Design And Application Of Voting System Under Multi-Level Full Information

With the rapid development of information technology,the traditional voting methods are difficult to meet the practical needs.This makes the voting behavior more formal than meaningful in voting decisions,so the limitations of voting methods are disclosed based on the study of previous literature and then combined with the actual situation.First,most of the traditional voting methods are comprehensive evaluation of candidates and thus ignore the existence of hierarchical differences among candidates;second,the existing voting methods can hardly reflect the subjective preferences of the decision-making group in a simple and effective way;third,in the face of the increasing complexity and increase of voting data,it is difficult for the traditional voting methods to fully realize the mining and analysis of the data.And in recent years,machine learning methods are widely used in various fields because they all show good performance in processing data.Therefore,this paper combines the support vector machine model with voting theory to construct a full-information voting learning model,and on this basis,we design a multi-level full-information voting learning system that meets the characteristics of the model and the practical needs.The voting method is to select the most satisfactory one from multiple alternatives,while the support vector machine model is to seek to an optimal hyperplane that maximizes the interval,thus achieving the distinction between positive and negative example points.Therefore,from a theoretical point of view,both are aimed at achieving the distinction between data(candidates),and this purpose provides an opportunity to combine the two.Therefore,in this paper,we analyze the voting theory and support vector machine method in depth,and construct a full-information voting feature space to realize the combination of the two,and then construct a full-information voting learning model.On top of this,in order to realize the voting learning model to recognize multi-category data,this paper also combines four multi-categorization strategies to construct a one-to-many all-information voting support vector machine(RIF-SVM),a one-to-one all-information voting support vector machine(OIF-SVM),a decision tree all-information voting support vector machine(DIF-SVM)and a directed acyclic graph all-information voting support vector machine(BIF-SVM),respectively.BIF-SVM).Six voting datasets were then used to validate the classification recognition performance of the models,and the results showed that the variation of voting size and voting data sample size did not significantly affect the model performance in the case of small and medium data sample sizes,and the BIF-SVM model had the best performance among the four types of models.Since the noise problem is unavoidable in practical applications,the robustness of the model is also analyzed by noiseresistance experiments.In the noise-resistance experiments,the performance of each model decreases significantly after adding 5% and 10% of noise data to the original voting data,and the more the proportion of noise data is added,the worse the performance is.Although the performance of the BIF-SVM model is still optimal,its performance has drawn close to the rest of the models and does not show absolute superiority.Therefore,the best-performing BIF-SVM model is selected as the learning model for the voting system in this paper.In order to apply the learning model in practice it is also necessary to set up the corresponding voting learning system according to the model characteristics.The design of the system is divided into three main modules,namely,information input module,information processing module and information integration module.Through the definition and design of these modules,the voting process and counting method are reconstructed,which not only can realize the multi-level examination of candidates,but also can easily and adequately reflect the preferences of all decision makers,and because the system uses machine learning methods to find the mapping relationship between the voting data set and excellent employees,the identification results of the system can largely improve the scientific and rational nature of voting decisions.It also reduces the subjective judgment of decision makers to a certain extent,thus improving the fairness of decision making.After the construction of the voting learning system,the system is also implemented through a real-life case study.In this case,the background of the application,the selection of indicators and the training of the model are introduced in detail,and finally the analysis and decision making are carried out based on the model results.In this part,we not only demonstrate the implementability of the multi-level full-information voting learning system,but also re-present the advantages of the system,so that the voting learning system constructed in this paper can not only provide reference ideas for further development of voting methods,but also provide reference opinions for real-life voting decision-making activities.