An Interdisciplinary Approach to Misinformation and Concept Drift in Historical Cannabis Tweets

This paper historically analyzes ten years of anglophone Cannabidiol (CBD) tweets using temporal topic modeling and proposes a classification method to detect commercial and non-commercial tweets to better understand topic changes over time. The topics in this conversation change as they affect, and are affected by, cannabis laws, policies and consumers. In the broader conversation on the effects of CBD, many claims of its effectiveness are exaggerated to the point of misinformation. Thus, analyzing the change in CBD topics over time has implications for how both advertising and misinformation regarding health spreads on social networks.