BiLSTM and dynamic fuzzy AHP-GA method for procedural game level generation

Area of Science:

• Artificial Intelligence
• Game Design
• Educational Technology

Background:

• The COVID-19 pandemic accelerated the adoption of game-based learning, increasing demand for game content.
• Generating game content and levels is traditionally time-consuming and expensive, requiring improvements in efficiency and cost-effectiveness.
• Procedural Content Generation (PCG) offers automated solutions for creating game levels, leveraging artificial intelligence techniques.

Purpose of the Study:

• To develop and evaluate a hybrid AI system for automated procedural game level generation.
• To address the need for efficient, cost-effective, and high-quality game content creation in educational settings.
• To compare the performance of the proposed hybrid method against existing multi-criteria decision-making (MCDM) approaches.

Main Methods:

• A hybrid system combining Bidirectional Long Short-Term Memory (BiLSTM) and Fuzzy Analytic Hierarchy Process-Genetic Algorithm (FAHP-GA) was developed.
• The proposed system was implemented in a custom-developed educational game to generate game levels.
• Performance evaluation involved comparisons with other MCDM methods and statistical analyses.

Main Results:

• The BiLSTM-based FAHP-GA hybrid method demonstrated effectiveness in procedural game level generation.
• The developed system successfully created game levels for the educational game case study.
• Statistical analyses confirmed the viability and performance of the proposed approach.

Conclusions:

• The BiLSTM-based FAHP-GA hybrid system is a viable and effective solution for procedural game level generation.
• This AI-driven approach offers a promising method to reduce the time and cost associated with game content creation.
• The study highlights the potential of hybrid AI techniques in enhancing educational game development.