Social Media Mining for Understanding Traffic Safety Culture in Washington State Using Twitter Data

Abstract

Traffic safety culture has emerged as a significant factor in support of the recognition of existing traffic safety policies and programs and as a contextual variable to describe high-risk behaviors of drivers. However, it is an arduous task to understand people’s beliefs and attitudes that collectively make up traffic safety-related influences. The growing acceptance of social media platforms such as Facebook and Twitter have spurred great interest in massive data collection and its use in conducting a comprehensive analysis of people’s viewpoints on a particular topic. This study applied social media mining to shed light on traffic safety culture in the state of Washington. To this end, the researchers collected traffic safety–related tweets over the past 4 years in Washington based on a set of keywords. After cleaning and reprocessing, the collected tweets were used in sentiment analysis using Linguistic Inquiry and Word Count (LIWC) to measure the public’s beliefs and attitudes toward the importance of traffic safety, possibility of zero fatalities, usefulness of traffic law enforcement, and six types of high-risk behaviors, including impaired driving, speeding, distracted driving, unrestrained vehicle occupants, teenage drivers, and older drivers. Next, the topic modeling technique was applied to discover important latent topics related to traffic safety culture. This research, which capitalizes on social media mining, overcomes the limitations of the conventional survey method, which are time-consuming and costly. The generated information may facilitate understanding of the barriers to preventing fatal traffic accidents in Washington and around the country and developing solutions to overcome them.