A Sentiment-driven SPSS Requirement Analysis Method Based on Online Reviews

In recent years, the rapid development of information and communication technologies (ICT) has enabled the prevailing digital transformation, bridging the gap between the physical and virtual worlds. Meanwhile, a continuously growing demand for more personalized products leads to the fact that enterprises must diversify their business model from the traditional product range. To adapt to this change, enterprises are adopting a new business model to offer not only physical products, but also personalized services, called smart product-service system (SPSS). Among the SPSS development process, innovation design is one of the most critical steps. How to obtain and analyze user requirements could significantly improve the efficiency and quality of the design process. Therefore, a sentiment-driven requirement analysis method based on online reviews is proposed. Firstly, ontologies are constructed to describe SPSS design knowledge. Secondly, requirements are automatically elicited from user reviews with natural language processing (NLP) and clustered according to their similarity. Next, sentiment analysis is conducted, and a representation method of sentiment is proposed named sentiment vector. Thirdly, according to the Kano model, requirements could be classified to support the further design process with designers. Finally, a case study is carried out to demonstrate the feasibility and advantages of the proposed novel method.