README_entity.md

entity --- The psynlp module for matching entities in text

The entity submodule contains some very basic funcationality for extracting entities from text.

A matcher is implemented in BasicEntityMatcher, which simply finds literal phrases using the Spacy matcher. Note that the context element of an Entity is not used by default.

Usage

from psynlp.entity import BasicEntityMatcher

entity_phrases = {}
entity_phrases['symptoms'] = ['boos', 'somber', 'slaapt slecht'] # also accepts phrases
entity_phrases['diagnoses'] = ['depressie', 'psychose']

bem = BasicEntityMatcher(entity_phrases, spacy_model="spacy_model_name")
bem.extract_entities("Deze patient is somber, heeft wellicht een depressie.")
>>> [(token_span=(3, 4), rule=symptoms, text=somber), (token_span=(8, 9), rule=diagnoses, text=depressie)]

API

BasicEntityMatcher::initialization

bem = BasicEntityMatcher(entity_phrases,
			 spacy_model,
			 case_sensitive=False)

Field	Description
entity_phrases	A dictionary of rule=>phrases pair, such as for instance under the Usage paragraph in this readme.
spacy_model	The spacy model to load (by default 2_include_embeddings is used).
case_sensitive	Whether to match phrases case sensitive

BasicEntityMatcher::functions

Function	Description	Returns
extract_entities(text)	Extract entities from a text	A set of Entity objects

EntityMatcher::interface

class CustomEntityMatcher(EntityMatcher):
	def extract_entities(self, text):
		return entities

Entity::initialization

e  = Entity(token_start,
		  token_end,
		  rule,
		  text)

Field	Description
token_start	The index of the first token that is matched
token_end	The index of the last token that is matched
rule	The rule of the dicationary that was matched
text	The exact text that was matched