Text & writing
Lexical text similarity
Compare token overlap and frequency with three transparent scores.
A clear answer. Here’s how.
Tokenises lowercase Unicode letters/numbers with internal apostrophes. Reports set Jaccard, set Dice and term-frequency cosine as percentages. Each text is capped at 100,000 characters and must contain a word. Scores ignore word order, meaning, synonyms and punctuation; they are descriptive lexical overlap, not semantic similarity or a plagiarism verdict.
See a worked example
Worked example
A fixed demonstration, separate from the values you enter above.
- First text
- The small blue boat crossed the blue lake. A small boat waited by the shore.
- Second text
- A blue boat crossed the small lake.
Example result
- Set Jaccard (%)
- 70
- Set Dice (%)
- 82.3529411765
- Token-frequency cosine (%)
- 87.2871560944
- Shared distinct tokens
- 7
Lowercase Unicode letter/number tokens; word order, meaning, punctuation and synonyms are ignored. Scores do not measure semantic similarity or plagiarism.