Reddit Text for Sleep and Depression Analysis via SleepDepNet

2.5 MB of user-generated text from Reddit communities related to sleep and mental health, supporting the SleepDepNet multi-task learning framework. The dataset, created by Akshi Kumar and last updated on 2026-05-07, is licensed under CC-BY-4.0 and available in CSV format. It was used to train models achieving F1-scores of 0.89 for sleep quality classification and 0.86 for depressive sentiment analysis.

Use Cases

• Training multi-task models for joint sleep and mental health analysis based on the described SleepDepNet framework.
• Analyzing linguistic patterns associated with sleep disturbances based on the described attention mechanisms and emotion-aware features.
• Developing interpretable models for emotional and sleep-related expressions based on the described attention mechanisms.
• Calculating a unified risk score (SleepDepScore) based on the described integration of model outputs.
• Benchmarking text classification performance for digital mental health monitoring based on the reported F1-scores.

Strengths

• Model performance is reported with specific F1-scores of 0.89 and 0.86 for the two classification tasks.
• The dataset is explicitly designed to support scalable analysis and interpretability via attention mechanisms.
• It is openly licensed under CC-BY-4.0, facilitating reuse and redistribution.

Limitations

• Row count is unknown, which may limit suitability assessment for large-scale training.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect source bias inherent to Reddit discourse and the specific communities sampled.

Provenance

Source

Data collected from Reddit communities related to sleep and mental health.

Collection Method

User-generated text collected for the SleepDepNet research study.

Time Range

null

Freshness

Last updated 2026-05-07 17:36:58; freshness should be verified.

Geography

null