Call it a match! AI integration with psychometrics in pairing models for better conversation

Abstract

This thesis explores the integration of psychometric data into AI-based customer-operator pairing models to enhance call center performance. Traditional models often rely heavily on historical data, overlooking human factors such as personality traits and work behaviors. To address this, a personality survey grounded in the Big Five Personality Traits, Core Self-Evaluations, Working Styles, Emotional Labor, and Customer Service Behaviors was conducted among call center operators. Aligning with the study’s first research question, factor analysis was applied to validate the accuracy of these theoretical models in representing operator personalities. Then, using supervised learning techniques – Logistic Regression, Random Forest, Gradient Boosting Machines, and XGBoost – models were trained and evaluated on datasets incorporating personality features. Results revealed a modest but not statistically significant improvement in F1 score and Recall. However, personality traits emerged among the top ten most important features, highlighting their potential to enhance model transparency and explainability. Beyond that, the inclusion of psychometric data fosters a better alignment between operators and customers, enhancing trust, reducing conflicts, and improving customer satisfaction. Additionally, leveraging operator strengths can optimize performance, mitigate burnout, and inform personalized training initiatives. Psychometric traits also provide predictive insights into critical service behaviors, contributing to a more human-centered approach to customer service. These findings underscore the value of psychometric data in complementing existing AI-driven pairing systems, paving the way for more tailored, transparent, and effective call center operations.

Date of Award	31 Jan 2025
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Nuno Filipe Loureiro Paiva (Supervisor)