Exploring Artificial Intelligence in Higher Education through Lightweight Natural Language Processing
Antonio Matas Terrón, José Manuel Ríos Ariza, Antonio Luque de la Rosa, José Jesús Sánchez Amate
Source abstract
The integration of Artificial Intelligence (AI) in education has prompted a wide array of perspectives among educational stakeholders, frequently captured through open-ended survey responses. However, approaches to analysing such data under non-ideal conditions remain limited. This study examines the feasibility of combining open-ended responses with "Light Natural Language Processing" techniques to explore the discourse on AI in higher education. Employing an exploratory proof-of-concept design, the research utilised two independent datasets comprising 31 Pedagogy students from Spain and 35 teachers from Peru, all of whom responded to open-ended questions regarding Artificial Intelligence. Semantic representations based on embeddings—derived from term weighting and truncated Singular Value Decomposition (SVD)—were employed alongside semantic anchors and exploratory visualisation to analyse heterogeneous texts. The results demonstrate that light semantic workflows can generate interpretable representations, support proximity-based inspection, and enable anchor-guided exploration, even with small sample sizes. Specifically, the analysed data reveal that while student discourse is oriented towards pragmatic utility and self-regulated learning, the teacher narrative is dominated by ethics and regulatory anxiety. The paper concludes by reflecting on the potential of this approach as a scalable tool for monitoring the educational climate in real time.
Evidence graph
No public relationships recorded yet.
Integrity note: This page is a factual metadata record created by deterministic ingestion. It is not a claim that the work moves a mathematical frontier or has been independently verified.