Mining sentiment tendencies and summaries from consumer reviews

Abstract

Consumer reviews are an invaluable aid for businesses in obtaining consumers’ feedback to facilitate their marketing campaigns. However, the rapidly increasing number of consumer reviews makes it difficult for businesses to obtain a comprehensive view of consumer opinions about their products, especially about the highlighted product features. This study presents a framework to provide an understanding of how consumers’ feedback changes over time and what concerns consumers most. The framework presented here contributes to consumer review analysis in three ways. First, a novel model is proposed to extract the feature words that are semantically relevant to each highlighted product feature. Second, consumers’ feedback on each highlighted product feature is converted into a sentiment tendency graph, which may reflect how the feedback changes over time. Third, consumers’ feedback is also summarized, which may reveal what consumers appreciate and what concerns them most. The experimental results show that the proposed model can effectively extract the feature words for each highlighted feature of the product. Moreover, both sentiment tendency graphs and summaries could complement each other and provide a more detailed picture for consumers and manufacturers.