https://github.com/swagath18/personalized-recommender
https://github.com/swagath18/personalized-recommender
faiss-vector-database gpt-4 gradio numpy openai python rag recommendation sentiment-classification transformer
Last synced: 6 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/swagath18/personalized-recommender
- Owner: Swagath18
- Created: 2025-05-15T22:18:23.000Z (9 months ago)
- Default Branch: main
- Last Pushed: 2025-05-15T23:30:27.000Z (9 months ago)
- Last Synced: 2025-07-25T03:47:34.379Z (7 months ago)
- Topics: faiss-vector-database, gpt-4, gradio, numpy, openai, python, rag, recommendation, sentiment-classification, transformer
- Language: Python
- Homepage:
- Size: 42 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# LLM-Powered Personalized Product Recommender
A context-aware recommendation system powered by **Retrieval-Augmented Generation (RAG)** and **GPT-4**, enhanced with **hybrid reranking** and a **feedback sentiment analysis system** using VADER.
## Workflow
## Features
- Semantic product search with FAISS + Sentence Transformers
- GPT-4 for natural language product recommendations
- Personalized reranking using user context + cosine similarity
- Multi-tab UI with Gradio (Recommendations + Feedback)
- Feedback logging with real-time sentiment analysis (VADER)
- Logged query, prompt, and model responses for auditability
## Stack
- Python
- SentenceTransformers (MiniLM-L6-v2)
- FAISS (vector search)
- OpenAI (GPT-4 API)
- VADER Sentiment Analysis
- Gradio (for UI)
- Pandas, Numpy, Scikit-learn
## Folder Structure
```
.
├── llmpoweredrag.py # Main app script
├── amazon_electronics_products.csv # Sample dataset
├── .env # API keys
├── recommendation_log.txt # GPT prompt/response logs
├── feedback_log.txt # Feedback + sentiment logs
└── README.md
```
## Setup Instructions
```bash
git clone https://github.com/yourusername/llm-personalized-recommender.git
cd llm-personalized-recommender
pip install -r requirements.txt
python llmragsenti.py
```
Make sure your `.env` file contains:
```
OPENAI_API_KEY=your-api-key-here
```
## Try These Test Cases
**Query:**
`best noise-cancelling headphones for Zoom calls`
**Context:**
```
prefers Bose or Sony
needs comfort for extended meetings
wants good microphone quality
```
Click "Recommend" and observe the GPT-4 output with ranked FAISS items.
Then test feedback like:
```
The suggestions made no sense. I didn’t find them helpful.
```
## Output Example
```text
Retrieved Products:
Sony WH-1000XM4 (Context Score: 0.874)
Bose QC35 II (Context Score: 0.832)
GPT-4 Recommendations:
1. Sony WH-1000XM4 - Superior noise cancellation...
2. Bose QC35 II - Excellent comfort and clarity...
```
## Sentiment Feedback Log
Each entry is saved with timestamp and label:
```
[2025-04-24 15:06:36] User Feedback: Loved it! | Sentiment: positive (0.89)
```
## License
MIT
## Future Improvements
- Add image-based product cards
- Deploy on Hugging Face Spaces or Streamlit Cloud
- Build dashboard with feedback insights