Before diving into the technical details of how chatting with documents works, let's discover what it is. There are many applications that provide such functionality, such as ChatGPT 4 and ChatwithPDF. These applications allow users to have interactive conversations with AI models while incorporating the additional context provided by documents.
Chatting with documents is an innovative approach that enhances the capabilities of conversational AI systems. By integrating documents into the conversation, users can provide specific information, references, or background knowledge to guide the AI's responses. This contextual information can be in the form of text, PDF files, web pages, or any other document format.
How chat with document works?
The chat with document feature works by turning a document into a chat or question-answering (QA) app. Here are the general steps involved:
- Splitting: The document is split into smaller sections of a specified size.
- Storage: The splits are stored, often with embeddings, in a storage system like a vectorstore.
- Retrieval: The app retrieves the relevant splits from storage, usually based on similarity to the input question.
- Output: An LLM (Language Model) generates an answer using a prompt that includes the question and the retrieved splits.
By following these steps, the chat with document feature allows users to interact with the document and get answers to their questions.
