Faulty Information Processing Dataset: Investigating Beck’s Cognitive Distortions in Clinical Chatbot Interactions

The Faulty Information Processing Dataset (FIPD) is designed to identify two cognitive distortions—Personalization and Overgeneralization—which play a critical role in the early diagnosis and prognosis of depression. The dataset consists of 2,265 textual messages, including 1,955 from Twitter and 310 from mental health blogs. It is categorized into two classes: Personalization and Overgeneralization. The FIPD is intended to support the understanding of Faulty Information Processing within clinical chatbot-generated interactions.