Dataset for modeling Beck's cognitive triad to understand depression

Area of Science:

• Psychology
• Computational Linguistics
• Data Science

Background:

• Depression is often characterized by a negative cognitive triad: a pessimistic view of oneself, the future, and the world.
• Understanding these subjective symptoms requires analyzing language used in personal narratives and social media.
• Existing datasets may not adequately capture the nuances of the cognitive triad in real-world online communication.

Purpose of the Study:

• To present and describe the Cognitive Triad Dataset (CTD), a novel resource for studying Beck's cognitive triad.
• To facilitate computational modeling of depression symptoms by providing labeled data from diverse online sources.
• To enable research on aspect detection and sentiment classification related to self, world, and future outlooks.

Main Methods:

• Compilation of 5886 messages from the Time-to-Change blog, Beyond Blue, and Twitter.
• Manual labeling of data by skilled annotators into six categories: self/world/future-positive and self/world/future-negative.
• Evaluation of the dataset on aspect detection and sentiment classification subtasks.

Main Results:

• The Cognitive Triad Dataset (CTD) contains a substantial volume of labeled social media messages.
• The dataset is structured to support analysis of the cognitive triad components.
• Initial evaluations demonstrate the dataset's utility for sentiment analysis and aspect detection.

Conclusions:

• The CTD provides a valuable resource for researchers investigating depression and cognitive biases.
• This dataset can advance the computational understanding of subjective depressive symptoms as expressed online.
• Future work can leverage the CTD to develop more accurate models for mental health monitoring via social media analysis.