Using Natural Language Processing (NLP) to Analyze Education Policies: A Systematic Review

As governments worldwide generate an unprecedented volume of education policy texts, ranging from curriculum frameworks to legislative reforms, there is growing demand for scalable and rigorous methods to analyze them. This systematic review examines how Natural Language Processing (NLP) has been applied to education policy analysis from 2015 to 2025, highlighting key methodological trends, theoretical foundations, and ethical considerations. Framed within a pragmatic research paradigm that values methodological pluralism and practical relevance, the study integrates policy analysis theories, including the policy cycle, top-down and bottom-up models, and policy learning, with computational text analysis techniques. Following PRISMA 2020 guidelines, a systematic literature search was conducted using Scopus, Web of Science, and arXiv, identifying 32 peer-reviewed studies applying validated NLP methods such as topic modeling, sentiment analysis, discourse analysis, and transformer-based summarization to education policy documents. A custom appraisal rubric evaluated methodological transparency, validation metrics, ethical safeguards, and reproducibility. The studies cover 12 countries and diverse policy contexts, though most focus on English-language texts. Findings reveal increasing sophistication in NLP applications, especially through models like BERT and GPT-4, with topic modeling dominating curricular and legislative domains, and sentiment analysis widely used for stakeholder feedback. Critical gaps persist in ethical oversight, multilingual inclusivity, and participatory design. Many studies overlook non-English corpora, limiting global generalizability and reinforcing linguistic inequities. This review offers a theoretically grounded, methodologically rigorous synthesis of NLP’s role in enhancing transparency, efficiency, and responsiveness in education policy analysis. It identifies future priorities including explainable and culturally adaptive models, investment in low-resource language tools, and interdisciplinary collaboration, charting a path toward equitable and evidence-based educational governance worldwide.