The Arabic language’s grammar, variety of dialects, and lack of annotated datasets make its Natural Language Processing (NLP) very challenging. This paper introduces a manual and automatic annotating platform that will help researchers and annotators in a simple approach. The platform includes CAMeL Lab tools, pre-trained models, and a multi-language pretrained RoBERTa model, as well as a customizable model using inference scripts, and the manual annotation module contains new features like tracking users, detecting label conflicts, and a majority voting feature to keep results consistent, and a new feature that annotates using OpenAI API. A new module is introduced, as well as the visualization module that displays histograms, heatmaps, and scatterplots to visualize annotations and extract insights. The platform has been evaluated using the System Usability Scale (SUS). The evaluation resulted in a SUS score of 77.5, which means users found the tool easy use and friendly, In summary, this system offers a practical and adaptable approach to improving Arabic NLP, In conclusion, the tool offers a practical and adaptable approach to improving Arabic NLP.
Danial et al. (Thu,) studied this question.