| In the era of big data,with the development of various businesses and the maturity of information technology,enterprises have gradually accumulated a huge amount of data in their information systems,which contains huge business value behind the data.However,in the process of enterprise development,low information sharing and mutual disconnection between information and business processes and applications are common among the information systems of various businesses,which easily lead to the formation of information silos in enterprises.Enterprises need huge technical data support to realize the procedural industrial processes and technologies,which requires not only data sharing by solving the problem of information silos within the enterprise,but also data integration with many different industrial enterprises.Therefore,to achieve data fusion within or between enterprises,some enterprises start to mine the correlation relationship between data and try to connect the accumulated huge amount of data to build a data space.In response to this phenomenon,this paper proposes a dynamic probe based multi-source heterogeneous data fusion model to realize the fusion between data,which is mainly as follows:The data fusion visualization model based on log analysis is proposed for the relational data widely used in enterprise business systems.By analyzing the full-link access logs of different systems,the dynamic association model between cross-domain data is established,and the quantitative results of association after data fusion are given in the form of forceoriented graph visualization,which can explore the evolution law of association relationship between data over time,discover the hot data in information systems and explore the data access differences between different users and other effective information,and provide visualization support for cross-domain data fusion,data space construction and business process optimization.The data fusion framework based on fuzzy logic is proposed.By using dynamic probes to collect key information such as global data structures of different application systems,data involved in system operation and various log files,a data association model for enterprise data space is constructed at three levels: schema,instance,and log,respectively.In the schema association level,the multi-dimensional similarity discrimination algorithm combined with semantic analysis is used to realize the similarity mining of schema data;in the instance association level,the self-coding algorithm combining feature vector similarity analysis and deep learning is used to complete the association matching of different data for structured data such as numeric and character type and unstructured data such as long text type;in the log association level,the equivalence relationship in logs is established by analyzing data access In the log association level,the association between different data is established by analyzing the equivalence relationships in the logs.In addition,to solve the problem that the similar values of data in multiple dimensions cannot be compared,a fuzzy logic-based inference model is used to construct a normalized value to measure the similarity of data by combining the similarity values of data in different dimensions,to realize data fusion based on fuzzy logic.Based on the two schemes proposed above,a data fusion visualization system based on dynamic probes is built.By deploying dynamic probes to collect and store the data to be analyzed in different dimensions,the two schemes based on log and fuzzy logic analysis are algorithmized as the root of data fusion analysis,and a data visualization system with front and back-end separation is built.The system is designed,developed,and deployed around data analysis to meet users' data fusion visualization needs,and the proposed scheme is verified to be of practical application value in the process of data fusion and data space construction through the built system. |
PDF Full Text Request