How to apply zero‐shot learning to text data in substance use research: An overview and tutorial with media data

Abstract

A vast amount of media‐related text data is generated daily in the form of social media posts, news stories or academic articles. These text data provide opportunities for researchers to analyse and understand how substance‐related issues are being discussed. The main methods to analyse large text data (content analyses or specifically trained deep‐learning models) require substantial manual annotation and resources. A machine‐learning approach called ‘zero‐shot learning’ may be quicker, more flexible and require fewer resources. Zero‐shot learning uses models trained on large, unlabelled (or weakly labelled) data sets to classify previously unseen data into categories on which the model has not been specifically trained. This means that a pre‐existing zero‐shot learning model can be used to analyse media‐related text data without the need for task‐specific annotation or model training. This approach may be particularly important for analysing data that is time critical. This article describes the relatively new concept of zero‐shot learning and how it can be applied to text data in substance use research, including a brief practical tutorial.

Associations between posting about alcohol on social networking sites and alcohol-induced blackouts in a sample of young adults not in 4-year college.

“…For example, there is evidence that dedicated algorithms can accurately identify alcohol (Bonela et al, 2024) and alcohol-related blackouts in text (Riordan et al, 2022), alcohol in images (Bonela et al, 2023), and substance use risk from Instagram profiles (Hassanpour et al, 2019). Large language models may also allow researchers to accurately screen social media posts without training a dedicated algorithm (which can also take a significant amount of time and computational resources; Riordan et al, 2024).…”