Target-Based Sentiment Analysis on DR Restaurant (test)
68.64PrecisionFull model
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Full modelComponents=Full Framework (E2E-TBSA)2018.11 | 68.64 | 71.01 | 69.8 | |
| LM-LSTM-CRFTagging Scheme=Unified, Backbone=LM-enhanced LSTM-CRF2018.11 | 68.46 | 64.43 | 66.38 | |
| Base model + BGComponents=Stacked LSTMs + Boundary Guidance2018.11 | 67.51 | 66.42 | 66.96 | |
| LSTM-CRF-1Tagging Scheme=Unified, Backbone=LSTM-CRF2018.11 | 66.1 | 66.3 | 66.2 | |
| Base model + BG + SCComponents=Stacked LSTMs + Boundary Guidance + Sentiment Consistency2018.11 | 63.95 | 69.65 | 66.68 | |
| CRF-unifiedTagging Scheme=Unified2018.11 | 63.39 | 57.74 | 60.43 | |
| Base model + BG + OEComponents=Stacked LSTMs + Boundary Guidance + Opinion Extraction2018.11 | 62.85 | 66.77 | 65.22 | |
| LSTM-unifiedTagging Scheme=Unified, Backbone=LSTM2018.11 | 62.8 | 63.49 | 63.14 | |
| NN-CRF-unifiedTagging Scheme=Unified2018.11 | 62.61 | 60.53 | 61.56 | |
| HAST-TNetTagging Scheme=Pipeline2018.11 | 62.18 | 73.49 | 67.36 | |
| NN-CRF-jointTagging Scheme=Joint2018.11 | 61.56 | 50 | 55.18 | |
| LSTM-CRF-2Tagging Scheme=Unified, Backbone=LSTM-CRF with CNN2018.11 | 61.56 | 67.26 | 64.29 | |
| Base modelComponents=Stacked LSTMs2018.11 | 61.48 | 66.16 | 63.73 | |
| NN-CRF-pipelineTagging Scheme=Pipeline2018.11 | 60.09 | 61.93 | 61 | |
| CRF-jointTagging Scheme=Joint2018.11 | 60 | 48.57 | 53.68 | |
| CRF-pipelineTagging Scheme=Pipeline2018.11 | 52.28 | 51.01 | 51.64 |