Token Classification Model
This model was trained to classify tokens in geological texts, identifying entities such as bacias, campos, estruturas físicas, fluidos da Terra, fósseis, minerais, and other domain-specific terms.
Metrics
Evaluation Results
The model was evaluated using precision, recall, and F1-score metrics on a validation dataset. Below are the results per entity:
- BACIA: Precision: 0.92 | Recall: 0.95 | F1-Score: 0.93 | Support: 581
- CAMPO: Precision: 0.94 | Recall: 0.84 | F1-Score: 0.89 | Support: 99
- ESTRUTURA_FISICA: Precision: 0.87 | Recall: 0.84 | F1-Score: 0.85 | Support: 396
- FLUIDODATERRA: Precision: 0.86 | Recall: 0.85 | F1-Score: 0.85 | Support: 339
- FOSSEIS: Precision: 0.86 | Recall: 0.76 | F1-Score: 0.81 | Support: 336
- MINERAIS: Precision: 0.90 | Recall: 0.87 | F1-Score: 0.88 | Support: 217
- NAO_CONSOLID: Precision: 0.84 | Recall: 0.69 | F1-Score: 0.76 | Support: 131
- PALEOAMBIENTE: Precision: 0.83 | Recall: 0.69 | F1-Score: 0.75 | Support: 486
- POÇO: Precision: 0.96 | Recall: 0.94 | F1-Score: 0.95 | Support: 104
- ROCHA: Precision: 0.91 | Recall: 0.94 | F1-Score: 0.92 | Support: 848
- TEXTURA: Precision: 0.84 | Recall: 0.72 | F1-Score: 0.78 | Support: 29
- UNIDADE_CRONO: Precision: 0.94 | Recall: 0.95 | F1-Score: 0.95 | Support: 1119
- UNIDADE_LITO: Precision: 0.91 | Recall: 0.88 | F1-Score: 0.90 | Support: 468
Aggregated Metrics:
- Micro Average: Precision: 0.90 | Recall: 0.88 | F1-Score: 0.89 | Support: 5153
- Macro Average: Precision: 0.89 | Recall: 0.84 | F1-Score: 0.86 | Support: 5153
- Weighted Average: Precision: 0.90 | Recall: 0.88 | F1-Score: 0.89 | Support: 5153
Intended Use
This model is designed for token classification tasks in geological texts, making it useful for applications in geology-related research, natural language processing (NLP) pipelines, or resource extraction.
Limitations
The model is trained specifically for geological texts and might not perform well for other domains. Fine-tuning may be required for use in other fields.
