Adaptive user modelling in car racing games using behavioural and physiological data.

Area of Science:

• Human-Computer Interaction
• Game Design
• Affective Computing

Background:

• Personalized content adaptation in video games can significantly boost user engagement.
• Procedural content generation (PCG) can create tailored game experiences, enhancing player motivation and immersion.
• Adaptive user models are crucial for dynamically adjusting game content to individual player skills and preferences.

Purpose of the Study:

• To propose an adaptive user modeling approach for personalized video game content adaptation.
• To utilize unobtrusive physiological data and game actions to identify player strengths and weaknesses in car racing games.
• To automatically generate tailored game content, specifically car racing tracks, to improve player experience and engagement.

Main Methods:

• Employing an adaptive user modeling approach based on Trace Theory.
• Utilizing machine learning to extract features from physiological data and in-game actions, clustering them into low-level primitives.
• Transforming primitives into higher-level abstractions (experience, exploration, attention) to inform track alteration decisions.

Main Results:

• The system successfully created personalized car racing tracks based on player data.
• Collected data from 52 users validated that the model provides statistically significant, personalized decisions.
• Tailored content variations correlated with user satisfaction, demonstrating effective feedback incorporation.

Conclusions:

• The proposed adaptive user model effectively personalizes game content using physiological and behavioral data.
• This approach enhances player engagement and driving experience in car racing games.
• The system demonstrates the capability to automatically incorporate user feedback into procedural content generation for improved gameplay.