<!-- Markdown version of https://atomasevic.xyz/software.html, generated by scripts/generate_structured_cv.rb -->

# Software

## [transforEmotion](https://github.com/atomashevic/transforEmotion)

R package

**transforEmotion** is an R package for multimodal sentiment and emotion analysis using transformer models. Recognizes emotions in text, images, and video. Includes dynamic exploratory graph analysis of emotion time series.

- Text, image, and video emotion recognition
- Zero-shot classification with transformer models
- Dynamic network analysis of emotion dynamics
- Open-source, cross-platform

**[GitHub](https://github.com/atomashevic/transforEmotion), [CRAN](https://cran.r-project.org/package=transforEmotion)**

## [YClient-Reddit](https://github.com/atomashevic/YClient-Reddit)

Python app

**YClient-Reddit** is a fork of YSocial, modified for Reddit-like LLM agent social simulations.  
YSocial builds a digital twin of an online social media platform using Large Language Models (LLMs).

### Key Modifications for Reddit-Like Simulation

- Adapted agent prompts and interaction logic to mimic Reddit-style posting, commenting, and voting
- Modified data structures and simulation parameters to reflect Reddit's community and thread-based organization
- Adjusted recommender systems and feed generation to better align with Reddit's content discovery mechanisms
- Additional configuration options and scripts for Reddit-specific experiments

**[GitHub](https://github.com/atomashevic/YClient-Reddit)**

## [YSocial hybrid human-AI community experiment](https://doi.org/10.5281/zenodo.22964196)

Dataset

Data from the YSocial field experiment on AI identity labels: platform activity, agent memory, agent interviews, and browser telemetry.

**[Zenodo](https://doi.org/10.5281/zenodo.22964196)**

## [YSocial-Reddit-Explorer](https://github.com/atomashevic/YSocial-Reddit-Explorer)

Flask app

Flask dashboard for inspecting YSocial agent-based simulations. Moderate content or users, run what-if scenarios, and track downstream effects.

- Browse threads, users, votes, and interactions
- Apply moderation and observe downstream effects
- Export subsets for analysis

**[GitHub](https://github.com/atomashevic/YSocial-Reddit-Explorer)**

## [MADOC Tools](https://github.com/atomashevic/pymadoc)

Python & R

Helpers for accessing the Multi‑Platform Aggregated Dataset of Online Communities (MADOC).

- Python: [pymadoc](https://github.com/atomashevic/pymadoc) ([PyPI](https://pypi.org/project/pymadoc/))
- R: [rMADOC](https://github.com/atomashevic/rMADOC)

## [Face of Populism](https://github.com/atomashevic/face-of-populism)

ML + datasets

Code and data for analyzing facial emotional expressions of political leaders. Image annotation study and processing pipeline.

**[GitHub](https://github.com/atomashevic/face-of-populism)**