{"sections":[{"heading":"Why datasets and software create adjudication problems","paragraphs":["A substantial and growing share of scientific research produces primary outputs that are not peer-reviewed journal articles in the traditional sense. Researchers in computational biology, bioinformatics, genomics, climate science, epidemiology, and machine learning often produce datasets, databases, software packages, and computational pipelines whose impact on their fields exceeds that of any individual publication. These researchers frequently encounter O-1A adjudications that treat their contributions skeptically because USCIS adjudicators are trained to look for peer-reviewed publications and named prizes, not for software repositories or database accession records.","The core regulatory problem is that 8 C.F.R. § 214.2(o)(3)(iv)(F) defines the scholarly articles criterion in terms of authorship of scholarly articles in professional journals or other major media. A dataset deposited in a public repository such as NCBI GenBank, figshare, the Harvard Dataverse, or Dryad, and a software package maintained as an open-source project on GitHub, do not obviously fit within this definition. The regulatory text, however, also permits comparable evidence where the criteria do not readily apply to the occupation, and this provision is the primary vehicle for presenting dataset and software contributions to USCIS in a form that satisfies the evidentiary requirements.","The comparable evidence provision at 8 C.F.R. § 214.2(o)(3)(iv) allows petitioners to submit evidence comparable to that described in the eight criteria when those criteria are not readily applicable. An O-1A petition for a computational scientist or data-driven researcher should explicitly invoke this provision in the cover letter, explain why the standard criteria do not fully capture the field's recognition framework, and then present the dataset or software contribution with the same analytical rigor that would accompany a peer-reviewed publication exhibit. Attempting to force a major software package into the scholarly articles criterion without invoking the comparable evidence provision is weaker than explicitly using the regulatory flexibility that already exists."]},{"heading":"What the original contributions criterion requires","paragraphs":["The original contributions of major significance criterion at 8 C.F.R. § 214.2(o)(3)(iv)(E) requires showing contributions to the field that are both original and of major significance—not simply original, and not simply recognized as contributions. For dataset and software-producing researchers, the major significance element is usually the most important to document because USCIS adjudicators may be uncertain how to assess significance in the absence of traditional markers like publication in a high-impact journal or receipt of a named award. The petition must provide the analytical framework for making this assessment, not leave it to the adjudicator.","A software package used by thousands of researchers across multiple disciplines—such as widely adopted bioinformatics tools, statistical computing libraries, or machine learning frameworks in their respective fields—constitutes an original contribution of major significance that expert declarations can support readily. The challenge is documentation: citation counts for companion papers if one exists, download statistics from package repositories such as CRAN download logs or conda-forge counters, GitHub star and fork statistics, and adoption in published papers that cite the software all contribute to this exhibit. Expert declarations should explain what the relevant community considers significance—that a tool used in a substantial fraction of published papers in a field is a central contribution, not a marginal one.","For datasets, the evidence of major significance comes from downstream use: the number of publications that used the dataset, policy documents or clinical guidelines that cited it, derivative datasets or models built on it, and the recognized scientific questions that became tractable only after the dataset was made available. Major genomics initiatives and large epidemiological cohorts represent datasets of self-evident significance. But smaller, specialized datasets can also satisfy the criterion if the petition demonstrates that they answered questions the field had been unable to address and that subsequent research built on them. This requires expert declarations from researchers who actually used the dataset, not just general characterizations of its quality or importance."]},{"heading":"Evidence that routinely satisfies the criteria","paragraphs":["Data-descriptor articles—publications in journals specifically designed for dataset documentation, such as Scientific Data from Nature Portfolio, GigaScience, Data in Brief, and PLOS ONE's data note format—are peer-reviewed publications that clearly qualify as scholarly articles. A well-cited data descriptor in Scientific Data for a major genomics, ecological, or climate dataset constitutes direct evidence of recognition by the scientific community. Data descriptors should be included in the scholarly articles exhibit with their citation counts and with expert commentary on the dataset's downstream impact, establishing a link between the publication and the original contributions criterion as well.","Software publications—articles in the Journal of Open Source Software, the Journal of Statistical Software, Bioinformatics Applications Notes, and similar peer-reviewed outlets—similarly qualify as scholarly articles. A Journal of Open Source Software article documenting a widely-used Python or R package carries citation counts and usage data that demonstrate real-world impact. Citation records for software publications should be pulled from Google Scholar, Web of Science, and Dimensions, as different databases index software papers with different levels of comprehensiveness. Including documentation from multiple sources provides a more complete picture of the software's citation impact and supports the original contributions exhibit simultaneously.","GitHub metrics—stars, forks, watchers, and download counts from package managers—serve as supporting evidence for the original contributions exhibit rather than as standalone evidence of extraordinary ability. These metrics document adoption but do not by themselves demonstrate that the adoption constitutes a major contribution to the field. Expert declarations are necessary to bridge the gap between adoption data and the legal standard of major significance. A declaration from a senior researcher explaining that the petitioner's package is the standard tool for a specific computational task within the relevant community—and that without it the community would need inferior alternatives or independent development—translates usage statistics into evidence that meets the regulatory standard."]},{"heading":"Evidence USCIS regularly discounts","paragraphs":["Testimonials from users of a software package or dataset gathered through online community forums, social media, or code review threads do not satisfy any of the eight evidentiary criteria and have limited value as supporting exhibits. USCIS has become more sophisticated about distinguishing between community appreciation evidenced by online comments and the kind of peer recognition that demonstrates extraordinary ability: expert declarations from recognized authorities, citations in peer-reviewed publications, and adoption by distinguished research programs. Self-reported user testimonials from individuals whose own credentials and field standing are unknown to the adjudicator carry minimal analytical weight in isolation.","GitHub statistics presented without expert context are regularly insufficient to demonstrate major significance. An adjudicator who sees a repository with several thousand stars has no framework for determining whether that represents significant community adoption in the relevant field or ordinary activity for an open-source project of its type. Without expert declarations contextualizing those numbers—explaining what the typical adoption level is for tools in the relevant specialty, what research programs have adopted the tool, and what methodological problems it addressed—the statistics are evidence of activity but not evidence of distinction within the community of researchers who constitute the petitioner's field.","Preprints alone—posted on bioRxiv, arXiv, SSRN, or medRxiv—do not satisfy the scholarly articles criterion as typically applied, because they have not undergone peer review. A preprint that has been downloaded and cited extensively may reflect significant community engagement, but the adjudicator will generally expect to see that it was subsequently published in a peer-reviewed outlet. Petitioners who have produced important preprints that remain unpublished should consider framing those preprints under the original contributions criterion—as contributions documented by citation and expert analysis—while separately pursuing peer-reviewed publication before the petition filing date if timing permits."]},{"heading":"How to present borderline evidence","paragraphs":["A dataset with modest download statistics but significant policy impact—for example, an environmental monitoring dataset incorporated into a state agency's regulatory framework—can satisfy the original contributions criterion if the policy adoption is well-documented. The petition should include the regulatory document citing the dataset, a letter from the relevant agency official confirming how the data was used, and an expert declaration explaining why policy-level adoption represents a significant scientific contribution beyond what citation counts alone would indicate. Framing the policy adoption explicitly under the major significance element of the criterion, rather than presenting it as background context, converts a potentially ambiguous exhibit into a persuasive one.","Invitations to present dataset or software work at major conferences—whether in computational biology, data science, or discipline-specific venues—can support the original contributions exhibit by providing evidence of recognized significance, even though presenting one's own work does not itself satisfy the judging criterion. An expert declaration should explain what an invited presentation at a relevant conference implies about the petitioner's standing in the field and how keynote invitations are typically extended within the community. This transforms an event that could be dismissed as self-promotion into evidence of community recognition of the petitioner's specific contributions.","Inclusion of a dataset or software package in a published systematic review, meta-analysis, or methodological benchmark study provides strong evidence that the contribution is recognized as significant by peer researchers who evaluated it specifically for inclusion in their work. These secondary publications—which analyzed the dataset or compared the software against competing tools and selected it—are a form of expert validation that USCIS can readily understand: another researcher examined the contribution and found it worthy of specific inclusion in their published work. These citations should be highlighted in the original contributions exhibit alongside total citation counts."]},{"heading":"Building and auditing the complete file","paragraphs":["A pre-filing audit for a computational scientist should assess evidence under each applicable criterion. First, identify all peer-reviewed publications including data descriptors and software papers with citation analyses. Second, identify dataset and software contributions with usage and citation data, and assess whether these are presentable under original contributions or comparable evidence. Third, identify awards or grants recognizing the work. Fourth, identify peer review service for journals, conferences, or grant panels. Fifth, assess critical role claims for academic or industry research positions. Sixth, run salary comparisons against BLS OEWS data for the relevant occupation code. For each criterion, the audit should assess whether the evidence would survive a skeptical adjudicator's review before the petition is assembled.","The cover letter is particularly critical in dataset and software petitions because it must accomplish more analytical work than in conventional academic petitions. Rather than simply mapping exhibits to criteria, it must explain why the petitioner's primary output form is relevant to the extraordinary ability standard, invoke the comparable evidence provision where appropriate, explain the significance metrics used in the field, and apply that framework to the specific petitioner. Adjudicators who receive a well-organized, analytically complete cover letter alongside a thoughtfully curated exhibit package make better decisions than those who receive a collection of documents with a brief cover letter that describes rather than analyzes the evidentiary record.","Petitions for computational scientists who work at the intersection of two fields, or in industry research rather than academic research, should identify the specific professional community whose standards will govern the extraordinary ability assessment. A machine learning researcher at a technology company who produces both research publications and widely-adopted open-source frameworks is best assessed as a researcher in the machine learning community—with relevant recognition forums including NeurIPS, ICML, ICLR, ACL, and EMNLP and relevant journals including JMLR and Transactions on Machine Learning Research—not as a software engineer. The petition's expert declarations should make this community identification explicit."]}],"article":{"title":"How to Document O-1A Extraordinary Ability When Your Primary Research Output Is a Dataset or Software Package","excerpt":"Researchers whose primary output is a dataset or software package often struggle to frame that work within O-1A evidentiary categories. This guide explains how download metrics, dependency counts, citation trails, and repository adoption data translate into persuasive original contributions and scholarly articles evidence.","category":"O-1 Strategy","date":"Sep 18, 2026","readTime":"8 min read"},"prev":{"title":"Filing an O-1A Extension After a Research Interruption: Evidence Strategy for Gaps in Activity","slug":"filing-an-o-1a-extension-after-a-research-interruption-evidence-strategy-for-gaps-in-activity"},"next":{"title":"O-1A for Dipterists: Building an Extraordinary Ability Petition for Fly Biologists and Medical Entomologists","slug":"o-1a-for-dipterists-building-an-extraordinary-ability-petition-for-fly-biologists-and-medical-entomologists"},"related":[{"title":"O-1A Industry Awards: Building the Extraordinary Ability Case","slug":"o-1a-industry-awards-building-the-extraordinary-ability-case"},{"title":"How to Present Collaborative Research in an O-1A Petition When Contributions Are Shared Across a Large Team","slug":"how-to-present-collaborative-research-in-an-o-1a-petition-when-contributions-are-shared-across-a-large-team"},{"title":"Filing an O-1A Extension After a Research Interruption: Evidence Strategy for Gaps in Activity","slug":"filing-an-o-1a-extension-after-a-research-interruption-evidence-strategy-for-gaps-in-activity"},{"title":"How to Document Extraordinary Ability When Your Primary Funding Source Is a Foreign National Research Foundation","slug":"how-to-document-extraordinary-ability-when-your-primary-funding-source-is-a-foreign-national-research-foundation"},{"title":"O-1A Petition Strategy for Researchers Transitioning from a Postdoctoral Position to an Independent Research Role","slug":"o-1a-petition-strategy-for-researchers-transitioning-from-a-postdoctoral-position-to-an-independent-research-role"},{"title":"How to Address a Prior O-1A Denial in a Renewed or Refiled Petition","slug":"how-to-address-a-prior-o-1a-denial-in-a-renewed-or-refiled-petition"}]}