Research On Evaluation Method Of Relational Network Visualization

The rapid development of the Internet has brought people a large amount of data,and the value implied in the data is naturally more difficult to find.Data visualization technology is one of the ways to solve the above problems.The data with relational connections can be characterized by relational network visualization.However,the visualization results need effective evaluation means.At present,there are many researches on qualitative and subjective evaluation of visualization,and few researches on quantitative and objective evaluation.Compared with the former,the quantitative and objective evaluation method can more accurately represent the evaluation results,and the real-time and repeatability of the experiment will also be improved.Therefore,aiming at the node characteristics in relational network visualization,this paper proposes an objective evaluation algorithm from the perspectives of information quantity,composition method and color psychology.Aiming at the influence of edge distribution and node location on relational network visualization,this paper proposes an evaluation method of edge distribution and location of nodes through the theories of conditional entropy and trisection composition.In the former,according to the edge distribution characteristics of nodes,the probability density function is designed to meet the requirements of conditional entropy in mathematical theory,and then the value of conditional entropy can be used to describe the edge distribution in visualization.Combining the three-step composition method with the two-step mobile search method,and according to the characteristics of nodes in the relationship network,the method of calculating the supply-demand ratio in the first step of the two-step mobile search method is improved.Finally,the concept of the prominence of nodes in the relationship network is obtained,and the value of the prominence is used as the evaluation basis of the location of nodes in the visible area.In view of the influence of node color on the visualization effect of relational network,based on color psychology,this paper evaluates the color from three aspects: hue,saturation and lightness,and divides it into hue,saturation and lightness.Finally,the two calculation methods are combined to obtain the overall evaluation method of color.Among them,the evaluation method of hue is based on the cold and warm degree of color,while the evaluation method of saturation and lightness is related to the stimulus a in pad emotion model.Finally,fuzzy analytic hierarchy process is used to combine them.Through the designed objective experiment and subjective experiment,the proposed evaluation method is verified,and the results show that the objective evaluation method is reasonable.The algorithm proposed in this paper can obtain an objective and theoretically supported relationship network visual evaluation result.If it is used instead of the subjective evaluation method,it can ensure the real-time and repeatability of relationship network visual evaluation.