Files
Abstract
The extensive use of social media platform in the context of the political election makes it an important data source to provide public opinion information. Such data helps to understand peoples perception of candidates and the spatial variation of the perceptions, which becomes a significant alternative indicator for the voting results in addition to traditional socioeconomic factors. Focusing on the event of 2016 U.S Presidential Election, this research utilizes geo-tagged Twitter data in Georgia and further analyzes the spatial pattern of public sentiments from Tweet contents. After examining the election voting result from socioeconomic characters and Tweet-based sentiments as well as their relationships, this research raised a framework for election event explanation and prediction at three different scenarios, associating the socioeconomic with sentiment variables to regression models to better explain and predict the county-level voting result.