Rule-Based Arabic Sentiment Analysis using Binary Equilibrium Optimization Algorithm.

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Artificial Intelligence

Background:

• The internet generates vast amounts of user comments, necessitating sentiment analysis (opinion mining) for understanding public opinion.
• While English sentiment analysis is well-researched, Arabic sentiment analysis has received limited attention.
• Machine learning models for sentiment analysis are often "black boxes," lacking user interpretability.

Purpose of the Study:

• To develop an interpretable, rule-based Arabic sentiment analysis approach.
• To introduce a novel binary equilibrium optimization metaheuristic algorithm for rule generation.
• To address the gap in Arabic sentiment analysis research and provide transparent models.

Main Methods:

• A classification rule-based approach was developed for Arabic sentiment analysis.
• A new binary equilibrium optimization metaheuristic algorithm was employed for generating classification rules.
• The proposed method was evaluated on the Opinion Corpus for Arabic (OCA).

Main Results:

• The approach generated a classification model comprising thirteen rules.
• The proposed method demonstrated superior classification accuracy compared to other white-box models.
• The rule-based system offers enhanced interpretability over traditional black-box models.

Conclusions:

• The proposed rule-based Arabic sentiment analysis approach, optimized by a novel metaheuristic, is effective and interpretable.
• This research contributes a valuable, transparent tool for analyzing sentiments in Arabic text.
• The findings highlight the potential of rule-based systems and metaheuristic optimization in specialized NLP tasks.