CODEWITHSUNDEEP
python-3.xnlpspacyentity-linking
Manimaran Paneerselvam
Manimaran Paneerselvam

Reputation: 68

Error running Spacy entity linking example

I was trying the entity linking example in spacy.

This is the information about spaCy in my system.

============================== Info about spaCy ==============================

spaCy version    2.2.2
Location         C:\Users\manimaran.p\AppData\Local\Continuum\anaconda3\envs\spacy\lib\site-packages\spacy
Platform         Windows-8.1-6.3.9600-SP0
Python version   3.7.3
Models

Using this example to train the entity linker and generating the knowledge base for the same with this example.

I can create a knowledge base with the available en_core_web_md, this is the output for the same.

# python "create kb.py" -m en_core_web_md -o pret_kb
Loaded model 'en_core_web_md'

2 kb entities: ['Q2146908', 'Q7381115']
1 kb aliases: ['Russ Cochran']

Saved KB to pret_kb\kb
Saved vocab to pret_kb\vocab

Loading vocab from pret_kb\vocab
Loading KB from pret_kb\kb
2 kb entities: ['Q2146908', 'Q7381115']
1 kb aliases: ['Russ Cochran']

When I try to train the entity linker with the knowledge base from above, I get this error.

# python "entity linker.py" ./pret_kb/kb ./pret_kb/vocab
Created blank 'en' model with vocab from 'pret_kb\vocab'
Loaded Knowledge Base from 'pret_kb\kb'
Traceback (most recent call last):
  File "entity linker.py", line 156, in <module>
    plac.call(main)
  File "C:\Users\manimaran.p\AppData\Local\Continuum\anaconda3\envs\spacy\lib\site-packages\plac_core.py", line 328, in call
    cmd, result = parser.consume(arglist)
  File "C:\Users\manimaran.p\AppData\Local\Continuum\anaconda3\envs\spacy\lib\site-packages\plac_core.py", line 207, in consume
    return cmd, self.func(*(args + varargs + extraopts), **kwargs)
  File "entity linker.py", line 113, in main
    sgd=optimizer,
  File "C:\Users\manimaran.p\AppData\Local\Continuum\anaconda3\envs\spacy\lib\site-packages\spacy\language.py", line 515, in update
    proc.update(docs, golds, sgd=get_grads, losses=losses, **kwargs)
  File "pipes.pyx", line 1219, in spacy.pipeline.pipes.EntityLinker.update
KeyError: (0, 12)

I did follow the instructions specified here. I used the en_core_web_md to create the knowledge base since I do not have a pre-trained model.

I did not write any custom code just trying to run this example, Can someone point me to the right direction.

Upvotes: 2

Views: 1548

Answers (1)

John G
John G

Reputation: 58

This was asked and answered in the following issue on spaCy's GitHub.

It looks like the script no longer worked after a refactor of the entity linking pipeline as it now expects either a statistical or rule-based NER component in the pipeline.

The new script adds such an EntityRuler to the pipeline as an example. I.e., 

# Add a custom component to recognize "Russ Cochran" as an entity for the example training data.
# Note that in a realistic application, an actual NER algorithm should be used instead.
ruler = EntityRuler(nlp)
patterns = [{"label": "PERSON", "pattern": [{"LOWER": "russ"}, {"LOWER": "cochran"}]}]
ruler.add_patterns(patterns)
nlp.add_pipe(ruler)

However, this can be replaced with your own statistical NER model.

Upvotes: 1

Related Questions