Open 4 billion parameter model improves scientific citation recovery accuracy
ATTRICITE: Training an Open 4B Model for Citation Recovery toward Faithful Attribution
Digital LibrariesArtificial Intelligence
Summary
Identifying the exact scientific paper an author is referring to in their writing can be tricky. This paper looks at a way to find the original source from the passage that includes a citation, helping ensure correct credit is given. The researchers created a new open-source AI model called ATTRICITE that gets better at this task than some larger models, even though it uses fewer computing resources. They also made a dataset from recent computer science papers to test how well this works over time.
What this means in practice
- •For academic search engines: Improve search results by accurately linking mention passages to original cited papers, enhancing paper recommendation relevance.
- •For document management systems: Automatically verify citations in digital scientific libraries to ensure faithful source attribution and reduce errors.
Authors
Yee Man Choi, Xuehang Guo, Songcheng Cai, Yimu Wang, Yi R. Fung, Qingyun Wang
Abstract
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.