Commit Β·
aaac9ac
1
Parent(s): 66c4727
Updated README
Browse files
README.md
CHANGED
|
@@ -19,8 +19,10 @@ Track how research concepts evolve over time using AI-powered semantic embedding
|
|
| 19 |
- π **Track concept evolution** through time windows
|
| 20 |
- π§ **Semantic embeddings** using sentence-transformers (MiniLM-L6-v2)
|
| 21 |
- π **Interactive timeline** visualization with Gradio
|
|
|
|
| 22 |
- π― **Kalman filter** constraints prevent unrealistic concept jumps
|
| 23 |
- πΎ **Local embeddings cache** - no API costs
|
|
|
|
| 24 |
- π **Works locally and on HuggingFace Spaces**
|
| 25 |
|
| 26 |
## Quick Start
|
|
@@ -103,9 +105,10 @@ Your app will be live at: `https://huggingface.co/spaces/YOUR-USERNAME/concept-t
|
|
| 103 |
|
| 104 |
### 3. View Results
|
| 105 |
|
| 106 |
-
- Timeline
|
| 107 |
-
-
|
| 108 |
-
- Statistics: high/moderate/low confidence counts
|
|
|
|
| 109 |
|
| 110 |
## How It Works
|
| 111 |
|
|
@@ -182,11 +185,12 @@ pytest tests/test_arxiv_client.py -v
|
|
| 182 |
concept_tracker/
|
| 183 |
βββ app.py # Gradio interface
|
| 184 |
βββ backend/
|
| 185 |
-
β βββ gradio_wrapper.py # Gradio event handlers
|
| 186 |
β βββ tracker.py # Main orchestrator
|
| 187 |
β βββ arxiv_client.py # ArXiv API wrapper
|
| 188 |
β βββ embedding_service.py # Embeddings + cache
|
| 189 |
β βββ kalman_tracker.py # Kalman filter logic
|
|
|
|
| 190 |
β βββ config.py # Configuration
|
| 191 |
β βββ models.py # Pydantic data models
|
| 192 |
β βββ utils/
|
|
@@ -254,7 +258,7 @@ concept_tracker/
|
|
| 254 |
- π² **Tree branching** with HDBSCAN clustering
|
| 255 |
- π² **Bidirectional tracking** (trace concepts to origins)
|
| 256 |
- π² **Multi-signal validation** (citations, author overlap)
|
| 257 |
-
-
|
| 258 |
|
| 259 |
## License
|
| 260 |
|
|
|
|
| 19 |
- π **Track concept evolution** through time windows
|
| 20 |
- π§ **Semantic embeddings** using sentence-transformers (MiniLM-L6-v2)
|
| 21 |
- π **Interactive timeline** visualization with Gradio
|
| 22 |
+
- πΊοΈ **t-SNE visualization** showing concept trajectory through embedding space
|
| 23 |
- π― **Kalman filter** constraints prevent unrealistic concept jumps
|
| 24 |
- πΎ **Local embeddings cache** - no API costs
|
| 25 |
+
- π₯ **Export results** to JSON or CSV
|
| 26 |
- π **Works locally and on HuggingFace Spaces**
|
| 27 |
|
| 28 |
## Quick Start
|
|
|
|
| 105 |
|
| 106 |
### 3. View Results
|
| 107 |
|
| 108 |
+
- **Timeline**: Concept evolution across time windows with expandable paper lists
|
| 109 |
+
- **Visualization**: t-SNE plot showing concept trajectory through embedding space
|
| 110 |
+
- **Statistics**: Total papers, time steps, and high/moderate/low confidence counts
|
| 111 |
+
- **Export**: Download results as JSON (full data) or CSV (flattened paper records)
|
| 112 |
|
| 113 |
## How It Works
|
| 114 |
|
|
|
|
| 185 |
concept_tracker/
|
| 186 |
βββ app.py # Gradio interface
|
| 187 |
βββ backend/
|
| 188 |
+
β βββ gradio_wrapper.py # Gradio event handlers + export
|
| 189 |
β βββ tracker.py # Main orchestrator
|
| 190 |
β βββ arxiv_client.py # ArXiv API wrapper
|
| 191 |
β βββ embedding_service.py # Embeddings + cache
|
| 192 |
β βββ kalman_tracker.py # Kalman filter logic
|
| 193 |
+
β βββ visualization.py # t-SNE visualization
|
| 194 |
β βββ config.py # Configuration
|
| 195 |
β βββ models.py # Pydantic data models
|
| 196 |
β βββ utils/
|
|
|
|
| 258 |
- π² **Tree branching** with HDBSCAN clustering
|
| 259 |
- π² **Bidirectional tracking** (trace concepts to origins)
|
| 260 |
- π² **Multi-signal validation** (citations, author overlap)
|
| 261 |
+
- β
**Export results** to JSON/CSV
|
| 262 |
|
| 263 |
## License
|
| 264 |
|