Real-Time Transportation Social Media Analytics Using Pulse (Pulse-T)
Real-Time Transportation Social Media Analytics Using Pulse (Pulse-T)
Project Duration: 24 months
Project Budget (Federal UTC Funds): $60,000
Project Budget (Cost-share): $30,000
Institution: Arizona State University
Abstract
As city planners and transportation system planners consider changes and upgrades to transportation systems and infrastructure, they require models that accurately reflect communities’ needs. Planners need access to advanced activity-travel demand analysis models that are responsive and sensitive to emerging transportation technologies; models are needed that not only provide insights into communities’ current travel demands and behaviors, but also help understand people’s attitudes and expectations toward a change — or a proposed change — in a community’s transportation infrastructure or transportation options. For example, using this system public sentiment can be tracked when accidents involving autonomous vehicles occur or when transportation milestones are achieved in this field.
However, the data on which the current models rely has limitations that prevent planners and policymakers from tapping into residents’ attitudes and perceptions widely, across the population and across time. Current models utilize surveys or opinion polls and yield a regimented set of responses to fixed questions. Moreover, the surveys reach a relatively small number of self-selecting individuals. They measure attitudes or self-reports of behavior at a single point in time, and to update them with new research topics or at new points in time is laborious and expensive. And, while some research requires datasets that extend over time, other, critical research requires real-time data that allows gauging current community sentiment around a topic.
Policymakers and researchers are increasingly recognizing the need to delve into emerging data sources like social media for use in transportation planning. Social media provides large amounts of rapidly refreshing rich data that differs by location and by geographic, demographic, and socio-economic factors, which, when subjected to machine learning algorithms and, powerful, innovative analytics, can help researchers recognize important patterns and model communities’ perceptions and sentiments, discovering how specific people are influenced by or are influencing other groups of people. Social media can also provide live data about the impact of a policy change; in particular, Twitter offers a large volume of publicly available data in which people and groups broadcast their feelings and preferences far more widely than what a survey instrument could capture.
In this project we build the Pulse-T, which will exponentially expand the access of TOMNET researchers and other organizations to an up-to-date, filtered data set of public opinion and discussions around virtually any transportation research area. Researchers and organizations will have user perceptions on transport demand at their fingertips, enabling them to take appropriate measures and actions and undertake planning projects much more effectively than is possible today.