Investigating Opinions on Public Policies in Digital Media: Setting up a Supervised Machine Learning Tool for Stance Classification

Supervised machine learning (SML) provides us with tools to efficiently scrutinize large corpora of communication texts. Yet, setting up such a tool involves plenty of decisions starting with the data needed for training, the selection of an algorithm, and the details of model training. We aim at establishing a firm link between communication research tasks and the corresponding state-of-the-art in natural language processing research by systematically comparing the performance of different automatic text analysis approaches. We do this for a challenging task – stance detection of opinions on policy measures to tackle the COVID-19 pandemic in Germany voiced on Twitter. Our results add evidence that pre-trained language models such as BERT outperform feature-based and other neural network approaches. Yet, the gains one can achieve differ greatly depending on the specific merits of pre-training (i.e., use of different language models). Adding to the robustness of our conclusions, we run a generalizability check with a different use case in terms of language and topic. Additionally, we illustrate how the amount and quality of training data affect model performance pointing to potential compensation effects. Based on our results, we derive important practical recommendations for setting up such SML tools to study communication texts.

Educational Data Mining: Sentiment analysis in a university domain during the pandemic

The new normality caused by the COVID 19 health emergency forced governments to adopt mandatory isolation rules, the measures modified social interaction in almost all scenarios. In the educational field, educational establishments in general closed their facilities and adopted technological tools to replace the physical classroom, thus inaugurating a period of emergency remote education (Hodges et al., 2020) (Portillo, S., Castellanos, L., Reynoso, O., & Gavotto, O., 2020) (Castaneda and Vargas, 2021). The present investigation analyzes the student perception about the implementation of emergency remote education in the Social Communication Career of the Central University of Ecuador in the academic period 2020 -2020, for which an opinion poll was applied to students. The qualitative instrument collected the opinion, and the corpus was processed in a Python bookstore. In general, a positive assessment of the new educational reality was evidenced, the use of platforms and social networks was highlighted, as well as innovative educational practices in: master classes, experimentation, tutorials and evaluations;Among the negative aspects, the low teacher/student interaction, the lack of didactics in digital environments, and the deepening of the technological gap were questioned.

Don’t Read the Comments: Examining Social Media Discourse on Trans Athletes

How are transgender athletes understood in popular discourse? This paper adapts and merges Glaser and Strauss’ 1967 Grounded Theory Method with computerized Automated Text Analysis to provide clarity on large-n datasets comprised of social media posts made about transgender athletes. After outlining the procedures of this new approach to social media data, I present findings from a study conducted on comments made in response to YouTube videos reporting transgender athletes. A total of 60,000 comments made on three YouTube videos were scraped for the analysis, which proceeded in two steps. The first was an iterative, grounded analysis of the top 500 “liked” comments to gain insight into the trends that emerged. Automated Text Analysis was then used to explore latent connections amongst the 60,000 comments. This descriptive analysis of thousands of datapoints revealed three dominant ways that people talk about transgender athletes: an attachment to biology as determinative of athletic abilities, a racialized understanding of who constitutes a proper “girl”, and perceptions of sex-segregated sports as the sole way to ensure fairness in athletic opportunities. The paper concludes by drawing out the implications of this research for how scholars understand the obstacles facing transgender political mobilizations, presents strategies for addressing these roadblocks, and underscores the importance of descriptive studies of discourse in political science research concerned with marginalization and inequality.

Desinformación rusa en Europa del Este. Marcos mediáticos de la vacunación en ro.sputnik.md

The news site ro.sputnik.md is the Romanian language version of the Sputnik news website platform, owned by the Russian government, one of the main channels used by the Kremlin to disseminate mis- and disinformation across Russian borders. To map the media frames and the lexical and discursive constructions, the research proposes a mixed methods content-based approach, where automated text analysis (frequency, co-occurrence, n-grams) is combined with thematic and discourse analysis. Six emphasis frames are identified in the corpus (N=1,165): Superiority of the Russian Sputnik V Vaccine, Fatal/Side Effects of EU Authorized Vaccines, Limitations of Individual Rights and Freedoms, EU and/or Romanian Authorities' Struggle, Children and Teenagers' Protection, and Big Pharma Conspiracy. The findings show that specific discursive patterns are associated with the negative news value: death, side effects (blood clot, thrombosis, coagulation), restrictions, and interdictions or warnings (serious, risk, negative, panic, etc.), while the conflict news value is associated with warfare vocabulary (defense, threat, battle, fire, gunpowder, etc.);and eliteness, with well-known actors (state leaders, European leaders, famous "conspirators") and countries (powerful international actors, meaningful neighbours).

Sustainability and Resilience Assessment in the Pandemic Emergency

The coronavirus pandemic has affected all production and service sectors. The crisis has led to the increased awareness of the fragility and vulnerability of our societies and our lifestyles. The speed at which it spreads has led first to a change of our behavior and our way of life but at the same time, it has also changed the way we consider our buildings and how we live in them. Considering the new visions aimed at sharing spaces, such as those of work with the coworking spaces or frontal lessons in the classroom, they must now be rethought. In tourism and catering, relations with customers are reviewed and public transport is radically rethought. In this context, the concept of resilience plays a key role. How resistant are we and our environment to crises like COVID-19? What is the significance of resilience in sustainability? Is there a link between the assessment of resilience and life cycle assessment? In this paper is represented a general framework to connect resilience and sustainability assessments.

La agenda de los grupos de interés frente a la COVID-19: el rastro digital en Twitter

Los grupos de interés están teniendo un rol fundamental en la resolución de los problemas generados por la pandemia COVID-19 porque abordan múltiples asuntos y representan a los grupos sociales en torno a un reto social sin precedentes. Múltiples organizaciones expresan cada día las preocupaciones y demandas de sus constituyentes, forjando de esta manera una agenda en la que se debate cómo hacer frente a los múltiples efectos de la pandemia en diversos sectores. Sin embargo, dada la dinámica de atención propia de la situación de emergencia y la diversidad de los temas, es difícil seguir la diversidad y complejidad de múltiples actores y asuntos. Este artículo describe la agenda de los grupos de interés en España a partir de las publicaciones en la red social Twitter de las 140 organizaciones más activas entre marzo de 2018 y marzo de 2021. Mediante la clasificación automática de texto es posible concluir que la atención agregada por tipos de grupos de interés a los principales asuntos de la agenda varía poco a partir del estallido de la crisis provocada por la pandemia. La atención a las dimensiones sanitarias, sociopolíticas y económicas relacionadas con la COVID-19 sigue patrones similares entre los diferentes tipos de grupos y es transversal a los asuntos de la agenda. Estos resultados demuestran que los grupos de interés continúan ejerciendo sus funciones de representación de intereses sin alterar significativamente su comportamiento en respuesta a la crisis.Alternate : Interest groups are playing a fundamental role in solving the problems generated by the COVID-19 pandemic, as they address multiple issues and represent social groups around an unprecedented social challenge. Multiple organizations express their constituents’concerns and demands every day, thus forging an agenda in which they debate how to deal with the multiple effects of the pandemic in various sectors. However, given the dynamics of attention inherent to the emergency situation and the diversity of issues, it is difficult to follow the diversity and complexity of multiple actors and issues. This article describes the agenda of interest groups in Spain, based on the publications on Twitter by the 140 most active organizations between March 2018 and March 2021. Using automated text classification, it is possible to conclude that the aggregate attention by types of interest groups to the main items on the agenda vary little after the outbreak of the crisis caused by the pandemic. Attention to the health, socio-political and economic dimensions related to COVID-19 follows similar patterns among the different types of groups and is transversal to the issues on the agenda. These results show that interest groups continue to carry out their interest representation function without significantly altering their behaviour in response to the crisis.