Social-path embedding-based transformer for graduation development prediction.

Area of Science:

• Educational Data Mining
• Social Network Analysis
• Machine Learning

Background:

• Student graduation development prediction is crucial for educational planning.
• Existing models often overlook the influence of social relationships on student outcomes.
• Integrating social network data can enhance prediction accuracy.

Purpose of the Study:

• To propose a novel deep learning model, Social-path Embedding-based Transformer Neural Network (SPE-TNN), for student graduation development prediction.
• To investigate the impact of social relationships on students' post-graduation choices.
• To improve the accuracy of predicting students' future development pathways.

Main Methods:

• Developed a Social-path Embedding-based Transformer Neural Network (SPE-TNN).
• Incorporated a Social-path selection layer to identify impactful social relationships.
• Utilized a Transformer layer for feature weight balancing and a Multi-layer projection layer for prediction.
• Employed social network data alongside academic achievement data.

Main Results:

• SPE-TNN effectively integrates social relationship data into student development prediction.
• The proposed model demonstrates superior performance compared to existing approaches on real-world datasets.
• Analysis confirmed the significant impact of social connections on graduation development.

Conclusions:

• Social relationships play a vital role in student graduation development prediction.
• SPE-TNN offers a robust framework for leveraging social network information in educational data mining.
• The findings provide valuable insights for educators and institutions to support student career planning.