busola-sho/bilingual-semantic-book-recommender

A cross-lingual semantic search app built with Python, LangChain, Chroma, Hugging Face Transformers, and Gradio. Search English books using queries in Yoruba or English. # Bilingual Search Semantic Book Recommender A simple web app to search English-language books using queries in Yoruba (a low-resource language) or English. Built using Python, LangChain, Chroma, and Gradio, this project demonstrates a cross-lingual semantic search workflow. ## Demo Note: Space may take ~2-3 minutes to wake on first visit. *Search results display the book title, ISBN, description, and cover image.* Test it here: huggingface.co ## Features - Enter a query in **Yoruba or English**. - Optional Yoruba → English translation for seamless search. - Returns the top-`k` most relevant English books. - Displays **book title, ISBN, description, and cover image**. - Built on **vector embeddings** for semantic search. ## Installation 1. Clone the repository ```bash git clone github.com cd semantic-book-recommender ``` 2. Create a virtual environment ```bash python -m venv .venv source .venv/bin/activate # macOS/Linux .venv\Scripts\activate # Windows ``` 3. Install dependencies ```bash pip install -r requirements.txt ``` 4. Create a .env file with your OpenAI API and HUGGINGFACE HUB API key: ```bash OPENAI_API_KEY=****** HUGGINGFACEHUB_API_TOKEN=***** ``` ## Project Structure ``` book-recommender/ │ ├── app.py # main entry point to run the UI ├── requirements.txt ├── README.md │ ├── data/ │ └── books_cleaned.csv # cleaned book dataset │ ├── notebooks/ │ └── data_exploration.ipynb # dataset exploration & cleaning │ ├── internal/ # core modules │ ├── embeddings.py # build/load vector embeddings │ ├── search.py # vector similarity search │ ├── translate.py # Yoruba → English translation │ └── gradio_ui.py # defines the Gradio interface ``` ## Usage After following the steps above (including ensuring the virtual environment is activated), run: ```bash python3 app.p …