How sirm00se/baseball-chatbot works
Baseball-Chatbot is a rules-focused RAG chatbot with a web interface and a documented older terminal client. The web workflow sends a question through FastAPI to the RAG engine, which rewrites it, retrieves rule chunks using embeddings and FAISS, fetches their metadata from SQLite, and asks Groq to generate an answer. Separate corpus-preparation paths scrape MLB glossary pages or chunk local rules documents, then build vector and SQL data. The excerpts show the web and retrieval links; some preparation details are supported by the README and source index rather than inspected call sites.
Main entry points: rag_engine.py RAG engine · rebuildVectorDatabase.py Database rebuild · vectordatabase.py Vector builder · AddBaseballRules.py Rules chunker
Components 25
Web interface
- Chat UI
static/chat.js - Chat page
static/index.html - FastAPI API
src/main.py
Answer retrieval
- RAG engine
src/rag_engine.py - Question rewrite
src/rag_engine.py - Query embedding
src/rag_engine.py - Vector search
src/rag_engine.py - FAISS index
databases/vector_index.faiss - Chunk lookup
src/rag_engine.py - Rules database
databases/baseball_vectors.db - RAG prompt
src/rag_engine.py
Rule corpus
- Glossary scraper
src/dataScraper.py - Rules chunker
src/AddBaseballRules.py - Rules documents
data/BaseballRules.txt - Terms document
data/Definition of Terms.txt - Rules CSV
data/baseballrules.csv - Vector builder
src/vectordatabase.py - Vector metadata
data/vector_metadata.csv - SQL conversion
src/SQLConversion.py - Database rebuild
src/rebuildVectorDatabase.py
Terminal client
- Terminal question loop
src/answerQuestion.py
Other components
- Chat user
- Groq LLM
- MLB glossary
- Ollama model
Connections 29
- Chat user → Chat UI: enters question
static/chat.js - Chat UI → FastAPI API: posts question
static/chat.js - FastAPI API → Chat page: serves page
src/main.py - FastAPI API → RAG engine: calls
src/main.py - RAG engine → Question rewrite: rewrites
src/rag_engine.py - RAG engine → Query embedding: embeds query
src/rag_engine.py - RAG engine → Vector search: searches vectors
src/rag_engine.py - Vector search → FAISS index: searches
src/rag_engine.py - RAG engine → Chunk lookup: looks up chunks
src/rag_engine.py - Chunk lookup → Rules database: queries
src/rag_engine.py - RAG engine → RAG prompt: builds prompt
src/rag_engine.py - RAG engine → Groq LLM: requests completion
src/rag_engine.py - Glossary scraper → MLB glossary: fetches glossary
src/dataScraper.py - Glossary scraper → Rules CSV: writes chunks
src/dataScraper.py - Rules documents → Rules chunker: reads rules
src/AddBaseballRules.py - Terms document → Rules chunker: reads terms
src/AddBaseballRules.py - Rules chunker → Rules CSV: appends chunks
src/AddBaseballRules.py - Vector builder → Rules CSV: loads chunks
src/vectordatabase.py - Vector builder → FAISS index: writes index
src/vectordatabase.py - Vector builder → Vector metadata: writes metadata
src/vectordatabase.py - SQL conversion → Vector metadata: loads metadata
src/SQLConversion.py - SQL conversion → Rules database: writes rules table
src/SQLConversion.py - Database rebuild → Glossary scraper: runs
src/rebuildVectorDatabase.py - Database rebuild → Rules chunker: runs
src/rebuildVectorDatabase.py - Database rebuild → Vector builder: runs
src/rebuildVectorDatabase.py - Database rebuild → SQL conversion: runs
src/rebuildVectorDatabase.py - Terminal question loop → Ollama model: requests answers
src/answerQuestion.py - Terminal question loop → FAISS index: searches
src/answerQuestion.py - Terminal question loop → Rules database: queries chunks
src/answerQuestion.py