Intelligent video interview agent used to predict communication skill and perceived personality traits

Hung Yue Suen, Kuo En Hung, Chien Liang Lin*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

The prediction of individual interpersonal communication skills and personality traits is a critical issue in both industrial and organizational psychology and affective computing. In this study, we invited 114 participants, including 57 interviewers and 57 interviewees, to collect the ground truth of interviewees’ communication skills and personality traits as perceived by real human interviewers in a structured behavioral interview setting. We develop an asynchronous video interview (AVI) platform with an artificial intelligence (AI) decision agent based on a TensorFlow convolutional neural network (CNN), called AVI-AI, that can be used to partially displace human raters’ work in the initial stage of employment screening and to successfully predict a job candidate’s communication skills and personality traits. The experimental results show that AVI-AI can predict not only a candidate’s interpersonal communication skills but also his or her openness, agreeableness, and neuroticism, as perceived by experienced human resource professionals. The interrater reliability values were all acceptable to support the ground truth assumption. However, our AVI-AI could not predict the conscientiousness and extraversion as perceived by the real human raters in this study.

Original languageEnglish
Article number3
JournalHuman-centric Computing and Information Sciences
Volume10
Issue number1
DOIs
Publication statusPublished - 2020 Dec 1

Keywords

  • Affective computing
  • Big five
  • Deep learning
  • Lens model
  • Personality computing

ASJC Scopus subject areas

  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Intelligent video interview agent used to predict communication skill and perceived personality traits'. Together they form a unique fingerprint.

Cite this