arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2609.14248cs.DLcs.AI

ATTRICITE:训练一个开放的4B模型用于引文恢复以实现忠实归因

ATTRICITE: Training an Open 4B Model for Citation Recovery toward Faithful Attribution

  • University of Waterloo(滑铁卢大学)
  • Vector Institute(向量研究所)
  • College of William and Mary(威廉与玛丽学院)
  • University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)

机构由 AI 辅助整理,请以论文原文为准。

Yee Man Choi, Xuehang Guo, Songcheng Cai, Yimu Wang, Yi R. Fung, Qingyun Wang

AI总结:

本文提出ATTRICITE,一个4B参数模型,通过工具使用进行引文恢复,在CITEALIGN数据集上以GRPO微调提升准确率至59.8%,优于更大模型,促进忠实归因研究。

AI中文摘要:

忠实的引文归因始于识别科学声明所意图引用的来源。我们通过引文恢复来研究这种来源识别能力:从包含引用的段落中恢复原作者所引用的论文。我们的评估采用已发表作者的引用作为可观察的人类归因信号,并将目标恢复作为迈向忠实归因进展的代理指标。我们引入了ATTRICITE,一个开放的4B参数模型,在CiteGuard检索环境中训练用于工具使用的引文恢复,以及CITEALIGN,一个从近期科学文献中抽取的包含7,607个实例的计算机科学数据集。为了进行受控评估,我们构建了CITEALIGN的一个包含709个实例的基准子集,其中包括来自2024年出版物的410个开发实例和来自2025年出版物的299个时间上留出的测试实例。在推理温度为0.7的三次运行中,GRPO微调将Qwen3-4B的目标匹配准确率从49.4%±1.5%提高到59.8%±0.2%,提升了10.4个百分点。尽管仅使用4B参数,ATTRICITE仍优于gpt-oss-20b,并接近GPT-5.4-mini(差距3.9个百分点),而Gemma 4 31B IT以72.0%±1.0%的成绩实现了最强的整体性能。我们发布了模型和收集流程(见https URL),以支持在不断演变的科学文献中针对忠实归因的引文恢复的可重复研究。

英文摘要:

Faithful citation attribution begins with identifying the intended source for a scientific claim. We study this source-identification capability through citation recovery: recovering the paper cited by the original author from a citation-bearing passage. Our evaluation adopts the published author's citation as an observable human attribution signal and uses target recovery as a proxy for progress toward faithful attribution. We introduce ATTRICITE, an open 4B-parameter model trained for tool-using citation recovery within the CiteGuard retrieval environment, together with CITEALIGN, a 7,607-instance computer-science dataset drawn from recent scientific literature. For controlled evaluation, we construct a 709-instance benchmark subset of CITEALIGN, comprising 410 development instances from 2024 publications and 299 temporally held-out test instances from 2025 publications. Across three runs at an inference temperature of 0.7, GRPO fine-tuning improves Qwen3-4B from 49.4%$\pm$1.5% to 59.8%$\pm$0.2% target-match accuracy, a gain of 10.4 percentage points. Despite using only 4B parameters, ATTRICITE outperforms gpt-oss-20b and comes within 3.9 points of GPT-5.4-mini, while Gemma 4 31B IT achieves the strongest overall performance at 72.0%$\pm$1.0%. We release the model and collection pipeline https://github.com/KathCYM/AttriCite to support reproducible research on citation recovery toward faithful attribution in a continually evolving scientific literature.

补充信息

↑