A model proposal for assessment of institution maturity level using artificial intelligence methods in higher education quality processes

The Higher Education Quality Council (YÖKAK) determines universities' quality maturity levels using the Rubric Assessment Key (RUB) included in the Institutional Internal Evaluation Report (KIDR). However, studies have shown that there are inconsistencies in the processes of determining universities' RRA levels, discrepancies between internal and external evaluation scores, and complete objectivity is not achieved. This thesis proposes an artificial intelligence-supported application model for evaluating quality assurance processes in higher education institutions. The primary objective of the research is to objectively classify institutions' quality maturity levels by analyzing texts in the KIDR using natural language processing (NLP). The research was conducted using datasets generated from four main areas: leadership and governance, education and training, research and development, and social contribution, based on the YÖKAK DDA structure. These datasets were compiled at the sentence level from the internal evaluation reports of 103 accredited universities and labeled according to five maturity levels. In this study, classification performances were compared using both machine learning algorithms (Naive Bayes, Logistic Regression, Support Vector Machines, Random Forest, XGBoost) and deep learning models (BERTurk, Electra_TR, RoBERTa, TURQUA, XLM-R DistilRoBERTa). Data preprocessing and modeling were performed in the Google Colab Pro+ environment using Python. The results show that the models exhibited accuracy between 80.58% and 93.63%, with deep learning models generally achieving higher performance. During the implementation phase of the research, a web-based decision support system was developed to visualize model outputs. This system allows users to instantly view maturity level estimates by entering text. In conclusion, this thesis presents an innovative model that combines text-based data analysis and artificial intelligence methods, contributing to the digital transformation of quality assurance processes in higher education institutions. The developed system facilitates the objective analysis of internal evaluation reports and provides an infrastructure that can be used as a decision support tool for quality management units.