A Systematic Literature Review of Deep Learning Algorithms for Personality Trait Recognition

Personality trait recognition has an essential role in the job screening process. The psychologies perform the analysis based on the survey, the handwriting of the participant, or conduct the interview. The process takes a lot of time and money. Consequently, it inspires that researchers develop a tool to help the screening process faster. The current review papers lack an explanation of deep learning algorithms in personality trait recognition. So we perform this study to classify the latest deep learning algorithms in personality trait recognition. Using Preferred Reporting Items for Systematic reviews and MetaAnalyses Method (PRISMA), we collected 25 key papers which are from Scopus, IEEE, Science Direct, Emerald Insight, and ACM. In this paper, we focus on discussing the problems, deep learning methods, and open issue in personality trait recognition. Based on the finding, the primary problem in personality trait recognition is the complexity of the text and audiovisual data. It induces the low performance of personality trait recognition. In detail, the text-based trait recognition has lower accuracy than the visual and audio based methods. Therefore we have a bigger opportunity to improve the performance of text-based personality recognition than audiovisual based. Moreover, the real experiment in audiovisual trait recognition still in the early stages and it can be explored in more detail.