Understanding the road rage behavior and implications: A textual approach using legal cases in Korea

Abstract

This study aims at exploring the road rage behavior using extreme legal cases of Korea. This study uses a topic modeling algorithm with a text embedding technique to classify the road rage by characteristics (i.e., type, behavior, damage, punishment, and context). We collected 39 representative cases of the road rage from precedent data in Korea. We used nominalizing keywords and a latent Dirichlet allocation (LDA) algorithm of topic modeling among text mining techniques to analyze each case’s criminal facts and punishment. After preprocessing by indexing with a coherence score, we set the appropriate number of topics. Through the study, it was confirmed that the model has the highest explanatory power with five topics. We further analyze the five topics for occurrence and subsequent road rage, method of realizing road rage, psychological state of the offenders during road rage, method of collecting evidence of road rage, and type and extent of the damage. By categorizing the keywords, we derive the road rage cases by type. Through this study, precedents dealing with criminal cases of road rage were analyzed to derive the types and causes of retaliatory driving. This result is expected to be used as basic data for the enactment of road rage policies and laws in the future.