Computationalizing Game Design Knowledge
Building AI/ML-friendly knowledge bases from game ontologies and design pattern datasets. Graph neural networks for knowledge discovery in game design.
Game design knowledge has accumulated across decades of scholarship — from Björk and Holopainen’s gameplay design patterns to formal game ontology frameworks. My central question is whether AI can meaningfully grasp and extend this body of knowledge. To explore this, I build graph-structured knowledge bases from game ontology frameworks and design pattern datasets, then apply graph neural network (GNN) techniques for knowledge discovery.
My paper AEGIS: Authentic Edge Growth In Sparsity for Link Prediction in Edge-Sparse Bipartite Knowledge Graphs (Liu & Tatar, 2025) develops an edge augmentation framework for link prediction in sparse bipartite graphs — directly applicable to the game design pattern network where many design relationships remain sparsely documented.
Complementing this, my systematic literature review Towards Computationally Creative Game Design in HCI (Liu, Cotton, Björk & Tatar, 2026, under review at ACM Games: Research and Practice) surveys 236 papers to map AI-assisted game design research, finding that computational creativity remains significantly underrepresented in the field and arguing for a more integrated, human-centred research direction.