According to a recent EU report, about 60% of enterprises requiring ICT specialists report hard-to-fill vacancies. It is estimated that the ICT labour market will experience a shortage of millions of ICT specialists in the next few years. Published data further reveal that two factors mainly cause this shortage: i) an insufficient number of students enrolled in ICT disciplines (a problem worsened by a gender gap); ii) high dropout rate with respect to other disciplines. E-learning is one obvious response to the lack of digital skills, to extend the possibility of career advancement for workers who have not been able to access higher levels of education before entering the world of work, for less wealthy students who leave far from Universities, and students with disabilities. However, the limited efficacy of training outcomes remained unchanged and also worsened during the recent pandemic emergency, thus demanding for novel and innovative approaches. Specifically, the integration of leading-edge IT solutions (such as machine learning and artificial intelligence) may pave the way towards a new model of learning in the digital world, focusing on customisation of learning solutions to meet learner needs and capacities, combining formal and informal learning (social, collaborative and experiential) and streamlining the specific ways of achieving a positive outcome. The objective of this project is to conduct research on the design and development of new methodologies, adaptive interactive systems, tools and applications, along with innovative solutions for user modelling and learning analytics and, in general, AI-based approaches to provide personalised support in contexts such as e-learning, e-training and e-assistance. The study will rely on publicly available data (e.g., XuetangX) on student logs, and data on student e-tivities and social interactions concerning the Computer Science degree in e-learning jointly held by Sapienza and Unitelma.
Advances in Personalised Learning are essential to cope with the various learning abilities of the students. Personalised learning reveals a drastic change in devising a syllabus that is adaptable to the students¿ needs instead of a traditional one that teachers organise before the course starts. The current lack of ad-hoc tailored studying pathways hinders students fully interacting with the course. An investment in automatic systems able to attain the interaction behaviours of each user providing them with assistance throughout the completion of the course is beneficial for both students and teachers/tutors. Although the idea of adaptive content in education has been present since the mid-1980s [13], its adoption is not spread throughout the online learning environments. Moreover, because there is a gap in the literature of adaptive learning, the study and creation of such systems are beneficial to lay the foundation of personalised and self-personalisable pathways in the telematic courses at Unitelma Sapienza. In detail, we aim for the first time to provide a large-scale analysis of personalised learning environments (PLEs) according to the following:
1. A thorough study of the current limitations of online learning environments (OLEs) to assess the need for tailored content in order to improve retention rates.
2. The realisation of a dynamic and time-related system capable of categorising students into behavioural profiles that give access to different learning paths. We aim to use this system to amplify the effects on the efficacy of the courses offered by Unitelma Sapienza.
3. An investigation of the relationship between learning paths and gender since women are underrepresented in the ICT sector and computer science degrees.
In conclusion, we believe that understanding and creating PLEs will help build the next generation learning environment. For this reason, we deem it fundamental to evaluate more carefully a prototypical system for a large collection of online courses (i.e. MOOCs and e-degrees) and discover the relationships between learning profiles and course/degree completion status.
[13] Bloom, B.S., 1984. The 2 sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational researcher, 13(6), pp.4-16.