Spaces:
Running
Running
|
Download README.md from Hugdrigo/researchlens: direct link, hf CLI and curl.
- Browser
- Download file 1.83 kB
-
https://huggingface.co/spaces/Hugdrigo/researchlens/resolve/main/README.md
- Command line
-
hf download hf://spaces/Hugdrigo/researchlens/README.md
-
curl -L -o README.md https://huggingface.co/spaces/Hugdrigo/researchlens/resolve/main/README.md
1.83 kB
| title: ResearchLens | |
| emoji: 🔎 | |
| colorFrom: blue | |
| colorTo: indigo | |
| sdk: static | |
| app_file: index.html | |
| pinned: false | |
| license: mit | |
| short_description: Multilingual document intelligence in the browser | |
| # ResearchLens | |
| A privacy-conscious, browser-first multilingual NLP application for academic, policy and business documents. | |
| This Static Space analyzes English, Spanish and Portuguese text without server-side compute or a generative API. It provides document metrics, an extractive summary, lexical signals and multilingual topic indicators. PDF extraction is performed in the browser with PDF.js. | |
| ## Why a Static Space? | |
| The portfolio uses scarce compute only where machine-learning inference materially adds value. ResearchLens relies on transparent analytical NLP, so moving it to a Static Space makes the project faster to share, cheaper to maintain and more privacy-friendly. | |
| ## Capabilities | |
| - paste text or upload PDF/TXT/Markdown | |
| - English, Spanish and Portuguese language detection | |
| - word, sentence and paragraph metrics | |
| - reading-time estimate | |
| - lexical diversity | |
| - extractive summarization | |
| - unigram and bigram lexical signals | |
| - multilingual topic scoring | |
| - fully client-side analysis | |
| ## Technical design | |
| - Vanilla JavaScript | |
| - deterministic NLP heuristics | |
| - PDF.js for browser-side PDF text extraction | |
| - no backend | |
| - no API keys | |
| - no external generative AI service | |
| ## Limitations | |
| - Scanned/image-only PDFs require OCR and may not extract text. | |
| - Language detection is heuristic. | |
| - Extractive summaries preserve source sentences but can miss nuance. | |
| - Topic signals are based on a curated lexicon rather than a learned classifier. | |
| - Results are analytical signals, not factual verification. | |
| ## License | |
| MIT for project code. PDF.js is loaded from its public distribution and remains subject to its own license. | |