Browsing by Subject "Text analysis"

Now showing 1 - 10 of 10

An exploratory content analysis of comments and thumbnails of YouTube mukbang videos using computational methods
(2024-05) Kim, Hyun Ji, M.A.; Love, Brad (Ph. D. in media and information studies); Jihye Lee
Mukbang continues to thrive on YouTube, originating as a trend but ultimately becoming ingrained in a cultural phenomenon. Nevertheless, the impact of Mukbang on individuals' eating patterns is a subject of debate. Mukbang videos can have negative consequences, such as weight gain and the development of eating disorders. Therefore, it is crucial to comprehend the viewers' perceptions of these videos. This study is exploratory research that utilizes computational methods to analyze comments and investigate the sentiment and emotion of the viewers. The mention of food cravings and YouTuber names is also being observed to find out if people focus on food or the YouTuber while they watch the video. Additionally, the visual attributes of the video thumbnails are examined to explore the visual characteristics of popular Mukbang videos. The results indicate that the dominant sentiment and emotion expressed by the people were 'positive' and 'happiness'. Additionally, the study reveals that the group that mentions craving or YouTubers showed a distinct sentiment from the group that does not. Visual analysis revealed that common characteristics observed in popular Mukbang videos include the use of warm colors and the inclusion of a person's face in their thumbnails.
Exploring how natural language reflects individual and group social dynamics
(2021-12-02) Seraj, Sarah; Pennebaker, James W.; Swann, William B; Gosling, Samuel D; Durland, Mike
Language can be a window into people’s thoughts, feelings, and life experiences. With the increasing use of online communication platforms, researchers now have more avenues to study people’s life events using real-time and real-world data. My dissertation attempts to identify the most important language markers for understanding people’s cognitive, social, and emotional lives. What are signs in language that predict a distressing life event and how do people cope with it in the months afterward? After identifying a large group of users on Reddit (N = 6,813) who had gone through emotional upheavals such as breakups, divorce, or other distressing life events, we tracked their language in the months before, during and after their upheaval (Chapter 2). In 2020, the world faced a global crisis: the COVID-19 pandemic. In the US, the pandemic was followed by the killing of George Floyd at the hands of police and the Black Lives Matter (BLM) protests of summer 2020, a time of national reckoning on police brutality. These two events naturally led to the question of how people dealt with collective upheavals compared to a personal crisis like a breakup and how the context of the pandemic (social isolation, people living in lockdowns) affected people’s response to the BLM movement. Would the two upheavals interact with each other in any way? A large-scale Reddit dataset (33.7 million posts, 1.37 million users) was used to study the two upheavals (Chapter 3). After identifying important language markers that help us understand people’s psychological state during personal and collective upheavals, we wanted to see if the same markers were important for understanding social dynamics outside of the context of upheavals. A group of individuals who were all part of the same work team were recruited to hold a series of one-on-one chats with everyone else on the team (N = 27; 198 conversations). The language markers that predict successful conversations were identified from the study (Chapter 4). The final chapter puts together the insights from the three different studies and highlights the contribution of each of them.
Leveraging pervasive data to study and support mother-infant dyads in the wild
(2023-04-20) Yao, Xuewen; De Barbaro, Kaya; Thomaz, Edison; Yang, Diyi; Julien, Christine; Li, Jessy; Barber, Suzanne
The ubiquity of mobile devices and wearable sensors coupled with fast-evolving machine learning algorithms has transformed people's daily life, specifically in the healthcare domain. These advancements can be leveraged to not only detect and infer people's behavior patterns with great precision but also provide "just-in-time" support in a seamless, non-intrusive, and cost-effective fashion. The first year of a child's life is a particularly challenging period for the mother, and also a vital period for child developments. I hypothesize that pervasive data, such as motion, audio, text data, collected online or in people’s daily life, can be leveraged to provide support to postpartum women and their families in need in the wild. In my dissertation, I developed models that can detect two clinically-relevant parent and infant behaviors in naturalistic home interactions, namely, infant crying and parent holding. Traditional methods by developmental scientists rely heavily on behavioral observations and self-reported data while computer scientists build models using data collected in controlled environments, such as lab and hospital. These methods limit researchers’ understanding of the natural variation in mother-infant interactions across families, and its specific impacts on child development. In my work, I leveraged data collected in longitudinal home environments and built detection models that provide objective, unobtrusive, and continuous measurements of parent holding and infant crying with accuracy 0.870 and 0.613 respectively. Additionally, I evaluated both models based on assessment scenarios specific to developmental science such as event-based accuracy and contingency analysis. Another piece of my work focuses on using natural language processing to understand the experience of postpartum women experiencing or at risk of postpartum mood and anxiety disorders and to provide them with empathy in the form of a conversational agent, chatbot. Specifically, I collaborated with Postpartum Support International (PSI) and obtained text transcripts between trained volunteers and support seekers. After analyzing 7014 conversations using a combination of human annotations, dictionary models and unsupervised techniques, I find stark differences between the experiences of "distressed" and "healthy" mothers in psychological states, concerns, and goals. Additionally, incorporating the insights from the descriptive analysis as well as empathy and open questions, I designed, built, and evaluated three chatbots that accept open-ended user input to provide postpartum women with support.
The narrative arc : exploring the linguistic structure of narrative
(2015-12) Blackburn, Katherine Geraldine; McGlone, Matthew S., 1966-; Pennebaker, James W.; Peña, Jorge; Dailey, Reñe; Vangelisti , Anita
Over the millennia, philosophers and scholars have theorized on the building blocks of narrative. Until now, a reliable methodology to quantitatively investigate narrative structure has not been unavailable. The Narrative Arc Theory (NAT) was developed to investigate whether or not a common structure existed for narratives. To do this, the Linguistic Inquiry Word Count (LIWC) text analysis program was used on a corpora composed of over 5,000 narratives. A five equal sized parts segmentation strategy was developed to track how narratives develop. Simple word count analyses of function and emotion words identified five common narrative dimensions that were shared across diverse narrative text.
Power in text : extracting institutional relationships from natural language
(2018-09-14) Shaffer, Robert Bradley; Elkins, Zachary, 1970-; Jones, Bryan; Jessee, Stephen; Wilkerson, John
How do legislators allocate policy-making authority? At least in the legal context, distribution-of-power arrangements are usually articulated in written documents. Unfortunately, extracting these relationships is difficult, leading scholars to restrict themselves to studies of single policy areas or to a small set of high-visibility laws. In this project, I address this limitation through a neural network-based approach that extracts power relationships from legal language in a scalable, valid fashion. I then apply this approach to study institutional design in enacted US legislation. Substantively, I demonstrate that policy preferences of executive and legislative actors exert surprisingly little influence on formal institutional design choices. For all but the most politically salient laws, implementation arrangements are structured by the policy area and issue under consideration rather than elite political preferences. This argument - which would not have been possible to test without the measurement tools I develop - highlights both the importance of the tools I develop and the need for scalable measurement techniques in political science.
Power in text : extracting institutional relationships from natural language
(2016-12) Shaffer, Robert Bradley; Jessee, Stephen A., 1980-
How do legislators allocate policy-making authority? Generally speaking, institutional design decisions involve a trade-off between efficiency and accountability, as legislators seek to simultaneously maximize bureaucratic effectiveness and ensure favorable policy outcomes. At least in the legal context, these design decisions are often articulated in textual documents (e.g. statutes and constitutions). Unfortunately, existing measurement schemes cannot capture the full range of institutional design technologies available to legislative actors. These limitations have prevented scholars from addressing important questions regarding the relationship between executive/legislative preference conflicts, background institutional context, and downstream design of legislation. In this paper, I develop a text-based measurement scheme intended to address these limitations, which I apply to an original dataset of American legislative texts.
The control environment and financial reporting quality : does "tone at the top" matter?
(2017-08) Lord, Jeanmarie Marcelle; Kinney, William R.; Chen, Shuping; Koonce, Lisa; Pennebaker, James; Robinson, John
Are the attitudes and cognitive style of top management related to an entity’s financial reporting quality? The answer is important because the control environment, which includes management’s philosophy and operating style under COSO, is presumed to be the foundation of internal control necessary for effective transaction level processing and for management’s judgments implementing financial reporting and disclosure requirements. Yet empirical tests of the presumption are impaired by the inability to measure management’s attributes. In this study, I extract linguistic markers of top management’s attitudes and cognitive style from their response to financial reporting queries from the Securities and Exchange Commission staff and I relate the markers to concurrent material misstatements in the financial statements under scrutiny. To minimize the influence of outside professional advisors, I use responses to the initial mandated comment letters for small business issuers whose statements contained material misstatements and an industry- and size-matched sample of firms without misstatements. Using factor analysis, I find a factor composed of lower analytical thinking and higher clout significantly improves the explanatory power of a logistic regression of concurrent material misstatement and reduce the Type 1 and Type 2 classification error rates by 5% and 25%, respectively. Results are consistent with management’s attitudes and cognitive style being associated with financial reporting quality, either directly through management’s accounting judgments or indirectly through internal controls.
Thinking beyond utility and practicality : art education discussion viewed through the lens of a three-function model
(2012-12) Lee, Elizabeth Rachel; Bolin, Paul Erik, 1954-; Mayer, Melinda M
This study was about language. Its purpose was to explore how a specific set of material culture ideas is represented in art education discussion through what is termed in this study “the three-function model.” The model states that all human-made objects, including images, perform multiple roles and/or serve multiple purposes, simultaneously, and without limit. These roles and functions of objects fall into three categories: technological (utilitarian); sociological (communicative); ideological (instructive). Discovering this model inspired two questions: (a) how might the three-function approach to the discussion of objects augment art education’s understanding and practice of Material Culture theory? (b) to what benefit might such an approach be integrated into art education practice? To answer these questions, I designed a two-stage analysis. First, the examination of literature written toward three audience groups (educator-oriented, practitioner-oriented, general audience) in order to identify three types of information (definitions, statements about objects, and statements about function) for the purpose of forming an overall understanding of how cohesive or disparate discussion appears to be within each audience group. Second, cross-analyzing the three information groups for the purpose of understanding the similarities of and differences between the discussions of the three audience groups. The results of this study suggest that the problem of multiple and contrary definitions for shared terminology may be restricted to only two important words: craft and art. Conceptual approaches employed by the writers included anthropological, philosophical, concrete, theoretical, advocate, and analytical. Although all 15 writers acknowledge the social nature of objects, and all employ the term function similarly, there are indeed gaps in art education discussion: social and ideological functions of craft and art objects that go unnoticed, and missed opportunities to explore those connections and their cultural relevance. The three-function model can provide names for the ideas we are talking around, but not quite about.
This video is sponsored : text and sentiment analysis of YouTube health-related vlog comments and brand endorsement effectiveness
(2020-12-11) Yang, Zihan, M.A.; Wilcox, Gary B.
With the rapid development of YouTube and social media influencers, influencer brand endorsements have received much industry and scholarly attention. For brand endorsements made by YouTube influencers, comments under videos provide a cue for measuring endorsement effectiveness. This thesis examines the comments of both sponsored and non-sponsored YouTube health-related vlogs, as well as consumers’ commenting behaviors under the vlogs and purchase intentions toward the endorsed products. Text and sentiment analysis and survey techniques are used to examine the linguistic style of comments, the commenting behaviors of consumers, and their purchase intentions. The analysis reveals that a narrative, external, and positive linguistic style is found in the comments of health-related vlogs, and consumers’ positive commenting behavior leads to higher purchase intentions of healthy products.
Understanding political leaders and political processes through language
(2020-05-08) Jordan, Kayla Nicole; Pennebaker, James W.; Gosling, Samuel; Swann, William; Hart, Roderick
Political leaders and processes have long been studied by many disciplines including political science, communication, and psychology. Developments in technology and methodology make it possible expand and refine our understanding of how people move in the political realm. The current work builds from existing multidisciplinary work to explore political questions using psychological text analysis. Beyond the content of what people say, how people use words like pronouns, articles, and prepositions can illuminate deeper psychological traits such as thinking style, authenticity, and confidence (Boyd, 2017; Kacewicz et al., 2014; Pennebaker et al., 2014; Tausczik & Pennebaker, 2010). By combining psychological text analytic methods with traditional questions and methods, a more holistic understanding of how people act and interact within the political sphere can be developed. This dissertation presents a set of three studies exploring political leaders and processes through the lens of leaders, citizens, and the media. The first published study examined trends in leadership traits over time with a focus on a narrative, confident communication styles (Jordan, Sterling, Pennebaker, & Boyd, 2019). The second published study explored how citizens during the 2016 presidential election talked about the candidates on Twitter and what impact that conversation had on people’s perceptions of the candidates (Jordan, Pennebaker, & Ehrig, 2018). The third study looked at the media as a mediator of perceptions of political candidates by analyzing coverage of major party presidential candidates in the last five U.S. election (Jordan, in preparation).