Multilingual News Framing Analysis

The repository contains the demo code for news framing analysis based on ChatGPT and LangChain.

The code applies the following tasks:

1- Idenitifies the article's language. This step is important for handling multilingualism.

2- It identifies the check-worthy entities that shapes the news narrative.

3- Based on detected entities, it identifies the article's stance towards entities and selects sentences from the article as evidences of its rationales.

The steps for running demo:

• Create a virtual environment

virtualenv -p python3.10 venv

• Install requirements

pip install -r requirements.txt

• Create a .env file

OPENAI_API_KEY=API_KEY
CHATGPT_MODEL=gpt-3.5-turbo-1106
OPENAI_ORG=ORG_KEY

• Run the Stremlit demo

streamlit run app.py

Acknowledgement

The work is supported by KID2.