Multi-Faceted Evaluation of Tool-Augmented Dialogue Systems

2025/10/22 by Hou, Zhaoyi Joey, Shourya, Tanya, Wang, Yingfan +3
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2510.19186

Abstract

Evaluating conversational AI systems that use external tools is challenging, as errors can arise from complex interactions among user, agent, and tools. While existing evaluation methods assess either user satisfaction or agents' tool-calling capabilities, they fail to capture critical errors in multi-turn tool-augmented dialogues-such as when agents misinterpret tool results yet appear satisfactory to users. We introduce TRACE, a benchmark of systematically synthesized tool-augmented conversations covering diverse error cases, and SCOPE, an evaluation framework that automatically discovers diverse error patterns and evaluation rubrics in tool-augmented dialogues. Experiments show SCOPE significantly outperforms the baseline, particularly on challenging cases where user satisfaction signals are misleading.

Citations

• Neural Models and Language Model Prompting for the Multidimensional Evaluation of Open-Ended Conversations
• Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy
• Evaluating LLM-based Agents for Multi-Turn Conversations: A Survey
• Automated test generation to evaluate tool-augmented LLMs as conversational AI agents
• Reward Models in Deep Reinforcement Learning: A Survey
• On the Benchmarking of LLMs for Open-Domain Dialogue Evaluation
• Planning, Creation, Usage: Benchmarking LLMs for Comprehensive Tool Utilization in Real-World Complex Scenarios
• ToolTalk: Evaluating Tool-Usage in a Conversational Setting
• MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use
• MINT: Evaluating LLMs in Multi-turn Interaction with Tools and Language Feedback
• Simple LLM Prompting is State-of-the-Art for Robust and Multilingual Dialogue Evaluation
• Counterfactually Auditable Lifecycle Certification for Autonomous Agents
• ToolAlpaca: Generalized Tool Learning for Language Models with 3000 Simulated Cases
• API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs
• G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment

Related

• Dialogue You Can Trust: Human and AI Perspectives on Generated Conversations
• ToolCritic: Detecting and Correcting Tool-Use Errors in Dialogue Systems
• TD-EVAL: Revisiting Task-Oriented Dialogue Evaluation by Combining Turn-Level Precision with Dialogue-Level Comparisons
• Don't Forget Your ABC's: Evaluating the State-of-the-Art in Chat-Oriented Dialogue Systems
• ChatChecker: A Framework for Dialogue System Testing and Evaluation Through Non-cooperative User Simulation