The effect of product distance on the eWOM in recommendation network

Abstract

The online Product Recommendation Networks (PRNs), connecting similar products with hyperlinks, have been widely implemented in user-generated content websites and ecommerce systems. With the PRNs as the virtual shelves, this paper explores the impact of the distance between products on the formation of product electronic Word-of-Mouth (eWOM). Employing an empirical book recommendation network of Amazon, the study one explores the effect of a focal product's neighborhood (nearby others) on its eWOM, and study two explores the eWOM similarity between product pairs that are at one, two and three clicks away from each other. The results reveal the significant role played by the product distance on the association of their eWOM. On one hand, a focal product's eWOM is largely influenced by that of its neighborhood. On the other hand, the good connectivity between two products, which is defined as the number of paths connecting them, is closely associated with the eWOM similarity between them. The findings suggest that the products should be considered as interactive collectives rather than separated individuals particularly in the eWOM studies.

Word-of-Mouth Engagement in Online Social Networks: Influence of Network Centrality and Density

“…They have access to a greater amount of information and can influence the opinion of others through their connections. It has also been found [6] that users who are part of high-density networks are more likely to be influenced by the positive or negative opinions and reviews of other users. The current research focuses on the influence of social network centrality and density on users' tendency to post positive or negative reviews about products or services, known as electronic Word of Mouth (eWOM).…”

Ready for Emerging Threats to Recommender Systems? A Graph Convolution-based Generative Shilling Attack

“…It has been indicated that recommender systems are vulnerable to malicious profile injections, which are also known as the shilling attacks [13]. With a large number of skewed ratings, injected fake user profiles can mislead the recommender system in measuring similarities [19] [29]. As a result, attackers can manipulate the recommendation results for their intended commercial interests.…”