Data-Driven Decision-Making to Rank Products According to Online Reviews and the Interdependencies Among Product Features

Abstract

The surge in online shopping has led to an increase in online customer reviews (OCRs), posing challenges for product selection based on product features and customer sentiment. This is where the combination of multicriteria decision-making (MCDM) and sentiment analysis (SA) methods come in. In this article, we propose a hybrid approach for product ranking that addresses challenges identified in previous studies. These challenges include accurately considering feature interdependencies, identifying hesitancy and uncertainty in consumer purchase decisions, and using a more robust method for ranking alternative products. In doing so, we utilize SA and unsupervised machine learning to extract features from OCRs. We employ a combination of association rule mining (ARM) and fuzzy cognitive maps (FCM) to calculate feature weights based on interdependencies among features. In addition, we formulate a decision matrix using sentiment orientation and intuitionistic fuzzy theory. The interval-valued intuitionistic fuzzy (IVIF) theory ensures reliable decision-making information. The IVIF-multiobjective optimization by ratio analysis plus full multiplicative form method (MULTIMOORA) is applied to rank alternative products. Using Amazon comments, five mobile phones are ranked to demonstrate the methodology. The proposed framework improves decision-making in product selection based on OCRs by considering feature interdependencies. Sensitivity analysis and comparisons with other MCDM methods evaluate its robustness. By addressing previous limitations and incorporating interdependencies among features, this comprehensive approach provides reliable decision-making in product selection based on OCRs.