text similarity engine
standard keyword searches miss paraphrased concepts. i built this engine to detect deep semantic overlap and paraphrased text.
- year
- 2025
- role
- creator
- stack
- typescript, python, transformer embeddings
combines a multi-stage analysis pipeline: fast exact string matching, n-gram shingle overlap, and transformer embeddings for deep sentence meaning.
displays side-by-side span highlights, allowing reviewers to visually inspect overlapping passages instead of trusting an opaque score.
built for academic integrity, research, and content analysis with open-source transparency from day one.
links
