Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/gauravooo2/text-to-sql-llm-ai-chatbot


https://github.com/gauravooo2/text-to-sql-llm-ai-chatbot

Last synced: 8 days ago
JSON representation

Awesome Lists containing this project

README 

        

# 🧠 AI Chatbot with Gemini LLM, LangChain, SQL, and Streamlit



This project is an AI-powered chatbot application built using **Google's Gemini LLM**, **LangChain**, **ChromaDB**, and **Streamlit**. The chatbot is specifically designed to retrieve answers based solely on a local SQL database of students' records. This ensures that the bot doesn't generate random or hallucinated answers but provides SQL query-based responses strictly from the dataset.



## 📝 Project Overview



This chatbot is a data retrieval tool that leverages **Google Gemini's Language Model (LLM)**, allowing it to convert natural language questions into SQL queries. The queries are then run against a local SQLite database, and the results are returned to the user via a simple and intuitive **Streamlit** UI. **LangChain** is used as part of the backend to help integrate the LLM and the SQL query generation process.



The core idea behind this project is to have a reliable, data-constrained chatbot that will **only answer based on the data available in the database**—it does not generate responses that are not tied to the underlying database. This makes it particularly useful for scenarios where data accuracy and consistency are paramount.



## 🌟 Key Features



1. **Gemini LLM Integration**: The app uses Google's Gemini LLM to process user inputs and convert them into SQL queries dynamically.

2. **SQL Query Generation**: The bot is capable of transforming user questions into precise SQL queries that retrieve the necessary information from the `STUDENT` table in an SQLite database.

3. **Data Accuracy**: The chatbot restricts its answers to the information available in the `student.db` SQLite database, preventing random responses.

4. **Streamlit UI**: A user-friendly interface built with **Streamlit** allows users to input queries and get responses directly from the chatbot.

5. **LangChain**: Leveraged to build a seamless pipeline between the language model and the database query execution.

6. **SQLite Database**: The data is stored in a local `SQLite` database (`student.db`), containing information about students such as their name, class, section, and marks.

7. **Prompt-Based AI**: The chatbot uses a well-defined prompt to guide its behavior, ensuring SQL query generation adheres to the defined schema of the database.



## ⚙️ Technologies Used



- **Google Gemini LLM**: A language model that processes user queries and generates responses.

- **LangChain**: Used to build robust LLM-powered applications that connect the language model to the database.

- **ChromaDB**: An optional component that helps manage and store embeddings or vectorized data (currently not fully utilized in this version but can be expanded).

- **SQLite**: A lightweight SQL database to store the student data.

- **Streamlit**: A Python framework to build interactive and user-friendly web apps.

- **Python**: Backend and core logic.



## 🛠️ Setup Instructions



To run this project locally or on a cloud platform like Render, follow the steps below.



### 1. Clone the Repository



```bash

git clone https://github.com/GauravOOO2/Text-To-SQL-LLM-AI-ChatBot.git

cd Text-To-SQL-LLM-AI-ChatBot

```



### 2. Install Dependencies



Make sure you have **Python 3.8+** installed. Then, install the required dependencies using `pip`:



```bash

pip install -r requirements.txt

```



### 3. Set Up Environment Variables



Create a `.env` file in the project directory and add your **Google API key** for Gemini LLM:



```bash

GOOGLE_API_KEY=your-google-api-key

```



### 4. Initialize the SQLite Database



You can either create the SQLite database (`student.db`) manually, or you can run the `sql.py` script to set up the database with sample student records.



To create the database with predefined student data:



```bash

python sql.py

```



### 5. Run the Streamlit Application



Run the Streamlit app locally using the following command:



```bash

streamlit run app.py

```



### 6. Deploy on Render



If deploying on Render, ensure you:



1. Set the start command as `streamlit run app.py`.

2. Include the `Procfile` in the root of your project with the following content:



```bash

web: streamlit run app.py

```



3. Configure the environment variable `GOOGLE_API_KEY` in the Render dashboard.



## 💻 Project Structure



```

.

├── app.py                # Main Streamlit app

├── sql.py                # Script to initialize SQLite database with sample student data

├── student.db            # SQLite database (optional, can be generated dynamically)

├── requirements.txt      # Project dependencies

├── .env                  # Environment variables (not included in repo)

├── Procfile              # Start command for Render deployment

└── README.md             # Project documentation (this file)

```



### File Descriptions



- **`app.py`**: Contains the main logic for the Streamlit web app, including user input handling, invoking the Gemini LLM, and displaying the SQL query results.

- **`sql.py`**: Initializes an SQLite database with student records, creating the table and populating it with data.

- **`student.db`**: The SQLite database where student data is stored. It contains the table `STUDENT` with columns for name, class, section, and marks.

- **`requirements.txt`**: Lists the Python packages required to run the project (Streamlit, google-generativeai, LangChain, etc.).



## 🔍 How It Works



1. **User Query Input**: The user enters a natural language question in the Streamlit app.

2. **LLM Processing**: The Gemini LLM takes the user question and converts it into an SQL query based on the pre-defined prompt.

3. **SQL Query Execution**: The SQL query is executed against the local SQLite database (`student.db`).

4. **Response**: The chatbot displays the query result as the response to the user. The bot **only retrieves and responds with data present in the `STUDENT` table** and does not generate any other information.



### Example Queries:

- **User**: *"Tell me all students studying in AI & ML class?"*

- **Bot**: *SELECT * FROM STUDENT WHERE CLASS='AI & ML';*

  

  **Response**: `Emily Davis, Michael Brown`



- **User**: *"How many students are there in total?"*

- **Bot**: *SELECT COUNT(*) FROM STUDENT;*

  

  **Response**: `10`



## 🚀 Future Enhancements



- **Expand Dataset**: Include more detailed data fields (e.g., grades, attendance, etc.) to make the chatbot more informative.

- **ChromaDB Integration**: Fully integrate ChromaDB for better vector search and embedding management, expanding the bot's ability to understand queries beyond simple SQL.

- **Enhanced UI**: Improve the frontend design with additional features like charts and visual data representations.

- **Error Handling**: Add more robust error handling for invalid or out-of-scope queries.



## 🧑‍💻 Contributing



Contributions are welcome! If you'd like to improve this project or add new features, feel free to open a pull request.



1. Fork the project.

2. Create a new feature branch.

3. Submit a pull request detailing your changes.