poltextlab_admin

On February 27, 2025, poltextLAB organized a full-day Generative Artificial Intelligence (GenAI) training for researchers of HUN-REN. The professional program, led by Miklós Sebők and Rebeka Kiss, aimed to provide a comprehensive overview of the application of generative artificial intelligence in scientific research, with a particular focus on methodological foundations, ethical considerations, and the integration of GenAI into specific research processes. More than 40 researchers from various HUN-REN research centers and institutes registered for the course. Participants came from institutions such as the Centre for Energy Research, the Centre for Economic and Regional Studies, the Biological Research Centre in Szeged, and the Centre for Social Sciences. Additionally, representatives from various other research institutes were also present, including the Institute for Nuclear Research, the Balaton Limnological Research Institute, the Institute of Earth Physics and Space Science, and the Institute of Experimental Medicine. Due to high demand, all available spots were filled within just a few hours. The full-day training was structured into three main blocks: The first part introduced participants to the fundamentals of generative artificial intelligence, including the functioning and limitations of large language models. The second block focused on the practice of prompt engineering, the selection of appropriate models, and cost comparisons. The final section centered on research applications, covering topic selection, literature analysis, and data analysis, while also addressing ethical and legal aspects such as data protection, intellectual property rights, and proper citation methods. The full training curriculum is available HERE.The photos of the training can be found HERE.

The Hungarian daily newspaper, Népszava interviewed Miklós Sebők, director of poltextLAB, on the global regulation of artificial intelligence (AI). The discussion covered the European Union’s AI Act, the differing approaches of the US and China, as well as the military and energy-related applications of AI. According to Miklós Sebők, regulating AI presents significant global challenges: “The situation is similar to the military use of nuclear energy after World War II.” While the European Union prioritizes ethical and data protection considerations, the United States and China focus more on innovation and economic growth. The full interview is available HERE.

The following interns have joined poltextLAB for the 2024-25 spring semester, participated in the introductory training session led by researcher Anna Takács, and have officially started their work:

On February 27-28, 2025, the workshop “Natural Language Processing with Large Language Models. From Bag of Words to Agents” was held at the University of Hildesheim. The event focused on the latest developments in large language models and natural language processing, providing participants with insights into cutting-edge NLP technologies, including transformers and AI-driven agents. During the workshop, poltextLAB’s AI models were also presented, showcasing innovative solutions for analyzing political and social science texts. The event featured interactive lectures and hands-on exercises, allowing participants to gain firsthand experience in AI-driven text analysis. Workshop materials, including presentations and exercises, are available at the following GitHub repository: https://github.com/TimBMK/LLM-workshop-HI.

The Second Budapest Winter Workshop took place on February 26-27, 2025, in Budapest, at the HUN-REN Centre for Economic and Regional Studies and Corvinus University, organized by ANETI Lab and the Center for Collective Learning (CCL). The event focused on the latest advancements in network science, artificial intelligence, and the social sciences. On the first day of the workshop, Miklós Sebők, head of poltextLAB, gave a presentation about the Babel Machine, titled “Leveraging Open Large Language Models for Multilingual Classification Tasks: The Babel Machine Approach.” His presentation explored the potential applications of open large language models in multilingual text classification tasks.

On February 25, 2025, Gennadii Iakovlev delivered a guest lecture as part of the poltextLAB V-Shift Lendület project. In his presentation, he introduced his research titled “Affective Elite Polarization in European Parliamentary Speeches: A Novel Measurement Approach Using Large Language Models”, which explores the application of large language models in measuring affective polarization in European parliamentary speeches. The event took place at the HUN-REN Centre for Social Sciences, where attendees gained insight into the latest methodological advancements in political speech analysis. The abstract of Gennadii Iakovlev’s research is available HERE.

The Networks and Large Language Models workshop was held on February 25, 2025, in Budapest at the HUN-REN Centre for Social Sciences. The event was jointly organized by poltextLAB and ANETILAB, focusing on the applications of large language models and network analysis in the social sciences. Speakers, including Orsolya Ring, István Üveges, Bálint Jeszenői, Miklós Sebők, Luca Gallo, and Balázs Lengyel, covered topics such as the diffusion of policy narratives using large language models, solving classification tasks with LLMs, and spatial and contagion processes in networks. The workshop aimed to showcase poltextLAB’s research and foster scientific dialogue on the applications of artificial intelligence and network analysis in the social sciences.    

Miklós Sebők participated in the “Trump’s Inauguration and Changes in the Global Order” English-language roundtable discussion, held on Friday, February 21, 2025, in Budapest. The event took place as part of the Hungarian Diplomatic Academy (MDA) and the Central European Lawyers Initiative (CEULI) joint public policy mentorship program, which provides Stipendium Hungaricum scholarship recipients with the opportunity to gain a deeper understanding of global political and legal processes. 

Researchers and data analysts now have access to the ParlLawSpeech dataset, representing a groundbreaking advancement in the analysis of political texts.  The collection includes more than 40,000 legislative bills, 28,000 adopted laws, and 3 million parliamentary speeches from seven countries (Austria, Czech Republic, Germany, Denmark, Spain, Croatia, Hungary) and the European Union. This database is particularly valuable for those studying democracy using modern Text-as-Data, Natural Language Processing (NLP), or Large Language Model (LLM)-based methods.  The dataset was developed within the framework of the Horizon2020 OPTED project, with support from institutions such as GESIS – Leibniz Institute for the Social Sciences, poltextLAB, and the Wissenschaftszentrum Berlin für Sozialforschung (WZB).  The project aims to create new opportunities for a deeper understanding of political discourse and legislative processes. 

A new study by Miklós Sebők, Csaba Molnár, and Anna Takács, titled “Levelling up quantitative legislative studies on Central-Eastern Europe: Introducing the ParlText CEE Database of Speeches, Bills, and Laws”, has been published in Intersections. East European Journal of Society and Politics. The study introduces ParlText CEE, a newly available database covering nearly 1.9 million text vectors and metadata from parliamentary speeches, bills, and laws for Czechia, Hungary, Poland, and Slovakia for the period from 1990–1991 to 2022–2024. The database standardises legislative and speech records, enhancing research accessibility and comparability. This resource offers valuable opportunities for scholars studying legislative processes and political discourse in Central-Eastern Europe. The study is available here.