基于在线产品评论和TextCNN的机械产品感性评价方法

Perceptual Evaluation Method of Mechanical Products Based on Online Product Review and TextCNN

机械产品感性评价是感性工学设计的重要内容。为了克服传统感性评价方法主观强、实时性差等问题,获取更为准确的评价,提出结合在线产品评论和TextCNN的机械产品感性评价方法。利用TextCNN的情感分析模型先判断评论的情感极性,并结合评论中程度副词得到感性词评价值,构建感性评价值与机械产品属性参数的非线性映射模型,用于预测产品属性参数与用户感性评价间的关系,为机械产品参数设计提供一定参考意见。案例以汽车网络平台数据为例,分析六组感性词映射模型预测性能,验证了该方法的有效性和可行性。

Perceptual evaluation of mechanical products is an important part of perceptual engineering design.In order to overcome the problems of strong subjectivity and poor real-time performance of traditional perceptual evaluation methods and obtain more accurate evaluation,a perceptual evaluation method of mechanical products combined with online product review and text convolutional neural networks(TextCNN)is proposed.The emotional analysis model of TextCNN is used to judge the emotional polarity of the comment,and the emotional word evaluation value is obtained by combining the degree adverbs in the comment.A nonlinear mapping model between the emotional evaluation value and the mechanical product attribute parameters is constructed to predict the relationship between the product attribute parameters and the user's emotional evaluation,as to provide some reference for the parameter design of mechanical products.Taking the data of automobile product network platform as an example,the prediction performance of six groups of perceptual word mapping model is analyzed,verifying the effectiveness and feasibility of this method.