Text analytics of social media: Sentiment analysis, event detection and summarization

In the last decade, large numbers of social media services have emerged and been widely used in people's daily life as important information sharing and acquisition tools. With a substantial amount of user-contributed text data on social media, it becomes a necessity to develop methods and tools for text analysis for this emerging data, in order to better utilize it to deliver meaningful information to users. ^ Previous work on text analytics in last several decades is mainly focused on traditional types of text like emails, news and academic literatures, and several critical issues to text data on social media have not been well explored: 1) how to detect sentiment from text on social media; 2) how to make use of social media's real-time nature; 3) how to address information overload for flexible information needs. ^ In this dissertation, we focus on these three problems. First, to detect sentiment of text on social media, we propose a non-negative matrix tri-factorization (tri-NMF) based dual active supervision method to minimize human labeling efforts for the new type of data. Second, to make use of social media's real-time nature, we propose approaches to detect events from text streams on social media. Third, to address information overload for flexible information needs, we propose two summarization framework, dominating set based summarization framework and learning-to-rank based summarization framework. The dominating set based summarization framework can be applied for different types of summarization problems, while the learning-to-rank based summarization framework helps utilize the existing training data to guild the new summarization tasks. In addition, we integrate these techneques in an application study of event summarization for sports games as an example of how to better utilize social media data. ^

Text Analytics of Social Media: Sentiment Analysis, Event Detection and Summarization

In the last decade, large numbers of social media services have emerged and been widely used in people's daily life as important information sharing and acquisition tools. With a substantial amount of user-contributed text data on social media, it becomes a necessity to develop methods and tools for text analysis for this emerging data, in order to better utilize it to deliver meaningful information to users. Previous work on text analytics in last several decades is mainly focused on traditional types of text like emails, news and academic literatures, and several critical issues to text data on social media have not been well explored: 1) how to detect sentiment from text on social media; 2) how to make use of social media's real-time nature; 3) how to address information overload for flexible information needs. In this dissertation, we focus on these three problems. First, to detect sentiment of text on social media, we propose a non-negative matrix tri-factorization (tri-NMF) based dual active supervision method to minimize human labeling efforts for the new type of data. Second, to make use of social media's real-time nature, we propose approaches to detect events from text streams on social media. Third, to address information overload for flexible information needs, we propose two summarization framework, dominating set based summarization framework and learning-to-rank based summarization framework. The dominating set based summarization framework can be applied for different types of summarization problems, while the learning-to-rank based summarization framework helps utilize the existing training data to guild the new summarization tasks. In addition, we integrate these techneques in an application study of event summarization for sports games as an example of how to better utilize social media data.

Online News Media Impact on Disaster Response: October 26, 2010 Indonesian Disaster Case Study

This paper reflects a research project on the influence of online news media (from print, radio, and televised outlets) on disaster response. Coverage on the October 2010 Indonesian tsunami and earthquake was gathered from 17 sources from October 26 through November 30. This data was analyzed quantitatively with respect to coverage intensity over time and among outlets. Qualitative analyses were also conducted using keywords and value scale that assessed the degree of positivity or negativity associated with that keyword in the context of accountability. Results yielded insights into the influence of online media on actors' assumption of accountability and quality of response. It also provided information as to the optimal time window in which advocates and disaster management specialists can best present recommendations to improve policy and raise awareness. Coverage of outlets was analyzed individually, in groups, and as a whole, in order to discern behavior patterns for a better understanding of media interdependency. This project produced analytical insights but is primarily intended as a prototype for more refined and extensive research.

The Caribbean Court of Justice: An investigation of the impact of elites on its creation and present structure as it relates to original jurisdiction

The primary focus of this dissertation is to determine the degree to which political, economic, and socio-cultural elites in Jamaica and Trinidad & Tobago influenced the development of the Caribbean Court of Justice's (CCJ) original jurisdiction. As members of the Caribbean Community (CARICOM), both states replaced their protectionist model with open regionalism at the end of the 1980s. Open regionalism was adopted to make CARICOM member states internationally competitive. Open regionalism was also expected to create a stable regional trade environment. To ensure a stable economic environment, a regional court with original jurisdiction was proposed. A six member Preparatory Committee on the Caribbean Court of Justice (PREPCOM), on which Jamaica and Trinidad & Tobago sat, was formed to draft the Agreement Establishing the Caribbean Court of Justice that would govern how the Court would interpret the Revised Treaty of Chaguaramas (RTC) and enforce judgments. ^ Through the use of qualitative research methods, namely elite interviews, document data, and text analysis, and a focus on three levels of analysis, that is, the international, regional, and domestic, three major conclusions are drawn. First, changes in the international economic environment caused Jamaica and Trinidad & Tobago to support the establishment of a regional court. Second, Jamaica had far greater influence on the final structure of the CCJ than Trinidad & Tobago. Third, it was found that in both states the political elite had the greatest influence on the development and structure of the CCJ. The economic elite followed by the socio-cultural elite were found to have a lesser impact. These findings are significant because they account for the impact of elites and elite behavior on institutions in a much-neglected category of states: the developing world.^

Implicit metaphysis of the Visuddhimagga : assāsa-passāsa as a vital animating force

This research examines how assasa-passisa and its surrounding concepts are discussed in Buddhaghossa's 5th century Theravada work, the Visuddhimagga (The Path of Purification) to determine if there is metaphysical use of the term in the text and to determine if the concept of assasa-passasa is similar to the better-known Indian concept of prana (metaphysical vital animating force), indicating whether Theravada Buddhism more closely resembles other Indian religions in terms of metaphysical content. Text analysis reveals how assasa-passasa is described in the Visuddhimagga as an animating vital force, suggesting that Theravada Buddhism has an implicit ontology similar to other Indian schools of philosophy. Secondarily, this paper argues that because assisa-passasa plays a similar role to prana in the Visuddhimagga, it is also operationally similar and could be functioning as the implicit intermediary between links in the chain of dependent co-arising-as the vehicle of paticcasamuppada.