https://github.com/vsancnaj/ai-agent-document-analyzer
This project is an AI-powered document analysis bot designed to process and extract information from PDF documents.
https://github.com/vsancnaj/ai-agent-document-analyzer
ai autonomous llm ollama
Last synced: 4 months ago
JSON representation
This project is an AI-powered document analysis bot designed to process and extract information from PDF documents.
- Host: GitHub
- URL: https://github.com/vsancnaj/ai-agent-document-analyzer
- Owner: vsancnaj
- Created: 2024-08-23T15:46:51.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2025-01-19T22:14:50.000Z (5 months ago)
- Last Synced: 2025-01-31T06:51:18.310Z (5 months ago)
- Topics: ai, autonomous, llm, ollama
- Language: Python
- Homepage:
- Size: 1.4 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# **AI-Agent Document Analyzer**
This project is an **AI-powered document analysis bot** that processes and extracts information from PDF documents. It uses **LlamaIndex**, **LangChain**, and **Ollama** models to parse, analyze, and interact with document content. The bot intelligently handles user queries, retrieving only **relevant information** from the PDFs.
## **🚀 Why I Built This**
1. **To deepen my understanding of LLMs** by building a practical, hands-on project.
2. **To implement Retrieval-Augmented Generation (RAG)**, reducing hallucinations by feeding the model curated, factual data.
3. **To create a tool I wish I had during my thesis**—an AI that could compare and summarize research papers interactively.
This project is still **a work in progress**. I am continuously enhancing the bot by adding new tools and improving the **clarity and readability** of responses. Running a **local Ollama model** ensures **privacy** and eliminates API costs, allowing unrestricted experimentation.
---
## **✨ Features**
- **📄 PDF Parsing** – Extracts and interprets text from PDFs using `LlamaParse`.
- **🔍 Intelligent Querying** – Uses embeddings and vector search to **find only the most relevant** sections of documents.
- **📊 Vector Database** – Stores and retrieves information efficiently.
- **🤖 ReAct Agent** – Selects the appropriate tools to answer user queries dynamically.
- **💬 Interactive Chatbot** – Engages in real-time conversations about the documents.
- **🔗 Smart Link Extraction** – Extracts **only** the links relevant to the user’s query, not all links.
---
## **⚙️ Prerequisites**
Before running the project, ensure you’ve set up the environment:
### **1️⃣ Install Ollama (No API Key Needed)**
- Download and install **Ollama** (free & local).
- Verify Ollama is running:
```sh
ollama list
```
### **2️⃣ Get a LlamaParse API Key (Free)**
- **Sign up for a key**: [LlamaParse](https://llamaindex.ai/)
- **Add it to your `.env` file**:
```plaintext
LLAMA_PARSE_API_KEY=your_api_key
```
### **3️⃣ Use Local Embeddings (No API Key Required)**
- Ensure the embedding model is accessible:
```sh
local:BAAI/bge-m3
```
---
## **📥 Installation**
1. **Clone the Repository**:
```sh
git clone https://github.com/your-username/AI-Agent-Document-Analyzer.git
cd AI-Agent-Document-Analyzer
```
2. **Install Dependencies**:
```sh
pip install -r requirements.txt
```
3. **Set Up Environment Variables**:
- Create a `.env` file.
- Add your **LlamaParse API key**.
4. **Prepare Data**:
- Place your **PDF documents** inside the `data` directory.
---
## **▶️ Usage**
Start the document analysis bot:
```sh
python main.py
```
## **📝 Example Queries**
"Summarize this document."
"What are the key concepts discussed?"
"What is the objective of this research paper?"
## **🛠️ Code Overview**
| Component | Functionality |
|---------------------------|----------------------------------------------------------|
| LlamaParse | Parses PDFs into a structured format. |
| VectorStoreIndex | Converts document text into embeddings for search. |
| Query Engine | Finds and retrieves only the most relevant sections. |
| ReAct Agent | Dynamically decides which tools to use based on queries. |
| Intent Classifier | Determines if a query needs general info or links. |
| Contextual Link Finder | Extracts only the relevant links based on user queries. |
## **🛠️ Future Improvements**
- Improve response formatting for better clarity.
- Enhance multi-document support.
- Integrate better summarization techniques for long documents.