GyameraYaw/ai_10022200141

RAG chatbot — Ghana elections & 2025 budget. Custom retrieval, hybrid search, Streamlit UI. # Manual RAG Chatbot — Ghana Election & Budget Assistant - **Student:** Yaw Acheampong Ahenkora Gyamera - **Index:** 10022200141 - **Deployed:** rag-ghana-chatbot.streamlit… --- ## Overview A fully manual Retrieval-Augmented Generation (RAG) chatbot that answers questions grounded in two Ghanaian datasets: - **Ghana Presidential Election Results (1992–2020)** — regional-level CSV - **Ghana 2025 Budget Statement** — full 252-page PDF **No LangChain, LlamaIndex, or pre-built RAG frameworks were used.** Every component — chunking, embedding, vector storage, retrieval, and prompt construction — is implemented from scratch in Python. --- ## Data Sources **Ghana Presidential Election Results (1992–2020)** Regional vote totals for all major presidential candidates across eight general elections. Covers parties including NDC, NPP, PNC, and CPP at the regional granularity. Does not contain constituency-level breakdowns or 2024 results. **Ghana 2025 Budget Statement** The full 252-page budget statement released by the Ministry of Finance, covering macroeconomic targets, sector-by-sector expenditure allocations, revenue measures, and fiscal policy narratives for the 2025 financial year. --- ## Project Structure ``` . ├── src/ # Core RAG modules (chunker, embedder, retriever, pipeline…) ├── scripts/ # Build index + experiment runners ├── experiments/ # Logged outputs from all experiments (Parts A–G) ├── docs/ # Architecture, design decisions, report ├── diagrams/ # Architecture diagram (PNG + drawio) ├── app/ # Streamlit UI └── data/ ├── raw/ # Source CSV + PDF (not committed) └── processed/ # Cached embeddings + chunk index (committed) ``` --- ## Setup ```bash # 1. Create and activate a virtual environment python -m venv .venv source .venv/bin/activate # Windows: .venv\Scripts\activate # 2. Install dependencies pip install -r requirements.txt # 3. Set your OpenAI API k …