preprint2026

- Submitted to arXiv Distribird

Literature-Informed Prior Distribution Design for Bayesian Model Calibration

A multi-agent LangGraph pipeline that turns scientific literature into informative Bayesian priors. Searches Semantic Scholar and OpenAlex in parallel, extracts numerical values, and fits a probability distribution via AIC model selection, improving MCMC sampling efficiency in 80% of cases across 20 parameters in ten scientific domains.

Süli, P. P., Eigner, G., Hollós, R.

With HUN-REN AI1Science

- HUN-REN AI Center

Try the tool Source

accepted2026

- 35th EAEEIE Conference, Ruse, Bulgaria When Graders Drift and Models Hallucinate

Auditing Human Grading with Local LLMs

A GDPR-compliant grading framework that runs four open-weight LLMs against teacher grades on six C# programming exams (159 submissions, 73 students). The best model reaches 75.3% subtask-level exact-match and ICC 0.974 with the human grader; the ensemble flags 122 likely teacher errors, 85% of which are stricter (not more lenient) than the model consensus. Accepted to the 35th Annual Conference of the European Association for Education in Electrical and Information Engineering.

Süli, P. P., Balázs, E., Eigner, G.

With Obuda University

- Biomatics & Applied AI Institute
- Physiological Controls Research Center accepted2026
- IEEE INES 2026, Budapest, Hungary Emergent Health Preferences in Large Language Models

A Time-Trade-Off Analysis Using EQ-5D-3L

Do LLMs hold coherent preferences over health states? We run the standard time-trade-off protocol on five LLMs across all 243 EQ-5D-3L states under 31 demographic personas (165,726 model responses)



and compare to established population norms. The paper characterises model-specific biases and the stability of model-derived utilities under persona conditioning. Accepted to the 30th IEEE Jubilee International Conference on Intelligent Engineering Systems (INES 2026), July 2–4, 2026.

Süli, P. P., Eigner, G., Zrubka, Z.

With Obuda University Research and Innovation Center

- Doctoral School of Applied Informatics in progress2026
- Poster at HAICON26
- In preparation Syxplain

A Neurosymbolic Framework for LLM-Guided Symbolic Regression and Automated Scientific Discovery

A neurosymbolic framework that pairs LLM-guided symbolic regression with automated scientific discovery, closing the loop from data to candidate equation to written explanation. Rather than only steering the search, it interprets the entire Pareto front and grounds every inference in real literature via deep formula research. P. P. Süli and R. Hollós contributed equally and share first authorship.

Süli, P. P., Hollós, R., Szabó, G., Eigner, G.

With HUN-REN AI1Science

- HUN-REN AI Center

Try the tool

in progress2026

- In preparation University Students and the Knowledge Transfer Value Chain

Organised Innovation Spaces and the role of Large Language Models





Explores how university students are reshaping the knowledge transfer value chain inside Organised Innovation Spaces under the influence of LLMs such as ChatGPT. Combines a literature review with an explorative LLM-prompting methodology over 25 student personas spanning geographies, education levels, and fields of study, surfacing usage and attitude patterns alongside concerns about superficial learning, reduced human interaction, and academic integrity.

Buday, V., Süli, P. P., Klofsten, M., Eigner, G., Ge, J., Ding, W., Ni, Q., Wang, F.-Y.

With Obuda University

- Doctoral School of Innovation and Management
- Linköping University
- Centre for Innovation and Entrepreneurship
- CASIA
- State Key Laboratory for Management and Control of Complex Systems

Author· distribird.streamlit.app

github

Distribird

Multi-agent LangGraph pipeline that turns scientific literature into informative Bayesian priors. Searches Semantic Scholar and OpenAlex in parallel, extracts reported numerical values, and fits a probability distribution via AIC model selection. Built with HUN-REN AI1Science.

- Python
- LangGraph
- LLM agents
- Bayesian inference

HUN-REN-AI1Science/Distribird Try the tool

Contributor· 6 merged PRs

github

SpotBugs

Six merged pull requests to the widely-used Java static-analysis tool, including two recent CERT Java detectors with full test cases: SER11-J (prevent overwriting of externalizable objects) and VNA04-J (ensure calls to chained methods are atomic). The atomicity detector connects to the published Acta Cybernetica paper on multithreading atomicity.

- Java
- Static analysis
- CERT Java
- AST visitors

spotbugs/spotbugs

📌 Research (Keylong)
🏢 KenguTech
📍 Keylong