Favi-Score: A Measure for Favoritism in Automated Preference Ratings for Generative AI Evaluation.
Pius von DänikenJan DeriuDon TuggenerMark CieliebakPublished in: ACL (1) (2024)
Keyphrases
- artificial intelligence
- evaluation measures
- user preferences
- machine learning
- expert systems
- intelligent systems
- data driven
- knowledge based systems
- knowledge representation
- similarity measure
- social networks
- utility function
- search engine
- fully automated
- evaluation method
- evaluation criteria
- evaluation methods
- discriminative learning