Text guides

Check Word Overlap and Frequency in Two Drafts

·

Use both scores when you need to know whether two drafts share the same vocabulary and whether the drafts emphasize the same words. Compare two small ingredient notes first so you can verify every count.

Text Similarity Checker

Count shared and exclusive words in the tea blend notes

Select Load tea blend example and Compare texts. Text A reads “Tea tea spice.” Text B reads “Tea spice spice mint.” With case folded, both texts contain tea and spice. Mint appears only in Text B.

The Jaccard score is 66.67% because two of the three distinct words are shared. Cosine similarity is 73.03% because the dot product is 4 and the vector lengths are the square roots of 5 and 6.

Count shared and exclusive words in the tea blend notes
WordText A countText B count
mint01
spice12
tea21

Paul Jaccard’s original account of set overlapNormalized dot product in the scikit-learn reference

Change a repeated word and compare the two scores

Change Text B to “Tea tea spice mint.” Compare again. The word set stays the same, so Jaccard remains 66.67%. The new frequency mixture matches Text A more closely and cosine rises to 91.29%.

Reorder the words in Text B without changing their counts. Both scores stay the same. Use Text Compare if the job is to find the line or position that moved.

Paul Jaccard’s original account of set overlapNormalized dot product in the scikit-learn reference

Inspect accents and open the full report

Compare “Café checks.” with “Cafe checks.” Only checks appears in both. An accent adds a distinct term. A precomposed é and an e followed by a combining acute accent still match after NFC normalization.

Filter Only in Text B to check exclusive terms. Download JSON report for the full vocabulary table, including terms outside the visible page and current filter. The a and b fields give frequencies, and cosine and jaccard are stored as proportions.

Unicode normalization formsECMAScript word segmentation

Common questions

Why is cosine unavailable when one text is empty?

An empty vocabulary vector has length zero. The cosine formula requires a positive length on each side.

How can two opposite statements get the same score?

The scores use vocabulary and frequency. Word order can reverse meaning while leaving both scores at 100%. Review the source statements for their meaning.

Sources

Put it into practice.

Related guides