Read how Russia and its allies have pushed OSCE/ODIHR into being more lenient. Not the kind of bias they like to complain about. Check our new paper in Democratization (OA). Original corpus data validated on expert assessments by Varieties of Democracy Institute. tandfonline.com/doi/full/10.10…

🚨ICYMI🚨 #Rstats backbone 2.0.0 can now identify significant edges in dense unweighted networks. Here's an example where the communities are obscured by a bunch of random noisy edges, but are clear in the backbone 👇

Election monitoring reports from OSCE/ODIHR correlate w expert assessment of #ElectoralIntegrity. But, there is a strong and positive bias towards Russia + allies, indicating defensive and lenient stances, say Michal Mochtak, A. Drnovsky & C. Lesschaeve tandfonline.com/doi/full/10.10…

How to estimate #China's #influence on the foreign policy of other countries? We suggest one way: using the #UN General Assembly #voting and #rhetoric. We also show that except #Serbia, no #CEE country shows shifts toward China. (OPEN ACCESS) link.springer.com/article/10.100…

#Figures of voting and rhetoric (syntax and semantics) similarity of the #CEE countries towards #China, #US, and #Germany attached here, charts for ALL countries worldwide freely available as #supplementary files (link in the article).

Thanks to co-authors Michal Mochtak and Kamil Liskutin for the collaboration!

🇨🇳Chinese diplomats more positive about the US in 2021, shows MFA press conference analysis In an update to their dataset, Richard Turcsanyi and Michal Mochtak analyse China’s foreign policy discourse in 2021. Check out the whole article: ceias.eu/chinese-diplom…

Looking at shitstorm in the #US-#China relations these days, it may be surprising that in 2021, #Chinese diplomatic #discourse of the US actually #improved after the Trump years. Check out the #update to our #MFA press conferences #dataset, now covering 2002-2021! Michal Mochtak

The CMFA PressCon Corpus reached its first big milestone - 1000 downloads! I think the recently released 2021 update is starting to kick in 👀doi.org/10.7910/DVN/BA… Richard Turcsanyi

Some of the highlights from the Slovene Language Technologies and Digital Humanities (JTDH 2022) conference: Peter Rupnik, Michal Mochtak and Nikola Ljubešić introducing ParlaSent-BCS Dataset of Sentiment-annotated Parliamentary Debates, also available at our repository. #ParlaMint

New article mapping the war discourse in #Bosnian Parliament. Analysis of two decades of #parliamentary debates show the central #conflict exists primarily alongside #Bosniak–#Serb grievances, redefining the three-sided stalemate of Bosnian politics. OA. tandfonline.com/doi/full/10.10…

The update to our #database of #Chinese MFA #press conferences is out, covering now the years 2002-2022! The database is entirely #OpenAccess, with almost 30k questions and answers coded according to various research needs. Michal Mochtak dataverse.harvard.edu/dataset.xhtml?…

New dataset on Hugging Face Hub: 🌐 ParlaSent Dataset! • 📊 Multilingual Sentiment Dataset • 🏛️ Focus: Parliamentary Debates • 🌍 Languages: Czech, English, Serbo-Croatian, Slovak, and Slovene • 📚 Detailed Annotations Explore Dataset here: 🔗 huggingface.co/datasets/class…

A major #update to our #Chinese MFA press conferences #dataset to include #SentimentAnalysis of each question and answer! Besides, another year has been added to cover the 2002-2023 period with more than 30 thousand questions and answers. doi.org/10.7910/DVN/BA… Michal Mochtak

How did #Chinese diplomacy talk about foreign countries in 2023? Based on our #data, the #US recorded the most #negative sentiment, while #developing countries were the most #positive – China’s #neighbours and #developed countries (minus the US) being in between. Michal Mochtak

📢 Exciting Publication in EJPR! 📢 Meet authdetect—a model for detecting authoritarian discourse in political speeches, trained on 77 years of UNGD data. Easily accessible in Google Colab for all researchers! 📊 🔗 ejpr.onlinelibrary.wiley.com/doi/10.1111/14…

🚀 Excited to announce v5 of the CMFA PressCon Corpus by me & Richard Turcsanyi ! 📅 Now covering Oct 2002 - Dec 2024 💬 33,199 Q&A dyads w/ speaker names, named entities, & sentiment. 📊 Ready for deep dives into Chinese diplomacy! 🗂️ doi.org/10.7910/DVN/BA…

New paper out! 📄 ParlaSent: Mapping Sentiment in Political Discourse with LLMs 🇪🇺💬 ✅ Beats sentiment dictionaries ⚡ Rivals GPT-4o in a zero-shot classification 🛠️ Comes with a tutorial — ready to use! 👉 tandfonline.com/doi/full/10.10…

As elections near, do parties fuel democracy or weaken it? My new piece in #Democratization reveals a systematic rise in authoritarian rhetoric: the Authoritarian Election Countdown Effect. Read OA: tinyurl.com/yuaadcjr.

A good colleague @uantwerpen is hiring a Postdoc in Computational Political Science (ERC DEMO-LIES, NLP + disinformation). 📅 Deadline: 16 Oct 2025 👉 uantwerpen.be/en/jobs/vacanc…