Emotions and Topics in Online WOM: Application of Latent Semantic Analysis

With the growth of internet usage, customers often search online for product information and have access to dozens or hundreds of product reviews from other customers. While it is clear that not all customer reviews are helpful, less is known about why certain online reviews are more helpful than others. Past research demonstrated that valence of a review affects the informational value of the contents and thus the perceived helpfulness of the review. However, in a purchase or information search situation, people encounter a variety of emotions which are likely to be expressed in the reviews. Potential customers read the reviews to collect or verify information and to see what other people think. Despite the fact that reviews contain emotions, few studies have investigated how emotions expressed in the review affect the helpfulness of the review. Do discrete emotions have differential informational value in this case? In this article, we build on cognitive appraisal theory to examine how discrete emotions (e.g., hope, happiness, anxiety, and disgust) embedded in the reviews affect the helpfulness votes of potential customers. We hypothesize that reviews containing emotions associated with certainty are more helpful and that reviews containing emotions associated with uncertainty are less helpful regardless of their valences. We adopt a quantitative content analysis approach to measure emotions in these reviews. Specifically, we use Latent Semantic Analysis (LSA) to measure the emotional contents of the reviews. Findings demonstrate