DUI

DUI maps and classifies drug-related terminology in social media using a hierarchical taxonomy and term embeddings to support epidemiological and semantic analyses of substance use and recovery language.

Key Features:

• Hierarchical taxonomy: A structured taxonomy of 108 addiction-related categories encompassing over 9,000 terms with semantically related category groupings.
• Corpus-derived embeddings: Term embeddings derived from a corpus of 7,472,545 Reddit posts authored by 1,402,017 users.
• Social media term identification: Identification of constituent drug-use terms in social media posts from platforms including Reddit and Twitter.
• Thematic analysis integration: Categories were established through thematic analysis combined with term embeddings.
• Semantic and exploratory analysis: Supports semantic analyses and exploratory investigation of language patterns with integrated visualization-based analyses.

Scientific Applications:

• Addiction epidemiology surveillance: Quantifying and monitoring substance use–related language patterns on social media for epidemiological studies.
• Linguistic analysis of drug-use jargon: Characterizing slang and jargon unique to drug-use subcultures to inform semantic and sociolinguistic research.
• Recovery-language monitoring: Analyzing discussions of recovery and treatment-related language in online communities.

Methodology:

The hierarchical taxonomy (108 categories, >9,000 terms) was established via thematic analysis combined with term embeddings computed from 7,472,545 Reddit posts by 1,402,017 users, and term identification in Reddit and Twitter posts uses these embeddings.