@inproceedings{imrattanatrai-etal-2026-learning,
title = "Learning from Dialogue Traces: Retrieval-Guided Tool Use and Response Generation for Task-Oriented Dialogue Agents",
author = "Imrattanatrai, Wiradee and
Fukuda, Ken and
Hamasaki, Masahiro",
editor = "Wang, Yimu and
Choi, Yee Man and
Wu, Di and
Ouyang, Siqi and
Azadani, Mozhgan Nasr and
Fung, Yi R. (May)",
booktitle = "Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently ({G}round{LM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.groundlm-1.71/",
pages = "981--1000",
ISBN = "979-8-89176-478-1",
abstract = "Effective task-oriented dialogue (TOD) requires agents to coordinate two tightly coupled decisions of when to invoke external tools and how to formulate a response that satisfies the user{'}s goal. However, their turn-level behavior remains weakly grounded under fixed zero-shot prompts. Meanwhile, existing retrieval-based TOD agents primarily retrieve task-relevant content from external knowledge sources, guiding which information to include but offering limited guidance on how to select tools and formulate responses in similar dialogue contexts. To mitigate this, we introduce \textbf{TRACE}, a \textbf{T}urn-level \textbf{R}etrieval \textbf{A}gent with \textbf{C}ontextual \textbf{E}xperiences for TOD{~}[{\ensuremath{<}}https://github.com/aistairc/TRACE-TOD{\ensuremath{>}}]. Instead of relying on retrieved factual records alone, TRACE retrieves goal-conditioned dialogue snippets as behavioral demonstrations at each turn. These snippets capture similar user goals, constraints, tool-use decisions, and response patterns to guide both tool calls and response generation. Experiments on MultiWOZ{~}2.1 and Schema-Guided Dialogue (SGD) show that TRACE improves Inform and Success over a competitive LLM agent and end-to-end fine-tuned models under both simulation-based and static-user evaluations. Human evaluation and retrieval-focused analyses further show that goal-conditioned snippets and retrieval quality drive these gains."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="imrattanatrai-etal-2026-learning">
<titleInfo>
<title>Learning from Dialogue Traces: Retrieval-Guided Tool Use and Response Generation for Task-Oriented Dialogue Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wiradee</namePart>
<namePart type="family">Imrattanatrai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ken</namePart>
<namePart type="family">Fukuda</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Masahiro</namePart>
<namePart type="family">Hamasaki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yimu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yee</namePart>
<namePart type="given">Man</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Di</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siqi</namePart>
<namePart type="family">Ouyang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mozhgan</namePart>
<namePart type="given">Nasr</namePart>
<namePart type="family">Azadani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yi</namePart>
<namePart type="given">R</namePart>
<namePart type="given">(May)</namePart>
<namePart type="family">Fung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-478-1</identifier>
</relatedItem>
<abstract>Effective task-oriented dialogue (TOD) requires agents to coordinate two tightly coupled decisions of when to invoke external tools and how to formulate a response that satisfies the user’s goal. However, their turn-level behavior remains weakly grounded under fixed zero-shot prompts. Meanwhile, existing retrieval-based TOD agents primarily retrieve task-relevant content from external knowledge sources, guiding which information to include but offering limited guidance on how to select tools and formulate responses in similar dialogue contexts. To mitigate this, we introduce TRACE, a Turn-level Retrieval Agent with Contextual Experiences for TOD [\ensuremath<https://github.com/aistairc/TRACE-TOD\ensuremath>]. Instead of relying on retrieved factual records alone, TRACE retrieves goal-conditioned dialogue snippets as behavioral demonstrations at each turn. These snippets capture similar user goals, constraints, tool-use decisions, and response patterns to guide both tool calls and response generation. Experiments on MultiWOZ 2.1 and Schema-Guided Dialogue (SGD) show that TRACE improves Inform and Success over a competitive LLM agent and end-to-end fine-tuned models under both simulation-based and static-user evaluations. Human evaluation and retrieval-focused analyses further show that goal-conditioned snippets and retrieval quality drive these gains.</abstract>
<identifier type="citekey">imrattanatrai-etal-2026-learning</identifier>
<location>
<url>https://aclanthology.org/2026.groundlm-1.71/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>981</start>
<end>1000</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Learning from Dialogue Traces: Retrieval-Guided Tool Use and Response Generation for Task-Oriented Dialogue Agents
%A Imrattanatrai, Wiradee
%A Fukuda, Ken
%A Hamasaki, Masahiro
%Y Wang, Yimu
%Y Choi, Yee Man
%Y Wu, Di
%Y Ouyang, Siqi
%Y Azadani, Mozhgan Nasr
%Y Fung, Yi R. (May)
%S Proceedings of the 1st Workshop on Grounding Language Models: Learning Faithfully and Efficiently (GroundLM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-478-1
%F imrattanatrai-etal-2026-learning
%X Effective task-oriented dialogue (TOD) requires agents to coordinate two tightly coupled decisions of when to invoke external tools and how to formulate a response that satisfies the user’s goal. However, their turn-level behavior remains weakly grounded under fixed zero-shot prompts. Meanwhile, existing retrieval-based TOD agents primarily retrieve task-relevant content from external knowledge sources, guiding which information to include but offering limited guidance on how to select tools and formulate responses in similar dialogue contexts. To mitigate this, we introduce TRACE, a Turn-level Retrieval Agent with Contextual Experiences for TOD [\ensuremath<https://github.com/aistairc/TRACE-TOD\ensuremath>]. Instead of relying on retrieved factual records alone, TRACE retrieves goal-conditioned dialogue snippets as behavioral demonstrations at each turn. These snippets capture similar user goals, constraints, tool-use decisions, and response patterns to guide both tool calls and response generation. Experiments on MultiWOZ 2.1 and Schema-Guided Dialogue (SGD) show that TRACE improves Inform and Success over a competitive LLM agent and end-to-end fine-tuned models under both simulation-based and static-user evaluations. Human evaluation and retrieval-focused analyses further show that goal-conditioned snippets and retrieval quality drive these gains.
%U https://aclanthology.org/2026.groundlm-1.71/
%P 981-1000
Markdown (Informal)
[Learning from Dialogue Traces: Retrieval-Guided Tool Use and Response Generation for Task-Oriented Dialogue Agents](https://aclanthology.org/2026.groundlm-1.71/) (Imrattanatrai et al., GroundLM 2026)
ACL