doc-chatbot 🤖

An AI chatbot featuring conversational memory, designed to enable users to discuss their CSV, PDF, DOCX and TXT data in a more intuitive manner. plus, you can put a dictionary as a fixed local knowledge base, and ... this chatbot will refuse to admit mistakes!📄

based on Robby the Robot from Forbidden Planet

By integrating the strengths of Langchain and OpenAI, Robby employs large language models to provide users with seamless, context-aware natural language interactions for a better understanding of their data.🧠

For better understanding, see my medium article 🖖 : Build a chat-bot over your CSV data

Quick Start 🚀

Based on Robby chatbot, more features added:

• multiple files supports, you can choose which file you want to ask about
• word files (.docx) supports
• local dictionary supports, if user query like "what is X", a local dict file will be used in the first place to answer directly, this can avoid the hallucination of AI and saves tokens used by context docs.
• Custom moderation support, which will analyze attitudes and topics that help keep the conversation going in the way you specify (main reason: I don't want chatbots to admit mistakes when users blame :)
• Some detail tweaks: token over limit bug fix, random spin text displayed while waiting for chatgpt response, custom chunk_size and block separator for text splitter, custom top_k for retriever, custom prompt with or without keywords, block score display, etc...

removed feature:

• CSV agent, because I don't need it, you can find it in the original project: https://github.com/yvann-hub/Robby-chatbot

TO-DO :

• enable print tokens utilizations for the conversation
• Add free models like vicuna and free embeddings
• Replace chain of the chatbot by a custom agent for handling more features | memory + vectorstore + custom prompt

Running Locally 💻

Follow these steps to set up and run the service locally :

Prerequisites

• Python 3.8 or higher
• Git

Installation

Clone the repository :

git clone https://github.com/chinesewebman/doc-chatbot.git

Navigate to the project directory :

cd doc-chatbot

Create a virtual environment :

python -m venv .venv
.\.venv\Scripts\activate

Install the required dependencies in the virtual environment :

pip install -r requirements.txt

Launch the chat service locally :

streamlit run src/Home.py

That's it! The service is now up and running locally. 🤗

local dict construction

the orignal dict file should be a .txt file, each phrase begin with a name following by a : and the explaination of it follows up, can cross multiple lines, ended in ### use new-dict.py to create or merge the dict file, expamples are in the comments. during the merge process, if there is conflicts between 2 dict files, the content comes from the first one will be chosen. at last, the file "user-dict.json" will be used during query, so rename your dict file to this name.

Contributing 🙌

Contributions are always welcome! If you want to contribute to this project, please open an issue, submit a pull request or contact the creator of Robby at barbot.yvann@gmail.com (: or me: chinesewebman@163.com