Extensible Multi-criteria Optimization Classifier for Prediction of Chinese Semantic Word-formation Patterns
Zhiwang Zhang, Xinseng Li, Hailin Zou, Shiyong Kang, Guangxia Gao · Procedia Computer Science · 2013
Data mining has become crucial in modern science and industry. Data mining problems raise interesting challenges for different research domains. For the supervised learning methods, owing to the class-overlapping, inconsistent, non- compatible, and contradictory problems in real world applications, the predictive performance of multi-criteria optimization classifier (MCOC) and other traditional data mining approaches will rapidly degenerate. In this paper we put forward an novel extensible MCOC (EMCOC) based on the dependent function: firstly the matter-element models of input data and test data are built, then the weighted dependent degree of data are computed according to the definition of the dependent function, and EMCOC based on the matter-element models is built for predicting the patterns of Chinese semantic word- formation. Our experimental results and comparison with MCOC show that our proposed approach can increase the separation of different patterns, the predictive performance of semantic pattern of a new compound word. In conclusion, we know that some extension-based methods can be used to effectively deal with such challenging problems in data mining.