Abstract
University teachers increasingly prioritize career satisfaction and self-realization, focusing on the evolving demands of the academic environment. Understanding these subjective experiences requires data-driven approaches that reveal more profound insights into their professional well-being. This study introduces the Bidirectional Encoder Representations from Transformers -Enhanced Career and Self-Realization Text Mining Framework (BE-CSTMF) to construct a predictive model for university teacher's self-realization and career satisfaction. The framework combines Enhanced BERT's advanced natural language processing capabilities to extract deep contextual insights from textual data. BE-CSTMF integrates these insights with XGBoost's robust predictive modeling to perform actionable analysis using text mining techniques on survey and academic data. Key patterns, sentiments, and influential factors related to career satisfaction and self-realization were identified using a dataset comprising textual responses from university teachers. Predictive modeling techniques enabled a comprehensive understanding of factors such as work-life balance, recognition, professional growth opportunities, and institutional support. The BE-CSTMF model reliably predicts university professors' self-realization and career happiness, revealing their professional well-being characteristics. This strategy provides concrete help for building targeted career and work satisfaction interventions. This study highlights the potential of combining cutting-edge natural language processing and machine learning techniques to develop tailored strategies for supporting academic professionals' career growth.