Data Extraction

[cbadenes]

Analyze information sources and extend the data received with added information according to the ontology.

[cbadenes]

Use of Open Refine Reconciliation API from OpenCorporates and OpenRefine to:

• match company names to legal corporate entities:
• elimination of duplicates
• company identification (reconciliation)

Some errors identified from results:

• company name may include location (e.g. 'ALLNEX NORWAY', 'ALLNEX USA'..)
• some of them are addresses (e.g. '31 CALLE 25-45 ZONA 12 COLONIA SANTA ELISA OFICINA 601 ' )
• hierarchies are not recognized (e.g. 'CLARIANT CORPORATION' , 'CLARIANT INTERNATIONAL AG')

Some suggestions:

• apply rules to clean input data
• extend input data with patent information?

[cbadenes]

Retrieval of company information from external resources. It involves: Named Entity Recognition (NER), Named Entity Disambiguation (NED), and Named Entity Linking (NEL).

• A Literature Overview
• DBpedia Spotlight: DBpedia entity retrieval through a web service offered as a Docker image or index in Lucene. It uses Apache OpenNLP to identify the entity mentions. A basic Python script has been developed.
• Wikidata entity fishing: entity recognition and disambiguation against Wikidata at document level, in particular for a PDF with layout positioning and structure-aware annotations. Online demo available
• spaCy Entity Linking