Research Of Clothing Kansei Evaluation Method Based On Word Vector Clustering

With the rapid development of the economy in recent years,the domestic clothing market is growing continuously.Consumers’ demand for clothing keeps increasing.Meanwhile,public consumption psychology keeps changing from rational to emotional gradually.To gain consumers’ favor in the increasingly fierce market competition,it is essential to pay more attention to their deep emotional needs.It is an important challenge to interpret,extract and predict the implicit emotional needs of consumers and use the result as the guidance of the design process.Based on men’s clothing and Kansei Engineering as the theoretical basis,this paper provides a new way to build Kansei evaluation method of clothing.It also constructs the prediction model and evaluation platform of emotional intention.First of all,this paper summarizes the traditional vocabulary for Kansei evaluation and methods for extracting them and then points out the defects of the traditional ways,which are high dependence of experts,high research cost and poor expansibility.Then,this paper raises the construction method of Kansei evaluation method based on word vector clustering.This method significantly reduces the manpower cost of research and also has good expansibility.Combined with the Kansei evaluation method obtained via word vector clustering,based on man’s suit and the DeepFashion open source dataset,this paper uses a larger scale of datasets,which makes up for disadvantages of small number of samples in the traditional model construction research.Two kinds of models,multiple linear regression and multilayer perceptron neural network,are used to predict the emotional scores of clothing.The model can help designers understand the relationship between design elements and consumers’ emotional cognition.Finally,this paper develops an Kansei evaluation platform,for collecting consumers’ emotional evaluation efficiently.In this paper,word vector clustering has greatly optimized the construction process of the Kansei evaluation method.The prediction model and the consumer oriented Kansei evaluation platform can help designers predict the emotional intention of clothing.They can also help designers understand the relationship between design elements and consumers’ emotional cognition.Thus,it can improve the design process and help design clothes that meet the emotional needs,which leads to optimization and upgrading of the clothing industry.