Abstract: To provide the effective assisting means for product's design assessment,four interface layout elements that influenced user perception were analyzed and twelve metrics of interface layout characteristics were proposed by around  the user perceptual prediction of product information interface.The reliability of the interface layout metrics system was verified,and the result showed that the layout metrics system met the basic set of four potential factors,but the multiple correlation was existed between the layout metrics and subjective evaluation which could improve the initial perceptual mapping model.The different prediction models were created by using two types of applied neural network and linear regression methods.Through contrasting the predicted deviation of test samples,the result showed that linear regression prediction model had higher degree of data fitting and the function relationship of perceptual prediction model AM=g{f(M)} was deducted.Theoretical basis of interface layout design and program evaluation were also provided.

Key words: interface layout, perceptual prediction, confirmatory factor analysis, neural network, linear regression, product design

摘要： 为了给产品的设计评价提供有效的辅助手段,围绕产品信息界面的用户感性预测,分析了影响用户感性的四个界面布局要素,提出12个界面布局特征的衡量指标;验证了界面布局指标体系的可靠性,结果表明布局指标体系符合4个潜在因子的基本设定,但布局指标与主观评价间存在多元相关性,从而改进了最初感性映射模型;运用神经网络和线性回归两类方法分别建立了预测模型,通过比对测试样本的预测偏差表明线性回归预测模型的数据拟合度更高,推导了感性预测模型的函数关系AM=g{f(M)},并结合实例对预测模型进行了验证。该研究成果为界面布局设计和方案评估提供了理论依据。

关键词: 界面布局, 感性预测, 验证性因子分析, 神经网络, 线性回归, 产品设计