Fine-Grained Opinion Mining–Based Conjoint Analysis for Automatically Extracting Consumer Preference Using the Modelled Product Review Dataset

Consumer preference describes the subjective individual taste of consumers to various kinds of products as measured by the utility of the product. Revealing consumer preferences is crucial for both companies and consumers. Unfortunately, most studies attempting to extract consumer preferences are based on discrete choice studies involving surveys of several respondents, which is considered to be time-consuming and costly. Another work that is based on online product reviews pays less attention to the semantics of words. Accordingly, this study proposes a novel platform to extract consumer preferences based on the discrete choice technique using datasets obtained from the product review domain. This study introduces SimplEx, a simple and robust extractor for aspect-based sentiment analysis. The platform also accounts for the contextual meaning of words in the step of sentiment extraction. At last, based on modelled product review data, the consumer preference is generated by making use of the conjoint analysis technique. We look into SimplEx’s accuracy, and on average, the performance is 0.92. Evaluation using linear and moving-average trendlines indicates that the proposed method outperforms the other two baselines. Moreover, the proposed methodology effectively discloses the product’s partial utility and offers a robust platform for deriving consumer preferences.