Research

Paper

AI LLM March 02, 2026

Let the Agent Search: Autonomous Exploration Beats Rigid Workflows in Temporal Question Answering

Authors

Xufei Lv, Jiahui Yang, Yifu Gao, Linbo Qiao, Houde Liu

Abstract

Temporal Knowledge Graph Question Answering (TKGQA) demands multi-hop reasoning under temporal constraints. Prior approaches based on large language models (LLMs) typically rely on rigid, hand-crafted retrieval workflows or costly supervised fine-tuning. We show that simply granting an off-the-shelf LLM autonomy, that is, letting it decide what to do next, already yields substantial gains even in a strict zero-shot setting. Building on this insight, we propose AT2QA, an autonomous, training-free agent for temporal question answering that iteratively interacts with the temporal knowledge graph via a general search tool for dynamic retrieval. Experiments on MultiTQ demonstrate large improvements: AT2QA achieves 88.7% Hits@1 (+10.7% over prior SOTA), including a +20.1% gain on challenging multi-target queries, showing that agentic autonomy can decisively outperform fine-tuning for temporal question answering. Code and the full set of sampled trajectories are available on https://github.com/AT2QA-Official-Code/AT2QA-Official-Code

Metadata

arXiv ID: 2603.01853
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-03-02
Fetched: 2026-03-03 04:34

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.01853v1</id>\n    <title>Let the Agent Search: Autonomous Exploration Beats Rigid Workflows in Temporal Question Answering</title>\n    <updated>2026-03-02T13:33:39Z</updated>\n    <link href='https://arxiv.org/abs/2603.01853v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.01853v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Temporal Knowledge Graph Question Answering (TKGQA) demands multi-hop reasoning under temporal constraints. Prior approaches based on large language models (LLMs) typically rely on rigid, hand-crafted retrieval workflows or costly supervised fine-tuning. We show that simply granting an off-the-shelf LLM autonomy, that is, letting it decide what to do next, already yields substantial gains even in a strict zero-shot setting. Building on this insight, we propose AT2QA, an autonomous, training-free agent for temporal question answering that iteratively interacts with the temporal knowledge graph via a general search tool for dynamic retrieval. Experiments on MultiTQ demonstrate large improvements: AT2QA achieves 88.7% Hits@1 (+10.7% over prior SOTA), including a +20.1% gain on challenging multi-target queries, showing that agentic autonomy can decisively outperform fine-tuning for temporal question answering. Code and the full set of sampled trajectories are available on https://github.com/AT2QA-Official-Code/AT2QA-Official-Code</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-03-02T13:33:39Z</published>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Xufei Lv</name>\n    </author>\n    <author>\n      <name>Jiahui Yang</name>\n    </author>\n    <author>\n      <name>Yifu Gao</name>\n    </author>\n    <author>\n      <name>Linbo Qiao</name>\n    </author>\n    <author>\n      <name>Houde Liu</name>\n    </author>\n  </entry>"
}