What is it about computational communication science?

Continuing with political language online, we seek to understand the relevance and divergence of news on the internet. Sounds trivial? Well, unfortunately, it isn't: What is "contemporary" news is decided upon by many rather than a few, it contains journalistically verified messages as well as mis- and disinformation and fake news. Jo(sephine) Lukito (Assistant Professor at the U of Texas at Austin’s School of Journalism and Media) guides us, Emese Domahidi (Professor at TU Ilmenau) and Mario Haim (Professor at LMU Munich), through the exciting and "hybrid" online news environment as well as through her own research investigating particularly the malicious political language within online public spheres. Of course, CCS plays a large role in that too, as Jo is a strong advocate of computational methods and especially of multi-platform research.

It is not very hard to find dispute, also harsh dispute, online. A phenomenon also called digital contention, this raises several questions such as why are controversies more pronounced on the web? Have people turned into a rude mob in recent years or does the web help the quarrelsome to become more present? Also, what does this mean for our research, the theories and methods we apply? On that, Emese Domahidi (Professor at TU Ilmenau) and Mario Haim (Professor at LMU Munich) talk with Christian Baden (Associate Professor at the Department of Communication and Journalism and the Smart Institute at the Hebrew U of Jerusalem) who is not only interested in the topic for his own research but who is also heading the oft-mentioned EU-funded OPINION network (https://www.opinion-network.eu/) that brings together scholars working to automatically detect and extract opinions from unstructued data.

Let's put on your legal suit and join Emese Domahidi (Professor at TU Ilmenau) and Mario Haim (Professor at LMU Munich) welcoming Natali Helberger (Distinguished Professor of Law & Digital Technology, with a special focus on AI at the U of Amsterdam). We talk about the difficulties that come with regulating newly emerging technology. We also talk about all kinds of upcoming EU regulations (such as the Digital Services Act, DSA, the Digital Markets Act, DMA, and the AI Act) and the challenges of these, but also about the differences to other jurisdictional systems. Finally, we put this into perspective of CCS, talking about what will likely change in the new future for researchers (take-home message: a lot!).

In this episode, Emese Domahidi (Professor at TU Ilmenau) and Mario Haim (Professor at LMU Munich) talk to Ágnes Emőke Horvát (Assistant Professor in Communication and Computer Science at Northwestern University where she leads the Lab on Innovation, Networks, and Knowledge, LINK) about what gender biases are, their origins and how prevalent these systematic misrepresantions are. Moving to Computational Communication Science, we then discuss how gender biases (and inequalities, more generally) affect our research, our data, tools, measures, and models. And we tackle the big question how potential routes forward could look like.

How to measure racism in news media is the main question in today's episode. Ahrabhi Kathirgamalingam looks into racist and discriminative language as well as dynamics of racism in some 30 years of German-speaking news media. As that's quite a lot of data, of course Ahrabhi also builds on CCS methods. Yet, in addition to the mere amount of data, coding racism also bears big questions of validity and ethics for coders and annotators -- an issue where CCS might also be able to help. In this episode hosted by Jana Bernhard, Ahrabhi talks us through dictionaries and the many options to construct and validate dictionaries in this area. Her research is part of her PhD project about which she is happily reachable via [email protected]. Also, some results were presented at the 2023 ICA in Toronto. Oh, and if you want to guest or host a future episode, please don't hesitate reaching out to us.

Today's CCS study is about the application and particularly the development of dictionaries to apply to quantitative text analyses. Anke Stoll (together with Lena Wilms and Marc Ziegele in this publication from 2023) developed a dictionary to detect German incivility. She did so through a combination of manual and automated approaches, through classic word lists and word embeddings. Hosted by Emese Domahidi, Anke takes us through her approach, the challenges, and of course the potentials she sees with these kinds of techniques. The journal article was just published in Communication Methods and Measures. Oh, and if you want to guest or host a future episode, please don't hesitate reaching out to us.

In today's episode, Frederic R. Hopp (⁠@Freddy_Hopp) discusses with Emese Domahidi (⁠@MissEsi⁠) and Mario Haim (⁠@DrFollowMario⁠) about morality. What's that, why does it affect our daily lifes and our social cohesion, what does it have to do with media content, and how can it be measured? CCS research offers a wide variety of tools to handle morality but also comes with quite a lot of challenges. Freddy takes us through them and discusses with us how research on morality is also affected by current societal developments.

Let's dive into another CCS study, together with Jana Bernhard. This time, hosted by Mario Haim, Jana talks about her approach to topic modelling through algorithmic embeddings to analyze political communication in Austria from 2012 to 2021. Jana and Mario discuss the potential need for more sophisticated methods, they explain the approach Jana has taken, and they discuss whether it was actually worth it. Her study is currently in preparation for publication but interested listeners can get a sneak peak at https://de.slideshare.net/secret/DDmhlvuMusUFY6. Oh, and if you want to guest or host a future episode, please don't hesitate reaching out to us.

Let's dive into a CCS study, together with Waqas Ejaz. In this episode, hosted by Valerie Hase, Waqas tells us about why and how he used topic modeling (LDA) for analyzing news coverage on climate change in a low-income country such as Pakistan. In that, and apart from data access, Waqas and Valerie discuss the sensitive decision of the appropriate number of topics in topic modeling.

In this first episode of the second season, Fabienne Lind (@FabienneLind) discusses with Emese Domahidi (@MissEsi) and Mario Haim (@DrFollowMario) about the English centrism in academia and how this affects our CCS research. This particularly includes the method of content analysis where we use pre-trained models and/or build on training data that have been affected by a largely western and English-speaking perspective. And we discuss multi-lingual text analysis and the many advantages as well as challenges this approach offers.