“Data curation” names a set of actions — organizing, describing, validating, and preserving research data so it stays findable and usable by someone other than its creator (see CASRAI’s Data curation entry for the full operational definition). This guide surveys the actual software and services people use to do that work, organized into three categories: curation workflows built into repository platforms, standalone tools for cleaning and validating data and its metadata, and institutional or shared curation services that supply the human expertise no piece of software replaces on its own.
No single tool performs all of data curation end to end. In practice, a researcher or data steward typically combines a cleaning/validation tool used before deposit with a repository’s own curation review at the point of deposit, sometimes backed by an institutional curation service for datasets that need specialist review.
Repository-integrated curation workflows
Most established repository platforms build some curation review directly into the deposit process, rather than treating curation as a separate step performed elsewhere.
Dataverse
Dataverse, the open-source repository platform originally developed at Harvard and now run as a broader open-source project, structures deposits around dataset-level and file-level metadata, versioning, and a review workflow that lets a curator (an institutional data steward, not necessarily the depositor) check submitted metadata and files before a dataset version is published. Harvard’s own instance, Harvard Dataverse, is one widely used deployment; many other universities run their own separately branded Dataverse installations. See CASRAI’s Data repository entry for how repository-level review fits into the broader curation picture.
Figshare (institutional deployments)
Figshare, commercially operated by Digital Science, offers curation review as part of its institutional (Figshare for Institutions) product tier — an institution’s data curation staff can review submitted metadata and files before a dataset is made public, distinct from the free public figshare.com tier, which does not include pre-publication curation review. See CASRAI’s Figshare data repository guide for the fuller feature and certification picture.
Dryad
Dryad, a non-profit data repository partnered with the California Digital Library, curates and reviews submissions as a condition of publication — its curation team checks that deposited files are usable, appropriately licensed, and adequately described before a dataset goes live, rather than publishing deposits automatically on submission.
DSpace
DSpace, open-source institutional repository software used by many university libraries (often for a mix of publications, theses, and datasets), supports configurable review/workflow steps — an item can be routed through one or more approval stages before it becomes publicly visible, which many institutions use as their curation checkpoint.
CKAN
CKAN, open-source data portal software originally developed by the Open Knowledge Foundation, is widely used for government open-data portals and by some research data catalogues. It provides dataset organization, versioning, and metadata-entry forms, though — unlike Dataverse, Figshare for Institutions, or Dryad — it does not itself impose a curation-review gate before publication; that has to be layered on by whoever administers a given CKAN instance.
Metadata and quality-checking tools
A second category of tools focuses specifically on the cleaning, validating, and describing work that happens to data and its metadata, independent of any particular repository.
OpenRefine
OpenRefine (formerly Google Refine) is a free, open-source tool for cleaning and transforming messy tabular data — finding and correcting inconsistent values, splitting or merging columns, reconciling entries against external identifier sets, and tracking every change made so the cleaning process itself stays auditable. It is one of the most commonly taught data-cleaning tools in academic library data-services instruction, precisely because it handles the “validating and cleaning” step of curation (per the DCC’s own framing of digital curation) without requiring programming skills.
Frictionless Data / Frictionless Framework
Frictionless Data, a project associated with the Open Knowledge Foundation, provides a lightweight specification (Table Schema, Data Package) and accompanying open-source tooling for validating that a tabular dataset actually matches its declared structure — correct data types, required fields present, values within expected ranges — before it is shared or deposited. This addresses a specific curation gap: a dataset can look complete while still containing type mismatches or missing values that only a structural validation pass catches.
Repository-side metadata validation
Beyond standalone tools, repositories that mint DataCite DOIs enforce their own metadata validation at the point of submission — checking a deposit against DataCite’s mandatory metadata properties (Identifier, Creator, Title, Publisher, PublicationYear, ResourceType) before a DOI can be issued. See CASRAI’s Core metadata entry for the full mandatory-property set. This functions as an automated curation gate even where no human curator reviews every deposit individually.
Institutional and shared curation services
Software alone does not perform curation; a substantial share of real curation work is done by people, and much of that capacity is organized institutionally or shared across institutions rather than built by any single university from scratch.
Data Curation Network
The Data Curation Network (DCN) is a US-based membership consortium of institutional and non-profit data repositories, hosted by the University of Minnesota Libraries and originally funded by the Alfred P. Sloan Foundation, that pools curation expertise across member institutions — member repositories share curator staff time so that a dataset requiring specialist domain knowledge (a particular file format, discipline, or data type) can be reviewed by a curator with the right expertise, rather than requiring every member institution to independently develop every kind of specialist curation capacity itself. The DCN also publishes shared curation primers and best-practice documentation used well beyond its own membership.
In-house institutional data curation services
Many research universities maintain their own data curation function inside the library or a research data office, typically staffed by data librarians or dedicated data curators/stewards, who review datasets before deposit, help researchers select an appropriate repository, and advise on metadata and file-format choices. See CASRAI’s Data stewardship and the data curator role guide for how this function is typically structured and staffed.
Large-file and bulk-transfer infrastructure
For datasets too large to curate practically through a web upload form, research computing infrastructure such as Globus (originally developed at Argonne National Laboratory and the University of Chicago) is commonly used to move large research data volumes between storage systems, computing clusters, and repositories as part of a curation workflow — it is a transfer tool rather than a curation tool itself, but it is frequently a practical prerequisite for curating and depositing large datasets at all.
Choosing among these tools
Which of the above actually matters for a given project depends on where in the data lifecycle the work sits. Before deposit, cleaning tools like OpenRefine and structural validators like Frictionless Data address data quality directly. At deposit, the choice of repository determines how much curation review happens automatically — a repository with institutional curation review (Dataverse with an active curator, Figshare for Institutions, Dryad) catches problems a self-service deposit into an unreviewed CKAN instance would not. Where a dataset needs specialist review no single institution can staff alone, a shared service like the Data Curation Network is the more realistic option than expecting a generalist repository curator to have the needed domain expertise. See CASRAI’s How to choose an open data repository guide for the repository-selection decision this guide’s tool categories feed into, and the research data lifecycle guide for how curation fits alongside planning, storage, and reuse.
Frequently asked questions
Is a data repository the same thing as a data curation tool?
No. A repository is where a dataset is deposited and stored; curation is the set of actions (organizing, describing, validating, preserving) applied to the data, which may happen before deposit, as part of the repository’s own review process, or both. Some repositories (Dataverse, Figshare for Institutions, Dryad) build meaningful curation review into their deposit workflow; others provide storage and metadata forms without an active review gate. See CASRAI’s Data repository entry for the fuller distinction.
Do I need special software to curate research data, or can I do it manually?
Small, simple datasets can be organized and described manually with care — consistent file naming, a written README, manual checks for missing values. Tools like OpenRefine become genuinely valuable once a dataset is large or messy enough that manual review is impractical or error-prone, and structural validators like Frictionless Data catch classes of error (type mismatches, out-of-range values) that manual review reliably misses at scale.
Are data curation tools free to use?
OpenRefine, Frictionless Data/Frictionless Framework, CKAN, DSpace, and the core Dataverse software are all free and open source. Figshare for Institutions is a paid institutional product; the public figshare.com tier is free but does not include curation review. Institutional curation services are typically funded through library or research-office budgets rather than charged per use to individual researchers.
Who actually performs curation review at most institutions?
Typically a data steward, data curator, or data librarian based in the university library or a research data office — see CASRAI’s data steward / data curator role guide. For specialist review beyond a single institution’s in-house expertise, some universities participate in shared services like the Data Curation Network rather than building every specialist capability internally.
