The Study And Application Of The Similarity Method In Examination Analysis System

With more and more examination types, an examination analysis system is needed to categorize, manage, analyze and compare different examinations and the analysis result is treated as the basis guiding our work and decision, which is very significant to the enrollment and examination management institutions. Due to complexity in the information of the examinees, in order to reduce the complexity of examination categorizing system and improve the efficiency of the systematic analysis, the Data Mining method is used to treat and analyze the data. The common methods in the Data Mining include statistical analysis, neural network, fuzzy theory and genetic algorithm. Due to the uncertainty in the information of the examinees, fuzzy methods are used to transform the data into categories.Based on the characteristics of interval-valued fuzzy sets, this research paper conducts a fuzzy treatment of the information about the examinees and gets n intervals of attributes which reflect a class of each examinee such as the interval of age, the interval of educational background, etc. These n intervals of attributes stand for different classes of the examinees. According to many similarity metrics algorithm of interval-valued fuzzy sets in the aspect of the correlative degree, the similarity is analyzed among different classes of the examinees, which is used to judge the similarity of the student sources among different examinations in order to meet our requirement of the examination analysis decision.After testing the presently common-used similarity methods, we find such methods as the traditional Dengfeng-Chuntian method, Michell method are proposed based on specific environment and hypothesis and the accuracy is influenced greatly by their preconditions and fail to meet the different requirements in accuracy of the system in different analysis process. Therefore, as the theoretical basis of our future research, we uniformly improve the presently common-used similarity methods by combining the details in the examination analysis system. Based on the presently used methods, we propose a new similarity method with its structured function, then we explain the newly structured function with examples and we can further structure the similarity function of the needed accuracy according to different requirements of the systems.