An Agent-Based Decision Fusion Regression Model for Analyzing Cyberspace Users' Engagement in Informatics Blogs

Abstract

Direct interactive communication technology, called a blog, enables users to post comments and share links to maintain interaction with fellow users. In modern information systems, a blog platform allows individuals or groups to write their opinions on a specific topic. Broad informatics blogs provide spaces for people to share ideas an exchange opinions on various issues. Driven informatics-based learning methods are used to detect and analyze users tendencies in these blog posts. Blog features exhibit a direct correlation with the extend to which bloggers adhere to both national cultural patterns and social and political developments in their respective countries. This paper proposes an agent-based decision fusion regression (ADFR) model for analyzing cyberspace users engagement in blog posts. The ADFR model is tested using the BlogFeedback dataset, which has been used in several previous research studies. The model results are compared with three machine learning algorithms: Linear Regression (LR), Decision Tree (c4.5), and Decision Forest (DF) for performing blog post analysis tasks. The R2 score reaches 0.0695 with ADFR, while DT shows only 0.0002, DF reveals 0.0104, and LR demonstrates 0.0544 in blog comment the highest level of predictive power and outstanding generalization potential.