Machine Learning Approaches for Classifying the Peace-War Orientations of Global News Organizations’ Social Media Posts
No Thumbnail Available
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Description
The study used the existing conceptualizations of peace and war journalism to
create supervised machine learning text classifiers trained and tested to
identify the war or peace orientations of news stories posted on social media.
Peace-oriented journalists promote peace initiatives, ignore differences, and
promote conflict resolution. In contrast, war-oriented journalists promote
differences between opposing parties and instigate violence as means to
resolving conflicts. Using Naïve Bayes, Logistic Regression, Decision Trees,
Random Forests, and Support Vector Machines (SVM), the study trained and
tested five computational models to detect the peace or war orientations of the
news posted on social media. The results indicate that Random Forest has
the highest predictive accuracy for predicting war or peace orientations of
online news stories. Naïve Bayes ranked the least accurate algorithm for
predicting peace or war orientations of online news stories.
Keywords
H Social Sciences (General), QA75 Electronic computers. Computer science