News

26 September, 2024 – Successful participation in the SLICES-SC Project

The poltextLAB team’s Babel Machine project has been highlighted as an open call success story among the SLICES-SC Open Call winners, who have benefitted from the SLICES infrastructure. Our participation in the SLICES-SC project lasted from April 2023 to August 2024, during which the team was able to use the imec research group iLab.t’s GPULab infrastructure. The GPULab is a testbed with 125+ GPUs with over 570.000+ cuda cores and 1.8TB+ GPU RAM for AI research and every research that needs GPUs. The flyer detailing the success story can be downloaded here: https://slices-sc.eu/an-open-call-success-story-i-the-babel-machine/

29 July, 2024 – Miklós Sebők among NRDI NREP Excellence programme winners

Miklós Sebők has been chosen as one of the winners of the first tier of the Excellence programme, belonging to the National Research, Development and Innovation Office’s National Research Excellence Programme. The “BABELGLOB: A State-of-the-Art AI Tool for the Global Analysis of Policy Frames” project has received 50 million HUF in funding. Further reading (in Hungarian): https://nkfih.gov.hu/hivatalrol/sajtokozlemenyek/ket-kutato-nyert

26 June, 2024 – Miklós Sebők presents at the the 4th Strategic Workshop of the HUN-REN Hungarian Research Network

On June 26, 2024, Miklós Sebők gave a presentation entitled “No-code AI text analytics at your fingertips: research time travels with the Babel Machine” at the 4th Strategic Workshop of the HUN-REN Hungarian Research Network. In the presentation he explained how the Babel Machine, developed by poltextLAB for free academic use, can replace time-consuming traditional content analysis. With this development, the research group aims to democratise access to artificial intelligence. The presentation is available at: https://www.youtube.com/watch?v=PwfIgXQU_Yg

26 June, 2024 – István Üveges presents at the MILAB RAG meetup

On June 26, 2024, István Üveges participated in the meetup organized jointly by the Artificial Intelligence National Laboratory (MILAB) and the Budapest Deep Learning Reading Seminar on the topic of Retreival Augmented Generation (RAG). Üveges’ presentation was on the application of RAG in social science, which he demonstrated through legal texts. The programme of the meetup and István Üveges’ presentation slides are available here: https://mi.nemzetilabor.hu/hu/esemenyek/budapest-deep-learning-reading-seminarral-kozos-meetup-temaja-retreival-augmented.

20 June, 2024 – Rebeka Kiss’s presentation on the quality of legislation

On June 20, 2024, Rebeka Kiss attended the Tag der Parlamentsforschung – Day of Parliamentary Research conference organised by the Österreichisches Parlament in Vienna. Rebeka Kiss presented co-authored work with Miklós Sebők on  “The concept of tailor-made laws and legislative backsliding in Central-Eastern Europe.” The programme of the conference is available here: https://www.parlament.gv.at/dokument/fachinfos/parlamentsforschung/Day-of-Parliamentary-Research-2024-Abstracts.pdf.

18 June, 2024 – Sean Theriault’s lecture on the policy agendas of the Holy See as a global power

On June 18, 2024, Sean Theriault,  guest speaker of the VSHIFT Momentum project at poltextLAB, gave a lecture at the HUN-REN Centre for Social Sciences Institute for Political Science with the title “The Policy Agendas of a Global Power: The Case of the Holy See.” Theriault is a professor at the University of Texas; his expertise include the U.S. Congress, the presidency, political parties, elections, and party ploarization. He is the author of five books and numerous articles in a variety of journals on subjects ranging from presidential rhetoric to congressional careers and the Louisiana Purchase to the Pendleton Act of 1883.The abstract of his lecture is available here: https://politikatudomany.tk.hu/esemeny/2024/05/speaker-series-sean-theriault.

11 June, 2024 – New publication on the Babel Machine by Miklós Sebők, Ákos Máté, Orsolya Ring, Viktor Kovács and Richárd Lehoczki in Social Science Computer Review

On June 11, 2024, a new article has been published by Miklós Sebők, Ákos Máté, Orsolya Ring, Viktor Kovács and Richárd Lehoczki  in the Social Science Computer Review. The title of the article is “Leveraging Open Large Language Models for Multilingual Policy Topic Classification: The Babel Machine Approach.” The publication is available here: https://journals.sagepub.com/doi/10.1177/08944393241259434. 

Call for Proposals: First MethodsNET Conference

Open Large Language Models in Comparative Research When: Thursday 31 October – Friday 1 November 2024Where: Louvain-la-Neuve, BelgiumOrganisers: Orsolya Ring – Miklós Sebők, HUN-REN Centre for Social Sciences, Budapest, Artificial Intelligence National Laboratory, CSS, Budapest  With this Call for Proposals, we solicit papers for a panel on Open Large Language Models in Comparative Political Research at the First MethodsNET Conference at UCLouvain. The conference will take place between 31 October and 1 November 2024. The proposed panel is focused on the use of open large language models (LLMs, such as BERT, XLM-RoBERTa etc.) in comparative political and communication research. While our primary interest is not in applications based on proprietary models (such as GPT4), we encourage submissions on the comparison of the performance of open vs. closed models on various tasks.  We are particularly interested in papers addressing the following problems: Optimal utilisation of existing LLMs: Papers showcasing the effective utilisation of current LLMs in research endeavours, including classification tasks, along with discussions on their potential benefits and challenges. Advancements in LLM technology relevant to social science: Papers about new technical developments in LLMs that are significant to the social science community. The advantages and disadvantages of using open fine-tuned LLMs vs. proprietary models (such as GPT4) in research, with particular attention to validation and replicability issues.  Original-Translated-Multilingual: the usage of LLMs in the social sciences, either in the original language or with data translated into English or multilingual.  We expect four papers to be selected, subject to the decision of the conference organisers. The final papers need to be uploaded by 9 October. Please note that participation is self-funded. The conference has a standalone fee of €220. However, the price is reduced if you become a MethodsNET member. Please see more information here. How to submit? Please submit your proposal on this form by 17 June 2024. Should you have any questions, please contact us at ring.orsolya[at]tk.hu and sebok.miklos[at]tk.hu. 

2-4 May, 2024 – poltextLAB at the COMPTEXT 2024 Conference

Members of poltextLAB attended the Sixth International Interdisciplinary Conference on the Quantitative and Computational Analysis of Textual Data, held in Amsterdam between 2-4 May, 2024. At the COMPTEXT conference, poltextLAB’s research was showcased in the following presentations: Miklós Sebők, Levente Pakot, Orsolya Ring, Csaba Molnár, Ákos Holányi:Drifting towards the East? An AI-supported analysis of the sentiment of Central-Eastern European parliaments towards great powers. Áron Buzogány, Miklós Sebők, Melinda Manczinger:Evoking and Contesting Expertise in Parliaments: An AI-supported Analysis of Climate Change Debates in Seven Countries. István Üveges, Orsolya Ring, Gabriella Szabó:Emotions over time. Are political debates more passionate than before? Orsolya Ring, László Kiss:Uncovering disinformation: analyzing the thematic patterns and emotional content of Hungarian fake news portals during the COVID-19 pandemic and the Ukrainian war. You can find the full conference program here.

20 April, 2024 – New publication on the AI analysis of American presidential speeches by Miklós Sebők and co-authors in Policy Studies Journal

On April 20, 2024, a new article has been published by Miklós Sebők (co-authored by  Amnon Cavari and Akos Mate) in the Policy Studies Journal. The title of the article is “Staying on the democratic script? A deep learning analysis of the speechmaking of U.S. presidents.” The publication is available here: https://doi.org/10.1111/psj.12534.