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.

Lower Aptian Comparative Stratigraphy of the Basco-Cantabrian Region (Spain) and Eastern Cordillera (Colombia): implications for local factors in the depositional record of Oceanic Anoxic Event 1a (OAE-1a)

An important episode of carbon sequestration, Oceanic Anoxic Event 1a (OAE-1a), characterizes the Lower Aptian worldwide, and is mostly known from deeper-water settings. The present work of two Lower Aptian deposits, Madotz (N Spain) and Curití Quarry (Colombia), is a multiproxy study that includes fossil assemblages, microfacies, X-ray diffraction bulk and clay mineralogy, elemental analyses (major, minor, trace elements), Rock-Eval pyrolysis, biomarkers, inorganic and organic carbon content, and stable carbon isotopes. The results provide baseline evidence of the local and global controlling environmental factors influencing OAE-1a in shallow-water settings. The data also improve our general understanding of the conditions under which organic-carbon-rich deposits accumulate. The sequence at Madotz includes four intervals (Unit 1; Subunits 2a, 2b and 2c) that overlap the times prior to, during and after the occurrence of OAE-1a. The Lower Unit 1(3m thick) is essentially siliciclastic, and Subunit 2a (20m) contains Urgonian carbonate facies that document abruptly changing platform conditions prior to OAE-1a. Subunit 2b (24.4 m) is a mixed carbonate-siliciclastic facies with orbitolinid-rich levels that coincides with OAE-1a δ13C stages C4-C6, and is coeval with the upper part of the Deshayesites forbesi ammonite zone. Levels with pyrite and the highest TOC values (0.4-0.97%), interpreted as accumulating under suboxic conditions, and are restricted to δ13C stages C4 and C5. The best development of the suboxic facies is at the level representing the peak of the transgression. Subunit 2c, within δ13C stage C7, shows a return of the Urgonian facies. The 23.35-m section at Curití includes a 6.3-m interval at the base of the Paja Formation dominated by organic-rich marlstones and shales lacking benthic fossils and bioturbation, with TOC values as high as 8.84%. The interval overlies a level containing reworked and phosphatized assemblages of middle Barremian to lowest Aptian ammonites. The range of values and the overall pattern of the δ13Corg (-22.05‰ to -20.47‰) in the 6.3m-interval is comparable with Lower Aptian δ13C stage C7. Thus, conditions of oxygen depletion at this site also occurred after Oceanic Anoxic Event-1a, which developed between carbon isotope stages C3 and C6. Both sites, Madotz and Curití, attest to the importance of terrigenous and nutrient fluxes in increasing OM productivity that led to episodic oxygen deficiency.