Paste two texts to see how similar they are. You get a similarity percentage from three standard measures, cosine (word frequency), Jaccard (shared vocabulary) and Levenshtein (edit distance), and every phrase the texts share is highlighted side by side.
Last updated:
0%
Overall similarity—Average of the three measures
Cosine (word TF)0%Word frequency vectors
Jaccard (vocabulary)0%Shared ÷ all distinct words
Levenshtein0%Edit distance
Shared phrases highlighted
A Text A
B Text B
Shared phrases
Advertisement
Advertisement
How to use the Text Similarity Checker
Paste the first text into Text A and the second text into Text B.
Choose options: ignore case, ignore punctuation, skip stop words and the minimum shared phrase length.
Press Check similarity (results also update as you type).
Read the scores: overall similarity, cosine, Jaccard and Levenshtein, each from 0% to 100%.
Review the highlights: shared phrases are marked in both texts and listed below, longest first.
The three similarity formulas
Measure
Formula
Looks at
Order matters?
Cosine similarity
(A · B) ÷ (‖A‖ × ‖B‖), where A and B are word count vectors
Which words are used and how often
No
Jaccard similarity
|A ∩ B| ÷ |A ∪ B| on the sets of distinct words
Shared vocabulary only
No
Levenshtein similarity
1 − distance ÷ length of the longer text
Character insertions, deletions and substitutions
Yes
Worked example. Text A is “red apples and green apples” and text B is “green apples and red pears”. The word counts are A = {red 1, apples 2, and 1, green 1} and B = {green 1, apples 1, and 1, red 1, pears 1}. The dot product is 1 + 2 + 1 + 1 = 5, ‖A‖ = √7 and ‖B‖ = √5, so cosine = 5 ÷ √35 ≈ 84.5%. The vocabularies share 4 words out of 5 distinct words, so Jaccard = 80%.
How to read the results
High cosine, low Levenshtein: the same words in a different order, which is typical of a rearranged or lightly paraphrased text.
High everything: the texts are close copies with small edits. The highlighted phrases show exactly which passages match.
Low Jaccard but some shared phrases: mostly different texts that reuse a few exact sentences or quotes.
Short texts swing a lot: with only a few words, one changed word moves every score by several percent.
Shared phrases are found by matching runs of consecutive words (word n-grams) and extending each match as far as it goes, so a long copied sentence appears once rather than as many overlapping pieces. Phrase matching ignores punctuation and line breaks between words.
Frequently asked questions
How is the similarity percentage calculated?
The tool computes three standard measures and shows each one. Cosine similarity compares how often each word is used, Jaccard similarity compares the sets of distinct words, and Levenshtein similarity is 1 minus the edit distance divided by the length of the longer text. The overall score is the average of the three.
What is cosine similarity?
Each text is turned into a vector of word counts (term frequency). Cosine similarity is the dot product of the two vectors divided by the product of their lengths. It ranges from 0% (no words in common) to 100% (the same words in the same proportions), and it ignores word order.
What is the difference between Jaccard and cosine similarity?
Jaccard only looks at which distinct words appear: shared words divided by all distinct words in both texts. Cosine also weighs how often each word appears, so repeated keywords matter more. Jaccard is usually the lower and stricter of the two.
What does Levenshtein distance measure?
It is the minimum number of single-character insertions, deletions or substitutions needed to turn one text into the other. Unlike the word measures, it is sensitive to order and spelling. For long texts it is computed on words instead of characters to stay fast, and the tool tells you when that happens.
Is this a plagiarism checker?
No. It compares only the two texts you paste and does not search the web or any database. It is useful for spotting reused passages, checking a paraphrase against its source or comparing two versions of a document.
Why does removing stop words change the score?
Stop words such as the, and, of and to appear in almost every English text, so they make unrelated texts look somewhat similar. Removing them lets cosine and Jaccard focus on meaningful words. Shared phrases always keep stop words so the phrases read naturally.
Are my texts stored or uploaded?
No. All calculations run in your browser with JavaScript. Nothing is sent to a server, and the texts are only kept in your own browser's local storage so they are still there if you reload.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.