What Is Copyright in Web Data? – Definition and Considerations
Learn how copyright applies to collected web content, what facts versus expression means, where database rights differ, and what to do in practice.
Copyright protects original expression. That distinction — expression rather than fact — is the single most useful thing to understand about how it applies to collected web data. Nothing here is legal advice.
1. Expression Is Protected; Facts Generally Are Not
A product's price is a fact. The paragraph describing the product is expression. A review's star rating is a fact; the review's text is expression.
That is why collecting structured attributes and republishing entire articles sit in very different positions, even though both were fetched the same way.
2. Compilations Complicate It
A selection or arrangement of facts can attract protection in its own right where the arrangement involved originality. Some jurisdictions additionally recognise database rights protecting substantial investment in compiling data, independent of any originality.
This is a genuine divergence between legal systems, and it means the answer to "may I copy this dataset" can differ by jurisdiction with the same facts.
3. What Fetching Does Not Settle
Being able to fetch a page says nothing about permitted use. Three separate questions follow the fetch:
| Question | Governed by |
|---|---|
| May I fetch it? | robots.txt, terms, access controls |
| May I store it? | Terms, and copyright for the copy itself |
| May I republish it? | Copyright |
| May I build a product from it? | Copyright, database rights, terms |
Collections frequently answer the first and never consider the rest.
4. Common Positions in Practice
- Extracting facts for internal analysis is the least contentious use.
- Republishing full text is the most clearly problematic.
- Substantial excerpts occupy a middle ground that depends on amount, purpose, and jurisdiction.
- Aggregating facts into a new dataset raises compilation and database-right questions.
- Training on collected content is actively contested and evolving.
5. Attribution Is Not a Licence
Crediting a source is good practice and does not by itself make copying permitted. Conversely, some uses are permitted without attribution.
The two are separate: attribution is an ethical and often contractual matter, permission is a legal one.
6. Exceptions Exist and Are Narrow
Most systems recognise exceptions — quotation, research, criticism, text and data mining in some jurisdictions — with conditions attached, and their scope differs substantially between countries.
Relying on an exception is a considered decision that benefits from advice, not a default assumption.
7. Practical Steps
- Distinguish facts you extract from expression you copy.
- Store the minimum expression the purpose requires.
- Keep provenance, so any item's source can be identified.
- Read the terms alongside
robots.txt. - Prefer APIs, whose terms usually state permitted use explicitly.
- Take advice before building a product on third-party content.
8. Summary
Copyright protects expression rather than facts, which is why extracting attributes and republishing text differ so much. Compilations and database rights add jurisdiction-dependent complexity.
Fetching settles none of it. Separate the questions of fetching, storing, republishing, and productising — and take advice where a use case approaches any of the contested boundaries.