TL;DR
This paper introduces a novel framework for developing serious educational games (SEGs) that streamlines the process of integrating learning materials into game content. By leveraging existing educational resources and procedural content generation, the authors demonstrate a method to reduce development costs and effort, as showcased in their case study, "Chem Dungeon."
What Problem Does This Paper Address?
Developing effective Serious Educational Games (SEGs) presents a significant challenge: seamlessly integrating learning materials into engaging game content. Traditional approaches often require extensive collaboration between education experts and game developers, leading to laborious, time-consuming, and costly development cycles. This bottleneck limits the scalability and widespread adoption of SEGs. The core problem is how to efficiently map "knowledge spaces" (learning materials) to "game content spaces" (playable game elements) without relying heavily on expert intervention.
Methods
The authors propose an alternative SEG development framework that addresses the mapping challenge by embedding annotated knowledge chunks into categorized game content. Key aspects of their methodology include:
- Knowledge Space Annotation: Learning materials are annotated using information from reliable educational resources (like syllabi) and by measuring content representation specifications (e.g., text length, image size). This minimizes the need for subjective expert input.
- Procedural Content Generation (PCG): The framework advocates for PCG to generate game content, which significantly lowers development costs and allows for flexible control over game elements.
- Categorization of Game Content: Game content is categorized based on difficulty (Easy, Medium, Hard) and similarity. This is achieved through a combination of rule-based approaches and clustering analysis (specifically, BIRCH algorithm).
- Rule-Based Mapping: A crucial step involves defining rules to map learning materials to game content clusters. The strategy prioritizes the number of games within a cluster and the variation of game content parameters to ensure appropriate deployment and prevent player boredom.
- SEG Game Engine: A game engine is designed to dynamically select and generate game content based on the player's mastery level and the specific learning material, creating a personalized experience.
Datasets
The paper does not explicitly mention or introduce new datasets for training. Instead, it relies on:
- Existing Educational Resources: Syllabi and other documented resources are used to annotate learning materials.
- Periodic Table Data: Used for chemical compound information in the "Chem Dungeon" case study.
- Existing Entertainment Game: The "Chem Fight" game (open-sourced) served as the basis for the "Chem Dungeon" SEG.
- Generated Game Content: The framework generates a large space of game content configurations (48,600 in the case study) based on defined parameters.
Key Insights and Ideas
- Decoupling Knowledge and Content: Separating learning materials from game content and then establishing a systematic mapping process is key to efficient SEG development.
- Leveraging Existing Resources: Utilizing readily available educational resources for annotation significantly reduces the reliance on expensive expert input.
- PCG as a Scalability Enabler: Procedural content generation is vital for creating a large and varied game content space efficiently, making SEG development more scalable.
- Categorization for Better Mapping: Difficulty and similarity categorization of game content provides a structured approach for mapping learning materials, ensuring appropriate challenges and variety.
- Player Experience Matters: The framework considers player experience by aiming to prevent boredom through varied game content and by adapting game difficulty to player mastery.
Key Results
The paper presents a case study, "Chem Dungeon," to demonstrate the effectiveness of their framework.
- User Survey: A user survey with 50 participants indicated that the SEG was generally enjoyable, with 65.2% reporting the games as "FUN."
- Learning Performance: Statistical analysis showed that the game contributed to improved learning outcomes, with 69.4% of players showing improved learning performance.
- Correlation: Gaming activities were found to correlate with learning outcomes, with learning sessions involving more activities.
- Qualitative Feedback: Players reported that the game content and rules elicited positive affective experiences and enjoyment.
Limitations and Future Work
- Case Study Specificity: The current case study is based on "Chem Fight," which was already designed with educational purposes in mind. The authors acknowledge that the framework's applicability to SEGs with different learning types (beyond recalling chemical compounds) needs further investigation.
- Adaptation for Different Learning Types: The framework's ability to adapt to various learning objectives and knowledge types requires further exploration.
- Personalized Adaptation: Future work includes developing a system to predict player experiences using gaming data to enable personalized adaptation of SEGs.
- Broader Application: The authors plan to investigate their approach with other PCG-based entertainment game platforms and diverse learning materials.
Summary updated