Research And Application Of Customer Decision Analysis For Taobao

The goods in the Taobao mall are all the eyes of the people and involve all aspects of people's lives.When the customer through the terminal login to Taobao mall,the first problem faced is how to choose a favorite shops;and when the buyers into the shops to buy goods,need to further understand the products through browse comments,but due to the large number of "cheating"comments,the comment data set showing a extreme a state of equilibrium,it is difficult to get effective information.Therefore,we need to establish a mathematical model of decision analysis for Taobao customers,make decision analysis on shops and goods,and provide decision support for customers.In order to solve the problem that the traditional TOPSIS decision analysis method is difficult to carry out the decision analysis under the condition that the weight cannot be obtained,the partial order set representation method is introduced into the TOPSIS model,and the robustness of the arrangement is further strengthened.The traditional prospect theory model does not take the subjective risk attitude of the decision-maker into consideration when the decision analysis is made.A decision making method based on prospect theory is put forward.The grey relational coefficient of multiple factors is introduced into the value function of prospect theory,and the optimal weight is obtained through building the model,and then the alternatives are sorted.In order to improve the classification problem of unbalanced reviews dataset,this paper optimizes from two aspects of sample sampling and classification algorithm.In the sample,a new sample set is generated by cyclic oversampling of the few samples of the decision boundary,and the importance of the boundary sample is improved.The positive and negative penalty coefficient and the mixed kernel function are introduced for the traditional epsilon-SVM in the problem of the hyperplane offset in the classification of the unbalanced data sets.Using the objective entropy method to select the penalty coefficient to improve the performance of the classification algorithm.In view of the problem that the proposed classification model is too difficult to determine when the parameters are too many,the method of particle swarm optimization is introduced in this paper.The mechanism of feedback adjustment is introduced on the basis of the traditional particle swarm optimization method,which improves the precision of parameter optimization.Using the two optimized decision analysis methods proposed in this paper,each alternative Taobao shop is sorted intelligently.The experimental results show that the two decision analysis methods proposed in this paper are consistent with the official evaluation results given by Taobao,which proves that the decision method of this paper has a certain value of use.The classification model proposed in this paper is used to classify the comments.The experimental results show that the F-measure value of the classification model in this paper is higher than the standard SVM algorithm.It is proved that the classification model of this paper can provide the decision support for the customer through the accurate classification.