Extract Named Entities from Text

Extracts named entities using spaCy's named entity recognition. Entity types follow the OntoNotes 5.0 annotation scheme as implemented in spaCy models.

Usage

nlp_get_entities(x, entity_types = nlp_entity_sets$all, return_list = FALSE)

nlp_get_dates(x)

nlp_get_dates_and_times(x)

nlp_get_named(x)

nlp_get_numbers(x)

nlp_get_organizations(x)

nlp_get_people(x)

nlp_get_places(x)

nlp_get_proper_nouns(x)

Arguments

x

A character vector, factor, data frame, or list containing text data

entity_types

Character vector of entity types to extract. Entity types are: "PERSON", "NORP", "FAC", "ORG", "GPE", "LOC", "PRODUCT", "EVENT", "WORK_OF_ART", "LAW", "LANGUAGE", "DATE", "TIME", "PERCENT", "MONEY", "QUANTITY", "ORDINAL", "CARDINAL". See nlp_entity_sets for collections of entity types.

return_list

Logical indicating whether to return a named list with entities grouped by type (TRUE) or a flat character vector (FALSE). Default is FALSE.

Value

If return_list = FALSE (default): A character vector of unique extracted entities of the specified type(s). If return_list = TRUE: A named list where names are entity types and values are character vectors of unique entities for each type.

References

spaCy Named Entity Recognition spaCy Model Documentation OntoNotes 5.0 Annotation Scheme

See also

The cleanNLP package provides access to spaCy and other NLP backends.

cnlp_annotate for spaCy-based text processing.

The entity package on GitHub (https://github.com/trinker/entity) provided inspiration for these functions.

Examples

text <- c("John Smith works at Microsoft in Seattle.",
          "The deal was worth $1.2 million in 2023.",
          "He was the first employee to make 100% of his 3rd quarter targets.")

nlp_get_entities(text)
#>  [1] "2023"            "his 3rd quarter" "Seattle"         "$1.2 million"   
#>  [5] "first"           "Microsoft"       "100%"            "John Smith"     
#>  [9] "John"            "Smith"          
nlp_get_people(text)
#> [1] "John Smith"
nlp_get_organizations(text)
#> [1] "Microsoft"
nlp_get_places(text)
#> [1] "Seattle"
nlp_get_numbers(text)
#> [1] "$1.2 million" "first"        "100%"        
nlp_get_entities(text, "MONEY")
#> [1] "$1.2 million"
nlp_get_dates(text)
#> [1] "2023"            "his 3rd quarter"
nlp_get_proper_nouns(text)
#> [1] "John"      "Smith"     "Microsoft" "Seattle"  

# Get entities grouped by type
nlp_get_entities(text, c("PERSON", "ORG", "MONEY"), return_list = TRUE)
#> $PERSON
#> [1] "John Smith"
#> 
#> $ORG
#> [1] "Microsoft"
#> 
#> $MONEY
#> [1] "$1.2 million"
#>