Negation Handling in Machine Learning-Based Sentiment Classification for Colloquial Arabic

Abstract

One crucial aspect of sentiment analysis is negation handling, where the occurrence of negation can flip the sentiment of a review and negatively affects the machine learning-based sentiment classification. The role of negation in Arabic sentiment analysis has been explored only to a limited extent, especially for colloquial Arabic. In this paper, the authors address the negation problem in colloquial Arabic sentiment classification using the machine learning approach. To this end, they propose a simple rule-based algorithm for handling the problem that affects the performance of a machine learning classifier. The rules were crafted based on observing many cases of negation, simple linguistic knowledge, and sentiment lexicon. They also examine the impact of the proposed algorithm on the performance of different machine learning algorithms. Furthermore, they compare the performance of the classifiers when their algorithm is used against three baselines. The experimental results show that there is a positive impact on the classifiers when the proposed algorithm is used compared to the baselines.

A Deep Neural Network Optimized by a Genetic Algorithm to Improve Arabic Sentiment Classification

“…For F1-F7 in sentiment-based features, we adopted the publicly available resources introduced [40] which contain 3400 labeled sentimental words and 580 sentiment-carrying compound phrases. Negation is also considered in this study; we employed the rule-based algorithm presented [41] to extract F10-F12 in linguistic-based features. The algorithm detects negation words such as ‫ال(‬ ‫مو،‬ ‫)مش،‬ and then tags the opinionated words that might be affected within a predefined window length of words.…”

Arabic Opinion Mining System for E-Service Based on Ontology and Machine earning

“…Negation words can be used to identify the scope of negation, and the sentiment of the base phrases within the scope of negation can be reversed [8] [9]. SVM, NB, and K-NN classifiers reported better improvement on the results after applying the exceptional negation algorithm [10] there are three baseline models, the first is baseline in which the simple uni-gram model is used without considering the negation problem. Secondly, a uni-gram model which consider a negation scope of five words that directly follow a negation term, where, each term within the scope will be tagged with the negation mark.…”

“…Secondly, a uni-gram model which consider a negation scope of five words that directly follow a negation term, where, each term within the scope will be tagged with the negation mark. The last one, is a uni-gram model for an inclusive negation scope that includes all the words that follow a negation term until the end of the sentence, where, each term within the scope will be tagged with the negation mark [10].…”

“…Increasing the polarity of the sentiment to include 10 levels of customer satisfaction depending on the customer satisfaction subjectivity of the features rates and the most important domain features hierarchy from very-negative (1), negative (2), weak negative (3), neutral towards negative (4), neutral towards slightly negative (5), neutral (6), neutral towards positive (7), weak positive (8) positive ( 9), very positive (10). The research would recommend Appling the system methodologies to other complicated languages like (e.g.…”

Negation detection techniques in sentiment analysis: A survey

“…As a result, detecting negation is critical for SA systems development and refinement and will increase classifier accuracy. Still, it is, at the same time, a significant conceptual and technical challenge (Hussein et al, 2018;Al-Harbi, 2020;Eremyan et al, 2021). This led to an interest in the research problem of automatic negation detection.…”