Advanced Similarity Checker
Compare two texts by lexical overlap and semantic meaning.
Loading…
About Advanced Similarity Checker
Two sentences can share almost no words and still say the same thing, which a plain word comparison misses. Paste two texts and the checker reports lexical similarity, based on shared three-word phrases, and semantic similarity, based on sentence embeddings, as separate figures before combining them. It also picks out the sentence from each text that scored closest and lists up to 12 matching phrases, so you can see what drives the numbers.
Features
- Separate Lexical Similarity and Semantic Similarity scores
- Combined Similarity weights meaning at 65% and word overlap at 35%
- Semantic score from the all-MiniLM-L6-v2 sentence-embedding model on our server
- Closest matching sentence pair with its own percentage
- Up to 12 matching phrases listed from the lexical comparison
Limitations
- Similarity is a signal, not proof of copying; the tool does not determine authorship and does not detect AI-generated text.
- Only the two texts you paste are compared, up to 50,000 characters each, and nothing is searched on the web.
- The closest sentence pair is chosen from the first 200 sentences of each text, split at full stops, question marks and exclamation marks.
- Both texts are sent to our server for the semantic score, and the result is held for a limited time and then cleared.
How it works
- Paste one text into Document A and the other into Document B.
- Select Compare to run the word-overlap and meaning comparisons.
- Read Lexical Similarity straight away; Semantic Similarity appears when the server returns its score.
- Check Combined Similarity and the bar beneath the figures for the overall picture.
- Look at the closest matching sentence pair and the matching phrases list to see what is behind the numbers.
Example input
Document A: "Our new app helps users track their daily water intake and stay hydrated."
Document B: "The application we built lets people monitor how much water they drink each day."Frequently asked questions
- Lexical similarity is the overlap of three-word phrases between the texts, so it rises only when wording is reused. Semantic similarity compares sentence embeddings, so a paraphrase can score high even with few words in common.
- Combined Similarity is 35% of the lexical score plus 65% of the semantic score, rounded to a whole number. It appears only after the server has returned the semantic score.
- It shows the one sentence from each text that has the highest embedding similarity, with that pair's own percentage. It is a quick way to find the passage pulling the overall score up.
- The Plagiarism Checker uses the same two scores but reports one Overall Similarity with highlights. The Document Similarity Checker measures word overlap only, with a 200,000 character limit, so it cannot see paraphrases.
- Not by itself. Two independent texts on the same topic often score high on meaning. Treat it as a reason to read the closest sentence pair and the matching phrases, and judge from those.
Related tools
- Plagiarism CheckerCompare two documents for potentially matching content.
- Document Similarity CheckerCompare two documents for overlapping passages.
- AI ParaphraserRephrase text at light, moderate, or strong intensity.
- Word CounterCount words, characters, sentences and get reading time.
- Compare PDFCompare two PDFs and see what changed.