The deficit of infrastructure quality of the United States demands groundbreaking of more infrastructure projects. Despite the potential economic and social benefits brought by these projects, they could also negatively impact the community and the environment, which could in turn affect the implementation and operation of the projects. Therefore, measuring and monitoring public acceptance is critical to the success of infrastructure projects. However, current practices such as public hearings and opinion polls are slow and costly, hence are insufficient to provide satisfactory monitoring mechanism.

Meanwhile, the development of state-of-the-art technologies such as social media and big data have provided people with unprecedented ways to express themselves. These platforms generate huge volumes of user-generated content, and have naturally become alternative sources of public opinion. This research proposes a framework and an analysis methodology to use big data from social media (e.g. the microblogging site Twitter) for project evaluation. The framework collects social media postings, analyzes public opinion towards infrastructure projects and builds multi-dimensional models around the big data. The interface and conceptual implementation of each component of the framework are discussed. This framework could be used as a supplement to traditional polls to provide a fast and cost-effective way for public opinion and project risk assessment.

This research is followed by a case study applying the framework to a real-world infrastructure project to demonstrate the feasibility and comprehensiveness of the framework. The California High Speed Rail project is selected to be the object of study. It is an iconic and controversial large-scale infrastructure project that faced a lot of criticism, complaints and suggestions. Sentiment analysis, the most important type of analysis on the framework, is discussed concerning its application and implementation in the context of infrastructure projects. A public acceptance model for social media sentiment analysis is proposed and examined, and the best measurement of public acceptance is recommended.

Moreover, the case study explores the driving force of the change in public acceptance: the social media events. Events are defined, evaluated, and an event influence quadrant is proposed to categorize and prioritize social media events. Furthermore, the individuals influencing the perceptions of these events, opinion leaders, are also modeled and identified. Three opinion leadership types are defined with top users in each type listed and discussed. A predictive model for opinion leader is also developed to identify opinion leaders using an a priori indicator. Finally, a user profiling model is established to describe social demographic characteristics of users, and each demographic feature is discussed in detail.