| """ |
| Módulo de extracción de eventos. |
| |
| Contiene reglas y lógica para extraer tuplas 5W1H (Who, What, When, Where, Why, How) |
| a partir del texto procesado. |
| |
| Incluye: |
| - ClassicExtractorV1: Extractor basado en reglas sintácticas (sin memoria) |
| - ClassicExtractorV2: Extractor con resolución de correferencias |
| - FineTunedExtractor: Extractor basado en LLM fine-tuned |
| - ZeroShotExtractor: Extractor zero-shot con LLM |
| - process_corpus: Función de procesamiento batch con checkpoints |
| """ |
|
|
| from .classic import ClassicExtractorV1, ClassicExtractorV2 |
| from .batch_processor import ( |
| process_corpus, |
| process_corpus_batched, |
| extract_events_from_doc, |
| parse_dbpedia_entities |
| ) |
|
|
| from .extraction_rules import ( |
| extract_5w1h_rules, |
| get_full_text, |
| extract_5w1h_with_offsets, |
| extract_5w1h_from_document, |
| get_subtree_span |
| ) |
| from .fine_tuned import ( |
| FineTunedExtractor, |
| build_prompt, |
| build_prompt_few_shot, |
| split_text_by_sentences_with_indices, |
| parse_5w1h_from_chunk, |
| parse_labeled_text_to_json |
| ) |
| from .zero_shot import ( |
| ZeroShotExtractor, |
| chunk_text_by_token_limit, |
| extract_json_from_output, |
| merge_chunk_annotations |
| ) |
|
|
