Detecting harmful content or hate speech on social media is a significant challenge due to the high throughput and large volume of content production on these platforms. Identifying hate speech in a timely manner is crucial in preventing its dissemination. We propose a novel stacked ensemble approach for detecting hate speech in English tweets. The proposed architecture employs an ensemble of three classifiers, namely support vector machine (SVM), logistic regression (LR), and XGBoost classifier (XGB), trained using word2vec and universal encoding features. The meta classifier, LR, combines the outputs of the three base classifiers and the features employed by the base classifiers to produce the final output. It is shown that the proposed architecture improves the performance of the widely used single classifiers as well as the standard stacking and classifier ensemble using majority voting. We also present results on the use of various combinations of machine learning classifiers as base classifiers. The experimental results from the proposed architecture indicated an improvement in the performance on all four datasets compared with the standard stacking, base classifiers, and majority voting. Furthermore, on three of these datasets, the proposed architecture outperformed all state-of-the-art systems.
Social media sites, which became central to our everyday lives, enable users to freely express their opinions, feelings, and ideas due to a certain level of depersonalization and anonymity they provide. If there is no control, these platforms may be used to propagate hate speech. In fact, in recent years, hate speech has increased on social media. Therefore, there is a need to monitor and prevent hate speech on these platforms. However, manual control is not feasible due to the high traffic of content production on social media sites. Moreover, the language used and the length of the messages provide a challenge when using classical machine learning approaches as prediction methods. This paper presents a genetic programming (GP) model for detecting hate speech where each chromosome represents a classifier employing a universal sentence encoder as a feature. A novel mutation technique that affects only the feature values in combination with the standard one-point mutation technique improved the performance of the GP model by enriching the offspring pool with alternative solutions. The proposed GP model outperformed all state-of-the-art systems for the four publicly available hate speech datasets.
1 January 2023 in Turkish Journal of Electrical Engineering and Computer Sciences
It is well known that classifiers trained using imbalanced datasets usually have a bias toward the majority class. In this context, classification models can present a high classification performance overall and for the majority class, even when the performance for the minority class is significantly lower. This paper presents a genetic programming (GP) model with a crossover-based oversampling technique for oversampling the imbalanced dataset for binary text classification. The aim of this study is to apply an oversampling technique to solve the imbalanced issue and improve the performance of the GP model that employed the proposed technique. The proposed technique employs a crossover operator for generating new samples for the minority class in an imbalanced text dataset. By using a combination of this crossover-based oversampling technique with GP, the performance was improved. It is shown that the proposed combination outperforms all GP applications that use the original dataset without resampling. Moreover, the performance of the proposed system surpassed GP approaches using the synthetic minority oversampling technique (SMOTE) and random oversampling. Further comparison with the state-of-the-art on five imbalanced text datasets in terms of F1-score shows the superior performance of the proposed approach.
1 September 2024 in MINAR International Journal of Applied Sciences and Technology
Stroke is a major cause of death and permanent disability globally. Effective preventive measures depend on early stroke risk prediction. This, study investigates a form of machine learning classifiers for predicting the risk of stroke and evaluates their performance in comparison to more conventional models, including Artificial Neural Networks (ANN), Logistic Regression (LR), Decision Trees (DT), Support Vector Machines (SVM), Naive Bayes (NB), K-Nearest Neighbors (KNN), AdaBoost, Gradient Boosting, and Genetic Programming (GP). A public dataset from DataHack comprising 4981 participants was used, with class imbalance addressed by using Synthetic Minority Over-sampling Techniques (SMOTE). The dataset was split into a training set with 80% and a testing set with 20%. According to the experimental results, all the proposed approaches performed well in all the presented categories. Moreover, the GP approach achieved the highest results among the proposed approaches, reaching 95.6% accuracy, a significant improvement over the 90% state-of-the-art accuracy. The GP approach delivers prediction accuracy by creating symbolic expressions that illustrate correlations between the probability of stroke and risk factors. This study highlights the potential of machine learning approaches in predictive healthcare applications while also emphasizing the need for improvements through hybrid approaches, parameter modifications, and practical medical integration.