LingMess: Linguistically Informed Multi Expert Scorers for Coreference Resolution LingMess is a linguistically motivated categorization of mention pairs into 6 types of coreference decisions and learn a dedicated trainable scoring function for each category. This significantly improves the accuracy of the pairwise scorer as well as of the overall coreference performance on the English Ontonotes coreference corpus. Please check the official repository for more details and updates. Training on OntoNotes We present the test results on OntoNotes 5.0 dataset. Model Avg. F1 SpanBERT large + e2e 79.6 Longformer large + s2e 80.3 Longformer large + LingMess 81.4 Citation If you find LingMess useful for your work, please cite the following paper:
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy