Most of the research on social media is conducted to comprehend and to exploit the presence of public opinion in it. Meanwhile, there has been limited research that explores the engagement process and interaction results from social media accounts. Therefore, this study aims to map the interaction between government and society in social media in principal-agent context. Exploration made from tweets posted by @KemenDesa and citizens' tweets which are mentioning @KemenDesa dated from 1 October 2014 to 31 September 2017. The results give an idea of the engagement process and the content of the interaction so that the relationship between them in the principal-agent context can be mapped. In addition, the results of the analysis can also be used to evaluate the use of social media by public organizations and communities as an alternative medium to communicate.

Instead of studying the virtual space using the Political Public Sphere concept, this study adopts the Literary Public Sphere point of view to examine and narrate the nature of a Cultural Public Sphere in social media. The researchers see interactivity in social media as an articulation of expression involving emotions and aesthetics (affective communication). Using the mixed method of Topic Modelling, Social Network Analysis (SNA), and Discourse Analysis in the case of the presence of the #JogjaOraDidol hashtag in Twitter, this study conclude that the Cultural Public Sphere has three dimensions of Public Sphere as introduced by Dahlgren (2005). The dynamic of inclusivity for anyone to express themselves and to engage in public issues discussions indicates that space is inclusive not only because of the technical support of the media but also because of the commitment of its users (structural dimension). The emergence of three virtual communities (fans, artists and activists) that develop a collective identity represents a subset of the real local population and demonstrates the ideal role taking of the representational aspect of Public Sphere. The interactional one is indicated by the discourse constructed using reflexive but straightforward symbols represent the interaction between users and the meaning that users do to the contents of the media used. Meanwhile, the real action show of the discourse develops virtually does not entrap the user in pseudo-empowerment. As an implication, using specific parameter, notably the hashtag identifies a social movement, policymakers can use data from social media in the agenda-setting process. Additionally, in the context of #JogjaOraDidol, soft data can also be used to evaluate the moratorium policy of granting the hotel's construction permit.

Social media has become one of the primary sources of data which available for policy analysts and policymakers. As the evidence, active Twitter users are sending 500 million tweets per day containing thoughts, opinions, pictures, and other information. Social media offers new challenges related to how the data is acquired and how to analyze it. Unfortunately, the state-of-the-art methods in text mining are still unable to interpret texts fully. Thus, in social media analysis, we can only make a conclusion based on the insight into an event. Therefore, we propose a hybrid method that combines text mining and qualitative methods for analyzing social media data. This research was composed based on a review of studies and experimental results on the data taken from the Twitter. The results show that both techniques can complement each other and give in-depth analysis of the data. Furthermore, the results can be employed to observe social media data in a faster, cheaper, and more precise way. More importantly, the results of this study serve as a basis for further development of a method to reveal the facts behind texts that obtained from social media.