another that explains emph{how to evaluate}. The first taxonomy identifies key components of LLM-based agents for multi-turn conversations and their evaluation dimensions, including task completion, response quality, user experience, memory and context retention, as well as planning and tool integration. These components ensure that the performance of conversational agents is assessed in a holistic and meaningful manner. The second taxonomy system focuses on the evaluation methodologies. It categorizes approaches into annotation-based evaluations, automated metrics, hybrid strategies that combine human assessments with quantitative measures, and self-judging methods utilizing LLMs. This framework not only captures traditional metrics derived from language understanding, such as BLEU and ROUGE scores, but also incorporates advanced techniques that reflect the dynamic, interactive nature of multi-turn dialogues.

Problem

• Evaluating LLM-based agents in multi-turn conversational settings
• Systematically reviewing 250 sources to establish evaluation foundations
• Developing taxonomies for what and how to evaluate conversational agents

Innovation

• PRISMA framework for systematic literature review
• Two taxonomies: what and how to evaluate
• Hybrid evaluation combining human and automated metrics