Dynamic visualization of MCTS-Tree construction in the GGP Base Package

The paper presents a dynamic visualization system for the search tree of the Monte Carlo Tree Search (MCTS) algorithm implemented in the General Game Playing Base Package (GGP Base Package). The main limitation of existing approaches is the lack of tools for observing the evolution of the search tree during its construction, which complicates debugging and analysis of MCTS agents’ behaviour. The methodology includes: a systematic evaluation of eight visualization tools against 12 criteria (universality, dynamism, interactivity, scalability, performance, etc.); the design of a four‑layer architecture (Java/GGP Base Package → Redis → ASP.NET Core → React + D3.js); and the implementation of an interactive mechanism for replaying tree evolution with step‑by‑step analysis. The comparative evaluation shows that the proposed system achieves an integral score of 0.752 according to the defined criteria and, among the considered tools, offers the most favourable combination of key properties relevant for debugging MCTS agents. Experimental studies on the ConnectFour (6×8) game demonstrate that the system provides smooth visualization (>50 FPS) for MCTS trees with up to 1000 nodes, supports arbitrary games in GDL format and, according to expert judgment, enables differences in the behaviour of MCTS algorithm modifications to be identified within seconds of visual inspection, whereas without visualization this would require time‑consuming manual comparison of final statistics. The results confirm that dynamic visualization of intermediate MCTS tree states offers additional opportunities for detecting hidden implementation defects and non‑trivial behavioural patterns of the MCTS algorithm that remain unobvious when only the final tree state is analysed. The proposed tool may be of practical interest to researchers, developers and educators in the field of General Game Playing.

1. Browne C.B., Powley E., Whitehouse D., et al. A survey of Monte Carlo tree search methods. IEEE Transactions on Computational Intelligence and AI in Games. 2012;4(1):1–43. https://doi.org/10.1109/TCIAIG.2012.2186810

2. Świechowski M., Godlewski K., Sawicki B., et al. Monte Carlo tree search: a review of recent modifications and applications. Artificial Intelligence Review. 2022;56(5):2497–2562. https://doi.org/10.1007/s10462-022-10228-y

3. Genesereth M., Thielscher M. General game playing. Cham: Springer; 2014. 213 p.

4. Scheiermann J., Konen W. AlphaZero‑inspired game learning: faster training by using MCTS only at test time. IEEE Transactions on Games. 2023;15(4):637–647. https://doi.org/10.1109/TG.2022.3206733

5. Trudeau A., Bowling M. Targeted search control in AlphaZero for effective policy improvement. In: AAMAS '23: Proceedings of the 2023 International Conference on Autonomous Agents and Multiagent Systems, 29 May – 02 June 2023, London, United Kingdom. New York: ACM; 2023. P. 842–850.

6. Li R., Mo A., Su G., et al. AlphaZero‑Edu: Democratizing Access to AlphaZero. arXiv. URL: https://doi.org/10.48550/arXiv.2504.14636 [Accessed 20th March 2026].

7. Zhao Y., Hu Ch., Liu J. Playing with Monte‑Carlo Tree Search [AI‑eXplained]. IEEE Computational Intelligence Magazine. 2024;19(1):85–86. https://doi.org/10.1109/MCI.2023.3328150

8. Spinner Th., Kehlbeck R., Sevastjanova R., et al. -generAItor: tree‑in‑the‑loop text generation for language model explainability and adaptation. ACM Transactions on Interactive Intelligent Systems. 2024;14(2):14. https://doi.org/10.1145/3652028

9. Guerra‑Gómez J.A., Buck‑Coleman A., Plaisant C., et al. TreeVersity: Comparing tree structures by topology and node's attributes differences. In: 2011 IEEE Conference on Visual Analytics Science and Technology (VAST), 23–28 October 2011, Providence, RI, USA. IEEE; 2011. P. 275–276. https://doi.org/10.1109/VAST.2011.6102471

10. Taylor H., Stella L. An evolutionary framework for Connect‑4 as test‑bed for comparison of advanced Minimax, Q‑learning and MCTS. arXiv. URL: https://doi.org/10.48550/arXiv.2405.16595 [Accessed 20th March 2026].

11. Ala'anzy M.A., Madiyarova A., Aigeldiyev A., et al. Connect‑4 AI: A comprehensive taxonomy and critical review of methods and metrics. Symmetry. 2026;18(2):293. https://doi.org/10.3390/sym18020293