EchoMind-BD: A Dataset of Online Behavior and Depression Patterns Among Bangladeshi Users

Published: 5 August 2026| Version 1 | DOI: 10.17632/cw6b4b6m9n.1
Contributors:
Md Mijanur Rahman,
,
,
,

Description

The dataset "EchoMind-BD: A Dataset of Online Behavior and Depression Patterns Among Bangladeshi Users" was created to support research on depression detection using social media data. For this dataset, we collected social media captions from Bangladeshi users through Google Forms. Along with the captions, participants answered several questions related to their emotions, thoughts, and online behavior. These responses helped us better understand the context behind the captions and assisted in the labeling process. After collecting the data, each caption was carefully reviewed and labeled manually. Based on the content of the caption and the responses provided by the participants, every sample was assigned to one of three categories: Depressed, Not Depressed, or Neutral. Manual labeling was chosen to ensure that the annotations were as accurate and meaningful as possible. The dataset contains real-world social media text written by Bangladeshi users, making it useful for developing and evaluating machine learning models for depression detection and mental health analysis. Since the data comes from actual users, there are natural variations in writing style, vocabulary, sentence structure, emotional expression, and language use. These characteristics make the dataset more representative of real-life online communication. Categories of the Dataset: Depressed: Captions that show signs of sadness, loneliness, hopelessness, emotional distress, or other indicators commonly associated with depression. Not Depressed: Captions that reflect positive emotions, motivation, happiness, confidence, or a generally healthy emotional state. Neutral: Captions that do not clearly express positive or negative emotions and mainly contain informational, descriptive, or emotionally neutral content. This dataset can be used for text classification, sentiment analysis, and mental health research, helping researchers build intelligent systems that can identify depression-related patterns from social media content.

Files

Institutions

Categories

Psychology, Computer Science, Artificial Intelligence, Public Health, Computational Linguistics, Data Science, Machine Learning

Licence