Sentiment Analysis on Yelp '13 (test)
68.3AccuracyHUAPA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HUAPAuser and product information=true2018.01 | 68.3 | 0.628 | |
| NSC+UPA(BiLSTM)user and product information=true2018.01 | 65.5 | 0.672 | |
| LSTM-GRNN2016.07 | 65.1 | — | |
| NSC+UPAuser and product information=true2018.01 | 65 | 0.692 | |
| fastTexthidden units=102016.07 | 64.2 | — | |
| LUPDRuser and product information=true2018.01 | 63.9 | 0.694 | |
| NSC+LA(BiLSTM)user and product information=false2018.01 | 63.8 | 0.691 | |
| Conv-GRNN2016.07 | 63.7 | — | |
| NSC+LAuser and product information=false2018.01 | 63.1 | 0.706 | |
| UPNN(NSC)user and product information=true2018.01 | 63.1 | 0.702 | |
| NSCuser and product information=false2018.01 | 62.7 | 0.701 | |
| Standard DR-AGGaggregation=hierarchical2018.06 | 62.1 | — | |
| Reverse DR-AGGaggregation=hierarchical2018.06 | 61.6 | — | |
| Max pooling2018.06 | 61.1 | — | |
| Self-attentionaggregation=hierarchical2018.06 | 61 | — | |
| Average pooling2018.06 | 60.7 | — | |
| SVM+TF2016.07 | 59.8 | — | |
| CNN2016.07 | 59.7 | — | |
| UPNN(full)2018.06 | 59.6 | — | |
| UPNN(CNN)user and product information=true2018.01 | 59.6 | 0.784 | |
| Cached LSTM2018.06 | 59.4 | — | |
| UPNN(np UP)2018.06 | 57.7 | — | |
| UPNN(CNN and no UP)user and product information=false2018.01 | 57.7 | 0.812 | |
| RNTN+Recurrent2018.06 | 57.4 | — | |
| RNTN+Recurrentuser and product information=false2018.01 | 57.4 | 0.804 | |
| Trigram+UPFuser and product information=true2018.01 | 57 | 0.803 | |
| Trigramuser and product information=false2018.01 | 56.9 | 0.814 | |
| TextFeature+UPFuser and product information=true2018.01 | 56.1 | 0.822 | |
| TextFeatureuser and product information=false2018.01 | 55.6 | 0.845 | |
| Paragraph Vectoruser and product information=false2018.01 | 55.4 | 0.832 | |
| SSWE+SVMuser and product information=false2018.01 | 54.9 | 0.849 | |
| AvgWordvec+SVMuser and product information=false2018.01 | 52.6 | 0.898 | |
| Majorityuser and product information=false2018.01 | 41.1 | 1.06 | |
| JMARSuser and product information=true2018.01 | — | 0.985 |