The popularity of social media and the rise of sarcasm in online conversations make sarcasm detection a crucial task in natural language processing (NLP), which still faces several challenges, including data imbalance and expression implicitness. This article presents a novel approach for sarcasm detection in Arabic short texts using deep learning techniques with a particular focus on addressing data imbalances. To ensure accurate and reliable analysis of Arabic short text data, care must be taken in considering and adapting existing techniques due to lack of standardization and changes in word forms. Thus, the proposed approach utilizes transformers to capture contextual knowledge and addresses the data imbalance problem by combining various data sampling and augmentation techniques. Using Arabic short text data, this article selects the most effective techniques for detecting sarcasm, which proposes a sampling technique that generates new samples from promising and representative words. The proposed approach was evaluated on several Arabic sarcasm detection datasets, including a new dataset we collected called ArSarcasticNews, a curated collection of Arabic short news texts for sarcasm detection. The experimental results across all datasets demonstrate promise. Specifically, the proposed model achieves F1 scores of 0.58, 0.60, 0.84, and 0.64 for the sarcastic class on ArSarcasm-v2, iSarcasmEval, IDAT data, and ArSarcasticNews, respectively. The study highlights the importance of understanding the context between sentences in identifying sarcasm in short Arabic texts and the potential of data augmentation techniques to improve model performance. Moreover, the use of the SHAP-based model provides valuable insight into the features used by the model to detect sarcasm and can aid in identifying potential biases in the model. Based on the findings, this study improves sarcasm detection and outperforms other approaches by analyzing dialectical short text and addressing data imbalances.

Business customers and consumers share their reviews online on social platforms such as Twitter. Therefore, Twitter data sentiment analysis is extremely useful for both research and commercial purposes. Manually analyzing reviews takes a long time and effort, hence, automatic sentiment analysis is required. In this article, we address aspect-based sentiment analysis for Arabic food delivery reviews using several deep learning approaches. In particular, we propose to use Transformer-based models (GigaBERT and AraBERT), Bi-LSTM-CRF, and LSTM, as well as a classical machine learning algorithm (SVM). We also present our dataset of food delivery service reviews, which we collected from Twitter. We annotated them and used them for training and evaluating our approaches. The experiments show that both GigaBERT and AraBERT outperformed the other models in all the tasks. The Transformer-based models received F1-scores of 77% in the aspect terms detection task, 82% in the Aspect category detection task, and 81% in the aspect polarity detection task, gaining 2%, 4%, and 4% over Bi-LSTM-CRF and LSTM in the first, second, and third tasks, respectively.

Question semantic similarity is a challenging and active research problem that is very useful in many NLP applications, such as detecting duplicate questions in community question answering platforms such as Quora. Arabic is considered to be an under-resourced language, has many dialects, and rich in morphology. Combined together, these challenges make identifying semantically similar questions in Arabic even more difficult. In this paper, we introduce a novel approach to tackle this problem and test it on two benchmarks; one for Modern Standard Arabic (MSA), and another for the 24 major Arabic dialects. We are able to show that our new system outperforms state-of-the-art approaches by achieving 93% F1-score on the MSA benchmark and 82% on the dialectical one. This is achieved by utilizing contextualized word representations (ELMo embeddings) trained on a text corpus containing MSA and dialectic sentences. This in combination with a pairwise fine-grained similarity layer, helps our question-to-question model to generalize on dialects while being trained only on our newly developed question-to-question MSA data.

Data streams are continuous flows of data points. Novel class detection is an important part of data stream mining. A novel class is a newly emerged class that has not previously been modeled by the classifier over the input stream. This paper proposes deep embedding for novel class detection - a novel approach that combines feature learning using denoising autoencoding with novel class detection. A denoising autoencoder is a neural network with hidden layers aiming to reconstruct the input vector from a corrupted version. A nonparametric multidimensional change point detection approach is also proposed, to detect concept-drift (the change of data feature values over time). Experiments on several real datasets show that the approach significantly improves the performance of novel class detection.

In recent years, mass atrocities, terrorism, and political unrest have caused much human suffering. Thousands of innocent lives have been lost to these events. With the help of advanced technologies, we can now dream of a tool that uses machine learning and natural language processing (NLP) techniques to warn of such events. Detecting atrocities demands structured event data that contain metadata, with multiple fields and values (e.g. event date, victim, perpetrator). Traditionally, humans apply common sense and encode events from news stories but this process is slow, expensive, and ambiguous. To accelerate it, we use machine coding to generate an encoded event. In this paper, we develop a near-real-time supervised machine coding technique with an external knowledge base, WordNet, to generate a structured event. We design a Spark-based distributed framework with a web scraper to gather news reports periodically, process, and generate events. We use Spark to reduce the performance bottleneck while processing raw text news using CoreNLP.

Streaming data is one of the attention receiving sources for concept-evolution studies. When a new class occurs in the data stream it can be considered as a new concept and so the concept-evolution. One attractive problem occurring in the concept-evolution studies is the recurring classes from our previous study. In data streams, a class can disappear and reappear after a while. Existing studies on data stream classification techniques either misclassify the recurring class or falsely identify the recurring classes as novel classes. Because of the misclassification or false novel classification, the error rates increases on those studies. In this paper we address the problem by defining a novel ensemble technique “class-based” ensemble which replaces the traditional “chunk-based” approach in order to detect the recurring classes. We discuss the details of two different approaches in class-based ensemble and explain and compare them in detail. Different than the previous studies in the field, we also prove the superiority of both “class-based” ensemble method over state-of-art techniques via empirical approach on a number of benchmark data sets including Web comments as text mining challenge.

Classifying instances in evolving data stream is a challenging task because of its properties, e.g., infinite length, concept drift, and concept evolution. Most of the currently available approaches to classify stream data instances divide the stream data into fixed size chunks to fit the data in m

Static data mining assumptions with regard to features and labels often fail the streaming context. Features evolve, concepts drift, and novel classes are introduced. Therefore, any classification algorithm that intends to operate on streaming data must have mechanisms to mitigate the obsolescence of classifiers trained early in the stream. This is typically accomplished by either continually updating a monolithic model, or incrementally updating an ensemble. Traditional static data mining algorithms futile in a streaming context (and often in a distributed sensor network) due to their need to iterate over the entire data set locally. Our approach -- named HSMiner (Hierarchical Stream Miner) -- takes a hierarchical decomposition approach to the ensemble classifier concept. By breaking the classification problem into tiers, we can better prune the irrelevant features and counter individual classification error through weighted voting and boosting. In addition, the atomic decomposition of feature inputs enables straightforward mapping to distributing the ensemble among resources in the network. The implementation proves to be fast and very memory conservative, and we emulate a distributed environment via signal-linked threads. We examine the theoretical and empirical analysis of our approach, specifically examining trade-offs of three different novel class detection variations, and compare these results to a similar method using benchmark data sets.