🧵 NEW PAPER - "Quantifying Narrative Similarity Across Languages" is live at SMR. We combine NLP w pairwise LLM evals to track ideas across media ecosystems and languages. We also figured out a way to eval unsupervised probs on widely understood supervised performance metrics.
Can norm-based information campaigns reduce corruption? by Aaron Erlich and Jordan Gans-Morse is now available in Early View. @aserlich@J_GansMorse
https://t.co/Kq1tnt2zMj
I know #AcademicTwitter is much diminished, but let's say you get the following email from your media relations office. Is this an appropriate email for them to send or not? (I am having a bad reaction, but trying to get input if my bad reaction is appropriate)
Come and join us for the next two sessions of our webinar series "Using Social Media as a Research Tool". For details and registration: https://t.co/74lrOOBWzG
Anne-Marie Singh, @aserlich, @serhii_bahlai, and I share part of our survey on crime and policing in wartime Ukraine. While Ukrainians are becoming more receptive toward toward guns in civilian hands, there is no rush to mass gun ownership or vigilantism. https://t.co/vHlXFF4TV3
@KyivIndependent published our latest article on Ukraine's gun culture. We do not find a mass shift toward civilian gun acquisition. Post-war Ukraine is unlikely to be a country of gun-slinging outlaws.
THX @MattLightCrim, @aserlich, and Anne-Marie Singh
https://t.co/aSsSwz1vHs
ChatGPT Outperforms Crowd-Workers for Text-Annotation Tasks
ChatGPT outperforms crowd-workers for several annotation tasks, including relevance, stance, topics, and frames detection, w/ 20x less cost.
https://t.co/FBNDvkYoYK
I like it when journals make all other reviewers' comments available by default. I learn from those reviewers, and it makes the experience much better!