Information Extraction
Information extraction automatically obtains structured data—entities, relations, events—from unstructured text. We clarify that it is an umbrella of several subtasks and not a synonym for NER, trace the evolution of its methods from rules to large language models, and its challenges.
Information extraction is the task of automatically obtaining structured information—entities, relations, events—from unstructured or semi-structured text. It turns free text into data that a machine can query, for example to populate a database or build a knowledge graph. It is important to understand it as an umbrella of several subtasks, not as a synonym for a single one.
Its subtasks
Information extraction groups several tasks. Named-entity recognition locates and classifies mentions such as people, organizations or places. Relation extraction identifies the link between two already-detected entities. Event extraction detects occurrences and their participants. And coreference resolution determines when different expressions—a name, a pronoun—refer to the same entity. Entity recognition is thus a subtask, not the whole process.
Uses and challenges
Extraction is not verification: structured data must retain its link to the source passage so another person can check it.
Pieces using this term
This article was produced with artificial intelligence under human editorial oversight.
