Complex Network Construction And Visual Representation Of Big Data Information Space

Statistical analysis methods based on complex network have become a research hotspot in the era of artificial intelligence of Big Data,and have actively promoted the development of visualization of Big Data.From visual organization of large-scale data,traditional visualization form is limited by information space constrains.As a representation with high dimensional and high complexity in big data visualization,the representation form of complex network arises,enabling macro perspective of large-scale data using network structure.Current research on visual representation of complex network are still at the stage of exploration due to the lack of specific construction algorithm for complex networks,mostly visual representation of complex network only involves the use of visualization tools,the representation and user experience are limited by the existing visualization tools.Therefore,following Tomas information visualization process,the information visualization of big data is investigated combining theoretical analysis with case calculation,to facilitate an completed reasoning process of complex network with data statistics analysis,data visualization and assessment of data interaction.Firstly,a double exposure between nodes construction algorithms is proposed for the complex network optimization analysis;Secondly,a complex network web visualization platform for user interaction is developed in JavaScript,and successfully applied to the subject system of Massachusetts Institute of Technology;Finally,cognitive performance evaluation of visual representation for the complex network is carried out,and a new experimental methods is proposed and design,the results of algorithm calculation are used as the standard for experimental performance evaluation,and the conclusions of experimental are feed back to the computational application of visual coding.The main content of the whole study is explained as follows:(1)Based on graph theory,scale-free complex network theory and other related theories,the complexity transformation of network structure for complex network is explored;data cleaning and correlation analysis are carried out for the complexity calculated of network space;a double exposure between nodes construction algorithms is proposed for the complex network optimization analysis,the information unit relationship matrix is established;a complex network of subject knowledge is constructed based on information nodes,information relationships and time attributes.(2)Several visual representation algorithms layout is investigated by means of Gephi and Cytospace,As an important practice result of this study,the complex network visualization platform is developed with Javascript programming(a web platform for interactive operation of users)based on the optimized results of algorithm;the T-test theory is innovatively transformed into the constraint calculation of the visualization threshold,and the threshold analysis of percoloation for information flows is performed without discarding the important relationships of information,the weight and quantity coding of nodes,and the weight,quantity,and distance coding of edges are defined.(3)Based on statistical analysis methods of complex network,calculation analysis for the subject space network is performed with the information flow of nodes,community division of information cluster,structure evolution and simplification of information cluster;Furthermore,the results of the list-based are displayed in the form of graphical visualization,combining the calculation analysis with an visual representation.(4)The assessment methods of user cognitive behavior and physiological characteristics are innovatively applied to the search performance evaluation of complex network information flow.According to the scale-free network theory and the Dijkstra algorithm,the path of information flow is given,then a perceptual layering experiment of visual coding of complex network is designed.With the behavioral index and physiological index of eye-tracking experimental equipment,it is proved that the search performance of complex network information flow is interfered by different visual coding,and several suggestions for using visual coding are provided.The predecessor node theory is proposed and its effectiveness used in decision making is verified with the experimental results.