Entity linking

[marcelmmm]

EntityLinker and Entity Linking Datasets

[alanakbik]

I think the dependency 'wikipediaapi' needs to be added to requirements.txt

[alanakbik]

I've added the missing requirements and also changed the default sentence splitters to Sektok since SpaCy requires additional modules to be installed that don't come with default Flair. In the AQUAINT_EL and IITB_EL corpora, you can now pass the sentence_splitter as an argument in case you want to use SpaCy instead.

However, unfortunately the change to SegTok broke AQUAINT_EL since it now sometimes splits a sentence at the wrong place and breaks up an entity. You can reproduce this error with:

corpus = AQUAINT_EL()

Could you take a look?

[marcelmmm]

Thank you for all this!!
I changed the AQUAINT_EL corpus. Now it works when using SegTok as well.

[alanakbik]

@marcelmmm thanks a lot for adding this!