List your company
ResearchFormerly WizWand

Named Entity Recognition on ONTONOTES (test)

89.71F1 ScoreBiLSTM-CRF Flair
63.283670.144377.00583.8657May 5, 2018Mar 8, 2019Jan 9, 2020Nov 11, 2020Sep 14, 2021Jul 18, 2022May 21, 2023
Updated 8mo ago

Evaluation results

MethodLinks
BiLSTM-CRF FlairLM embeddings or extra data=true, Backbone=Flair2019.0689.71——
Merge and LabelLM embeddings or extra data=true, Backbone=BERT2019.0689.2——
BiLSTM-CRF with CVTLM embeddings or extra data=true2019.0688.81——
This workLexical features=true, Gazetteers=false, Capitalization features=true, Word embeddings=true, Character-level embeddings=true, Language Model embeddings=false, Lexical Similarity features=true2018.0687.95——
BiLSTM-CRF lexLM embeddings or extra data=true2019.0687.95——
Merge and LabelContextual Embeddings=false2019.0687.59——
Strubell et al.Lexical features=false, Gazetteers=true, Capitalization features=true, Word embeddings=true, Character-level embeddings=true, Language Model embeddings=false, Lexical Similarity features=false2018.0686.99——
BiLSTM-CRFContextual Embeddings=false, variant=Strubell et al.2019.0686.99——
ID-CNNContextual Embeddings=false2019.0686.84——
Shen et al.Lexical features=false, Gazetteers=false, Capitalization features=true, Word embeddings=true, Character-level embeddings=true, Language Model embeddings=false, Lexical Similarity features=false2018.0686.52——
Chiu and NicholsLexical features=false, Gazetteers=true, Capitalization features=true, Word embeddings=true, Character-level embeddings=true, Language Model embeddings=false, Lexical Similarity features=false2018.0686.28——
BiLSTM-CRFContextual Embeddings=false2019.0686.28——
Ratinov and RothLexical features=true, Gazetteers=true, Capitalization features=true, Word embeddings=true, Character-level embeddings=false, Language Model embeddings=false, Lexical Similarity features=false2018.0684.88——
Durrett and KleinLexical features=true, Gazetteers=true, Capitalization features=true, Word embeddings=false, Character-level embeddings=false, Language Model embeddings=false, Lexical Similarity features=false2018.0684.04——
Finkel and ManningLexical features=true, Gazetteers=true, Capitalization features=true, Word embeddings=false, Character-level embeddings=false, Language Model embeddings=false, Lexical Similarity features=false2018.0682.42——
Passos et al.Lexical features=true, Gazetteers=true, Capitalization features=true, Word embeddings=true, Character-level embeddings=true, Language Model embeddings=false, Lexical Similarity features=false2018.0682.24——
CPLLType=Ours2023.0581.47——
Clean dataMethod Category=Upper Bound2023.0581.47——
LW lossMethod Category=PLL2023.0580.45——
Token-level VotingMethod Category=Voting2023.0580.12——
Entity-level VotingMethod Category=Voting2023.0579.3——
Yang et al.Input=Gold segmentation, Reference Year=2016, semi-supervised=true, discrete features=true2018.0576.472.9880.15
Word-based LSTM-CRF (+char+bichar)Input=Gold segmentation2018.0575.7778.6273.13
PRODEN-mlpMethod Category=PLL2023.0575.36——
Che et al.Input=Gold segmentation, Reference Year=2013, semi-supervised=true2018.0575.0277.7172.51
Wang et al.Input=Gold segmentation, Reference Year=2013, semi-supervised=true2018.0574.3276.4372.32
Lattice LSTMInput=No segmentation2018.0573.8876.3571.56
Char-based LSTM-CRF (+bichar+softword)Input=No segmentation2018.0571.8174.3669.43
Word-based LSTM-CRF (+char+bichar)Input=Automatic segmentation2018.0571.773.3670.12
Word baselineInput=Gold segmentation2018.0569.5276.6663.6
Yang et al.Input=Gold segmentation, Reference Year=20162018.0568.5765.5971.84
Word baselineInput=Automatic segmentation2018.0565.6372.8459.72
SeqcrowdMethod Category=Annotator2023.0565.34——
Char baselineInput=No segmentation2018.0564.368.7960.35