Explainable Minesweeper: A Logic-Based Puzzle Generation System Through Constraint Satisfaction Problem-Based Inference

Key Points

• The research aims to create a system that automatically generates solvable minesweeper puzzles using logical inference.
• Utilized constraint satisfaction problem (CSP) for puzzle generation.
• Implemented two stages of logical reasoning: Simple Pattern and Advanced Pattern.
• Conducted simulation tests on small game boards (5×5 to 6×6) to assess success rates.
• Developed a hint system to explain the inference process in natural language.
• Achieved over 99% success rate in generating puzzles on small boards (5×5 to 6×6).
• Identified need for improvement on boards larger than 7×7.
• Confirmed that all generated puzzles can be solved using only logical reasoning.

Abstract