Automated blog feedback prediction with Ada-Boost classifier

Abstract

Automated analysis of social media documents has a tremendous impact in our day to day life since we extensively use social media to share our thoughts, feelings, tastes etc. However, the automatic social media analysis is still a very challenging task due to the massive amount of social media documents as well as the uncontrolled, dynamic and rapidly-changing content of social media documents. To automate social media analysis, this paper presents an automatic feedback prediction model based on novel Ada-Boost learning algorithm for blog documents considering realistic scenario. In this approach, an Ada-Boost classifier is applied to the numerous features extracted from crawled blog document to predict whether someone comments on a blog document or not in the next 24 hours of its publication in blogs. The evaluation results of the experiments conducted on the publicly available benchmark blog feedback data set indicate that the proposed technique is efficient both in terms of feedback prediction accuracy and computational time; the proposed approach yielded the maximum feedback prediction rate of 91.4%.