本研究旨在應用模糊理論截矩陣（a-cut），以及察覺的模糊邏輯模式（fuzzy logic model of perception）測度，改進傳統詮釋結構模式（interpretive structural modeling）受限於二元資料的限制，並應用於心理計量之實例分析。研究者首先說明詮釋結構模式的理論，其次引入察覺的模糊邏輯模式，說明非二元關係資料亦可進行詮釋結構分析。研究者以分數減法概念為實例，根據試題反應理論分析所得的反應機率資料，轉成概念屬性矩陣，進行模糊關係矩陣詮釋結構模式分析，以瞭解受試者的分數減法概念之結構。實例分析中，發現不同能力值的受試者，其分數減法概念結構各有其特徵；而且在傳統計分相同的情形下，其分數減法的概念結構亦有所不同。本研究的方法論及實例可提供心理計量或課程單元等有關複雜元素的系統化決定。最後，研究者根據研究心得與發現，提出未來有關應用及進一步研究之建議。

The purpose of this study is to improve the Interpretive Structural Modeling (ISM) and apply the improved ISM in the analysis of concept structure. Due to the constraint of binary-relationship in ISM, the improved model is based upon the a-cut of fuzzy theory and fuzzy logic model of perception (FLMP). The improved ISM can expand the applications of the source data format. Therefore, in this paper, there is an empirical data analysis for subtraction of fraction. From the empirical data analysis of improved ISM, the concept structure for each task-taker can appear clearly.In the empirical data analysis, the graphs of improved ISM for different ability are varied. We can realize the concept structure of each task-taker by his improved ISM graph. On the other hand, it is also clear that the improved ISM graphs are different for task-takers of the same score, which is the same number of correct items. Besides the data of psychometrics, the improved ISM can also apply in the other domains. For example, the data of research about management, curriculum is suitable for improved ISM. Finally, based upon the findings of improved ISM, some recommendations for empirical study and future research are suggested.