Abstract:
This study introduces an innovative Artificial Intelligence (AI)-based system
designed to address the challenges encountered in Sri Lankan school-level chess
tournaments, which are often overseen by teachers or non-FIDE titled arbiters.
The primary objective of this system is to automate the detection of threefold
repetitions, illegal moves, and missed moves, thereby enhancing decisionmaking
accuracy and promoting fairness in tournament play. To achieve this, the
system utilizes custom rule-based algorithms that achieve 100% accuracy in
detecting illegal moves and threefold repetition, alongside a trained
Convolutional Neural Network (CNN) developed from a dataset of 176,998
unique chess positions across 10,000 games to predict missing or incorrect
moves in the sequence, facilitating effective record sheet correction. Arbiters
can input move sequences using algebraic notation, enabling efficient
processing by the system. The system significantly reduces the time required for
arbiters to adjudicate complex situations, minimizing decision-making errors
that could impact tournament outcomes. Furthermore, the system cultivates a
respectful and rule-abiding competitive environment, contributing to equitable
decision-making in tournaments. Limitations of the research include the reliance
on a specific dataset for move prediction, which may affect the generalizability
of results. The implementation of Optical Character Recognition (OCR) for
scanning record sheets could further enhance efficiency compared to manual
entry. Additionally, sufficient hardware performance is required to operate the
CNN effectively, particularly in offline contexts. Future research may explore
the integration of this technology in resource-constrained environments and the
development of low-cost alternatives to improve accessibility and
implementation. In conclusion, this AI-driven system enhances the accuracy,
fairness, and organization of chess tournaments, alleviating the burden on
arbiters and leading to improved competitive outcomes. The findings hold
potential for broader implications in chess tournament management, ensuring
more reliable and efficient decision-making processes.
