A system for intergroup prejudice detection: The case of microblogging under terrorist attacks

摘要

Intergroup prejudice is a distorted opinion held by one social group about another, without examination of facts. It is heightened during crises or threat. It finds expression in social media platforms when a group of people express anger, resentment and dissent towards another. This paper presents a system for automated detection of prejudiced messages from social media feeds. It uses a knowledge discovery framework that preprocesses data, generates theory-driven linguistic features along with other features engineered from textual content, annotates and models historical data to determine what drives detection of intergroup prejudice especially during a crisis. It is tested on tweets collected during the Boston Marathon bombing event. The system can be used to curb abuse and harassment by timely detection and reporting of intergroup prejudice.