hejer2004/car_chatbot

An intelligent AI-powered car chatbot that helps you choose the best car for your preferences. It understands English, French, and Tunisian Dialect. Built with Flask, Groq (Llama 3), and Scikit-Learn. # 🚗 Intelligent Automotive Chatbot Welcome to the **Intelligent Automotive Chatbot**, an advanced AI-powered assistant designed to understand your car questions in **English**, **French**, or **Tunisian Dialect**. Unlike basic bots, this system uses a **Large Language Model (LLM)** to grasp complex intents and an **Out-Of-Domain (OOD) Detector** to stay focused on its automotive expertise. --- ## 🏗️ Architecture & Flow (How it works?) The application follows a robust "Retrieve & Generate" architecture. Here is the exact journey of a user request: ### 1. 📥 Input & Language Processing - **File**: `app2.py` → `language_processor.py` - **Action**: The user speaks. The system detects the language. - **Magic**: If the user speaks in **Tunisian** (e.g., *"nheb karhba rkhissa"*), it is automatically translated to English (*"I want a cheap car"*) before being processed. ### 2. 🛡️ The Guardian (OOD Detection) - **File**: `ood_detector.py` - **Action**: Before using expensive resources, the system checks: *"Is this really about cars?"* - **Technology**: Uses **TF-IDF Vectorization** + **One-Class SVM**. - **Result**: - "Pizza recipe?" ❌ Blocked immediately. - "Price of a Toyota?" ✅ Passes to the Brain. ### 3. 🧠 The Brain (Intent Analysis) - **File**: `llm_analyzer.py` - **Technology**: **Groq API** (Llama-3.3-70B Model). - **Action**: The LLM analyzes the text to extract structured data (JSON). - **Example**: - *Input*: "A cheap BMW" - *Output*: `{"intent": "search", "entities": {"make": "bmw", "sort_order": "price_asc"}}` ### 4. 🔧 The Engine (Query Execution) - **File**: `smart_query_builder.py` - **Action**: Takes the structured JSON and builds a complex **Pandas** query to filter the `data/cars2.csv` file. - **Capabilities**: Handles price, size, transmission filters, and generates a detailed catalog-style display. --- ## 🚀 Installation & Launch 1. **Install dependencies**: ```bash pip install -r requirements.txt ``` 2. **Configure usage environment**: Create a …