How dev-harsh01/capstone_project works
This metadata-first recommender turns a natural-language dataset request into structured requirements, fetches Kaggle metadata, retrieves likely candidates, then scores and explains the top five. Users can run that workflow through either the CLI or Streamlit; both expose the ranked results. The core stages are supported by excerpts and README descriptions. The graph omits a persistent store because discovery uses metadata only; Kaggle is the external metadata source.
Main entry points: cli.py CLI · streamlit_app.py Streamlit explorer · kaggle_client.py Metadata client · requirements_parser.py Requirement parser
Components 10
User interfaces
- CLI
rag_dataset_recommender/cli.py - Streamlit explorer
rag_dataset_recommender/streamlit_app.py
Recommendation workflow
- Requirement parser
rag_dataset_recommender/requirements_parser.py - Metadata client
rag_dataset_recommender/kaggle_client.py - Hybrid retrieval
rag_dataset_recommender/retrieval.py - Ranking and explanations
rag_dataset_recommender/recommender.py - Suitability scoring
rag_dataset_recommender/scoring.py
Domain models
- Dataset and request models
rag_dataset_recommender/models.py
External services
- Kaggle API
Other components
- Dataset requester
Connections 14
- Dataset requester → CLI: submits request
README.md - Dataset requester → Streamlit explorer: enters request
rag_dataset_recommender/streamlit_app.py - CLI → Requirement parser: parses request
rag_dataset_recommender/cli.py - Streamlit explorer → Requirement parser: parses request
rag_dataset_recommender/streamlit_app.py - CLI → Metadata client: fetches metadata
rag_dataset_recommender/cli.py - Streamlit explorer → Metadata client: fetches metadata
rag_dataset_recommender/streamlit_app.py - Metadata client → Kaggle API: lists datasets
rag_dataset_recommender/kaggle_client.py - CLI → Hybrid retrieval: retrieves candidates
rag_dataset_recommender/cli.py - Streamlit explorer → Hybrid retrieval: retrieves candidates
rag_dataset_recommender/streamlit_app.py - CLI → Ranking and explanations: ranks candidates
rag_dataset_recommender/cli.py - Streamlit explorer → Ranking and explanations: ranks candidates
rag_dataset_recommender/streamlit_app.py - Ranking and explanations → Suitability scoring: scores datasets
rag_dataset_recommender/recommender.py - Metadata client → Dataset and request models: creates records
rag_dataset_recommender/kaggle_client.py - Requirement parser → Dataset and request models: creates requirements
rag_dataset_recommender/requirements_parser.py