An emotion-oriented product evaluation model based on BP neural network and improved dempster-shafer theory.

In the process of user emotion-oriented product development, emotion perception has become a crucial dimension in evaluating product design. Aiming to address the issues of subjectivity bias and insufficient efficiency in multi-source information fusion in existing methods, this study proposes a hybrid product evaluation framework that integrates a back-propagation (BP) neural network and an improved Dempster-Shafer (D-S) evidence theory. The method constructs input vectors by combining product design features and user features. It utilizes multilayer BP networks to learn the nonlinear mapping relationship between product attributes and user emotions, thereby generating preliminary emotion perception probability assignments (BPA). To further address the issue of conflicting expert opinions, a weighting strategy based on the average distance of the evidence is introduced, and the D-S evidence fusion process is optimized. Ultimately, the improved evidence theory model can effectively unify the judgments of multiple experts and output a unified emotional evaluation result. The empirical study, with the sofa as the research object, shows that the proposed method can effectively identify key design elements and achieve quantitative evaluation in the emotional dimension. Through comparative experiments, this method demonstrates higher stability and adaptability in multi-source data processing and fusion accuracy, as well as good scalability, providing quantitative decision support for user-centered product design practices.