🧠 Real-Time Intelligent Search Assistant

A high-performance Retrieval-Augmented Generation (RAG) chatbot capable of conducting real-time web research to answer user queries. Unlike static RAG systems, this project dynamically searches the web using the Serper Dev API, processes live content, and utilizes Groq's fast inference engine to provide up-to-date answers with source citations.

---

🚀 Key Features

• Real-Time Web Search: Integrates Serper Dev API to perform live Google searches for up-to-the-minute information (e.g., "Who won the 2024 T20 World Cup?").
• High-Speed Inference: Utilizes Groq's LPU (Language Processing Unit) for near-instantaneous LLM generation.
• Dynamic RAG Pipeline:
  • Search: Fetches relevant URLs based on the user query.
  • Scrape & Load: Extracts text content from search results using WebBaseLoader.
  • Chunking: Splits text into manageable segments using RecursiveCharacterTextSplitter.
  • Vectorization: Embeds text segments into a FAISS vector store using HuggingFace embeddings.
  • Retrieval: Finds the most relevant chunks to answer the question.
• Performance Monitoring: Logs detailed latency metrics for every stage (Search, Content Fetch, Embedding, LLM Inference).
• Retriever Evaluation: Includes a built-in evaluation module to benchmark the accuracy of the retrieval system against expected source URLs.

---

🛠️ Tech Stack

• Language: Python 3
• Orchestration: LangChain (RetrievalQA Chain)
• LLM Engine: Groq (via GROQ_API_KEY)
• Search Engine: Serper.dev (Google Search API via SERPER_API_KEY)
• Vector Store: FAISS (Facebook AI Similarity Search)
• Embeddings: HuggingFace (sentence-transformers)

---

⚙️ Installation & Setup

1. Clone the Repository

   git clone [https://github.com/yourusername/Real-Time-Question-Answering-System.git](https://github.com/yourusername/Real-Time-Question-Answering-System.git)
   cd Real-Time-Question-Answering-System

2. Install Dependencies Ensure you have the required Python libraries installed:

   pip install langchain langchain-groq faiss-cpu google-search-results sentence-transformers numpy pandas

3. Configure API Keys You need API keys for Groq (LLM) and Serper (Search). You can set them in your environment variables or directly in the notebook execution block:

   import os
   os.environ["SERPER_API_KEY"] = "your_serper_api_key"
   os.environ["GROQ_API_KEY"] = "your_groq_api_key"

---

📊 Performance Metrics

The system automatically logs execution time for each component of the RAG pipeline to help identify bottlenecks:

Component	Metric Captured
Search	Time taken to query Serper API and retrieve URLs.
Content Fetch	Time taken to scrape and load data from the retrieved URLs.
Text Split	Time taken to chunk the loaded text.
Embedding	Time taken to generate embeddings and build the FAISS index.
LLM Inference	Time taken by Groq to generate the final answer.

Example Log Output:

[Search] Time taken: 1.185s [Content Fetch] Time taken: 0.218s [LLM Inference] Time taken: 0.697s [Total Research Query] Total time: 2.968s

---

🧪 Evaluation

The project includes a Researcher class with an evaluate_retriever method. This allows you to benchmark the retrieval accuracy using a labeled test dataset.

Test Data Format:

TEST_DATA = [
    {
        "question": "Who won the 2024 T20 World Cup?",
        "expected_source": "[https://www.espncricinfo.com/](https://www.espncricinfo.com/)..."
    },
    {
        "question": "What are the features of the new iPhone 16?",
        "expected_source": "[https://www.apple.com/](https://www.apple.com/)..."
    }
]