Aleksandar Tomašević
@atomashevic.bsky.social
Computational Social Science.
Assistant professor at University of Novi Sad.
🕸🐍🐧👨🏻💻 #RStats (random ordering).
www.atomasevic.com
Assistant professor at University of Novi Sad.
🕸🐍🐧👨🏻💻 #RStats (random ordering).
www.atomasevic.com
Rough first version of the paper is on ArXiv: arxiv.org/abs/2508.21740
Full paper with robustness analysis up in a few weeks.
Full paper with robustness analysis up in a few weeks.
Operational Validation of Large-Language-Model Agent Social Simulation: Evidence from Voat v/technology
Large Language Models (LLMs) enable generative social simulations that can capture culturally informed, norm-guided interaction on online social platforms. We build a technology community simulation m...
arxiv.org
September 1, 2025 at 12:40 PM
Rough first version of the paper is on ArXiv: arxiv.org/abs/2508.21740
Full paper with robustness analysis up in a few weeks.
Full paper with robustness analysis up in a few weeks.
Fails: Identity persistence, long-term coherence, true preferential attachment. ❌
Memoryless setup prevents identity formation; network core is more difuse, no agent differentiation in stylistic alignment.
Memoryless setup prevents identity formation; network core is more difuse, no agent differentiation in stylistic alignment.
September 1, 2025 at 12:40 PM
Fails: Identity persistence, long-term coherence, true preferential attachment. ❌
Memoryless setup prevents identity formation; network core is more difuse, no agent differentiation in stylistic alignment.
Memoryless setup prevents identity formation; network core is more difuse, no agent differentiation in stylistic alignment.
100% AI generated :)
March 17, 2025 at 4:10 PM
100% AI generated :)
If you want to take a quick look at (subset) of MADOC, Python and R packages are available:
pypi.org/project/pyma...
github.com/atomashevic/...
pypi.org/project/pyma...
github.com/atomashevic/...
pymadoc
Python package to download and combine parts of MADOC dataset
pypi.org
January 25, 2025 at 8:40 AM
If you want to take a quick look at (subset) of MADOC, Python and R packages are available:
pypi.org/project/pyma...
github.com/atomashevic/...
pypi.org/project/pyma...
github.com/atomashevic/...
Dataset is available on Zenodo:
zenodo.org/records/1463...
Preprint describing MADOC is on ArXiv:
arxiv.org/abs/2501.12886
zenodo.org/records/1463...
Preprint describing MADOC is on ArXiv:
arxiv.org/abs/2501.12886
MADOC: Multi-Platform Aggregated Dataset of Online Communities
The Multi-platform Aggregated Dataset of Online Communities (MADOC) is a comprehensive dataset that facilitates computational social science research by providing a unified, standardized dataset for c...
zenodo.org
January 25, 2025 at 8:40 AM
Dataset is available on Zenodo:
zenodo.org/records/1463...
Preprint describing MADOC is on ArXiv:
arxiv.org/abs/2501.12886
zenodo.org/records/1463...
Preprint describing MADOC is on ArXiv:
arxiv.org/abs/2501.12886
But package is primarily built for #Rstats users with limited or zero knowledge of #Python.
The package mimics and simplifies the usage of transformers library in Python, so you can perform inference, run small tasks in a few lines of R code without worrying what is happening in the background.
The package mimics and simplifies the usage of transformers library in Python, so you can perform inference, run small tasks in a few lines of R code without worrying what is happening in the background.
March 28, 2024 at 3:48 PM