Roman Urdu Hate Speech Detection Using Transformer-Based Model for Cyber Security Applications.

Social media applications, such as Twitter and Facebook, allow users to communicate and share their thoughts, status updates, opinions, photographs, and videos around the globe. Unfortunately, some people utilize these platforms to disseminate hate speech and abusive language. The growth of hate speech may result in hate crimes, cyber violence, and substantial harm to cyberspace, physical security, and social safety. As a result, hate speech detection is a critical issue for both cyberspace and physical society, necessitating the development of a robust application capable of detecting and combating it in real-time. Hate speech detection is a context-dependent problem that requires context-aware mechanisms for resolution. In this study, we employed a transformer-based model for Roman Urdu hate speech classification due to its ability to capture the text context. In addition, we developed the first Roman Urdu pre-trained BERT model, which we named BERT-RU. For this purpose, we exploited the capabilities of BERT by training it from scratch on the largest Roman Urdu dataset consisting of 173,714 text messages. Traditional and deep learning models were used as baseline models, including LSTM, BiLSTM, BiLSTM + Attention Layer, and CNN. We also investigated the concept of transfer learning by using pre-trained BERT embeddings in conjunction with deep learning models. The performance of each model was evaluated in terms of accuracy, precision, recall, and F-measure. The generalization of each model was evaluated on a cross-domain dataset. The experimental results revealed that the transformer-based model, when directly applied to the classification task of the Roman Urdu hate speech, outperformed traditional machine learning, deep learning models, and pre-trained transformer-based models in terms of accuracy, precision, recall, and F-measure, with scores of 96.70%, 97.25%, 96.74%, and 97.89%, respectively. In addition, the transformer-based model exhibited superior generalization on a cross-domain dataset.

Analysis of Risk Factors in Global Software Development: A Cross-Continental Study Using Modified Firefly Algorithm.

In today's competitive world, software organizations are moving towards global software development (GSD). This became even more significant in times such as COVID-19 pandemic, where team members residing in different geographical locations and from different cultures had to work from home to carry on their tasks and responsibilities as travelling was restricted. These teams are distributed in nature and work on the same set of goals and objectives. Some of the key challenges which software practitioners face in GSD environment are cultural differences, communication issues, use of different software models, temporal and spatial distance, and risk factors. Risks can be considered as a biggest challenge of other challenges, but not many researchers have addressed risks related to time, cost, and resources. In this research paper, a comprehensive analysis of software project risk factors in GSD environment has been performed. Based on the literature review, 54 risk factors were identified in the context of software development. These were further classified by practitioners into three dimensions, i.e., time, cost, and resource. A Pareto analysis has been performed to discover the most important risk factors, which could have bad impact on software projects. Furthermore, a modified firefly algorithm has been designed and implemented to evaluate and prioritize the pertinent risk factors obtained after the Pareto analysis. All important risks have been prioritized according to the fitness values of individual risks. The top three risks are "failure to provide resources," "cultural differences of participants," and "inadequately trained development team members."

Risk Prediction by Using Artificial Neural Network in Global Software Development.

The demand for global software development is growing. The nonavailability of software experts at one place or a country is the reason for the increase in the scope of global software development. Software developers who are located in different parts of the world with diversified skills necessary for a successful completion of a project play a critical role in the field of software development. Using the skills and expertise of software developers around the world, one could get any component developed or any IT-related issue resolved. The best software skills and tools are dispersed across the globe, but to integrate these skills and tools together and make them work for solving real world problems is a challenging task. The discipline of risk management gives the alternative strategies to manage risks that the software experts are facing in today's world of competitiveness. This research is an effort to predict risks related to time, cost, and resources those are faced by distributed teams in global software development environment. To examine the relative effect of these factors, in this research, neural network approaches like Levenberg-Marquardt, Bayesian Regularization, and Scaled Conjugate Gradient have been implemented to predict the responses of risks related to project time, cost, and resources involved in global software development. Comparative analysis of these three algorithms is also performed to determine the highest accuracy algorithms. The findings of this study proved that Bayesian Regularization performed very well in terms of the MSE (validation) criterion as compared with the Levenberg-Marquardt and Scaled Conjugate Gradient approaches.