Research On Hybrid Attribute Data Knowledge Acquisition Method Based On Granular Computing

With the development and application of the Internet of Things,a vast amount of data has been formed in various industries.How to filter data noise from the massive data to acquire valuable knowledge has become a hot research topic in information science.In addition,the data in the information system is constantly changing.How to use the relationship between the old data and new data to update the information efficiently is also one of the important issues in the field of information science.Granular computing can effectively update the knowledge in a dynamic information system and capture the potential value contained in the data by using a divide-and-conquer strategy.Based on the granular computing,neighborhood rough set model and incremental learning technology,this paper studies the incremental attribute reduction algorithm under the hybrid dynamic information system.The main work of this paper is as following:(1)Aiming at how to carry out effective attribute reduction when the attributes change in the hybrid dynamic information system,this paper analyzes the metrics of the hybrid attribute data.Then the incremental update of the knowledge granularity is explored when the new attributes are added into the hybrid information system.Based on the incremental update,this paper presents an incremental attribute reduction algorithm for hybrid information system.Finally,the effectiveness of the presented algorithm is verified by the UCI data set.(2)Aiming at how to carry out effective attribute reduction when the sample set changes in the hybrid dynamic information system,this paper uses the representation form of knowledge granularity in the neighborhood model and splits the information system into multiple sub-information systems.Then the incremental mechanism and calculation method of the knowledge granularity are given when the sample set of the information systems is increased or decreased.Based on the incremental mechanism,an incremental attribute reduction algorithm of the hybrid information system with the sample set changing is presented.Finally,the effectiveness of the presented algorithm is verified by the UCI data set.(3)Aiming at how to improve the efficiency of the attribute reduction algorithm when the sample changes in the massive hybrid dynamic information systems.Thispaper analyzes the theoretical results of the existing parallel reduction algorithm and uses the incremental attribute reduction algorithm as the theoretical basis to find the appropriate Key/Value pairs and the Map/Reduce functions.Then,a parallel incremental attribute reduction algorithm which provides an effective method for attribute reduction of the information system is designed.The effectiveness of the designed method is verified by the experimental comparison.In this paper,the knowledge granularity in the hybrid information system is analyzed by using the neighborhood rough set model and the granular computing theory.Then the knowledge granularity increment methods are discussed when the attribute set and the sample set change in the information system.Based on the increment method,this paper puts forward an incremental attribute reduction algorithm which can provide the theoretical support for the rapid update of attribute reduction results in hybrid information systems.The effectiveness of the designed dynamic attribute reduction algorithms is verified by the experimental comparison.The research in this paper extends the application range of knowledge-based granularity reduction algorithm and provides a new method for improving the efficiency of attribute reduction in dynamic hybrid information system.