Abstract
The need for designing Arabic text mining systems for the use on social media posts is increasingly becoming a significant and attractive research area. It serves and enhances the knowledge needed in various domains. The main focus of this paper is to propose a novel framework combining sentiment analysis with subjective analysis on Arabic social media posts to determine whether people are interested or not interested in a defined subject. For those purposes, text classification methods-including preprocessing and machine learning mechanisms-are applied. Essentially, the performance of the framework is tested using Twitter as a data source, where possible volunteers on a certain subject are identified based on their posted tweets along with their subject-related information. Twitter is considered because of its popularity and its rich content from online microblogging services. The results obtained are very promising with an accuracy of 89%, thereby encouraging further research.