Knowledge-based systems

Consumers are increasingly caring about air quality, and the air purifier market is facing fierce competition. In the era of e-commerce, it is possible to improve product design and enhance competitiveness by mining consumer demands for product functions from online reviews. However, the existing aspect-level sentiment analysis methods fail to deal with the over-segmentation of multi-word aspects, which is likely to cause the omission of aspects. Moreover, there is still a gap between sentiment analysis and demand recognition. To overcome these limitations, we propose an approach based on fine-grained sentiment analysis and the Kano model to extract consumer demands for product attributes from online reviews. Specifically, a recognition method based on part-of-speech rules is presented to identify multi-word product attributes. Inspired by the Kano model, extraction rules are designed to identify consumer demands for product attributes from the results of sentiment analysis. Finally, online reviews of air purifiers in Chinese market crawled from T-mall.com are used to illustrate the proposed approach. The correlation results show that there exists significantly positive correlation between product sales and the extracted attractive and one-dimensional product attributes. This indirectly demonstrates the effectiveness of the proposed method. Another dataset of refrigerators is further used to check the robustness of our proposed approach, and the results further demonstrate the effectiveness. • A framework for consumer preferences extraction from online reviews is proposed. • Presents a rule-based method to correctly recognize multi-word product attributes. • Designs rules to identify one-dimensional and attractive demands for product attributes. • Experiments on the online reviews of air purifiers demonstrate the effectiveness of the method.

Customer preferences extraction for air purifiers based on fine-grained sentiment analysis of online reviews