In recent years, computer language area has witnessed important evolvement with applications in different domains. Machine Translation MT technology, considered as a subfield, has received important development with different approaches and techniques. Although, many MT systems and tools that support Arabic already exist; however, the quality of the translation is moderate and needs some improvement. In addition, the high demand for effective technologies to process and translate information from/to Arabic motivated the researchers in Arabic Machine Translation (AMT) to propose new approaches and solutions following the mainstream method, notably neural machine translation (NMT). In this paper, we provide a comprehensive review and compare different NMT approaches mainly for Arabic-English (and English-Arabic) machine translation research works. The discussed approaches address different linguistic and technical challenges and problems while demonstrating great success compared to traditional methods. The results of this work can serve the researchers and professional to be up-to-date and provide them with the necessary resources for modelling and improving of the AMT. These resources include corpora, toolkits, techniques and new models. The obtained results outline various findings, critics, and open issues in this area.

Neural machine translation (NMT) has shown impressive performance when trained on large-scale corpora. However, generic NMT systems have demonstrated poor performance on out-of-domain translation. To mitigate this issue, several domain adaptation methods have recently been proposed which often lead to better translation quality than genetic NMT systems. While there has been some continuous progress in NMT for English and other European languages, domain adaption in Arabic has received little attention in the literature. The current study, therefore, aims to explore the effectiveness of domain-specific adaptation for Arabic MT (AMT), in yet unexplored domain, financial news articles. To this end, we developed a parallel corpus for Arabic-English (AR-EN) translation in the financial domain to benchmark different domain adaptation methods. We then fine-tuned several pre-trained NMT and Large Language models including ChatGPT-3.5 Turbo on our dataset. The results showed that fine-tuning pre-trained NMT models on a few well-aligned in-domain AR-EN segments led to noticeable improvement. The quality of ChatGPT translation was superior to other models based on automatic and human evaluations. To the best of our knowledge, this is the first work on fine-tuning ChatGPT towards financial domain transfer learning. To contribute to research in domain translation, we made our datasets and fine-tuned models available.

Generating automatic visualization from natural language texts is an important task for promoting language learning and literacy development for young children and language learners. However, translating a text into a coherent visualization matching its relevant keywords is a challenging problem. To tackle this issue, we proposed a robust story visualization pipeline ranging from NLP and relation extraction to image sequence generation and alignment. First, we applied a shallow semantic representation of the text where we extracted concepts including relevant characters, scene objects, and events in an appropriate format. We also distinguished between simple and complex actions. This distinction helped to realize an optimal visualization of the scene objects and their relationships according to the target audience. Second, we utilized an image generation framework along with different versions to support the visualization task efficiently. Third, we used CLIP similarity function as a semantic relevance metric to check local and global coherence to the whole story. Finally, we validated the scene sequence to compose a final visualization using the different versions for various target audiences. Our preliminary results showed considerable effectiveness in adopting such a pipeline for a coarse visualization task that can subsequently be enhanced.

Multimedia repository is helpful for educational activities since it offers several illustrations that facilitate the learning process and the text understanding. In this paper, we propose to build a multimedia repository from collected images using object extraction techniques. Then, we associate Arabic captions to all extracted objects. These extracted objects are used to compose new scenes that could illustrate efficiently the most important events in an Arabic story. We extend, thereby the notion of image composing to our approach as the task of constructing new images based on a set of toolboxes and a set of extracted objects as our multimedia repository of common animal behaviors. Our preliminary results show that the composed scenes using single objects provided a fair understanding of the main events of the stories as well as a coherent visual layout of all single objects. The diversity and the precision of the single objects images for the domain of animals have shown a great impact on composing new scenes either manually or dynamically.