Product ranking using hierarchical aspect structures

摘要

Product-related comparative reviews are one of the most helpful information sources for consumers to rank the competing products and make purchase decisions. The comparative reviews are even more valuable if they are concerning specific aspects. However, the fact that very few comparative reviews express direct opinions on product aspects makes aspect-based product ranking a difficult task. In this paper, we present a novel hierarchical aspect-based product ranking approach. We first mine aspect-based pairwise comparative opinions from both user reviews on multiple review websites and community-based question answering pairs containing product comparison information. Next, we use our hierarchical structure-based model to propagate and reassign the aspect-based comparative opinions by using the parent-child and sibling relations between aspects in the product aspect hierarchy. The structure-based model helps to address the data sparsity issue of very few or no comparative reviews for some aspects. Finally, we employ graph-based ordering algorithms to consolidate these reassigned pairwise opinions into listwise comparison results. Experiments on a set of candidate electronic products prove that the proposed approach is effective for aspect-based product ranking.