A Clustering-Based Framework for AI User Profiling in Intelligent Systems

Abstract

Artificial intelligence (AI) systems increasingly depend on user modeling to provide adaptive functionality and personalized interactions. Although prior studies have examined AI adoption using descriptive analysis or supervised prediction models, comparatively limited attention has been given to robust unsupervised profiling of AI users across multiple usage contexts. This study proposes a clustering-based framework for profiling AI users in professional environments using a multidimensional dataset derived from structured survey responses. The framework combines data preprocessing, unsupervised learning, dimensionality reduction, and cluster validation to generate interpretable user profiles from behavioral and contextual variables. Using a dataset of 62 participants, the analysis shows that a two-cluster solution provides the most suitable segmentation of the data. The identified clusters represent two distinct user profiles: low-integration users and high-integration users. The results indicate that the primary separation between users is driven by task-oriented AI usage in professional and practical contexts rather than by general familiarity with AI tools. The study further shows that structured survey data can be transformed into computational user models suitable for intelligent systems design. By linking user segmentation to personalization needs, the proposed framework contributes to data mining, user modeling, and adaptive AI system development.