RESUMEN

The establishment of a reasonable teacher evaluation indicator system has always been a research hotspot in teacher evaluation. Simplifying and essential indicators are the foundation for maintaining the speed and accuracy of teacher evaluation. Therefore, optimizing indicators in a convincing and interpretable manner becomes highly important. Due to the complex causal relationships and fuzzy uncertainties among teacher evaluation indicators, this paper proposes a method that combines the Triangular Fuzzy Decision-making Trial and Evaluation Laboratory Model (Fuzzy-DEMATEL) with Backpropagation Neural Network (BP) to optimize the complexity of assessment systems and identify key indicators, thereby establishing a precise and rational teacher evaluation index system. DEMATEL allows us to simplify and analyze the complex causal relationships among assessment indicators, mapping them into a causal relationship diagram. Fuzzy logic effectively handles the fuzzy and uncertain relationships among the indicators, converting fuzzy information into computable forms. The BP neural network is a data training model that, from an objective data perspective, compensates for subjective errors, thereby optimizing our indicator results. In addition, we conducted empirical and comparative research using relevant data from the TIMSS 2019 dataset, and found that this method can reduce the original indicator quantity by approximately 28 %-30 %, compared to methods such as Multi-Criteria Decision Making (MCDM), the results are superior and the indicators are more accurate.