Date of Award

5-16-2024

Publication Type

Thesis

Degree Name

M.Sc.

Department

Computer Science

Supervisor

Ziad Kobti

Abstract

In the dynamic landscape of social network analysis, the accurate prediction of implicit relationships presents a pivotal challenge. This paper introduces an innovative solution, the Relation Temporal Graph Convolutional Network with Confidence (R-CTGCN), specifically designed to address the intricate task of predicting implicit relations within evolving social networks. R-CTGCN unifies timestamp temporal embeddings, confidence metrics, and polynomial features within a comprehensive graph neural network framework, aiming to capture the evolving dynamics of networks and enhance predictive accuracy. Experimental evaluations conducted on diverse datasets, including Epinions and Enron, showcase R-CTGCN's superior performance compared to both baseline models and contemporary state-of-the-art methods. The emphasis on the roles of confidence and polynomial features underscores their significance in implicit relationship prediction. The outcomes contribute substantively to the understanding of predicting implicit relationships, positioning R-CTGCN as a robust tool tailored for complex social network scenarios. On the Enron dataset, R-CTGCN improved AUC by 11.5\%-18.06\% over competitors like Mirror and R-GCN, showcasing greater predictive accuracy. Ranking Score (RS) gains ranged from 0.5\% to 8.41\%, reflecting superior ranking precision. Notably, accuracy increased by up to 27.35\%, with corresponding enhancements in precision, recall, and F1 scores. Similarly, on the Epinions dataset, R-CTGCN achieved AUC improvements up to 10.82\% and RS gains up to 6.68\%, along with significant accuracy improvements. These results highlight R-CTGCN's robustness in predicting dynamic social network relationships, setting new benchmarks in relational learning.

Download

Share

COinS