The Undergraduate Games Corpus: A Dataset for Machine Perception of Interactive Media

• DOI: 10.1609/aaai.v35i1.16071
• Publication Date: 2022-09-08T17:53:55Z
• Authors: Barrett R. Anderson; Adam M. Smith
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Machine perception research primarily focuses on processing static inputs (e.g. images and texts). We are interested in machine of interactive media (such as games, apps, complex web applications) where audience choices have long-term implications for the experience. While there is ample AI methods task playing games (often just one game at a time), this work difficult to apply new in-development or use non-playing tasks such similarity-based retrieval authoring assistance. In response, we contribute corpus 755 structured metadata, spread across several platforms (Twine, Bitsy, Construct, Godot), with full source assets available appropriately licensed redistribution research. Because these were sourced from student projects an undergraduate development program, they reference timely themes their content represent variety levels design polish rather than only representing past commercial successes. This could accelerate understanding while anchoring that freshly-developed intended legitimate human experiences (rather lab-created testbeds). validate utility by setting up novel predicting tags relevant player experience code, showing representations better exploit structure outperform text-only baseline.