Making learning visible in architectural internships: NLP-supported learning evidence for work-based learning support
Rui FuPurpose
This paper examines how natural language processing (NLP)-supported learning evidence generated from internship logs can make workplace learning in architectural internships more visible and more actionable for higher education teachers.
Design/methodology/approach
A qualitative implementation study with embedded case analysis was conducted in a five-year architecture programme at a university with an applied education mission in China. The NLP-supported evidence-generation process was calibrated using internship logs from 289 students across five cohorts between 2020 and 2024 and was applied to all 47 students in the 2025 cohort. Qualitative inquiry comprised two semi-structured student group interviews involving 12 students in total and a separate teacher group interview involving four teachers. Two student cases and one programme-level case were examined in detail.
Findings
Four types of learning evidence were generated: Practical Competency Profile, Key Confusion Identification, Internship Experience Perception and Career Interest Exploration. The two student cases show how learning evidence helped teachers identify a developmental interest beyond routine tasks and respond to sustained difficulty in drawing revision. The programme-level case shows how aggregated evidence informed discussion of curriculum support for sustainable design. Learning evidence provided clearer and traceable process materials for teacher interpretation, student-teacher dialogue and programme-level reflection, supporting more timely and contextually grounded pedagogical judgement.
Research limitations/implications
The study concerns one architecture programme and three explanatory cases. Further research should examine the mechanism across disciplines, institutions and longer-term learning outcomes.
Practical implications
The mechanism can help teachers organise continuous internship logs, prepare focused support dialogues and identify recurring curriculum support issues from workplace learning records.
Social implications
The study supports transparent and supportive use of educational data by keeping teacher mediation, participant protection and non-assessment use central.
Originality/value
The paper contributes a learning-evidence mechanism that links workplace experience, reflective logs, teacher mediation and programme-level feedback. It shows how NLP can strengthen work-based learning support by organising reflective texts into interpretable evidence for teacher review and pedagogical dialogue.