Hello to all of you! I'm thrilled to present a project I created. Known as DocQuery, its purpose is to simplify the process of working with PDFs. I made DocQuery because I've always found navigating through PDF documents to be a little difficult. Thanks to some neat AI technology, you can use this tool to ask questions about your PDFs and receive prompt responses. DocQuery can be used by both professionals and students to identify important insights in reports or specialized information in study materials. Now let's get started and see how this project can completely change the way you use PDFs!

Inspiration

I developed DocQuery because I found organizing digital documents to be challenging. My goal is to make document management easier and provide users with instant access to insightful information. I'm transforming how people engage with their papers using AI and innovation. Come along with me as I open up new possibilities and change document interaction.

How This Works

1. Upload Document: Users start by uploading a PDF document to the system. The document is stored securely in an Amazon S3 bucket.

2. Indexing with OpenAI API: Upon upload, the system generates an index vector for the document using the OpenAI API. This index vector is a representation of the document's content and is crucial for accurate query responses.

3. Storage in Pinecone Vector Database: The index vector is stored in the Pinecone vector database, ensuring efficient retrieval and comparison during query processing.

4. User Query: Once the document is indexed, users can pose questions based on the content of the PDF. These queries are processed in real-time.

5. Query Processing: To answer the user's query, the system first identifies similarities between the query and the indexed documents stored in the Pinecone database. This narrowed-down set of documents serves as the context for the subsequent step.

6. AI Response Generation: The system utilizes a GPT model, taking into account the identified context (similar documents), previous conversations, and the original user query. The model generates a response tailored to the user's query and context.

7. Storage and Presentation: The generated response from the AI model is stored in the database for future reference and is promptly displayed to the user. This ensures a seamless and efficient interaction flow.

Features

• Real-time AI chatbot functionality for querying PDF documents.

• Utilization of advanced AI technologies for accurate and context-aware responses.

• Modern user interface for an enhanced user experience.

• Secure document storage using Amazon S3.

• Efficient indexing and retrieval using Pinecone vector database.

We welcome any feedback or suggestions for improving this project. Feel free to open an issue on GitHub with your thoughts and ideas.