Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Label-Free Subjective Player Experience Modelling via Let's Play Videos

Dave Goel, Athar Mahmoudi-Nejad, Matthew Guzdial

TL;DR

This work tackles PEM without annotated ground truth by mining Let's Play videos to infer affect from players’ vocal cues and training a CNN on frame sequences. It validates the approach through a human-subject experiment using an Angry Birds clone with physiological sensors and post-hoc surveys, showing the model tracks self-reported affect and aligns with some EDA signals while avoiding heavy annotation. The contributions include a full annotation-free pipeline, a dataset of Let's Play Angry Birds content, and an empirical evaluation that supports the feasibility of scalable PEM development for game design. The study highlights potential for rapid PEM deployment and AI-directed adaptation, while noting sensor-related challenges and the need for broader validation across games.

Abstract

Player Experience Modelling (PEM) is the study of AI techniques applied to modelling a player's experience within a video game. PEM development can be labour-intensive, requiring expert hand-authoring or specialized data collection. In this work, we propose a novel PEM development approach, approximating player experience from gameplay video. We evaluate this approach predicting affect in the game Angry Birds via a human subject study. We validate that our PEM can strongly correlate with self-reported and sensor measures of affect, demonstrating the potential of this approach.

Label-Free Subjective Player Experience Modelling via Let's Play Videos

TL;DR

This work tackles PEM without annotated ground truth by mining Let's Play videos to infer affect from players’ vocal cues and training a CNN on frame sequences. It validates the approach through a human-subject experiment using an Angry Birds clone with physiological sensors and post-hoc surveys, showing the model tracks self-reported affect and aligns with some EDA signals while avoiding heavy annotation. The contributions include a full annotation-free pipeline, a dataset of Let's Play Angry Birds content, and an empirical evaluation that supports the feasibility of scalable PEM development for game design. The study highlights potential for rapid PEM deployment and AI-directed adaptation, while noting sensor-related challenges and the need for broader validation across games.

Abstract

Player Experience Modelling (PEM) is the study of AI techniques applied to modelling a player's experience within a video game. PEM development can be labour-intensive, requiring expert hand-authoring or specialized data collection. In this work, we propose a novel PEM development approach, approximating player experience from gameplay video. We evaluate this approach predicting affect in the game Angry Birds via a human subject study. We validate that our PEM can strongly correlate with self-reported and sensor measures of affect, demonstrating the potential of this approach.
Paper Structure (17 sections, 3 figures, 3 tables)

This paper contains 17 sections, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. System Overview
  4. Let's Play Videos
  5. Postprocessing
  6. Model
  7. Human Subject Study Methodology
  8. Sensors
  9. Gameplay
  10. Survey
  11. Results
  12. Discussion
  13. Limitations
  14. Future Work
  15. Conclusions
  16. ...and 2 more sections

Figures (3)

  • Figure 1: System Overview of our Player Experience Model pipeline
  • Figure 2: Screenshots of levels 1, 2 and 3 from the human study
  • Figure 3: Scatterplot depicting the model output and EDA values for User 8 for Level 1, 2 and 3. The blue points represent the EDA output while the red points represent the model output.