Needs-based analysis of online customer reviews

A new approach to supplement traditional methods for assessing rapidly changing user needs, model the knowledgebase of online customer reviews as a matrix of reviews relating customer needs to product attributes and searches for hyper-rules relating hyperedges.

Abstract

Needs-based analysis lies at the intersection of product marketing and new product development. It is the study of why consumers purchase and what they do with those purchases. In a world of mass-customization and one-to-one marketing, anticipating the customer's needs is a key competitive advantage. In this paper, we consider a new approach to supplement traditional methods for assessing rapidly changing user needs. We model the knowledgebase of online customer reviews as a matrix of reviews relating customer needs to product attributes. In a hierarchical two-stage process, we first use association rules to cluster related attributes and needs into hyper-edges. In a second application of association rule mining, we search for hyper-rules relating hyperedges. The method is demonstrated on 10,500 customer reviews over two unrelated product domains, digital cameras and vacuum cleaners.