Volume 9, Issue 5

Insufficient information is often pointed out as one of the main problems with bug reports as most bugs are reported manually, they lack detailed information describing steps to reproduce the unexpected behavior, leading to increased time and effort for developers to reproduce and fix bugs. Current bug reporting systems lack support for self-hosted systems that cannot access third-party cloud environments or Application Programming Interfaces due to confidentiality concerns. To address this, we propose Watson, a Typescript framework with a minimalist User Interface developed in Vue.js. The objectives are to minimize the user’s effort to report bugs, simplify the bug reporting process, and provide relevant information for developers to solve it. Watson was designed to capture user’s interactions, network logs, screen recording, and seamlessly integration with issue tracker systems in self-hosted systems that cannot share their data to external Application Programming Interfaces or cloud services. Watson also can be installed via Node Package Manager and integrated into most JavaScript or TypeScript web projects. To evaluate Watson, we developed an Angular-based application along with two usage scenarios. First, the users experimented the application without using Watson and once they found a bug, they reported it manually on GitLab. Later, they used the same application, but this time, whenever they detect another bug, they reported it through Watson User Interface. Watson, as stated by the experiment participants and the evidences, is useful and helpful for development teams to report issues and provide relevant information for tracking bugs. The identification of bug root causes was almost three times more effective with Watson than manual reporting.

This manuscript offers an in-depth analysis of Explainable Artificial Intelligence (XAI), emphasizing its crucial role in developing transparent and ethically compliant AI systems. It traces AI’s evolution from basic algorithms to complex systems capable of autonomous decisions with self-explanation. The paper distinguishes between explainability—making AI decision processes understandable to humans—and interpretability, which provides coherent reasons behind these decisions. We explore advanced explanation methodologies, including feature attribution, example-based methods, and rule extraction technologies, emphasizing their importance in high-stakes domains like healthcare and finance. The study also reviews the current regulatory frameworks governing XAI, assessing their effectiveness in keeping pace with AI innovation and societal expectations. For example, rule extraction from artificial neural networks (ANNs) involves deriving explicit, human-understandable rules from complex models to mimic explainability, thereby making the decision-making process of ANNs transparent and accessible. Concluding, the paper forecasts future directions for XAI research and regulation, advocating for innovative and ethically sound advancements. This work enhances the dialogue on responsible AI and establishes a foundation for future research and policy in XAI.

In recent years, the utilization of tangible educational materials has attracted attention on educational settings. They provide hands-on learning experiences for beginners. This trend is especially notable in the field of programming education. Such educational materials are employed in many institutions worldwide. They liberate learners of programming from programming languages that are confined in a small computer screen. On the other hand, in the school setting, classroom time is limited. When instructing more than thirty students, it is hard for instructors to provide adequate guidance for everyone. To address this problem, we have developed a classroom support system for programming education that complements the use of tangible educational materials. With this system, instructors can monitor the real-time progress of each student during the class and analyze which parts of the program many students find challenging. Based on these analytical results, instructors can provide appropriate instructions for individual students and effectively conduct the class. This system is suitable for programming education in high schools. It quantifies each student’s ability of programming and track the progress of each student. We administered a questionnaire to both the students and the instructor. The results of the questionnaire show our system is well received by both students and the instructor. Even though our system demonstrates some usefulness for programming beginners, we are aware that our system has some serious limitations such as our rigid model answers.