← Projects
Transform long texts into concise summaries using advanced AI algorithms
Input Text
Summary
Your summary will appear here...
Generating summary...
0
Original Words
0
Summary Words
0%
Compression Ratio
0s
Time Taken
README.md
Transform long texts into concise summaries using four different AI algorithms -- both extractive and abstractive methods. Paste text or upload a file, pick a method, and compare results with compression metrics.
Summarization Methods
| Method | Type | How It Works |
|---|---|---|
| TextRank | Extractive | Graph-based sentence ranking (similar to PageRank) |
| LSA | Extractive | Latent Semantic Analysis via singular value decomposition |
| Luhn | Extractive | Keyword frequency scoring |
| BART | Abstractive | facebook/bart-large-cnn generates new summary sentences using beam search |
Extractive methods select and return existing sentences from the source. The abstractive method generates entirely new text.
Metrics
- Original and summary word counts
- Compression ratio (percentage reduction)
- Processing time
Tech Stack
- Flask -- API backend (
POST /api/summarize) - Hugging Face Transformers -- BART model for abstractive summarization
- Sumy -- LSA, TextRank, and Luhn extractive algorithms
- NLTK -- sentence tokenization
- PyTorch -- model inference runtime
API
| Method | Path | Description |
|---|---|---|
| GET | /text-summary/ | Main page |
| POST | /text-summary/api/summarize | Summarize text |