Automatic short answer grading (ASAG), a hot field of natural language understanding, is a research area within learning analytics. ASAG solutions are conceived to offload teachers and instructors, especially those in higher education, where classes with hundreds of students are the norm and the task of grading (short)answers to open-ended questionnaires becomes tougher. Their outcomes are precious both for the very grading and for providing students with “ad hoc” feedback. ASAG proposals have also enabled different intelligent tutoring systems. Over the years, a variety of ASAG solutions have been proposed, still there are a series of gaps in the literature that we fill in this paper. The present work proposes GradeAid, a framework for ASAG. It is based on the joint analysis of lexical and semantic features of the students’ answers through state-of-the-art regressors; differently from any other previous work, (i) it copes with non-English datasets, (ii) it has undergone a robust validation and benchmarking phase, and (iii) it has been tested on every dataset publicly available and on a new dataset (now available for researchers). GradeAid obtains performance comparable to the systems presented in the literature (root-mean-squared errors down to 0.25 based on the specific tuple $$\langle $$⟨ dataset-question $$\rangle $$⟩ ). We argue it represents a strong baseline for further developments in the field.

At first glance, for those who start out in it the academic environment may seem attractive, but they soon experience the difficulties inherent in this type of career. At the same time, the academic sector is crucial to the social, cultural, and economic development of any country. Given this important role, it is fundamental for the decision makers to guarantee the best return on investment made into this sector. The good health of workers has important implications for the quality of their lives since it affects their level of productivity at work, and it is especially relevant for research programmes, where most of the work is intellectual. In the present research, we have analysed the health of workers without tenure in the Italian academic environment, i.e. PhD students and short term contract researchers, in order to understand which factors have the most relevant impact on their state of health. 699 participants (398 females, 301 males) completed an online questionnaire that included both ad hoc Likert-scales and open-ended questions. Our results, elaborated through Structural Equation Modelling and Text Mining techniques, show how researchers experience high levels of anxiety both from the characteristics of the academic environment and from the career advancement system. Specifically, both job-related factors (i.e. perception of fairness, professional growth, and safety perception) and relational factors (i.e. relationships with supervisors and colleagues) predict the anxiety of non-tenured researchers. Furthermore, women researchers show a high level of anxiety compared with male researchers. Policy implications of our findings are provided.

The dropout phenomenon in higher education refers to students leaving their programs before completing their degrees. Despite recent improvements in graduation rates, Italy remains among the lowest in OECD countries, with a graduation rate of 45%, well below the OECD average of 69%. High dropout rates, particularly during the first two years of study, are influenced by both institutional factors, such as the decentralization of teaching and the expansion of degree programs, and student-specific factors, including academic performance and geographic distance from universities. Although reforms like the Bologna Process’s 3+2 model have aimed to improve retention, challenges persist. This paper employs multinomial logistic regression analysis, combined with finite mixture models, to explore the complex factors driving student dropout, offering insights for early intervention strategies and institutional reforms aimed at mitigating this issue.

Due to COVID-19, higher education institutions all over the world transitioned to online learning. The sudden and forced transition to this new learning methodology pushed the Universities to rapidly adequate to the needs, upgrading their digital platforms to comply with the new requirements. In the same way, teachers had to adapt their teaching to fit the new medium’s potentials and limitations. The final receivers of this striking change, the students, had to adequate to the novelty approach, though this process has not been painless. Several difficulties, challenges and opportunities arose in this transition process for students, and the full digital class delivery also stressed them emotively. This study explores University of Foggia students’ perceptions of the emergency online learning. The factors analyzed involved their perception about the University implementation of the online class delivery, their consideration about the future of online learning and their emotional impact in attending courses with this methodology. Quantitative and qualitative data were collected from 3,140 participants. The findings present how students have globally appreciated online learning, but they mostly prefer a blended learning approach. Furthermore, by using segmentation variables, differences emerged among the participants’ groups, indicating online learning can have great potentials, but more complex and integrated approaches are needed to fulfill the different learning needs.

Textual data analysis is critical for monitoring changing themes over time. To overcome challenges posed by data richness, graph theory emerges as a tool for investigating word-topic associations. We present an approach to clustering co-occurrence word networks that prioritises network similarity quantification over time. Addressing theoretical and network geometrical constraints, a statistical framework for manifold data analysis facilitates the grouping of semantic networks, partitioning the observed time frame into periods, and identifying dominant topics in each period via tensor decomposition. The analysis of Brexit-related tweets demonstrates the efficacy of modern methods for identifying social media patterns on public discourse.