A dynamic model for self-organization management and sentiment polarity analysis among college students through multi-text semantic enrichment

In recent years, student management methods have evolved from traditional disciplinary approaches to comprehensive quality education, emphasizing holistic development and personalized guidance. The application of information technology has made management more intelligent, with smart systems and big data analytics providing services more closely aligned with student needs. With the continuous development of artificial intelligence and deep learning technologies, NLP techniques have also made significant strides. This paper introduces an innovative SC-BGRU-ATT sentiment polarity analysis framework for analyzing student comment data using NLP techniques. This framework begins with the extraction of textual data using word embedding techniques, followed by further feature processing through Text-CNN and BI-GRU with a self-attention mechanism. Finally, a classification module is employed to identify the sentiment polarity of the texts. The test results on public datasets show that the framework achieves over 80% accuracy in identifying two types of sentiment polarity, and over 90% accuracy on self-built datasets. This significant performance demonstrates the potential of our framework in advancing sentiment polarity analysis. Moreover, by accurately capturing and analyzing students’ sentiment and opinions, this approach supports more effective self-organization and management of students. The innovation of this study lies in the integration of multiple models and the enhancement of relevant modules’ performance, leading to an improvement in sentiment polarity analysis. This technological advancement not only refines sentiment polarity analysis but also provides a scalable and adaptable framework for analyzing student sentiment in educational settings. The improved model performance offers robust, data-driven support for decision-making in student management, enabling institutions to better understand and respond to student needs. Furthermore, the modular design of our approach allows for seamless future enhancements, including integration with transformer-based models and multimodal data processing, ensuring continued advancements in sentiment analysis applications.