Integrate Wikidata to SciSpaCy EntityLinker

lubianat / ann

A repository for brainstorming and prototyping ideas related to the eLifeSprint project Annotate them all (https://sprint.elifesciences.org/annotate-them-all/)

Apache License 2.0

14 stars 0 forks source link

Integrate Wikidata to SciSpaCy EntityLinker #8

Open lubianat opened 4 years ago

lubianat commented 4 years ago

SciSpaCy is a package linked to the spaCy package for Natural Language Processing in python.

The code has an "Entity Linker" that links entities to different databases of concepts, such as the Unified Medical Language System.

One nice integration would be to link Wikidata to their framework.

It is kinda of a side project, but might improve the integration of Wikidata to the scientific NLP community.

arupgsh commented 4 years ago

There is an article in Medium suggesting that SpaCy 3 will introduce Wikidata based named-entity recognition. Source: https://medium.com/@mgalkin/spacy-irl-2019-and-wikidata-based-ner-64a799c17823

jvfe commented 4 years ago

There is an article in Medium suggesting that SpaCy 3 will introduce Wikidata based named-entity recognition. Source: https://medium.com/@mgalkin/spacy-irl-2019-and-wikidata-based-ner-64a799c17823

Looks like it's quickly becoming a reality! Take a look:

Their entity linking 101 shows linking to wikidata entities (e.g. Q7259)
Also, this pull request on their main repository and the issue it references.

I haven't looked into it very carefully but it does seem promising.