Behavioral Patterns Ontology

Although understanding the formation and expression of human behavioral patterns is central to psychology and other social sciences (e.g., sociology, economics), the lack of a standardized vocabulary has stunted progress in these fields. Establishing a formal ontology of human behavior would enable researchers to operate within a consistent and unified framework both within and across disciplines. This study introduces the Behavioral Patterns Ontology (BPO), a foundational step toward the formalization of psychological concepts through structured behavioral patterns. While existing ontologies in the behavioral and social sciences address domains such as addiction, behavior change interventions, mental functioning, emotions, and social entities, each is tailored to a highly specific purpose and does not systematically capture the learning and conditioning mechanisms through which behaviors are formed, reinforced, and modified over time. The BPO moves beyond domain-specific applications and provides a transferable framework for representing behavioral patterns across contexts. The BPO consists of 162 classes and 21 object properties interconnected through 149 hierarchical relationships and defined by 644 axioms, providing a structured representation of behavioral processes. To demonstrate its practical applicability, we used the BPO to annotate TikTok videos related to illicit/recreational drug-seeking behavior in order to better understand the observable behavioral patterns associated with drug-seeking activities on social media platforms. We then analyzed the annotated data using a Social Media Analysis Dashboard, developed by our adjacent team, which enabled the visualization and interpretation of human behavioral patterns at the level of individual videos. The annotations were next queried using ontology-driven SPARQL analyses, with results organized across multiple use cases, demonstrating that the BPO accurately captured learning mechanisms, substance-specific behavioral patterns, and the drivers and contextual settings of drug-seeking behavior on the TikTok platform. Lastly, a machine learning (ML) model was implemented as a preliminary feasibility analysis designed to classify videos into three behavioral categories: drug-seeking behavior, avoidance behavior, and no behavior. This experiment serves to evaluate the predictive potential of ontology-driven annotations under limited data conditions and thereby demonstrates the practical viability of integrating structured semantic modeling with data-driven classification approaches.