Deep Search for Academic Research 🕵️

The Deep Search Academic project is designed to streamline the research process, providing users with a powerful tool to explore academic literature efficiently. This application acts as a personal research assistant, enabling users to delve into new topics, extract relevant papers, and pose questions without the time-consuming task of manual searching.

Key Features:

• Smart Search: Input a research topic along with optional academic domains such as arxiv.org or ieeexplore.ieee.org. The application then comprehensively searches the web for pertinent PDF documents.
• Automatic Downloading: The tool automatically finds and downloads the relevant papers, saving them to the local system for easy access.
• Text Extraction: It extracts text content from the downloaded PDFs, prepping it for further analysis by the AI model.
• Advanced Indexing with RAPTOR: Utilizing a sophisticated RAPTOR indexing method, the application creates a multi-level summary tree from the extracted text. This ensures a thorough understanding of academic papers, from granular details to overarching concepts, facilitating superior answers.
• Conversational QA: Once the index is established, users can interact through a chat interface (powered by Streamlit) to ask complex questions and receive synthesized responses based on the analyzed documents.
• Export Findings: Results from the Q&A session can be exported as a formatted PDF or a Mermaid diagram that visualizes the research process, parameters, and resources utilized.

Technology Stack:

• Backend Logic: Implemented using LangGraph to create a resilient pipeline.
• Indexing: Custom RAPTOR method for intelligent, multi-level information retrieval.
• AI Models: Supports local models via Ollama (including Llama 3) and cloud models via Google Gemini.
• Frontend: Developed with Streamlit for a responsive web interface.

Target Users:

This tool is particularly beneficial for:

• Students & Academics: Ideal for drafting literature reviews, enabling quick identification of key themes and streamlined argumentative development.
• Data Scientists & Engineers: Perfect for exploring new machine learning frameworks or algorithms, facilitating swift comprehension through access to essential literature.
• Curious Minds: An engaging way for anyone interested in subjects such as quantum computing or cellular biology to immerse themselves in research.

Feedback and contributions are encouraged as the project continues to evolve, aiming to simplify the research experience for all.