Inspectra is a powerful web-based tool designed to detect plagiarism, paraphrased content, and semantic similarity between documents. It uses advanced NLP models like Sentence-BERT, BLEU, ROUGE, and keyword-based web scraping to identify both direct and paraphrased overlaps — going beyond traditional copy-paste detection.
- 🔍 Semantic Similarity using Sentence-BERT (
paraphrase-MiniLM-L6-v2) - 🧠 BLEU & ROUGE metrics for sentence-level evaluation
- 📊 Plagiarism Percentage Estimation based on similarity matrices
- 🌐 Web Scraping Detection – checks if the content exists online using:
- KeyBERT for keyword extraction
- Google + Wikipedia search scraping
- Cosine similarity with scraped web content
- ✨ Highlighting of matching/paraphrased sentences across documents
- 📎 Supports
.txtand.pdffiles
| Layer | Tool/Library |
|---|---|
| Backend | Python, Sentence Transformers, NLTK, ROUGE |
| Web Scraping | BeautifulSoup, Requests, KeyBERT |
| Similarity Engine | Cosine Similarity, ROUGE-L |
| Frontend | Streamlit |
| Preprocessing | Numpy, Regex, TF-IDF |