README.md

KB_AIchallenge

The repository for KB 6th Future Finance A.I. Challenge.

NOTE: This repository supports OpenAI API.

• Please your OpenAI API key into main.py file.

Proposal

1. We created a glossary of financial terms to make complex language easier to understand.
2. Generative AI and RAG to easily recreate financial terms and deliver personalized financial content.

1. Framework

• For more details, please see Description.pdf

2. Demonstration

Dataset Preparation

• Download and add the datasets into data folder to reproduce our experimental results.
• Data info :
  • 공공데이터포털 한국산업은행 금융관련 용어
  • 한국은행 경제금융용어 700선
  • 금융감독원 금융용어사전
  • Bank Marketing

Usage

1. Preprocessing

Data preparing:

• Directory: ./crawling/

python deposit_crawling.py

python foreign_deposit_crawling.py

python loan_crawling.py

python other_crawling.py

python trust_crawling.py

Data preprocess:

• Directory: ./dataset/

python preprocess.py  

python text_preprocess.py

2. QnA Task

Arguments

• --gpt_ver : ChatGPT version that you can use (default: gpt4)
• --question : Financial terms you may not know the definition of (ex. "기회비용의 정의가 뭐야?")

Using in Python

python main.py --gpt_ver <gpt_ver> --question <question> 

• If you want to use it on the web, contact @SongJSeop.

Directory and codes

.
+-- data
+-- assets
+-- crawling
|       +-- deposit_crawling.py
|       +-- foreign_deposit_crawling.py
|       +-- loan_crawling.py
|       +-- other_crawling.py
|       +-- trust_crawling.py
+-- dataset
|       +-- mydata_preprocess.py
|       +-- preprocess.py
|       +-- text_preprocess.py
+-- modules 
|       +-- Cosine_Similarity.py
|       +-- Embedding.py
|       +-- Openai_utils.py
|       +-- recommender.py
|       +-- utils.py
|       +-- Web_Research.py
+-- kb_backend
+-- kb_frontend
+-- main.py
+-- LICENSE
+-- README.md