Very happy to see the text-as-data reading group series kicking off for the winter term again🥳- the line-up sounds amazing! 🤓

.@sixtespana is very capable of English when it comes to chargeable additional arrangements to your contract. Suddenly, you only receive emails in Spanish. Online translation: we are held responsible for a damage that we haven't caused. Akward fraudulent schemes by @sixtespana

Finally, we learn about setting up a supervised machine learning tool for stance classification from @CV_ChristinaVi, @devnull90, Marcus Maurer, Olivier Quiring, & @IGurevych https://t.co/FF3Wu4H6F2

(2) Does media trust matter in the link between using different news sources and opinion polarization? It does ☝️ Key finding of our study: generalized media trust acts as a shield against all too strong opinion effects.

Letzte Woche gab's die aktuellen Ergebnisse zum Medienvertrauen in Deutschland in 2022. Wer wissen will, 'was davor geschah ' ➡️➡️➡️ die gesammelten Befunde über mehrere Jahre Medienvertrauensforschung

Couldn't ask for more: thank you so much for welcoming me so warmly @kkvk7 @FrauLena2006 & Simone. I am looking forward to spending the summer term in beautiful Düsseldorf as an interim professor for communication science @HHU_de .

Bei den Superlativen zur Beschreibung des #Warnstreik von #verdi #EVG haben sich viele Medien heute gegenseitig übertroffen. Aber welche Folgen hat das hinter verschlossenen Türen? Meine 2cents zur Bedeutung von Medien in Tarifkonflikten bei @DLFmedien

Happy to share a new publication in @djeditorialteam about what constitutes #news from an #audience or lay perspective. 50 free online copies are available at: https://t.co/SHJ4fmDJNW

326 Verhandlungsexpert:innen in der #Tarifpolitk haben @Oliver_Quiring @CV_ChristinaVi @LeneSchaaf befragt: Gewerkschaften und Arbeitgeber sahen Medien zwar weder als Chance noch als Risiko, vermuten aber Einflüsse etwa auf Arbeitsklima und Ausgang der #Tarifverhandlung

Free copies available here: https://t.co/WLHMyDQyL1 Give me a shout, if you want to learn more about our work👐

5/6 🏋️Different configurations of training data are consequential, too: above a minimum threshold, we find partial compensation of quality with quantity. We tried to further enhance performance by relying on existing labelled training data. Yet, transfer was difficult...

4/6🗣️🇩🇪🇬🇧🐦Performance gains are not only to expect with appropriate language input (in our Case: German), but especially after pre-training with Tweets on Covid-19 to learn type of text characteristics.

3/6 🖥️We compare models relying on SVM, CNN, and BERT algorithms. In line with research from NLP : BERT outperforms SVM and CNN by far. But BERT can also yield different perfomance results depending on pre-training.

2/6 🎯Setting up a SML classifier to identify stances on Covid-19 governmental measures in Tweets. 🛠️ we set up different classifiers in a comparative fashion. The classifiers represent different algorithms & different constellations of training data. Here is what we found...

More than happy to see this interdisciplinary work \w @devnull90 & others published! #sml #BERT #stanceclassification #ccs Are you thinking about setting up a supervised machine learning classifier to analyze communication texts? Our work provides plenty of advise... a🧵1/6

Für heute haben @_verdi und @djv zum #Warnstreik beim #ÖRR aufgerufen. Mit @CV_ChristinaVi habe ich mir angesehen, wie Journalist*innen eigene (vs andere) Tarifkonflikte darstellen. Spoiler: Sie berichten pos. über Arbeitskampf & Forderungen und kritischer über Arbeitgeber*innen

Appropriate stage for @OleKelm presenting his study on political consumerism. Congrats to this interesting decomposition of the concept and the robust modelling of the influencing factors. It is 'not only about shopping decisions' #ecrea22

Some of the first results from our large scale survey of negotiators in German trade conflicts. How do they perceive the role of the media and how do they incorporate both, their public and their bargaining activities? Watch out for our presentations at #ecrea22 @ifp_mainz

Vielen Dank für die spannenden Diskussionen und das Interesse an unserem(/w @gei2s) Auswertungsansatz #CoWaD für Nachrichtenwellen im der Krisenberichterstattung. Tolles neues Format - die Tool Demo - auf der #methoden22.

Spannender Impuls von @ankekat1000 bei der #methoden22 zu den Aussichten der Kommunikationswissenschaft auf automatisierte Inhaltsanalysen.