Plagiarism Checker — Academic Text

Detect semantic similarity between academic documents using sentence embeddings and vector search. Unlike keyword-matching checkers, this catches paraphrased and restructured text.

How it works

1. Chunks suspect document into sentences/paragraphs
2. Embeds each chunk using all-MiniLM-L6-v2 (90MB, downloads on first run)
3. Searches a FAISS vector index of reference documents for nearest neighbors
4. Returns matches at three severity levels: HIGH, MODERATE, LOW
5. Computes overall originality percentage

Usage

Default thresholds

• HIGH: ≥ 0.85 cosine similarity (near-verbatim)
• MODERATE: ≥ 0.70 (suspicious paraphrase)
• LOW: ≥ 0.50 (topically related)