Analysis of 4,000+ UK news articles to uncover disparities in how Palestinian and Israeli casualties are reported.
The project used Named Entity Recognition, sentiment scoring, dependency parsing, and casualty data to reveal structural issues in moral distance and narrative framing — offering an interactive, data-driven critique with a particular focus on the BBC.
Project outline
This project investigates disparities in empathy and moral framing in how UK media outlets report on Israeli and Palestinian casualties. Using a corpus of over 4,000 news articles, it combines Named Entity Recognition (NER), dependency parsing (via SpaCy), attribute-level sentiment analysis, and structured data on casualty numbers.
The analysis is informed by methodologies from the NLTK Handbook (NLP with Python) and state-of-the-art academic research on attribute-level sentiment. The pipeline blends automated parsing with human-in-the-loop coding to capture subtle framing effects not easily detected by off-the-shelf sentiment models.
Key features:
- Named Entity Recognition and dependency parsing to map entities and sentiment relationships
- Attribute-level sentiment analysis for fine-grained framing detection
- Human-in-the-loop validation of articles
- Geopolitical media framing analysis
- Awarded top marks (90%) by assessor
Media
