ScholarWorks@숙명여자대학교

초록

This study examines the impact of a company's ESG-related social media communication strategy on social media engagement of consumers and the publics. Using the latest deep learning techniques, especially BERTopic, a transfer learning natural language processing method, 39,042 ESG-related reports were extracted from the 30 largest companies based on domestic market capitalization. By analyzing the sentences, key keywords related to the environment(E), society(S), and governance(G) were extracted. Then, based on the extracted keywords, we classified posts related to the environment, society, and governance based on 22,400 posts extracted from the official Instagram accounts of 30 companies. The number of likes and comments for each E, S, and G post was calculated as to measure the engagement. As a result of the model inference, environmental(E) and governance(G) posts had a negative relationship with the number of likes, but social(S) posts had a positive relationship. In relation to the number of comments, all types of posts on the environment(E), society(S), and governance(G) had a positive relationship.

키워드