Research

Paper

AI LLM March 10, 2026

ESAinsTOD: A Unified End-to-End Schema-Aware Instruction-Tuning Framework for Task-Oriented Dialog Modeling

Authors

Dechuan Teng, Chunlin Lu, Libo Qin, Wanxiang Che

Abstract

Existing end-to-end modeling methods for modular task-oriented dialog systems are typically tailored to specific datasets, making it challenging to adapt to new dialog scenarios. In this work, we propose ESAinsTOD, a unified End-to-end Schema-Aware Instruction-tuning framework for general Task-Oriented Dialog modeling. This framework introduces a structured methodology to go beyond simply fine-tuning Large Language Models (LLMs), enabling flexible adaptation to various dialogue task flows and schemas. Specifically, we leverage full-parameter fine-tuning of LLMs and introduce two alignment mechanisms to make the resulting system both instruction-aware and schema-aware: (i) instruction alignment, which ensures that the system faithfully follows task instructions to complete various task flows from heterogeneous TOD datasets; and (ii) schema alignment, which encourages the system to make predictions adhering to the specified schema. In addition, we employ session-level end-to-end modeling, which allows the system to access the results of previously executed task flows within the dialogue history, to bridge the gap between the instruction-tuning paradigm and the real-world application of TOD systems. Empirical results show that while a fine-tuned LLM serves as a strong baseline, our structured approach provides significant additional benefits. In particular, our findings indicate that: (i) ESAinsTOD outperforms state-of-the-art models by a significant margin on end-to-end task-oriented dialog modeling benchmarks: CamRest676, In-Car and MultiWOZ; (ii) more importantly, it exhibits superior generalization capabilities across various low-resource settings, with the proposed alignment mechanisms significantly enhancing zero-shot performance; and (iii) our instruction-tuning paradigm substantially improves the model's robustness against data noise and cascading errors.

Metadata

arXiv ID: 2603.09691
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-03-10
Fetched: 2026-03-11 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.09691v1</id>\n    <title>ESAinsTOD: A Unified End-to-End Schema-Aware Instruction-Tuning Framework for Task-Oriented Dialog Modeling</title>\n    <updated>2026-03-10T13:59:02Z</updated>\n    <link href='https://arxiv.org/abs/2603.09691v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.09691v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Existing end-to-end modeling methods for modular task-oriented dialog systems are typically tailored to specific datasets, making it challenging to adapt to new dialog scenarios. In this work, we propose ESAinsTOD, a unified End-to-end Schema-Aware Instruction-tuning framework for general Task-Oriented Dialog modeling. This framework introduces a structured methodology to go beyond simply fine-tuning Large Language Models (LLMs), enabling flexible adaptation to various dialogue task flows and schemas. Specifically, we leverage full-parameter fine-tuning of LLMs and introduce two alignment mechanisms to make the resulting system both instruction-aware and schema-aware: (i) instruction alignment, which ensures that the system faithfully follows task instructions to complete various task flows from heterogeneous TOD datasets; and (ii) schema alignment, which encourages the system to make predictions adhering to the specified schema. In addition, we employ session-level end-to-end modeling, which allows the system to access the results of previously executed task flows within the dialogue history, to bridge the gap between the instruction-tuning paradigm and the real-world application of TOD systems. Empirical results show that while a fine-tuned LLM serves as a strong baseline, our structured approach provides significant additional benefits. In particular, our findings indicate that: (i) ESAinsTOD outperforms state-of-the-art models by a significant margin on end-to-end task-oriented dialog modeling benchmarks: CamRest676, In-Car and MultiWOZ; (ii) more importantly, it exhibits superior generalization capabilities across various low-resource settings, with the proposed alignment mechanisms significantly enhancing zero-shot performance; and (iii) our instruction-tuning paradigm substantially improves the model's robustness against data noise and cascading errors.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <published>2026-03-10T13:59:02Z</published>\n    <arxiv:comment>Published at International Journal of Machine Learning and Cybernetics (IJMLC)</arxiv:comment>\n    <arxiv:primary_category term='cs.CL'/>\n    <arxiv:journal_ref>Int. J. Mach. Learn. &amp; Cyber. 17, 127 (2026)</arxiv:journal_ref>\n    <author>\n      <name>Dechuan Teng</name>\n    </author>\n    <author>\n      <name>Chunlin Lu</name>\n    </author>\n    <author>\n      <name>Libo Qin</name>\n    </author>\n    <author>\n      <name>Wanxiang Che</name>\n    </author>\n    <arxiv:doi>10.1007/s13042-025-02823-6</arxiv:doi>\n    <link href='https://doi.org/10.1007/s13042-025-02823-6' rel='related' title='doi'/>\n  </entry>"
}