<![CDATA[Irrational or deviant thinking is a cognitive condition characterized by distortion in perception and reasoning. Such cognitive distortions are often reflected in an individual's speech and writing. Detecting distorted thinking at an early stage is crucial, as it can help mitigate the risk of severe depression. Cognitive Behavior Therapy (CBT) is one of the most widely studied approaches in psychotherapy research for depression. It has been recognized as an effective method for addressing cognitive distortions, depression, and negative thought patterns. Recent advancements in online CBT, particularly those incorporating Natural Language Processing (NLP) techniques, have significantly improved the diagnosis and treatment of cognitive distortions. Numerous studies have explored detecting and classifying cognitive distortions using machine learning models. Cognitive distortion detection is a form of short-text classification that presents a notable challenge – the limited availability of features that effectively capture a text's meaning or intent. Despite these challenges, BERT remains a consistently effective model across various text classification tasks. This study proposes a novel model for detecting cognitive distortions by introducing a new approach that combines sentence-level features from IndoBERT with keyword features derived from the class-based TF-IDF framework. The integration of these two feature sets demonstrated promising results, achieving an average accuracy of 0.787 and an F1 score of 0.769. These values represent improvements of 3.39% and 3.45%, respectively, compared to the IndoBERT-based detection model only. These findings highlight the potential of the proposed model as a valuable early detection tool to support online CBT programs.]]>
Copyrights © 2025
