This paper introduces an innovative approach to designing data requirements for the development of learning analytics tailored to diverse, constructionist, open learning tools. Leveraging the Repertory Grid Technique, we systematically elicited and analyzed stakeholders' personal constructs to inform evidence-based user-centered design specifications. Through requirements elicitation workshops with educators, we uncovered essential information pertaining to teacher perceptions about student interactions with exploratory learning tools and their alignment with 21st-century skills. The process yielded concepts in relation to tool-specific indicators of learner interactions, forming a comprehensive design specification for a customizable learning analytics tool. Initial feedback from stakeholders and learning analytics experts indicates strong alignment with implementation needs and a coherent synthesis of diverse perspectives. Future steps include evaluating the effectiveness of the designed tools through real-world implementation and practitioner feedback, further validating the efficacy of this design approach.
As technology advances and public awareness of cybersecurity grows, protecting data and promoting cybersecurity best practices is essential. Educational chatbots were developed to address this need, but current systems respond artificially and lack user-friendliness, resulting in a bad user experience. This paper presents Cybot, a chatbot specifically designed to tackle this gap in cybersecurity education for university students. Cybot can answer cybersecurity questions, offer human-like interactions, provide assessments to improve students' understanding and performance and provide emotional support with its motivational personality. Additionally, it decreases the workload of academic staff due to its constant availability and instant replies. An experiment was conducted on 30 university students to evaluate Cybot's performance. The results showed positive feedback as users reported that Cybot was effective and a supportive study partner who always responded to their inquiries at any time.
E-learning integrates information and communication technologies on technological platforms to facilitate access to learning, as well as make it interactive and personalized. Within e-learning, intelligent tutoring systems (ITS) use artificial intelligence to adapt content and activities to the needs of each student, perform academic monitoring and facilitate interaction with virtual learning objectives, allowing for interaction between teachers, students, and intelligent tutors. Smart learning environments (SLE) integrate ITS to create innovative, motivating, and effective learning experiences, based on educational theories and models for the teaching-learning process under the online modality. This article presents a systematic literature review (SLR) on these topics, with the objective of identifying the trends, challenges, and research opportunities in this field. Educational theories that support e-learning and promote student interaction with virtual platforms are analyzed. Student-centered ITS models focused on learning styles, and the integration of data mining and learning analytics, as well as strategies for the implementation and validation of intelligent systems in SLEs. The design of a self-adaptive architecture in the Moodle platform is proposed, for the academic monitoring of students in the online modality with the use of autonomous agents. Finally, it is concluded that the research should be geared towards adaptive learning, to perform academic monitoring focused on the student and their learning styles within the SLE.
A Learning Experience Platform (LXP) is an extension of a Learning Management System (LMS) with the emphasis on personalized learning support and learner motivation. In this paper extensions for the well established LMS Moodle are proposed, that include, beyond others, AI supported learning recommendations and user based feedback mechanisms to transform Moodle to Moodle LXP. The architecture of this Moodle LXP provides plug-ins for three types of recommendations, i.e. content-based and two collaborative recommendations. Furthermore, a separate module was developed that is responsible for data storage and calculation of recommended learning resources. Various metadata such as description, number of clicks of learning resources, rating, and feedback of the resources are used for the calculation. The extensions have been integrated into several lectures for three semesters. The system operates effectively and the students use the plug-ins for learning. To learn more about the quality of the recommendations, a mechanism for evaluating the results of the content-based recommendation using feedback from the lecturers was developed.
Datasets are crucial for data-driven decision-making in businesses and organizations, allowing for the optimization of processes and the identification of improvement opportunities. They are foundational for training machine learning models to recognize patterns and make predictions, encompassing various forms of data from simple text and numbers to complex images and graphs. Furthermore, datasets are vital for educating future data scientists and machine learning engineers, offering hands-on experience that enhances analytical skills and practical knowledge. However, a gap in university curricula regarding dataset creation skills is noted, which is essential for ensuring the quality of datasets and the robustness of decision processes. This paper proposes to address this educational gap by developing methodologies for teaching dataset creation within university settings, mindful of the distinct skill levels of students compared to professionals.
