Abstract:Sentiment analysis is a task in natural language processing aiming to identify the overall polarity of reviews for subsequent analysis. This study used the Arabic speech-act and sentiment analysis, Arabic sentiment tweets dataset, and SemEval benchmark datasets, along with the Moroccan sentiment analysis corpus, which focuses on the Moroccan dialect. Furthermore, the modern standard and dialectal Arabic corpus dataset has been created and annotated based on the three language types: modern standard Arabic, Moroccan Arabic Dialect, and Mixed Language. Additionally, the annotation has been performed at the sentiment level, categorizing sentiments as positive, negative, or mixed. The sizes of the datasets range from 2000 to 21,000 reviews. The essential dialectal characteristics to enhance a sentiment classification system have been outlined. The proposed approach has involved deploying several models employing the supervised approach, including occurrence vectors, Recurrent Neural Network-Long Short Term Memory, and the pre-trained transformer model Arabic bidirectional encoder representations from transformers (AraBERT), complemented by the integration of Generative Adversarial Networks (GANs). The uniqueness of the proposed approach lies in constructing and annotating manually a dialectal sentiment corpus and studying carefully its main characteristics, which are used then to feed the classical supervised model. Moreover, GANs that widen the gap between the studied classes have been used to enhance the obtained results with AraBERT. The classification test results have been promising, enabling a comparison with other systems. The proposed system has been evaluated against Mazajak and CAMelTools state-of-the-art systems, designed for most Arabic dialects, using the mentioned datasets. A significant improvement of 30 points in F NN has been observed. These results have affirmed the versatility of the proposed system, demonstrating its effectiveness across multi-dialectal, multi-domain datasets, as well as balanced and unbalanced ones.

Learning to Play Chess from Textbooks (LEAP): a Corpus for Evaluating Chess Moves based on Sentiment Analysis

Aspect-based Sentiment Evaluation of Chess Moves (ASSESS): an NLP-based Method for Evaluating Chess Strategies from Textbooks

SentiMATE: Learning to play Chess through Natural Language Processing

Learning Chess With Language Models and Transformers

Chess as a Testbed for Language Model State Tracking

SEntiMoji: An Emoji-Powered Learning Approach for Sentiment Analysis in Software Engineering

EK-Chess: Chess Learning System Based on Top-Level Chess Expert Knowledge Graph

Mastering Chess with a Transformer Model

Amortized Planning with Large-Scale Transformers: A Case Study on Chess

Evaluation of Sentiment Analysis in Finance: From Lexicons to Transformers

SL-TeaE: An Efficient Method for Improving the Precision of Teaching Evaluation.

Chess Rating Estimation from Moves and Clock Times Using a CNN-LSTM

Predicting User Perception of Move Brilliance in Chess

Explore the Reasoning Capability of LLMs in the Chess Testbed

ELM-Based Active Learning Via Asymmetric Samplers: Constructing a Multi-Class Text Corpus for Emotion Classification

Text sentiment analysis from students' comments using various machine learning and deep learning techniques: a comparative analysis

Large-scale Analysis of Chess Games with Chess Engines: A Preliminary Report

Contextual Sentence Analysis for the Sentiment Prediction on Financial Data

Application of Sentiment Analysis to Language Learning

Annotation and evaluation of a dialectal Arabic sentiment corpus against benchmark datasets using transformers