An Ontology-based Framework for the Modelling and Online Detection of Obsessive Compulsive Disorder
No Thumbnail Available
Date
2024-11
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Cardiff University
Abstract
In the contemporary digital landscape, the prevalence and impact of Obsessive- Com- pulsive Disorder (OCD) discourse in online platforms have garnered increasing signif- icance. This thesis presents an integrated framework aimed at detecting and classi- fying OCD in online discourse by harnessing the synergy between ontology develop- ment and machine learning. The primary objective is to enhance the understanding and identification of OCD-related content within the vast and varied landscape of on- line forums.
The research begins with the construction of a comprehensive ontology, named OCD, specifically designed to encapsulate the multifaceted aspects of OCD. This ontology is developed to represent the complex interplay of OCD symptoms, behaviors, and related mental health concepts. Drawing upon insights from medical literature, psy- chological studies, and existing biomedical ontologies, the OCD ontology provides a structured, hierarchical representation of OCD, enabling systematic identification and categorisation of OCD-related terms. Consequently, it furnishes a rich semantic framework that facilitates accurate interpretation of online discourse.
In addition to ontology development, the thesis explores machine learning method- ologies, particularly focusing on the classification of OCD-related posts on online plat- form. A variety of classification models are employed to analyse and categorise online content. Leveraging the OCD ontology as a foundational reference for feature extrac- tion and semantic analysis, these models are trained and evaluated on a corpus of OCD forum posts. The classification process is designed to discern various OCD manifestations, such as obsessions and compulsions, thereby offering a granular un- derstanding of the disorder’s portrayal in digital communication.
The outcomes of this thesis carry significant implications for mental health profes- sionals, online community moderators, and researchers. The developed framework and methodologies represent a pioneering tool for monitoring, understanding, and addressing OCD in the digital space.
Description
This thesis explores the intersection of ontology development and machine learning to detect and classify Obsessive-Compulsive Disorder (OCD) discourse in online platforms. By addressing the growing significance of OCD-related discussions in digital spaces, the research presents an innovative framework designed to enhance the understanding and identification of OCD content in online forums. The study begins with the creation of a specialized ontology, aptly named the OCD ontology, which encapsulates the diverse symptoms, behaviors, and associated mental health concepts of OCD. Drawing from medical literature, psychological studies, and existing biomedical ontologies, this hierarchical structure provides a robust semantic framework for systematic analysis and categorization. Complementing this, machine learning methodologies are employed to classify OCD-related posts, leveraging the ontology for feature extraction and semantic analysis. The models are trained on a curated dataset of OCD forum posts, enabling granular identification of OCD manifestations, including obsessions and compulsions. The outcomes offer a groundbreaking tool for mental health professionals, researchers, and online community moderators, facilitating improved monitoring, understanding, and management of OCD in digital environments.
Keywords
OCD, Machine Learning, Knowledge graph, Ontology
Citation
ISSN