What to Check Before Using Datashare for Your Own Document Research

Before you commit to Datashare, verify four things: whether you can actually get in (account, invitation, or local install), whether it handles your document types and languages, what the site says about data handling and terms, and whether its scale and security model match your research. Datashare is an ICIJ project built for searching large document sets, so it fits investigative and archival work better than small one-off file reviews.

Access: hosted site vs. local install

Datashare is open-source software from the International Consortium of Investigative Journalists (ICIJ). The practical question is which entry point you need.

• Hosted access at datashare.icij.org is intended for journalists and partners working with ICIJ-related document sets. Access is typically granted through ICIJ rather than open self-registration, so confirm your eligibility before planning around it.
• A local installation is the route for your own documents. You run Datashare on your own machine or server, and your files stay in your environment.

Check before starting:

1. Whether you qualify for hosted access, or whether you should install locally.
2. The hardware you have — Datashare is designed for large corpora, so disk space and memory matter more than for a simple file search tool.
3. Who will administer the instance if you install it for a team.

Document types and languages

Datashare is built for text-heavy collections: PDFs, Office documents, emails, and scanned files that need OCR before they become searchable. It supports multilingual search, which matters if your corpus mixes languages.

Before uploading anything, test a small sample:

• Run OCR on a few scanned pages and check the extracted text quality.
• Search for a term you know appears in a document and confirm it is found.
• Try a query in each language present in your set.

If your material is mostly images, audio, or structured databases, Datashare is a weaker fit than a purpose-built tool.

Terms, privacy, and data handling

Read the site's terms and privacy information before uploading sensitive material. Two distinctions matter:

• Hosted use means your documents sit on infrastructure you do not control. Confirm what the terms say about retention, access, and deletion.
• Local use keeps files in your environment, but you are then responsible for security, backups, and access control.

For confidential sources or leaked material, treat the hosting question as a decision, not a detail. If the terms do not clearly address your sensitivity requirements, default to a local install you control.

Scale and sensitivity fit

Datashare earns its place when a collection is too large to review by hand and you need cross-document search, entity extraction, and filtering. It is overkill for a handful of files.

Use this quick check:

A practical first test

Pick 20–50 representative documents, including at least one scanned file and one in a second language. Index them, run three searches you can verify by hand, and check OCR output on the scanned file. If results are accurate and the interface fits your workflow, scale up. If OCR quality or language handling fails at this size, it will not improve on a larger corpus.

The main risks are assuming hosted access is open when it is not, and uploading sensitive files before reading the terms. Resolve both before you invest time in indexing.