Imbalanced Text Classification based on Corporate Culture by using Support Vector Machine, Case Study: PT XYZ, Indonesia

Abstract

Corporate culture has received much attention over the last two decades because of its effects and potential impact on organizational effectiveness. Researchers and consultants are interested in the concept of corporate culture and how the values and philosophies of a business affect the achievements and behavior of its employees. It is considered that corporate culture and organizational strategy are related, particularly in the implementation of a certain plan within an organization. This study aims to develop a text classification model based on corporate culture by using the support vector machine for an imbalanced data set. The data set is message comments retrieved from a social media platform owned by the company and labeled according to five corporate cultures, i.e., integrity, collaboration, innovation, respect, and excellence. We found that the SVM model developed for Innovation value gives the best results with accuracy, and the F-1 score of the test set are 0.961 and 0.961, respectively.