SentiCoref
A South Slavic news dataset for document-level entity-level sentiment analysis. It contains multi-entity, multi-stance corpora and is used to evaluate ELSA and model routing strategies.
Timeline (last 90 days)
On the SentiCoref dataset, an "oracle" experiment shows that if all errors among the 10% most uncertain predictions could be corrected without harming correct predictions, Macro-F1 could improve by 9.14 points.
The encoder-averaging strategy helps on AspectBench and NewsMTSC but is harmful on SentiCoref.
On SentiCoref, routing 30% of records to the encoder-assisted LLM showed an improvement of 1.02 points, but the confidence interval includes zero.
All Facts (3)
On the SentiCoref dataset, an "oracle" experiment shows that if all errors among the 10% most uncertain predictions could be corrected without harming correct predictions, Macro-F1 could improve by 9.14 points.
50%UnverifiedThe encoder-averaging strategy helps on AspectBench and NewsMTSC but is harmful on SentiCoref.
50%UnverifiedOn SentiCoref, routing 30% of records to the encoder-assisted LLM showed an improvement of 1.02 points, but the confidence interval includes zero.
50%