#AIFLL24
The first afternoon session. Using Large Language Models in Authorship Analysis. Using Perplexity - exponentiated log-likelihood of sequence of tokens occurring.
#AIFLL24 #AuthorshipAnalysis
June 26, 2024 at 1:07 PM
Appraisal for intelligence analysis: a critical comparative analysis of the utility of Appraisal versus LIWC for forensic text analysis, by Madison Hunter.
Turns out LIWC is not accurate on forensic data. Don't take it at face value. Can point to useful things in data but don't rely on it.
#AIFLL24
June 27, 2024 at 11:23 AM