Journal of Digital Information Management Hsuanwei Michelle Chen, Patricia C. Franks, Lois Evans 14 5 2016 http://dline.info/fpaper/jdim/v14i5/jdimv14i5_2.pdf As social media becomes an important platform for organizations to use to interact with users, the ability to understand user opinions in social media communications has gained increased attention. One of the most popular approaches for exploring user opinions is sentiment analysis, which employs natural language processing, statistics, or machine learning to extract the sentiment of a text unit in terms of positive or negative attitudes. However, the effectiveness, interpretation, and accuracy of sentiment analysis rely heavily on the context in which it is conducted. In this paper, we investigate three sentiment analysis techniques for Twitter use by governments with their citizens, including a lexicon-based approach, a machine learning-based approach, and a hybrid approach. Our results reveal that, while each technique is developed based upon different rationales, the results are statistically robust and comparable. The study provides new insights into sentiment analysis in the context of government uses of social med