<![CDATA[This study aims to develop the Try Calligraphy serious game equipped with a personalized recommendation system to assist players in selecting the most suitable Arabic calligraphy style (khat) based on their performance. The primary objective of this research is to optimize learning personalization by implementing a User-Based Collaborative Filtering approach that predicts the most appropriate handwriting styles for new players based on similarity to prior users. Performance data consisting of final scores generated from decoration, neatness, and completion time are recorded and compared to construct player similarity profiles. The system calculates predicted scores for untested calligraphy styles using cosine similarity and subsequently recommends the top three styles with the highest estimated performance potential. Two experimental scenarios were conducted to assess predictive performance. The results show Mean Absolute Error (MAE) values of 16.08 and 13.92, indicating a moderate level of accuracy. These findings suggest that while the system is capable of providing relevant and targeted recommendations, additional training data and improved similarity parameter design can further enhance predictive precision. Usability evaluation using the GUESS-18 instrument involved ten respondents and produced average scores above 3.7 across all constructs, reflecting positive user perceptions in terms of usability, aesthetics, enjoyment, and personal engagement. Overall, the system demonstrates that the integration of User-Based Collaborative Filtering in a serious game environment can enhance personalized learning, increase user involvement, and support the digital preservation and education of Islamic calligraphy art.]]>
