5.3
JPT Research Digest
MBTI classification of companies using online reviews with Sentence-BERT
An, B., Shin, D., & Lee, H. (2022, November 11). MBTI classification of companies using online reviews with Sentence-BERT. Accelerating Digital Transformation in Immersive Economy, 35-51. |
Using the MBTI® system for personality detection from text has been popular in research for several years and covered in past issues of the Research Digest. Using a computer to analyze text from social media posts to determine a person's psychological type has been attempted using several different methods. One of those methods is known as BERT (Bidirectional Encoder Representations from Transformers), a machine learning framework for natural language processing.
In this study, An et al. used a Sentence-BERT (SBERT) algorithm to try and determine the personality type of an organization rather than an individual. The dataset used was comprised of 1,329,264 employee reviews from 1,241 companies listed on Glassdoor, a global platform that gathers reviews from users who have previously been or are currently employed with an organization. To find a company's personality type, the study analyzed the similarities between two types of SBERT vectored text: Glassdoor company reviews and descriptions of an organization's character.
The first finding was that ENTJ, ENTP, and ENFP were found to be the top three organizational personality types, with ENTJ occupying more than 34% of companies. Also, EJ and NT types of decision-making temperament were dominant. Furthermore, companies belonging to the same industry tended to have similar organizational characteristics. The authors hope to explore more companies and industries internationally in the future; a cross-cultural examination may indicate differences in culturally favored types, such as ENTJ being favored in American culture.
ARTICLE PERMALINK: https://www.myersbriggs.org/research-and-library/journal-psychological-type/mbti-classification-of-companies-using-online-reviews-with-sentence-bert/
ARTICLE COMMENTS: