Harnessing Social Media to measure Traffic Safety Culture: A Theory of Planned Behavior approach

The challenges facing today’s transportation instigate proactivity. Fostering a cohesive and integrated road safety ecosystem ensures safer road environments that in turn promote safer behaviors. In this regard, psychology theories and data analysis are powerful instruments in understanding traffic safety cultures and guiding targeted road safety initiatives. Studies backed up with theoretical bases such as the theory of planned behavior help in measuring traffic safety cultures through beliefs identification, allowing better support for the design of these initiatives. However, traditional methods of carrying out such studies often rely on surveys and questionnaires which can be cumbersome. As a result, in this work, we investigate a more efficient alternative. We present a general framework to automate the extraction and monitoring of these beliefs from naturally occurring discussions in social media and news platforms. Using text mining and natural language processing, we attempted to discern insights, beliefs, and opinions regarding traffic safety culture. We present an exploratory comparative analysis that combines sentiment analysis, topic classification, and beliefs classification. Our best performing models achieved F1-scores of 0.80 for sentiment classification, 0.61 for topic classification, and 0.72 for belief classification. The results and interpretations obtained throughout this study can be leveraged in building behavior paradigms to predict and influence the way road users act by pruning traffic safety cultures through implementing the right ideologies.