List your company
ResearchFormerly WizWand

Aspect Extraction on Restaurant (test)

84.97F1 ScoreWDEmb+B+CRF
68.943673.104377.26581.4257Mar 22, 2016Jul 30, 2016Dec 7, 2016Apr 16, 2017Aug 24, 2017Jan 1, 2018May 11, 2018
Updated 8mo ago

Evaluation results

MethodLinks
WDEmb+B+CRF2016.0384.97
RNCRFHand-crafted features=Yes2016.0384.93
RNCRF-OOpinion labels in training=No, Hand-crafted features=Yes2016.0384.25
RNCRFHand-crafted features=No2016.0384.05
SemEval-12016.0384.01
SemEval-22016.0383.98
LSTMHand-crafted features=Yes2016.0382.99
RNCRF-OOpinion labels in training=No, Hand-crafted features=No2016.0382.73
LSTMHand-crafted features=No2016.0381.15
CRF-2Dependency information=Explicit2016.0378.37
CRF-1Dependency information=None2016.0377
DE-CNNEmbedding=Double2018.0574.37
DE-OOD-CNNEmbedding=Double (Out-of-domain)2018.0574.2
DE-CNN-CRFEmbedding=Double, Layer=CRF2018.0574.1
MIN2018.0573.44
DE-LSTMEmbedding=Double, Architecture=LSTM2018.0572.94
BILSTM-CNN-CRFEmbeddings=Dependency-based2018.0572.5
NLANGP2018.0572.34
DE-Google-CNNEmbedding=Double (Google)2018.0572.1
GloVe-CNNEmbedding=General (GloVe)2018.0572.08
Domain-CNNEmbedding=Domain-specific2018.0571.75
LSTM2018.0571.26
MaxPool-DE-CNNEmbedding=Double, Pooling=Max pooling2018.0571.12
CRF2018.0569.56