Associate Rule and Word Embedding Mining of Cultural Differences in Online Review Content
(A2024-117999)

AUTHORS

Tuck Siong Chung, ESSEC Business School

ABSTRACT

This study employs associate rule mining and word embedding techniques such as Word2vec and Doc2vec to explore how consumers from different cultural backgrounds differ in the words used when writing reviews. Using review document vectors derived from an airline quality website we found that reviews by Chinese and American consumers are different based on their word content. Based on the associate rules mined, both groups of reviewers seem to be concerned with service with the American reviewers predominantly concerned with flight time while the Chinese reviewers are with the cabin service environment. Word embedding analysis indicates that Chinese reviewers are more specific in their descriptions of the factors affecting the service delivery process and service quality. The American reviewers on the other hand tend to compare one airline versus another and to use more generic criteria for evaluating service quality. In addition to demonstrating the cultural differences in how consumers write reviews, the methodology used in this paper provides an avenue for analysing, detecting and potentially adjusting for the reviews’ cultural biases.