Up-to-date information on the seminar can be found on GitHub: https://github.com/FUB-HCC/seminar_critical-social-media-analysis/wiki

Content

People are gathering in social media platforms in order to connect, represent, debate and purchase. Accordingly, data sourced from these platforms can and is widely used to create knowledge on contemporary social interaction, practice and culture. In this seminar, students are introduced to critical approaches to social media analysis using and experimenting with various methods emanating from qualitative social sciences and data sciences. The thematic focus of analysis lies on the characterization and evaluation of debates concerning scientific issues including the climate crisis, COVID-19 and associated conspiracy theories on YouTube and Twitter. Throughout the course of the seminar, experts from these scientific domains will be invited to suggest and discuss possible entry points and topics to be further investigated by the students.

While students will experiment with various tools for data extraction, visualization and analysis, the main objective of the seminar is to enable a critical evaluation of methods and their contribution to knowledge creation concerning digitally-mediated social interaction. This includes the entanglement of approaches such as Grounded Theory (qualitative coding), digital ethnography and machine learning. In particular, data science methods promise new investigative opportunities and a scalability to larger datasets, which are common in the analysis of social media data. Students will learn how to make data science methods productive, while at the same time grounding their investigation in empirically-observable social practice by use of qualitative methods. To do so, students will be introduced to human-centered research approaches pushed forward by the HCC Research Group at FU.

Here you can find our Code of Conduct.

Literature

Allgaier, Joachim. „Science and Environmental Communication on YouTube: Strategically Distorted Communications in Online Videos on Climate Change and Climate Engineering“. Frontiers in Communication 4 (2019). https://doi.org/10.3389/fcomm.2019.00036.

Baumer, Eric P. S., David Mimno, Shion Guha, Emily Quan, and Geri K. Gay. „Comparing Grounded Theory and Topic Modeling: Extreme Divergence or Unlikely Convergence?“ Journal of the Association for Information Science and Technology 68, Nr. 6 (Juni 2017): 1397–1410. https://doi.org/10.1002/asi.23786.

Chen, Nan-Chen, Margaret Drouhard, Rafal Kocielnik, Jina Suh, and Cecilia R Aragon. „Using Machine Learning to Support Qualitative Coding in Social Science: Shifting The Focus to Ambiguity“. ACM Transactions on Interactive Intelligent Systems 9, Nr. 4 (2018): 21.

Geiger, R Stuart, and David Ribes. „Trace Ethnography: Following Coordination through Documentary Practices“. In 2011 44th Hawaii International Conference on System Sciences, 1–10. Kauai, HI: IEEE, 2011. https://doi.org/10.1109/HICSS.2011.455.

Marres, Noortje, and Carolin Gerlitz. „Interface Methods: Renegotiating relations between digital social research, STS and sociology“. Sociological Review, 2015. http://research.gold.ac.uk/11343/.

Muller, Michael, Shion Guha, Eric P.S. Baumer, David Mimno, and N. Sadat Shami. „Machine Learning and Grounded Theory Method: Convergence, Divergence, and Combination“. In Proceedings of the 19th International Conference on Supporting Group Work - GROUP ’16, 3–8. Sanibel Island, Florida, USA: ACM Press, 2016. https://doi.org/10.1145/2957276.2957280.

Pfeffer, Jürgen, Katja Mayer, and Fred Morstatter. „Tampering with Twitter’s Sample API“. EPJ Data Science 7, Nr. 1 (Dezember 2018): 1–21. https://doi.org/10.1140/epjds/s13688-018-0178-0.

Schäfer, Mike S. „Online Communication on Climate Change and Climate Politics: A Literature Review“. WIREs Climate Change 3, Nr. 6 (2012): 527–43. https://doi.org/10.1002/wcc.191.

Student Project Results

Group 1: Quantitative and Qualitative Analysis of User Behavior Based on YouTube Videos Dealing With Climate Change

User behavior analysis based on social network data is an extensive research field usually affected by noise and misguiding elements. Millions of videos on one of the biggest and fastest-growing websites in the world - YouTube, often trigger long discussions in their comments section. Those interactions might count with the presence of trolls, bot accounts, human bot, or entities, which for various reasons, try to deviate the sentiment of the ongoing discussions. This paper focuses on trying to identify such patterns under videos of a highly controversial conspiracy theory about so-called "chemtrails". We want to determine user groups with the same opinion commenting on the same videos, find out if they also appear with similar comments in other videos of the topic, and what their intentions could be. To do so, we will leverage two different approaches for qualitative and quantitative analysis to successfully classify behavioral patterns on social media interactions, enabling us to answer the aforementioned questions.

Group 1b: What are the topics and practices of the most prolific commenters under videos related to climate change?

In the following paper we show a way of collecting a large amount of YouTube comments using some custom scripts and the YouTube API. Afterwards we show the results of some statistical insights into the data set and provide more in depth analysis of the 200 most frequent commenters.

Group 2: The end of political apathy?

A Mixed Methods Analysis of the general attitude towards young individuals that publicly engage in political topics.

A case study for climate change

Adults often complain about the allegedly unpolitical youth. However, with Fridays For Future, there exists now a very political global youth movement. We examined the public opinion on this development, i.e. we analyzed if people are pleased about the young age of such activists or if they are criticizing them and why. We did this by looking into YouTube comments under videos featuring young politically active people like the globally famous climate activist Greta Thunberg and the probably most popular climate change skeptic Naomi Seibt. For the analysis we used an empiric mixed methods approach. We found out that the attitudes towards the political engaged individuals of the younger generation are very polarized in a way that age-related comments either accuse inexperience, lack of education or assume that the youth is being used, or defend the courage and commitment that it takes to stand up and raise their voice. Index Terms—Ageism, Climate Change, Digital Humanities, Mixed Methods, Online Media, Youth Climate Activism, YouTube

Group 3: How is climate change discussed dierently on YouTube at various moments in time

YouTube is the most popular internet video platform. Every day more than one billion users watch more than a billion hours of video collectively. YouTube's wide reach and high popularity has inspired scholars of all elds to explore YouTube as a research medium. One of its advantages is that it is suitable for exploring any media content, be it educational or entertainment content, advertising or news. In addition, variations in presentation format are possible, from images and text to audiovisual materials and lms, allowing content to be created by every user regardless of their technical capabilities. Finally, as YouTube allows its users to post comments on presented video material that is publicly available, it is possible to explore comments on the displayed content. Many people use YouTube as a source of information on technology and science (Leon and Bourk 2018). But there is often controversy surrounding online video and science videos are no exception. Many scientic disputes resolved in the academic community continue to circulate in the social media and even in political debates. This is the case with climate change (Leon and Bourk 2018, Chapter 4). When scientic issues are presented as controversial, in most cases this is done from a political or social perspective rather than from a scientic perspective. In the following, we present a project in which we investigate comments on YouTube videos containing materials about climate change and try to analyse how different social and political events in uence the debate on such a controversial topic.

Group 4: Critical Discourse Studies (CDS) of Social Media: Taking a Greta Thunberg’s Speech on YouTube as a Case Study

Nowadays Social media has become an important information platform in our society. People use social media to connect to eachother, express their opinions, share their news, and discuss hot issues of society. With the development of social media, The influence on society and people’s daily life has become stronger. This research takes a speech of a Swedish environmental activist Greta Thunberg on YouTube as an example. Taking critical discourse analysis as a theoretical perspective. With qualitative analysis, we found the attitudes and intention behind the video. At the same time, by analyzing 27,324 online comments of the video, we found that the netizens have different attitudes toward Greta and environmental protection. With machine learning methode and Quantitative analysis, and using descriptive statistical methods. The authors believe that social media is playing an increasingly important role in expressing public opinions. This research helps us understand the attitudes of Internet users on the topic such as environmental protection.

Group 5: Alt-Right Influence in YouTube Comment Sections

In the context of the seminar "Critical social media analysis using mixed methods" we worked on identifying members of the altright in the comment section of YouTube videos using a natural language processing pipeline. As the alt-right drastically increased their activity on social media platforms and also YouTube, we think it is important to have means and tools to identify possible attempts by them to influence discussions. We tried to use the presence of dog whistles as an indicator for alt-right participation. To achieve this goal, we analyzed YouTube comment data with the supplied pipeline of the seminar, clustered similar comments and then checked them against a list of known dog whistles if the resulting clusters can be labelled with one of these dog whistles. We also investigated clusters close to the identified dog whistle clusters to see if we can identify new dog whistles.

Group 6: YouTube Platform Interventions discussed in Videos about Climate Change Skepticism

The aim of this study is to investigate how different types of platform interventions are discussed in the comment sections of YouTube videos in the context of climate change skepticism. The work focuses on exposing explanation and justification patterns to real and perceived platform interventions, determining the volume of comments debating the intervention methods as well as the user’s opinion on the relationship between interventions and free speech, with particular focus on platform credibility. An elaboration of the underlying theoretical background combined with a detailed literature review was carried out to accurately highlight the necessities, implementation strategies and consequences of interventions on social media platforms. To answer the research question a preliminary quantitative analysis followed by an extensive qualitative analysis was performed. The comment sections of ten precisely selected YouTube videos were considered where the average proportion of comments actually dealing with platform interventions was determined to be 3% corresponding to an average number of 99 comments of interest per video. A central component of the criticism on platform interventions consists in complaints about the YouTube search algorithm and the accessibility of specific videos via the search function. This resentment becomes especially evident by the observable trend of moving controversial content and entire communities to other platforms. It is further noticeable that the commenting users have fundamentally different perceptions of YouTube’s role as a social platform. A considerable part of the users understand intervention methods as a form of censorship and an infringement against their right of self-expression and accuse YouTube to follow a hidden agenda. In contrast, other users accept and justify the execution of platform interventions and request to delete unacceptable content more quickly. Overall, a positive correlation between the evaluation of the platform’s credibility and the evaluation of platform interventions was observable. As a third question, we have a theoretical discussion about the possibility to have an automatic sanctioning of alt-right members based on our results. After analyzing the comments, we were able to identify known dog whistles, which in most cases also appeared in the same clusters. It turned out that the alt-right clusters separate from other clusters, which made it possible for us to label them and then examine them more closely. Nevertheless, it was not possible for us to identify new dog whistles. But the pipeline can be used as a starting point for further research and with more data, new dog whistles could probably be identified.

Group 8: Geoengineering discourse on YouTube: A machine learning based approach for the analysis of discourses

Concurrent with the growing public interest in online discussions on social media platforms, as well as the growing importance of the topic of climate change, YouTube serves as an open platform for the general public to participate in social debates. Besides the academic and scientific debate on geoengineering as a method to alleviate global warming, a more marginal discourse takes place in which the word "chemtrail" is interchangeably used with "geoengineering" to describe the contrail left by aeroplanes. Chemtrail activists claim that some “global elites” ally with governments to use chemtrails for weather manipulation for some secret on-going programs. Public debates around chemtrail conspiracy theories have received little attention in the academic field, despite recent calls for greater understanding of the different ways in which the idea of geoengineering is understood and widely discussed by the general public. This paper works on gaining insights into the public discourse around chemtrails and geoengineering with the Grounded Theory approach supported by machine learning algorithms. Our study has two main contributions: (1) describing the discursive patterns from both (with or against scientific consensus) camps around the topic of geoengineering ; (2) implementing Mutual Information Score to enhance in-cluster structure for qualitative analysis.

Group 9: Analyzing conicting views on the COVID-19 Vaccination eorts in Social Media from an Establishment/Anti-Establishment standpoint using Machine Learning

During the seminar "Critical Social Media Analysis using Mixed Methods" in the Winter Term 2020/21 at Free University Berlin, we made ourselves familiar with the idea of interdisciplinary research between Computer Scientists and Digital Media researchers. In this context we learned on the one hand, how to work with technical tools tailored for Social Media like "YouTube Data Tools" and a natural language processing (NLP) pipeline while we on the other hand regularly read research papers in order understand Social Science methods or to get ideas for interesting research questions and design. The goal of the seminar was that we would ultimately design our own research papers in the eld of Digital Media research. Due to our unfamiliarity with NLP and scientic research in general, we decided to stay close to the methods that we learned during this seminar, but apply them to a research eld and research questions that interested us both. Due to the rising political and social tensions which we have witnessed in the recent period, both overseas in the United States and \at home" in Europe, we first had the idea to analyze the differences between left and right Establishment Critique. We felt that in some way the anger expressed in very dierent political movements seemed to be rooted in a rising dissatisfaction with the status quo in the West. But we encountered the problem that in the often "messy" real-world we did not have the possibility to study data that is neatly divided into the categories \establishment critique left" and \establishment critique right". We therefore decided to rather contrast \the establishment" and \the anti-establishment". Firstly, we tried to clarify for ourselves what we meant by these terms: for us (and based on some denitions we have looked up on this topic) the establishment can be seen as the current power structure in society and the people still having faith in the current order and acting in accordance with it, while the anti-establishment is a heterogenous collection of people or rather opinions held by people who distrust the current power elites and show some rebellion against them. Due to the very new and recent character of the COVID-19 pandemic, the increased importance of institutions like the WHO, the government, scientists etc. and the interesting conflict between the ideology of liberalism and pandemic response of governments, we decided to study the conflicting establishment/anti-establishment views of netizens on the topic of COVID-19. We further decided to focus on COVID- 19 vaccination due to availability of research on the topic of vaccine skepticism. In the following paper, we are going to draft a research design by starting with giving some context on the characterization of the establishment /antiestablishment divide and vaccination skepticism, formulating research questions that we want to answer, proposing a way of answering the research questions and ultimately presenting the results and insights we obtained on the basis of our research.