AI Summary & Key Takeaways
AI InsightUltimately, this framework provides a scalable, data-driven methodology that transitions ISAR analysis from static compliance auditing to proactive strategic management.
📌 3 Key Takeaways
Abstract
While Institutional Self-Assessment Reports (ISARs) are critical for higher education quality assurance, their manual evaluation is highly subjective and time-consuming. This methodological gap prevents university administrations from transforming retrospective text data into predictive governance tools. Addressing this gap, this study proposes an AI-supported analytical framework to evaluate the ten-year ISARs of Ankara University (2015-2024). The dataset combines unstructured qualitative narratives with specific quantitative Institutional Performance Indicators (IPIs), including staff size, student enrollment, and project budgets. Using Latent Dirichlet Allocation (LDA) for topic modeling and lexicon-based sentiment analysis, the research identifies key thematic shifts and the emotional tone of the reports. These text mining results were cross-referenced with quantitative IPIs using K-Means clustering, Isolation Forest for anomaly detection, and Long Short-Term Memory (LSTM) networks. The analysis reveals a distinct evolution in institutional priorities: an initial focus on administrative processes shifted toward research, followed by a recent emphasis on societal contribution and education. Notably, 2020 emerged as a statistical anomaly, capturing the systemic shock of the pandemic. Furthermore, LSTM projections forecast continued growth in selected IPIs for 2025. Ultimately, this framework provides a scalable, data-driven methodology that transitions ISAR analysis from static compliance auditing to proactive strategic management.