Item Recommendation for Word-of-Mouth Scenario in Social ECommerce

Abstract

By allowing users to share things with their friends, social commerce, which
differs from conventional e-commerce in which consumers buy products based on their own
initiative searching or platform suggestions, turns a social network into an inclusive
environment to conduct business. A user (sharer) may send a link to a product to friends with
whom they are linked on social media. The sharer might gain commission from the site after
the recipient buys the goods. The platform may help sharers drive sales by offering product
options that are more likely to be bought during social sharing. To the best of our knowledge,
this job of producing sharing ideas is brand-new and has never been investigated. We
characterise it as item recommendation for word-of-mouth scenario. In this research, we offer
a TriM (short for Triad-based word-of-mouth recommendation) model that can
simultaneously capture the sharer's impact and the receiver's interest, two key elements that
affect the receiver's decision to purchase the product. Our suggested TriM-Joint further
enhances recommendation performance by using joint learning on two portions of interaction
data to overcome the data sparsity problem. We demonstrate via trials that our suggested
models outperform state-of-the-art models, with improvements of at least 7.4% and 14.4%,
respectively.