A data-driven method for user satisfaction evaluation of smart and connected products

Abstract

As user-designer interactive products, smart, connected products (SCPs) have been highly focused. User satisfaction, being an important indicator to judge the performance of products, might be discussed through online customer reviews in the era of Big Data. Therefore, this paper tries to assess user satisfaction of SCPs with the aid of online reviews and then proposes a novel textual data-driven evaluation method. We first mine core attributes and evaluations by topics extraction and sentiment analysis instead of market survey approaches. Combining frequency weights and position weights, we then introduce an integrated approach to obtain the final importance of attributes. Specifically, a new frequency-oriented formula and a new position score-based formula are defined to calculate frequency and position weights, separately. Entire and partial user satisfaction degrees are subsequently analyzed by probabilistic linguistic-regret theory-multiplicative multi-objective optimization by ratio analysis (PL-RT-MULTIMOORA) and a formula with weighted rejoicing–regret values, respectively. Finally, the designed decision-making method is successfully applied to evaluate user satisfaction of smart speakers. Some useful suggestions are given to customers and manufacturers in the light of entire and partial user satisfaction degrees.