News Sentiment Analysis

This project analyses the evolution of sentiment in news headlines over time, utilising Python and Pandas for data manipulation, and Hugging Face Transformers models for sentiment analysis, emotion analysis, and keyword analysis. For more details, you can view the Project Proposal, Final Presentation, and Final Report.

Data Sources

The dataset comprises of news headlines spanning from 2007 to 2021, sourced from Kaggle:

• Australia: ABC News

• Ireland: The Irish Times

• India: Times of India

• England: Daily Mail

• United States: The New York Times

Usage

Step 1 - Filtering Data

1. Run python filterGBandUS.py to filter data specific to the United Kingdom and the United States.
2. Run python filterAll.py to filter data in the whole dataset.

Step 2 - Sentiment Analysis

1. Run python sentiment/analysis.py for sentiment analysis.
2. Run python sentiment/graph.py to generate visualisations based on the sentiment data.

Step 3 - Emotion Analysis

1. Run python emotion/analysis.py for emotion analysis.
2. Run python emotion/graph.py to visualise the emotion analysis results.

Step 4 - Keywords Analysis

1. Run python keywords/analysis.py, python keywords/analysis2.py, and python keywords/combine.py for keyword analysis.
2. Run python keywords/graph.py to visualise the keyword analysis results.