Online choice decision support for consumers: Data-driven analytic hierarchy process based on reviews and feedback

As online shopping flourished, consumers in their shopping can refer to rich product descriptions and a large amount of review information. For the scenario of consumer online choice decision among candidate products characterized by limited attributes, we refer to it as an online multi-attribute decision-making problem. To address the challenge of online choice decision support for consumers, we propose a data-driven analytic hierarchy process (AHP). The data-driven AHP includes extracting attributes of candidate products, calculating attribute values, attribute-weight learning, interaction-based preference revision process, and product ranking. In particular, we develop an Exp-strategy for attribute-weight learning, which helps learn the attribute weights of consumers who provide reviews as a reference for an end consumer. This learning method can handle dynamic online reviews without the problem of information overload. In addition, we design the interaction-based preference revision process to help the end consumer identify his attribute weights and make a choice decision.