With the increasing demand for a personalized product and rapid market response, many companies expect to explore online user-generated content (UGC) for intelligent customer hearing and product redesign strategy. UGC has the advantages of being more unbiased than traditional interviews, yielding in-time response, and widely accessible with a sheer volume. From online resources, customers’ preferences toward various aspects of the product can be exploited by promising sentiment analysis methods. However, due to the complexity of language, state-of-the-art sentiment analysis methods are still not accurate for practice use in product redesign. To tackle this problem, we propose an integrated customer hearing and product redesign system, which combines the robust use of sentiment analysis for customer hearing and coordinated redesign mechanisms. Ontology and expert knowledges are involved to promote the accuracy. Specifically, a fuzzy product ontology that contains domain knowledges is first learned in a semi-supervised way. Then, UGC is exploited with a novel ontology-based fine-grained sentiment analysis approach. Extracted customer preference statistics are transformed into multilevels, for the automatic establishment of opportunity landscapes and house of quality table. Besides, customer preference statistics are interactively visualized, through which representative customer feedbacks are concurrently generated. Through a case study of smartphone, the effectiveness of the proposed system is validated, and applicable redesign strategies for a case product are provided. With this system, information including customer preferences, user experiences, using habits and conditions can be exploited together for reliable product redesign strategy elicitation.

中文翻译：

---

基于本体的细粒度情感分析的智能产品重新设计策略

随着对个性化产品和快速市场响应的需求不断增加，许多公司希望探索在线用户生成内容 (UGC) 以实现智能客户听力和产品重新设计策略。UGC 的优点是比传统采访更公正，产生及时的响应，并且可以通过庞大的数量广泛访问。从在线资源中，可以通过有前景的情绪分析方法来利用客户对产品各个方面的偏好。然而，由于语言的复杂性，最先进的情感分析方法仍然不能准确地用于产品重新设计的实践。为了解决这个问题，我们提出了一个集成的客户听力和产品重新设计系统，它结合了客户听力情感分析的稳健使用和协调的重新设计机制。涉及本体和专家知识以提高准确性。具体来说，首先以半监督的方式学习包含领域知识的模糊产品本体。然后，通过一种新颖的基于本体的细粒度情感分析方法来利用 UGC。提取的客户偏好统计数据被转换为多层次，用于自动建立机会景观和质量表。此外，客户偏好统计以交互方式可视化，通过这些数据同时生成具有代表性的客户反馈。通过智能手机案例研究，验证了所提出系统的有效性，并为案例产品提供了适用的重新设计策略。有了这个系统，包括客户偏好、用户体验、