Veterans, PTSD and Social Media: Towards Identifying Trauma Text Categories using Grounded Theory

Document Type

Article

Publication Date

1-2020

Publisher

Association for Computing Machinery (ACM)

Source Publication

GROUP '20: Companion of the 2020 ACM International Conference on Supporting Group Work

Abstract

Text classification using machine learning can be applied in various contexts such as in classifying research papers, identifying relevant news stories, and detecting fake reviews. Training an algorithm to perform such tasks generally requires a dataset with predefined labels. Valid labels for texts in a given domain can be predefined by domain experts. However, when it comes to free-form text from messaging applications and social networking sites it is difficult to predict what labels may be extracted from the text. Grounded theory provides a method by which concepts that emerge from data can be expressed as categories and properties. These categories and properties can then be arranged in a hierarchical class label structure that can be used to build a dataset for training models. This study focuses on text related to veterans with post traumatic stress disorder and identifies a hierarchical class label structure, with the future goal of applying this to prevent crisis situations.

Comments

"Veterans, PTSD and Social Media: Towards Identifying Trauma Text Categories using Grounded Theory," published as a part of the conference GROUP '20: Companion of the 2020 ACM International Conference on Supporting Group Work (January 2020): 115-118. DOI.

Link to Full Text

Share

COinS