arXiv:2603.16073v2 Announce Type: replace Abstract: Scientific papers advance $\textit{claims}$ that later work supports, extends, or sometimes refutes. Yet existing methods for citation and claim analysis capture only fragments of this dialogue. In this work, we make these interactions explicit at the level of individual scientific claims. We introduce $\texttt{ClaimFlow}$, a claim-centric view of the NLP literature, built from $1{,}617$ ACL Anthology papers $(1979 - 2025)$ that are manually annotated with $5{,}689$ claims and $4{,}871$ cross-paper claim relations, indicating whether a citing
The proliferation of AI-generated content and the increasing complexity of scientific literature necessitates better tools for provenance and understanding the evolution of claims.
This work introduces a novel method for explicitly tracking the lineage and interactions of scientific claims, which is crucial for evaluating research impact and identifying foundational knowledge in AI.
The ability to systematically trace claim evolution across papers will enable more robust scientific evaluation and potentially accelerate knowledge discovery by highlighting key contributions and refutations.
- · NLP researchers
- · Scientific publishers
- · AI ethics researchers
- · Poorly substantiated research
- · Undetected duplicated claims
Improved methodologies for meta-analysis and systematic reviews in specific scientific domains.
Development of AI systems that can independently evaluate the validity and evolution of scientific arguments.
Enhanced trust in scientific findings due to transparent claim lineage, potentially influencing public policy and investment decisions.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL