A Arabic Language: Sentiment Analysis (Opinion Mining) using Sentic Computing Models, Tools and Techniques

Abstract

Emotion analysis and sentiment extraction from Arabic texts present a significant challenge in the field of affective computing, given the distinctive linguistic and semantic features of the Arabic language. Arabic is characterized by complex morphological and syntactic structures, multiple meanings of words, and a rich use of rhetoric and figurative language, in addition to the existence of diverse local dialects alongside Modern Standard Arabic. These characteristics make emotion assessment a delicate process requiring sophisticated natural language processing tools, such as Farasa and CAMEL Tools, Arabic emotion dictionaries, and deep learning models like Word2Vec, BERT, and AraBERT. These tools are used to classify texts at multiple levels (full text, sentence, and context), identify positive, negative, and neutral emotions, and handle abbreviations and emojis commonly used in social media, the implementation has been carried out using python programming language environment.

The study hypotheses were the accuracy of sentiment analysis and opinion extraction is positively correlated with the use of sophisticated sentiment computing models (such as deep learning). Furthermore, in Arabic texts, sentiment analysis systems that utilize specialized dictionaries and grammars yield more accurate results than those relying solely on statistical methods; hence the question to what extent the linguistic, semantic, and contextual features characterize the Arabic language and influence emotion analysis and opinion extraction?

What are the most common affective computing models, tools, and techniques used in emotion analysis? This

study have shown that integrating semantic, contextual, and rhetorical analysis with deep learning techniques enhances the accuracy of emotion extraction from Arabic texts, especially short texts such as tweets and posts.

The results moreover highlight the highly significant of establishing comprehensive databases that include Modern Standard Arabic and various dialects, and applying emotion analysis at multiple levels. This facilitates the use of these technologies in diverse fields such as marketing, social monitoring, digital education, and improving user experience. Despite the remarkable progress in tools and models, obstacles remain related to dialectal diversity, figurative language, and unstructured data, which open the door to developing more accurate and effective models for emotional computing in Arabic.