🌍 Ecoinvent Search Agent

A AI-powered chat application that helps users search and explore the Ecoinvent database (only publicly available meta-data, so no license needed).

This tool combines Semantic Search (to find relevant data even if keywords don't match exactly) with a Large Language Model (Llama 3 via Groq) to provide natural language answers.

Or use Ecoagent.tech for the refined version.

✨ Features

• Natural Language Search: Ask questions like "What are the impacts of steel production?" instead of searching for exact codes.
• Semantic Matching: Uses SentenceTransformers (BERT) to find relevant database entries based on meaning, not just keywords.
• AI Synthesis: Uses the Groq API (Llama 3.3 model) to read the data and explain it to you in plain English.
• Interactive Chat: Built with Streamlit for a clean, chat-like interface.

🛠️ Prerequisites

Before running this project, ensure you have the following:

1. Python installed (Version 3.8 or higher recommended).
2. Groq API Key: You need an API key to run the LLM. Get one for free at console.groq.com.

🚀 Setup & Installation

Follow these steps to get the app running on your computer.

1. Clone or download the Repository

2. Create a Virtual Environment

It is best practice to keep dependencies isolated.

Windows:

python -m venv venv .\venv\Scripts\activate

Mac/Linux:

python3 -m venv venv source venv/bin/activate

3. Install Dependencies

Install the required Python packages listed in requirements.txt:

pip install -r requirements.txt

4. Configure API Keys

Streamlit handles secrets (like API keys) using a specific file.

Create a folder named .streamlit in your project root.

Inside that folder, create a file named secrets.toml.

Add your Groq API key to the file:

.streamlit/secrets.toml GROQ_API_KEY = "gsk_..." ⚠️ Important: Never commit your secrets.toml file to GitHub!

🏃‍♂️ How to Run

Once everything is installed and configured, run the app with:

streamlit run app.py A browser window should automatically open pointing to http://localhost:8501.

📂 Project Structure

app.py: The main application code containing the UI, the search logic, and the LLM integration.

data/: Directory to store the dataset.

.streamlit/secrets.toml: Configuration file for API keys (ignored by Git).

requirements.txt: List of Python libraries needed.

🧠 How it Works

Embedding: When the app starts, it loads the dataset and the BERT model.

User Query: When you ask a question (e.g., "electric car battery"), the app converts your text into numbers (vector embedding).

Vector Search: It compares your query numbers with the numbers in the database to find the most mathematically similar records.

LLM Response: The app sends the found data + your question to the Llama 3 model, which writes a human-readable answer.

🤝 Contributing

Pull requests are welcome