Practice vocabulary for building knowledge graphs: entity extraction, relation extraction, entity linking, coreference resolution, and open knowledge graphs like Wikidata and DBpedia.
0 / 5 completed
1 / 5
What is 'entity extraction' in the context of knowledge graph construction?
Entity extraction (also called Named Entity Recognition, NER) is the NLP task of detecting spans in text that refer to real-world entities and assigning them types. It is the first step in populating a knowledge graph from text sources.
2 / 5
How would you explain 'entity linking' to a colleague?
Entity linking (also called entity disambiguation) maps a text mention to a specific knowledge base entry. 'Apple' could refer to the fruit or the company — entity linking resolves this ambiguity using context.
3 / 5
What problem does 'coreference resolution' solve during knowledge graph construction?
Coreference resolution groups all mentions in a document that refer to the same entity. Without it, 'Tesla' and 'the company' in the same paragraph might generate separate, unconnected nodes in the graph.
4 / 5
What is Wikidata in the context of open knowledge graphs?
Wikidata is one of the largest open knowledge graphs, containing hundreds of millions of statements about entities worldwide. It is machine-readable, multilingual, and freely licensed, making it a common seed for domain-specific knowledge graph construction.
5 / 5
What does 'populating a knowledge graph' typically involve?
Populating (or building) a knowledge graph covers the full construction pipeline: sourcing data, extracting entities and relations, resolving entity identity, mapping to an ontology, and loading triples into the graph. Each step has its own set of NLP and data engineering vocabulary.