Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games

Accurate and interpretable prediction plays a vital role in natural language processing (NLP) tasks, particularly for enhancing user trust and model transparency. However, existing models often struggle with poor adaptability and limited interpretability when applied to dynamic language prediction t...

Full description

Saved in:
Bibliographic Details
Main Authors: Jun Zhao, Jintian Ji, Robail Yasrab, Shuxin Wang, Liang Yu, Lingzhen Zhao
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:Algorithms
Subjects:
Online Access:https://www.mdpi.com/1999-4893/18/7/427
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850077174125559808
author Jun Zhao
Jintian Ji
Robail Yasrab
Shuxin Wang
Liang Yu
Lingzhen Zhao
author_facet Jun Zhao
Jintian Ji
Robail Yasrab
Shuxin Wang
Liang Yu
Lingzhen Zhao
author_sort Jun Zhao
collection DOAJ
description Accurate and interpretable prediction plays a vital role in natural language processing (NLP) tasks, particularly for enhancing user trust and model transparency. However, existing models often struggle with poor adaptability and limited interpretability when applied to dynamic language prediction tasks such as <b>Wordle</b>. To address these challenges, this study proposes an interpretable reinforcement learning framework based on an Enhanced Deep Deterministic Policy Gradient (Enhanced-DDPG) algorithm. By leveraging a custom simulation environment and integrating key linguistic features word frequency, letter frequency, and repeated letter patterns (rep) the model dynamically predicts the number of attempts needed to solve <b>Wordle</b> puzzles. Experimental results demonstrate that Enhanced-DDPG outperforms traditional methods such as Random Forest Regression (RFR), XGBoost, LightGBM, METRA, and SQIRL in terms of both prediction accuracy (MSE = 0.0134, R<sup>2</sup> = 0.8439) and robustness under noisy conditions. Furthermore, SHapley Additive exPlanations (SHAP) are employed to interpret the model’s decision process, revealing that repeated letter patterns significantly influence low-attempt predictions, while word and letter frequencies are more relevant for higher attempt scenarios. This research highlights the potential of combining interpretable artificial intelligence (I-AI) and reinforcement learning to develop robust, transparent, and high-performance NLP prediction systems for real-world applications.
format Article
id doaj-art-5aa4f781eb9544e2b333e882fd95f04a
institution DOAJ
issn 1999-4893
language English
publishDate 2025-07-01
publisher MDPI AG
record_format Article
series Algorithms
spelling doaj-art-5aa4f781eb9544e2b333e882fd95f04a2025-08-20T02:45:52ZengMDPI AGAlgorithms1999-48932025-07-0118742710.3390/a18070427Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based GamesJun Zhao0Jintian Ji1Robail Yasrab2Shuxin Wang3Liang Yu4Lingzhen Zhao5Department of Public Foundation, Wannan Medical College, Wuhu 241000, ChinaDepartment of Medical Information, Wannan Medical College, Wuhu 241000, ChinaMRC Biostatistics Unit, University of Cambridge, Cambridge CB2 1TN, UKDepartment of Medical Information, Wannan Medical College, Wuhu 241000, ChinaDepartment of Public Foundation, Wannan Medical College, Wuhu 241000, ChinaDepartment of Public Foundation, Wannan Medical College, Wuhu 241000, ChinaAccurate and interpretable prediction plays a vital role in natural language processing (NLP) tasks, particularly for enhancing user trust and model transparency. However, existing models often struggle with poor adaptability and limited interpretability when applied to dynamic language prediction tasks such as <b>Wordle</b>. To address these challenges, this study proposes an interpretable reinforcement learning framework based on an Enhanced Deep Deterministic Policy Gradient (Enhanced-DDPG) algorithm. By leveraging a custom simulation environment and integrating key linguistic features word frequency, letter frequency, and repeated letter patterns (rep) the model dynamically predicts the number of attempts needed to solve <b>Wordle</b> puzzles. Experimental results demonstrate that Enhanced-DDPG outperforms traditional methods such as Random Forest Regression (RFR), XGBoost, LightGBM, METRA, and SQIRL in terms of both prediction accuracy (MSE = 0.0134, R<sup>2</sup> = 0.8439) and robustness under noisy conditions. Furthermore, SHapley Additive exPlanations (SHAP) are employed to interpret the model’s decision process, revealing that repeated letter patterns significantly influence low-attempt predictions, while word and letter frequencies are more relevant for higher attempt scenarios. This research highlights the potential of combining interpretable artificial intelligence (I-AI) and reinforcement learning to develop robust, transparent, and high-performance NLP prediction systems for real-world applications.https://www.mdpi.com/1999-4893/18/7/427interpretable artificial intelligencereinforcement learningnatural language processingenhanced deep deterministic policy gradient<b>Wordle</b> prediction
spellingShingle Jun Zhao
Jintian Ji
Robail Yasrab
Shuxin Wang
Liang Yu
Lingzhen Zhao
Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
Algorithms
interpretable artificial intelligence
reinforcement learning
natural language processing
enhanced deep deterministic policy gradient
<b>Wordle</b> prediction
title Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
title_full Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
title_fullStr Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
title_full_unstemmed Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
title_short Interpretable Reinforcement Learning for Sequential Strategy Prediction in Language-Based Games
title_sort interpretable reinforcement learning for sequential strategy prediction in language based games
topic interpretable artificial intelligence
reinforcement learning
natural language processing
enhanced deep deterministic policy gradient
<b>Wordle</b> prediction
url https://www.mdpi.com/1999-4893/18/7/427
work_keys_str_mv AT junzhao interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames
AT jintianji interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames
AT robailyasrab interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames
AT shuxinwang interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames
AT liangyu interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames
AT lingzhenzhao interpretablereinforcementlearningforsequentialstrategypredictioninlanguagebasedgames