Data mining-driven communication strategy optimization model for parent-teacher social platforms

Abstract

Effective communication between parents and teachers is essential for student success; however, challenges persist in maintaining timely, personalized, and responsive interactions through digital social platforms. This research proposes a data mining–driven communication strategy optimization model utilizing deep learning (DL) techniques to enhance message relevance, engagement, and sentiment alignment in parent-teacher communication. Data were collected over 6 months from parent-teacher social platforms such as ClassDojo and Bloomz, used by various schools, comprising 67,800 communication threads (messages, timestamps, sentiment tags, response times, and read receipts). The raw data underwent preprocessing steps, including text cleaning and normalization. A White Shark Optimizer–driven Highway Long Short-Term Memory (WSO-HWLSTM) model is introduced to learn temporal patterns in sequential conversations and predict optimal message formats (in terms of length, tone, and timing) to enhance engagement. The White Shark Optimizer (WSO) is employed to select features and fine-tune the critical hyperparameters of the HWLSTM model for optimal performance. Results show that the WSO-HWLSTM model outperformed traditional algorithms for predicting high-engagement responses, achieving performance metrics consistently above 96% across all evaluation folds. The findings demonstrate that applying intelligent data mining and DL techniques to digital educational communication can yield personalized, actionable recommendations, thereby fostering stronger parent-teacher collaboration. By automatically adjusting messaging strategies based on learned patterns, schools can improve responsiveness, trust, and the flow of information among stakeholders.