← Projects

Engineering case study / 2023–2025

Metaverse behavior & synthetic data

Connecting behavior definitions to avatar-motion data.

Behavior definition · Survey · Motion & data production

Behavior data production overview
Research overview

Research question

Studying abusive behavior in a virtual world starts with deciding what a behavior means and how it can be represented. A motion sequence alone does not provide that definition. The dataset needs a connection between the behavior being studied and the examples used to represent it.

My contribution

I contributed to behavior definitions, a participant survey, and synthetic-data production. My work involved motion capture, motion generation, Blender, and Unity Perception within the research team's broader workflow.

From a concept to usable data

This work connected three layers: defining the behavior, gathering participant perspectives, and producing avatar-motion examples. My contribution crossed these layers rather than focusing only on the final model.

Motion capture and generated motion provided material for synthetic scenes. Blender and Unity Perception supported the production workflow. Keeping the intended behavior connected to the produced example was the central concern across these tools.

Outcome and takeaway

The research contributed to a conference presentation on constructing a dataset for abusive behavior detection in the metaverse. My contribution focused on behavior definitions and data production.

The takeaway is that dataset construction is part of the research argument. The categories, the way examples are produced, and the meaning assigned to them need to remain connected.