entitiescan

Entitiescan is a software framework designed to extract, classify, and link named entities from unstructured text, enabling the creation of structured representations such as knowledge graphs. It supports entity recognition across multiple languages and domains, and can be used to augment data catalogs, search systems, and analytics pipelines.

Core capabilities include named-entity recognition, entity classification into categories (persons, organizations, locations, dates, products), and entity linking to external knowledge bases. It processes documents, webpages, emails, and streaming text, outputting structured data in JSON, RDF, or CSV. It supports customizable models and a plugin architecture for domain-specific types.

Architecture and deployment: Entitiescan is built as a modular pipeline with an ingestion layer, processing engine, and storage adapters. It includes a model zoo with pre-trained models, tooling for custom domains, and interfaces for batch and streaming processing. Deployments are offered on-premises or in the cloud, with privacy-preserving options and governance controls.

History and usage: The project emerged from collaborations between academic researchers and industry practitioners to address information extraction and knowledge graph construction. It is maintained by a community of contributors and organizations, with documentation and tutorials to support integration into analytics stacks.

Applications and limitations: Common use cases include compliance monitoring, competitive intelligence, media analysis, and enterprise search. Limitations include dependence on model quality, language coverage, and domain adaptation needs; handling sensitive information requires appropriate access controls and compliance with data protection laws.

See also: Named-entity recognition, entity linking, knowledge graph, information extraction, NLP pipelines.