NKU-AI-Assistant

The NKU AI Assistant is your go-to helper for campus life at Nankai University. Whether you have academic questions or need assistance with daily tasks, the assistant is here to offer you considerate help, much like a friendly senior student.

Running Locally

This repository contains the backend part of the NKU-AI-Assistant project, a FastAPI project. If you need to run the complete NKU-Smart-Assistant project locally, you must run both the frontend and backend projects. The method for running the frontend project is detailed in NKU-Smart-Assistant. Below, we introduce how to run the backend locally.

1. Clone the repository

git clone
cd NKU-Smart-Assistant-API

2. Download data and model files
   Please download the files1_faiss_index folder, the files_faiss_index folder, and the models from the Google Drive link. Ensure that the project structure matches the following:

nku-smart-assistant-api
├─ .env
├─ chat
├─ edgedriver
├─ file
├─ files1_faiss_index
│  ├─ index.faiss
│  └─ index.pkl
├─ files_faiss_index
│  ├─ index.faiss
│  └─ index.pkl
├─ models
│  └─ m3e-base-huggingface
│     ├─ 1_Pooling
│     │  └─ config.json
│     ├─ config.json
│     ├─ gitattributes
│     ├─ model.safetensors
│     ├─ modules.json
│     ├─ pytorch_model.bin
│     ├─ README.md
│     ├─ sentence_bert_config.json
│     ├─ special_tokens_map.json
│     ├─ tokenizer.json
│     ├─ tokenizer_config.json
│     └─ vocab.txt
├─ NewsGet
├─ README.md
├─ requirements.txt
├─ server.py
├─ static
├─ videoinfo
├─ webscrap

3. Configure the .env file

# Neo4j Database 
NEO4J_URI=""
NEO4J_USERNAME=""
NEO4J_PASSWORD=""

# OpenAI
OPENAI_API_KEY=""
OPENAI_API_BASE=""

# Tavily
TAVILY_API_KEY=""

# LangSmith 
LANGCHAIN_TRACING_V2=true
LANGCHAIN_ENDPOINT=""
LANGCHAIN_API_KEY=""

# GROQ
GROQ_API_KEY=""

4. Set up the Python environment

conda create -n nku-smart-assistant-api python==3.9.13
pip install -r requirements.txt

5. Run the server.py file

uvicorn server:app --host 0.0.0.0 --port 8000

You can debug the FastAPI endpoints on the /docs port. Since I’ve already deployed the backend project to a cloud server, if you need to run it locally, make sure to modify the frontend by replacing the request URL in the submitUserMessage function in lib/chat/action.ts.