Digital Reactions, Discourse and Sentiment on Educational Reform: A Text Mining Study of the Türkiye Century Maarif Model on X

Abstract

The Türkiye Century Maarif Model, introduced by the Ministry of National Education in 2024, represents a major curriculum reform. Although previous research has examined teachers, administrators, and teacher candidates mainly through surveys and interviews, spontaneous digital reactions remain underexplored. This study investigates discourse surrounding the Maarif Model on X. Posts containing five reform-related hashtags were collected through the X API and analyzed in Python using a KDD-informed text-mining framework. The analysis combined BERT-based sentiment classification, word-frequency analysis, and Latent Dirichlet Allocation topic modelling. Data preparation included duplicate removal, relevance screening, and bot/noise filtering, while automated sentiment labels were interpreted cautiously because no independently coded gold-standard validation set was available. Positive posts constituted the largest sentiment category across all hashtag-based subsets, suggesting substantial symbolic support for the reform. However, negatively oriented posts contained more prominent references to projects, targets, administrative expectations, and implementation-related concerns. Topic patterns also linked the reform discourse to curriculum, teacher roles, student-centered approaches, institutional coordination, and local implementation. Broader hashtags were associated with relatively more critical or contested expressions, whereas institutionally aligned hashtags contained stronger positive discourse. These findings suggest that social media can provide an early source of feedback on large-scale educational reforms, although it should not be treated as representative public opinion. The study illustrates how computational analysis of digital discourse can complement traditional policy-evaluation methods.