Financial Natural Language Processing on FinDATA
0.25TSAFinancialBERT
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FinancialBERTLearning approach=STL2023.05 | 0.25 | 84.96 | 83.53 | 85.9 | 67.52 | 75.59 | |
| BERT-casedLearning approach=STL2023.05 | 0.232 | 86.57 | 86.19 | 85.43 | 71.3 | 76.73 | |
| Y-FinBERTLearning approach=STL2023.05 | 0.2275 | 85.62 | 86.55 | 85.66 | 65.45 | 74.75 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=Only Sentiment2023.05 | 0.2159 | 86.69 | — | — | — | — | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=w/o FSRL2023.05 | 0.2156 | 85.91 | 87.41 | 86.11 | — | 76.53 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=Full FinDATA2023.05 | 0.2151 | 87.06 | 87.51 | 86.52 | 69.88 | 77.8 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=w/o Number2023.05 | 0.2083 | 86.49 | — | — | 71.08 | 78.26 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=w/o CD2023.05 | 0.2077 | 86.36 | 87.49 | 85.63 | 71.32 | — | |
| BERT-uncasedLearning approach=STL2023.05 | 0.2069 | 86.08 | 87.09 | 85.69 | 70.89 | 76.7 | |
| P-FinBERTLearning approach=STL2023.05 | 0.2054 | 86.61 | 87.67 | 85.74 | 72.66 | 77.12 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=w/o Sentiment2023.05 | — | — | 87.79 | 86.49 | 70.6 | 78.4 | |
| MTLLearning approach=MTL, Backbone=P-FinBERT, MTL Subset=Only Number2023.05 | — | — | 87.25 | 85.7 | — | — |