学术文献库

In social commerce, users dialogue with each other on the topics related to the providers' products. However, the language customers use may vary from the language vendors use on their e-commerce websites and product descriptions. This situation can lead to possible misunderstandings in the social dialogue between customers, and incidental costs in the dialogue between customers and vendors. One possible reason for this difference is that words used by customers may have different meanings compared to those used by product description writers. We present a novel approach to measure this potential lexical-semantic gap for various e-commerce domains using an information-theoretical approach based on a large corpus of user reviews and product descriptions. Additionally, we use neural word embeddings to identify words with the highest semantic drift between reviews and descriptions as a tool to construct a ground truth for the task. We found that low levels of lexical-semantic gap are related to better customer satisfaction. Besides, this work is a step towards a better understanding of the effect of the use of language in e-commerce and social commerce. Potential applications of this technology could lead to better communication between customers and improvements in customer satisfaction, services, and revenue. Furthermore, this study opens up perspectives for applications in other domains with relationships beyond client-vendor, such as citizen-government and patient-healthcare system.