Compare two versions.
Paste or load the before and after content to establish the exact differences.
DiffFind FEATURES
Semantic Diff for Meaningful Document and Text Comparison
A traditional diff shows which words changed. A semantic diff helps you understand what those changes mean. DiffFind first runs a deterministic lexical comparison to identify the exact changes, then offers optional AI semantic analysis to explain what changed and why it matters.
How it works
01 02 Add semantic analysis.
Use optional AI analysis to explore the meaning and impact of the changes.
03 Review the source evidence.
Read the explanation alongside the original text and the deterministic diff.
Exact changes first, meaning second
- Lexical comparison identifies the exact changes.
- Semantic analysis helps explain what those changes mean.
- Deterministic comparison remains the source of truth.
- AI is optional and supplements the exact comparison.
- Semantic analysis can point out a change in meaning even when the wording differs.
- You can inspect the underlying differences instead of receiving an unsupported AI conclusion.
Example: one revision, two views
An illustrative example of a revised contract clause.
Original text
Either party may terminate this agreement with 30 days’ written notice.
Revised text
The customer may end this agreement with 30 days’ written notice; the supplier may terminate only for cause.
Exact lexical changes
- “Either party” changed to “The customer”.
- “terminate” changed to “end” for the customer.
- The clause “the supplier may terminate only for cause” was added.
Semantic interpretation
Termination is no longer mutual: the supplier has lost the right to end the agreement without cause. This is a change in rights, not just a rewording.
Why both views are useful
The lexical view shows exactly where the text changed and cannot be argued with. The semantic view tells you which of those changes deserve attention. Like any AI output, the interpretation should be checked against the exact changes.
More examples
Illustrative examples of the feature.
Different wording, essentially the same meaning
BEFORE
The service retries the request three times before failing.
AFTER
Before returning an error, the service makes three retry attempts.
The wording changed substantially, but both versions describe three retries before failure.
One character, a significant behavior change
BEFORE
Requests timeout after 30 seconds.
AFTER
Requests timeout after 3 seconds.
Removing the “0” changes the timeout from 30 seconds to 3 seconds: one character changes how long a request may wait.
Semantic diff compared with a basic text diff
- A basic diff marks changed lines or words. It cannot tell a harmless rewrite from a change in obligation.
- Different wording that means the same thing still shows up as a change in a text diff. Semantic analysis can note that the meaning is unchanged.
- One small edit, such as a deleted digit, can change behavior. Semantic analysis can flag it while the lexical diff pinpoints it.
Semantic diff across formats
Start with exact changes in Text Compare, then add semantic analysis. The same approach applies to Word documents and PDFs, and to structured data such as API responses: compare two URLs, compare cURL responses or compare XML files. To find an idea inside a document rather than compare two versions, use Semantic Search.
Frequently asked questions
What is a semantic diff?
A semantic diff goes beyond showing which words changed and helps explain what the changes mean. In DiffFind it is an optional AI analysis layered on top of an exact lexical diff.
How is semantic comparison different from a text diff?
A text diff shows exact additions, removals and changes. Semantic comparison helps you judge whether those changes affect meaning, including when the wording differs but the meaning is the same.
Does semantic analysis replace lexical comparison?
No. The deterministic lexical comparison remains the source of truth. Semantic analysis supplements it, and you can always inspect the underlying differences.
Is AI required to use DiffFind?
No. The exact comparison works without AI. Semantic analysis is optional.
Can the AI be wrong?
Yes. AI explanations can be incomplete or mistaken, so review them against the exact differences and the source text when a detail matters.
Prefer to compare on your computer?
DiffFind Desktop for macOS and Windows runs locally and is free to use. See the download options.
TRY IT WITH YOUR OWN TEXT