The Algorithm Research And Implementation Of Knowledge Classification Based On Rough Set And SOM Neural Network

With the economic development and social progress, information and knowledge in the life come into an explosive growth, if not effectively deal with, it will have a lot of data waste。Information increases while the valuable information becomes more difficult to obtain. It becomes an important research topic in the information age that how to deal with complex data and extract valuable information.In the field of computational-intelligence, there are a large number of mature theories and methods for the automatic identification and classification of knowledge. As a powerful tool in data analysis, Rough set has unique advantages in solving such problems such as knowledge classification compared with other data mining techniques, the use of rough set classification does not require prior knowledge, it can also be excavated better knowledge of the rules or to generate easy to understand rules from the less data while the method is simple and easy to operate. Meanwhile as a powerful tool to study the complexity, neural network technologyhas demonstrated the superiority of its extraordinary in recent years in pattern recognition and classification, identification filtering, automatic control, forecasting, etc., especially to handle any type of data, which many traditional methods can not match with.Through a combination of theory and methods to resolve practical issues is one of the main directions of research. Therefore, this study uses rough set and SOM neural networks and their advantages in dealing with massive data classification application and advantages of the discrete data, filtering, clustering analysis, data reduction, rule extraction and other steps in the flexible use of various algorithms, through the analysis of rough set theory and knowledge of SOM neural network research and application in knowledge classification to achieve a new combination of knowledge classification model.In the new model of knowledge classification, knowledge classification is divided into three areas, including the key attributes of the original data extraction and the discretization of continuous data, , attribute reduction of rough set data and the use of dimension reduction properties of SOM accumulation in low-latitude data class analysis. Meanwhile, in the classification model for the new knowledge in rough set attribute reduction , an aspect of a faster reduction algorithm is proposed which is innovative and efficient Thesis, the paper use C + + programming language to realizes basically the entire knowledge classification model of the three parts, and use part of the geophysical monitoring indicators data from the Central Geological Prospecting Fund Remote Monitoring Platform into the total program implementation to obtain the corresponding knowledge classification results. Through the results analysis, the new knowledge classification model of rough set and SOM neural network is proved to effective, a contrastive analysis of efficiency and feasibility in new algorithm and other traditional algorithms is in detailed, shows that this algorithm is effective.