Abstract: To meet the ability requirements of the enterprises to be able to analyze and predict customer preferences with the rapid change of customer preference for products,a customer preference identification and prediction method for product improvement was proposed.The Long Short Term Memory (LSTM) model was constructed to predict the emotion value and importance during product design iteration,and the prediction accuracy was calculated.Through the product design improvement model based on the emotional change mode of product features,the change mode of each feature was judged,and the product features to be improved and the improvement priority were defined.Finally,the online review of DJI Mini 2 UAV was taken as an example to verify the effectiveness of the method.

Key words: online reviews, product design improvement, customer preference, preference prediction

摘要： 针对顾客产品偏好快速变化对企业分析和预测顾客偏好能力的要求,提出一种面向产品改进的顾客偏好分析与预测方法,首先构建长短期记忆网络模型,预测产品设计迭代期间的情感值和重要度,并计算预测准确度;然后通过基于产品特征情感变化模式的产品设计改进模型判断各个特征的变化模式,明确待改进的产品特征及改进优先级;最后以DJI Mini 2无人机的在线评论为例验证了方法的有效性。

关键词: 在线评论, 产品设计改进, 顾客偏好, 偏好预测