上海交通大学学报 ›› 2019, Vol. 53 ›› Issue (7): 830-837.doi: 10.16183/j.cnki.jsjtu.2019.07.009

• 学报(中文) • 上一篇    下一篇

结构方程模型与人工神经网络结合的用户感知建模方法

颜波,褚学宁,张磊   

  1. 上海交通大学 机械与动力工程学院, 上海 200240
  • 出版日期:2019-07-28 发布日期:2019-08-02
  • 通讯作者: 褚学宁,男,教授、博士生导师,E-mail: xnchu@sjtu.edu.cn.
  • 作者简介:颜波(1993-),男,云南省保山市人,硕士生,主要研究方向为智能化与数字化设计.
  • 基金资助:
    国家自然科学基金资助项目(51875345,51475290,51075261)

User Perception Modeling by Combining Structural Equation Model and Artificial Neural Network

YAN Bo,CHU Xuening,ZHANG Lei   

  1. School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
  • Online:2019-07-28 Published:2019-08-02

摘要: 针对现有研究方法难以描述用户使用产品过程中会产生的多种感知之间的非线性关系和影响路径而导致用户感知建模不够真实准确的问题,提出了一种结构方程模型(SEM)与人工神经网络(ANN)相结合的用户感知建模方法.该方法首先利用SEM确定用户感知之间的因果关系和影响用户感知的主要因素;然后将SEM分析的结果转换为ANN模型的拓扑结构,建立结构化的神经网络模型,利用BP(Back Propagation)算法训练模型得到各网络节点间的连接权重,实现了用户感知建模;最后以智能手机用户感知建模为例验证了方法的有效性.分析结果表明,SEM-ANN模型具有良好的拟合优度和可解释性,并能准确地定量表达用户感知之间的相互关系和影响用户感知的因素.

关键词: 产品设计; 用户感知; 用户体验; 结构方程模型; BP神经网络; 智能手机

Abstract: It is difficult for the existing research methods to describe the nonlinear relationship and influence path among the users’ multiple perception constructs during the product usage. This may lead that the user perception model is not real and accurate enough. Therefore, a new method combining structural equation model (SEM) with artificial neural network (ANN) is proposed for user perception modeling. Firstly, based on the results of SEM analysis, main factors that influence user perception and the causal relationship between user perception constructs are identified; Then, the result of SEM analysis is converted to the topology of the ANN model, so that a structured artificial neural network model for user perception is established, in order to get the connection weights between the network nodes the BP (back propagation) algorithm is used to train the model; Finally, the validity of the proposed method is demonstrated by a case study of smart phone user perception modeling, the results show that the SEM-ANN model with good goodness of fit and interpretability can more accurately and quantitatively express the relationship between user perception constructs and the factors that influence user perception constructs.

Key words: product design; user perception; user experience; structural equation model; BP neural network; smart phone

中图分类号: