srtk.entity_linking package

srtk.entity_linking.linker_base module

By extending LinkerBase, you can easily generalize srtk.linker to other knowledge graphs.

class srtk.entity_linking.linker_base.LinkerBase

Bases: object

Base class for entity linking

abstract annotate(text: str, **kwargs)

Annotate a text with the entities in the knowledge graph

The returned dictionary should at least have the following fields: - question: The input text - question_entities: The entity ids of the entities in the text

Parameters:

• text (str) – The text to annotate

• kwargs – Extra arguments

Returns:

The annotated text with linked entities

Return type:

dictionary

srtk.entity_linking.wikidata module

class srtk.entity_linking.wikidata.WikidataLinker(endpoint, wikimapper_db, service='rel')

Bases: LinkerBase

Link entitiy mentions to Wikidata entities using the REL endpoint

annotate(text, **kwargs)

Annotate a text with the entities in the Wikidata knowledge graph

Parameters:

text (str) – The text to annotate

Returns:

A dictionary with the following keys:

question: The input text question_entities: The Wikidata ids of the entities in the text spans: The spans of the entities in the text entity_names: The names of the entities in the text not_converted_entities: The entities that are not converted to Wikidata ids

Return type:

dict

annotate_rel(text, token=None)

Annotate using a local REL service. Check https://rel.readthedocs.io/en/latest/tutorials/e2e_entity_linking for setup instructions.

Parameters:

text (str) – The text to annotate

Returns:

annotation results

Return type:

dict

annotate_tagme_wat(text, token)

Annotate using WAT or REL online services

Parameters:

• text (str) – The text to annotate

• token (str) – The token to access the service

Returns:

annotation results

Return type:

dict

srtk.entity_linking.dbpedia module

Link entitiy mentions to DBpedia entities with the DBpedia Spotlight endpoint.

Know more about DBpedia Spotlight at https://www.dbpedia-spotlight.org/.

class srtk.entity_linking.dbpedia.DBpediaLinker(endpoint)

Bases: LinkerBase

Link entitiy mentions to DBpedia entities with the DBpedia Spotlight endpoint

annotate(text, **kwargs)

Annotate a text with the entities in the DBpedia knowledge graph

Parameters:

text (str) – The text to annotate

Returns:

A dictionary with the following keys:

question: The input text question_entities: The DBpedia entities in the text spans: The spans of the entities in the text similarity_scores: The similarity scores of the entities in the text

Return type:

dict