Modified Seagull Optimization with Deep Learning for Affect Classification in Arabic Tweets

Abstract

Arabic is one of the world's most widely spoken languages, and there is a huge amount of digital content available in Arabic. By the categorization of Arabic documents, it becomes easier to search and access specific content of interest. With the increasing quantity of user-generated content on social media platforms and online forums, text classification becomes important for content filtering and moderation. Text classification on an Arabic corpus has broad applications, ranging from information retrieval and content moderation to sentiment analysis and machine translation. It enables efficient organization, analysis, and utilization of Arabic text data, contributing to various industries and domains. Therefore, this study develops a Modified Seagull Optimization with Deep Learning based Affect Classification on Arabic Tweets (MSGODL-ACAT) technique. The goal of the MSGODL-ACAT approach lies in the recognition and categorization of effects or emotions that exist in Arabic tweets. At the preliminary level, the MSGODL-ACAT technique preprocesses the input data to make the Arabic tweets into a meaningful format. Next, the Glove technique is used for the word embedding process. Moreover, the MSGODL-ACAT technique makes use of the deep belief network (DBN) method for affect categorization. At last, the MSGO algorithm is used for the optimal hyperparameter tuning of the DBN method which in turn enhances the classification results. The experimental evaluation of the MSGODL-ACAT approach is evaluated using Arabic tweets databases. The experimental outcomes signify the effectual performance of the MSGODL-ACAT algorithm over other current approaches.