Thesis (M.A.)--Ohio University, November, 2009. / Title from PDF t.p. Includes bibliographical references.

Thesis (Ph.D.)--Ohio University, June, 2009. / Release of full electronic text on OhioLINK has been delayed until June 1, 2013. Includes bibliographical references (leaves 222-242)

Thesis (Ph.D.)--Ohio University, June, 2009. / Title from PDF t.p. Release of full electronic text on OhioLINK has been delayed until June 1, 2013. Includes bibliographical references (leaves 222-242)

Thesis (M.A.)--University of Akron, Dept. of Sociology, 2009. / "May, 2009." Title from electronic thesis title page (viewed 11/18/2009) Advisor, Rebecca J. Erickson; Faculty Reader, Clare L. Stacey; Department Chair, John F. Zipp; Dean of the College, Chand Midha; Dean of the Graduate School, George R. Newkome. Includes bibliographical references.

The rise and growth of online social networks have spurred tremendous changes in our understanding of human behavior. Social scientists and companies have devised new tools to analyze the vast amounts of data obtained from these networks. Such advances have had two major consequences. First, it has allowed firms to significantly improve their segmentation and targeting strategies. Second, it also modified how problems are conceptualized. For example, books, academic papers, or webpages are now being studied under methods developed for social network analysis. This dissertation contributes to both applications. Essays 1 and 2 describe efficient targeting strategies in situations where access to information or computing power is costly. Although existing “seeding” methods have been quite successful in social networks, they often do not account for firms' limited computing power or assume that firms are omniscient. Essay 3 focuses on the art industry by conceptualizing paintings as items connected to each other in a network through their visual similarities. Indeed, we still do not perfectly understand what makes art financially valuable and even major auction houses are at awe when paintings are sold at prices multiple times higher than what they expected. In particular, we aim to quantify how an art piece's visual features and historical importance may impact prices and assess how auction houses and their marketing efforts may modify how art is evaluated and valued. This dissertation has three essays. In the first essay, we analyze how the friendship paradox, which states that your friends have more friends than you, may be generalized to situations where relationships are asymmetric. Indeed, the result assumes symmetric relations: if two people are friends, then each is the other's friend. For social networks that satisfy this assumption (e.g., Facebook), the friendship paradox implies that firms can potentially achieve faster and more widespread diffusion of information by seeding it with the friends of a group of people than with people in the group itself. We generalize the result to allow one-sided (leader/follower) relations and examine the implications for seeding in social networks where messages can be sent only by a leader to his/her followers. We obtain necessary and sufficient conditions under which the highest number of followers is obtained by seeding with (1) leaders, (2) followers, and (3) individuals chosen by ignoring the distinction between leaders and followers. We examine the seeding implications of the results for a subset of Twitter users. The second essay furthers our understanding of the friendship paradox and relates it to beta centrality and eigenvector centrality. We generalize the results to asymmetric relations, define two beta centrality measures and relate them to the singular vectors of the associated directed graph. Our first generalization shows that the expected number of k removed friends is no smaller than the expected number of k-1 removed friends when k is an even number. Such a relation does not necessarily exist when k is an odd number. As k increases to infinity, the limiting value of the expected number of k removed friends converges to the largest eigenvalue of the associated undirected graph. We interpret beta centrality to be a weighted sum of an infinite series of the numbers of k removed friends. It approaches eigenvector centrality when the weighting parameter becomes arbitrarily close to the inverse of the limiting value of the expected number of k removed friends. We further generalize these results to asymmetric relations (say, between followers and leaders) that can be represented by directed graphs. We show that the last person in a randomly selected alternating sequence of 2k+1 leaders and followers (followers and leaders) has no fewer followers (leaders) than the last person in a randomly selected alternating sequence of 2k followers and leaders (leaders and followers). As k increases to infinity, the expected number of leaders of the last person in a randomly selected sequence of 2k alternating leaders and followers converges to a value proportional to the largest singular value of the associated directed graph. Similarly, the expected number of followers of the last person in a randomly selected sequence of 2k alternating followers and leaders converges to a (different) value proportional to the largest singular value of the associated directed graph. We show that there is a reciprocal relation between the limiting expected values of leaders and followers. We generalize beta centrality to asymmetric relations and relate the limiting values of beta centrality scores for followers and leaders to the singular vectors of the associated directed graph. The third essay focuses on the art market. Auction houses hold auctions regularly throughout the year. However, once or twice a year, art investors and wealthy consumers attend highly selective marquee events: day and evening sales. Those carefully designed and highly marketed events often generate a lot of excitement for connoisseurs as most paintings get sold for tremendous amounts of money. But what makes those paintings special? We investigate how art is evaluated across those three types of auctions. Specifically, we build a deep learning model to summarize the paintings into a low dimensional representation space where each factor encodes a specific feature of the paintings’ aesthetics and further utilize those components to create “network” variables that will determine how influential and creative a painting is. We use those predictors in hedonic regression models to study how art returns differs across the three types of sales and subsequently analyze whether the paintings are evaluated differently. In particular, we find that paintings sold in evening sales generated an annualized return of 14.33% in the period 1999-2018 - more than three times the returns of paintings sold in regular or day auctions. Finally, we adopt a propensity score matching approach to create a homogeneous population of paintings - based on their likelihood to be auctioned in an evening sale - to assess the causal impact of being featured in an evening sale and find that such highlight increases a painting's price by almost $6 million.

South Africa, since 1994, has developed both politically and technologically resulting in an opening of communications both locally and globally. The 2009 national elections had been earmarked as a 'make or break' milestone for the political and social future of the young democracy. This election occurred amidst media analysts‘ concerns for the level of freedom of expression allowed to traditional forms of the South African media. New media, however, is not at present subject to the same regulations. Although a few cases of slander relating, for example, to Facebook have occurred, ephemeral cyber space appears to enjoy a greater degree of freedom of expression than the press and broadcast media. As a result the ability of these traditional forms of media to function effectively as a public sphere may be questioned, and some theorists claim that the Internet may offer an alternative medium for this function. This thesis looks at the potential of online communities to facilitate democratic expression by analysing a Facebook newsgroup text at the time of the election. In my exploration of the Facebook newsgroup Help us stop Jacob Zuma from becoming SA's next President I have analysed the text using two qualitative approaches. The critical discourse analysis traces competing South African discourses relating to the myths of the inherent violence of black men and the inherent racism of whites, the topics of crime and violence, Jacob Zuma and South Africanness. This approach‘s theoretical guidelines enforced a more objective view of the text, although interpretive methods in general grapple with subjectivity at a more observable level than do quantitative methods. The ethnographic hermeneutic component of the research is aimed at "making the obscure plain" (Blaikie, 1993: 28, cited in Neuman, 1997: 68) in the text, as well as documenting the inner workings of the online community and its relation to South African issues at the time of the national election. The findings are then measured against public sphere theory from Habermas' conception of the bourgeois public sphere to revisionist accounts (Fraser, 1997 and McKee, 2005) / Thesis (M.A.)-University of KwaZulu-Natal, Pietermaritzburg, 2010.

Automatic identification of speaker state is essential for spoken language understanding, with broad potential in various real-world applications. However, most existing work has focused on recognizing a limited set of emotional states using cues from a single modality. This thesis describes my research that addresses these limitations and challenges associated with speaker state identification by studying a wide range of speaker states, including emotion and sentiment, humor, and charisma, using features from speech, text, and visual modalities. The first part of this thesis focuses on emotion and sentiment recognition in speech. Emotion and sentiment recognition is one of the most studied topics in speaker state identification and has gained increasing attention in speech research recently, with extensive emotional speech models and datasets published every year. However, most work focuses only on recognizing a set of discrete emotions in high-resource languages such as English, while in real-life conversations, emotion is changing continuously and exists in all spoken languages. To address the mismatch, we propose a deep neural network model to recognize continuous emotion by combining inputs from raw waveform signals and spectrograms. Experimental results on two datasets show that the proposed model achieves state-of-the-art results by exploiting both waveforms and spectrograms as input. Due to the higher number of existing textual sentiment models than speech models in low-resource languages, we also propose a method to bootstrap sentiment labels from text transcripts and use these labels to train a sentiment classifier in speech. Utilizing the speaker state information shared across modalities, we extend speech sentiment recognition from high-resource languages to low-resource languages. Moreover, using the natural verse-level alignment in the audio Bibles across different languages, we also explore cross-lingual and cross-modality sentiment transfer. In the second part of the thesis, we focus on recognizing humor, whose expression is related to emotion and sentiment but has very different characteristics. Unlike emotion and sentiment that can be identified by crowdsourced annotators, humorous expressions are highly individualistic and cultural-specific, making it hard to obtain reliable labels. This results in the lack of data annotated for humor, and thus we propose two different methods to automatically and reliably label humor. First, we develop a framework for generating humor labels on videos, by learning from extensive user-generated comments. We collect and analyze 100 videos, building multimodal humor detection models using speech, text, and visual features, which achieves an F1-score of 0.76. In addition to humorous videos, we also develop another framework for generating humor labels on social media posts, by learning from user reactions to Facebook posts. We collect 785K posts with humor and non-humor scores and build models to detect humor with performance comparable to human labelers. The third part of the thesis focuses on charisma, a commonly found but less studied speaker state with unique challenges -- the definition of charisma varies a lot among perceivers, and the perception of charisma also varies with speakers' and perceivers' different demographic backgrounds. To better understand charisma, we conduct the first gender-balanced study of charismatic speech, including speakers and raters from diverse backgrounds. We collect personality and demographic information from the rater as well as their own speech, and examine individual differences in the perception and production of charismatic speech. We also extend the work to politicians' speech by collecting speaker trait ratings on representative speech segments of politicians and study how the genre, gender, and the rater's political stance influence the charisma ratings of the segments.

The influx of new data-intensive applications, such as machine learning and artificial intelligence, in high performance computing (HPC) and data centers (DC), has driven the design of efficient interconnection networks to meet the requisite bandwidth of the growing traffic demand. While the exponentially-growing traffic demand is expected to continue into the future, the free scaling of CMOS-based electrical interconnection networks will eventually taper off due to Moore’s Law. These trends suggest that building all-electrical interconnects to meet the increased demand for low latency, high throughput networking will become increasingly impractical going forward. Integrating optical interconnects capable of supporting high bandwidth links and dynamic network topology reconfiguration offer a potential solution to scaling current networks. However, the insertion of photonic interconnection networks offers a massive design space in terms of network topology and control plane that is currently under-explored. The work in this dissertation is centered around the study and development of control plane challenges to aid in the eventual adoption of optically-enabled reconfigurable networks. We begin by exploring Flexspander, a novel reconfigurable network topology that combines the flexible random expander networks construction with topological-reconfigurability using optical circuit switching (OCS). By incorporating random expander graph construction, as opposed to other more symmetric reconfigurable topologies, Flexspander can be built with a broader range of electrical packet switch (EPS) radix, while retaining high throughput and low latency when coupled with multi-path routing. In addition, we propose a topology-routing co-optimization scheme to improve network robustness under traffic uncertainties. Our proposed scheme employs a two-step strategy: First, we optimize the topology and routing strategy by maximizing throughput and average packet hop count for the expected traffic patterns based on historical traffic patterns. Second, we employ a desensitization step on top of the topology and routing solution to lower performance degradation due to traffic variations. We demonstrate the effectiveness of our approach using production traces from Facebook's Altoona data center, and show that even with infrequent reconfigurations, our solution can attain performances within 15\% of an offline optimal oracle. Next, we study the problem of routing scheme design in reconfigurable networks, which is a more under-studied problem compared to routing design for static networks. We first perform theoretical analyses to first identify the key properties an effective routing protocol for reconfigurable networks should possess. Using findings from these theoretical analyses, we propose a lightweight but effective routing scheme that yields high performance for practical HPC and DC workloads when employed with reconfigurable networks. Finally, we explore two fundamental design problems in the optical reconfigurable network design. First, it investigates how different OCS placement in the physical network topology lead to different tradeoffs in terms of power consumption/cost, network performance, and scalability. Second, we investigate how network performance is affected by different reconfiguration periods to understand how frequency of topology reconfiguration affects application performance. Taken together, the work in this dissertation tackles several key challenges related to efficient control plane for reconfigurable network designs, with the goal of facilitating the eventual adoption of optically-enable reconfigurable networks in high performance systems.

At the heart of the archive lie the questions of what will be repeated and what comes first, questions that ripple through curriculum studies and qualitative research. Whether social media platforms like Facebook or the monuments of white supremacists, archives increasingly mediate relationships with the past and generate monumental controversies. Hung up on archival exclusions and surplus values—the anarchive—this study considered three different archives: a monument dedicated to Harriet Tubman, a prominent social media platform, and two reading groups dedicated to process philosophy and affect studies. Studying the anarchive involved a mixture of ethnographic methods and speculative practices, like fictocriticism, reading groups, and assemblage art. The way these archives came together affected what they did. Rather than static receptacles, they affected and were affected by novel assemblages. Thus, anarchiving—attunement and experimentation with the archive’s virtuality—entailed taking on responsibility for what those archives might yet become.

In a time of extreme political polarization in America, it is necessary to understand how political attitudes are shaped and how and why political perspectives shift. This study explored political polarization and the search for common ground among American voters in order to evaluate 1) political orientation and associations with current and shifting political attitudes, perceptions, behaviors, and learning; 2) political efficacy and associations with political participation; 3) contributions of individual events, experiences, sources and/or social interactions to shifting political perspectives; and 4) the potential role of observational learning and experiential learning as it relates to political attitudinal change. Through a mixed-methods design, this study utilized both quantitative and qualitative methods in two phases in anticipation of the November 2020 U.S. presidential election. First, a survey was administered nationwide on Facebook in July and August of 2020. Second, a small number of participants that indicated that their political views had shifted during the Trump presidency were interviewed in October 2020 to discover more in-depth responses regarding both the formation and shifts of their political ideologies and attitudes. Quantitative statistical analysis from the survey (n = 1,313) revealed that 1) political orientation was significantly associated with a variety of current and shifting political attitudes, political participation, cognitive perceptions and emotions, social media and news media usage, and learning; and 2) personal and collective political efficacy was significantly associated with increased political participation. Qualitative analysis from the semi-structured interviews (n = 16) utilizing the Framework Method indicated that 1) social interactions with role models and media sources informed shifts in political attitudes, especially in regard to family; 2) deeper understandings of shifting political attitudes emerged when viewed through the observational learning subprocesses of attention, retention, reproduction, and motivation; 3) shifting political attitudes were associated with cognitive perceptions and emotions of partisan ingroups and partisan outgroups; 4) deeper understandings of shifting political attitudes emerged when viewed through the experiential learning stages of returning to the experience, attending to emotions, and re-evaluating the experience; and 5) barriers to shifting political attitudes were related to the cognitive perceptions of reason and emotion, whereas rewards for shifting political attitudes were related to increased political participation and learning. The study’s conclusions reaffirmed that there is widespread political polarization in the United States but that there are certain pathways towards common ground through civility and respect, intentional listening, and through the sharing of personal and emotional stories and lived experiences. Those who did shift their views shared an openness to learning through the observation of role models and sources (e.g., family, friends, mentors or teachers, religious leaders, media) and through reflection on past experiences. Aspects of political common ground were obtained not by agreeing on a contentious issue, but by acknowledging that multiple sides of an argument could be valid. It’s about how we approach the problem rather than how we solve the problem. Political orientation is much more complex than the dichotomous portrayal of left versus right.