Tuberculosis (TB) is is an infectious disease that generally attacks the lungs and causes death for millions of people annually. Chest radiography and deep-learning-based image segmentation techniques can be utilized for TB diagnostics. Convolutional Neural Networks (CNNs) has shown advantages in medical image recognition applications as powerful models to extract informative features from images. Here, we present a novel hybrid method for efficient classification of chest X-ray images. First, the features are extracted from chest X-ray images using MobileNet, a CNN model, which was previously trained on the ImageNet dataset. Then, to determine which of these features are the most relevant, we apply the Artificial Ecosystem-based Optimization (AEO) algorithm as a feature selector. The proposed method is applied to two public benchmark datasets (Shenzhen and Dataset 2) and allows them to achieve high performance and reduced computational time. It selected successfully only the best 25 and 19 (for Shenzhen and Dataset 2, respectively) features out of about 50,000 features extracted with MobileNet, while improving the classification accuracy (90.2% for Shenzen dataset and 94.1% for Dataset 2). The proposed approach outperforms other deep learning methods, while the results are the best compared to other recently published works on both datasets.

Big Data are highly effective for systematically extracting and analyzing massive data. It can be useful to manage data proficiently over the conventional data handling approaches. Recently, several schemes have been developed for handling big datasets with several features. At the same time, feature selection (FS) methodologies intend to eliminate repetitive, noisy, and unwanted features that degrade the classifier results. Since conventional methods have failed to attain scalability under massive data, the design of new Big Data classification models is essential. In this aspect, this study focuses on the design of metaheuristic optimization based on big data classification in a MapReduce (MOBDC-MR) environment. The MOBDC-MR technique aims to choose optimal features and effectively classify big data. In addition, the MOBDC-MR technique involves the design of a binary pigeon optimization algorithm (BPOA)-based FS technique to reduce the complexity and increase the accuracy. Beetle antenna search (BAS) with long short-term memory (LSTM) model is employed for big data classification. The presented MOBDC-MR technique has been realized on Hadoop with the MapReduce programming model. The effective performance of the MOBDC-MR technique was validated using a benchmark dataset and the results were investigated under several measures. The MOBDC-MR technique demonstrated promising performance over the other existing techniques under different dimensions.

Different social media platforms have become increasingly popular in the Arab world in recent years. The increasing use of social media, however, has also led to the emergence of a new challenge in the form of abusive content, including hate speech, offensive language, and abusive language. Existing research work focuses on automatic abusive content detection as a binary classification problem. In addition, the existing research work on the automatic detection task surrounding abusive Arabic content fails to tackle the dialect-specific phenomenon. Consequently, this has led to two important issues in the automatic abusive Arabic content detection task. In this study, we used a multi-aspect annotation schema to tackle the automatic abusive content detection problem in Arabic countries, based on the multi-class classification task and the dialectal Arabic (DA)-specific phenomenon. More precisely, the multi-aspect annotation schema includes five attributes: directness, hostility, target, group, and annotator. We specifically developed a framework to automatically detecting abusive content on Twitter using natural language processing (NLP) techniques. The developed framework used different models of machine learning (ML), deep learning (DL), and pretrained Arabic language models (LMs) using the multi-aspect annotation dataset. In addition, to investigate the impact of the other approaches, such as multi-task learning (MTL), we developed four MTL models built on top of a pretrained DA language model (called MARBERT) and trained on the multi-aspect annotation dataset. Our MTL models and pretrained Arabic LMs enhanced the performance compared to the existing DL model mentioned in the literature.

The increasing use of social media has led to the emergence of a new challenge in the form of abusive content. There are many forms of abusive content such as hate speech, cyberbullying, offensive language, and abusive language. This article will present a review of abusive content automatic detection approaches. Specifically, we are focusing on the recent contributions that were using natural language processing (NLP) technologies to detect the abusive content in social media. Accordingly, we adopt PRISMA flow chart for selecting the related papers and filtering process with some of inclusion and exclusion criteria. Therefore, we select 25 papers for meta-analysis and another 87 papers were cited in this article during the span of 2017–2021. In addition, we searched for the available datasets that are related to abusive content categories in three repositories and we highlighted some points related to the obtained results. Moreover, after a comprehensive review this article propose a new taxonomy of abusive content automatic detection by covering five different aspects and tasks. The proposed taxonomy gives insights and a holistic view of the automatic detection process. Finally, this article discusses and highlights the challenges and opportunities for the abusive content automatic detection problem.

In today’s world, news outlets have changed dramatically; newspapers are obsolete, and radio is no longer in the picture. People look for news online and on social media, such as Twitter and Facebook. Social media contributors share information and trending stories before verifying their truthfulness, thus, spreading rumors. Early identification of rumors from social media has attracted many researchers. However, a relatively smaller number of studies focused on other languages, such as Arabic. In this study, an Arabic rumor detection model is proposed. The model was built using transformer-based deep learning architecture. According to the literature, transformers are neural networks with outstanding performance in natural language processing tasks. Two transformers-based models, AraBERT and MARBERT, were employed, tested, and evaluated using three recently developed Arabic datasets. These models are extensions to the BERT, Bidirectional Encoder Representations from Transformers, a deep learning model that uses transformer architecture to learn the text representations and leverages the attention mechanism. We have also mitigated the challenges introduced by the imbalanced training datasets by employing two sampling techniques. The experimental results of our proposed approaches achieved 0.97 accuracy. This result demonstrated the effectiveness of the proposed method and outperformed other existing Arabic rumor detection methods.