Foundation Models for Information Extraction

Abstract In the chapter we consider Information Extraction approaches that automatically identify structured information in text documents and comprise a set of tasks. The Text Classification task assigns document to one or more pre-defined content categories classes. This includes many subtasks such as language identification, sentiment analysis, etc. Word Sense Disambiguation attaches predefined meaning each word document. Named Entity Recognition identifies named entities An entity is any object concept mentioned an referred by proper name. Relation aims relationship between extracted from text. covers coreference resolution, linking, event extraction. Most demanding joint extraction relations Traditionally, relatively small Pre-trained Language Models have been fine-tuned these yield high performance, while larger Foundation achieve scores with few-shot prompts, but usually not benchmarked.