Abstract: To address the limitation of traditional teaching evaluation methods in effectively handling the fuzziness and uncertainty embedded in students' textual feedback, a Novel Integration Fuzzy Model (NIFM) that combines text mining, sentiment analysis, and intuitionistic fuzzy sets is proposed. In this model, specific content related to evaluation indicators is first extracted from unstructured comment texts through Chinese word segmentation and keyword mapping. Subsequently, a lexicon-based sentiment analysis approach is adopted, and two sentiment quantification strategies, namely frequency statistics and intensity weighting, are designed. The results of the sentiment analysis are then converted into intuitionistic fuzzy numbers, by which the complex attitudes of students towards various teaching aspects are precisely quantified using the three dimensions of membership, non-membership, and hesitation degrees. Laplace smoothing and a custom intensity smoothing technique are introduced to ensure computational stability. The experimental results indicate that this model is capable of surpassing simple satisfaction statistics, thus providing a quantifiable decision-support tool for the fine-grained assessment and continuous improvement of teaching quality. The main contributions of this paper include: (1) methods for the acquisition, processing, and fusion of student and expert evaluation information are proposed; (2) a dynamic weighting method based on the reliability of evaluation results is designed; (3) the effectiveness of the proposed method is proved through case studies and comparisons with existing Multiple-Attribute Group Decision-Making (MAGDM) methods.