Multi-modal online review driven product improvement design based on scientific effects knowledge graph
Ruiwen Wang,Jihong Liu,Mingrui Li,Chao Fu,Yongzhu Hou
DOI: https://doi.org/10.1080/09544828.2023.2301229
2024-01-08
Journal of Engineering Design
Abstract:Online reviews serve as significant channels for users to express their preferences, constituting an essential data source for enterprises to identify product requirements. However, with the widespread adoption of smartphones, the act of capturing spontaneous photographs has become a habitual practice for the majority, resulting in the increasing prevalence of supplementary visual expressions within online reviews. Therefore, an important research question emerges: How can product requirements be effectively extracted from multimodal online reviews and subsequently translated into product design proposals? In this paper, we establish a framework, seamlessly integrating aspect-based sentiment analysis, product requirement identification, and requirement mapping based on a scientific effect knowledge graph. Firstly, we conduct aspect term extraction on the online reviews, followed by aspect sentiment classification. Subsequently, we delve deeper into the analyzed results obtained from aspect-based sentiment analysis to identify preferences in product requirements. Finally, we employ requirement mapping based on a scientific effect knowledge graph to generate proposals for product design improvements. To validate the efficacy of our approach, we conducted experiments and the results demonstrate that our method outperforms alternative approaches, while the requirement mapping based on a scientific effect knowledge graph efficiently facilitates the realisation of product design improvements.
engineering, multidisciplinary