What ArcXiv is—and what it is not

ArcXiv helps readers search and browse arXiv records, save papers, build learning material, and compare research signals. It is a discovery interface, not a publisher, journal, peer-review service, or substitute for reading the original work.

An ArcXiv page does not show that arXiv, Cornell University, a paper’s authors, or a venue endorses ArcXiv or any ranking, lesson, summary, or assessment shown here.

Public paper records are built primarily from descriptive metadata supplied through arXiv interoperability: identifiers, titles, author strings, abstracts, categories, submission and revision dates, comments, journal references, DOI values, and license identifiers when available. ArcXiv may normalize whitespace or formatting for display; it does not become the source of record by doing so.

Imports and indexes can lag behind the official record. For the current version, withdrawal status, author-provided wording, or a definitive identifier, follow the link on the paper page to arXiv.org.

ArcXiv does not publish or serve paper PDFs, TeX source, or full paper bodies. Original records and download links lead to arXiv. Copyright and reuse terms for an e-print belong to that work and its rights holders, not to ArcXiv.

Some research-signal jobs may process an arXiv PDF transiently. The PDF and extracted body are not retained after that job; concise derived reasons and source locators may be stored. The exact distinction between abstract-based and full-paper signals is documented in the research-signals methodology.

arXiv states that descriptive metadata about its e-prints is available under the CC0 1.0 Universal dedication. See the arXiv API Terms of Use for the source terms.

CC0 for descriptive metadata is not a blanket license for paper text, figures, PDFs, code, datasets, or other linked material. A paper’s arXiv record identifies its submission license when one is available. Check that record and the rights holder’s terms before reuse.

ArcXiv interface copy, generated explanations, assessments, and other product output are not represented here as part of arXiv’s CC0 metadata. Contact ArcXiv before assuming a reuse permission that is not expressly shown.

Feeds, search ordering, related-paper choices, research signals, lessons, and cited answers are ArcXiv product layers. They may be computed algorithmically or generated with AI; they are not fields supplied by arXiv or statements from paper authors.

Unless a page expressly identifies a human review, do not assume one occurred. Read the editorial, AI, and corrections policy for labels, limits, and how to report an error.

• arXiv API Terms of Use
• arXiv licenses and reuse guidance
• arXiv name and logo guidelines
• ArcXiv contact and reporting channel