An Adaptive Model for Identification of Influential Bloggers Based on Case-Based Reasoning Using Random Forest

Abstract

Bloggers play a role in individual decision making of users in online social networking platforms. Their capability of addressing a wide audience gives them influence over their audience, which companies seek to exploit. Identification of influential bloggers can be seen as a machine learning (ML) task and different ML techniques can help in classifying the professional blogger. In this paper, we propose a predictive and adaptive model named as Influential Blogger based Case-Based Reasoning (IB-CBR) model for the recognition of unseen influential bloggers. It incorporates self-prediction and self-adaptation (self-management) capabilities which are the essence of an automated system. The integration of Random Forest is found contributing to the efficiency of the IB-CBR model as compared to Nearest-Neighbor, and Artificial Neural Network. The performance of the proposed IB-CBR model is evaluated against other ML techniques by using standard performance measures on a standard blogger's dataset. It is observed that our proposed model exhibits 88-95% Accuracy and 94-97% True Positive Rate in the prediction and adaptation of professional bloggers, respectively, in three iterations of the proposed model. What's more, the IB-CBR model achieved 91-96% (increasing) F-measure, 91-98% (increasing) ROC AUC, and 36-11% (decreasing) False Positive Rate due to adaptivity. The IB-CBR model performed well when it is compared with other ML techniques using different standard datasets.