Rule-based entity recognition for non-default english models

[transfluxus]

How to reproduce the behaviour

Using the 1. example of entityruler-usage
Using the entity ruler

Trying it out with the following 3 models:

for model in ["en_core_web_sm", "en_core_web_md", "en_core_web_lg"]:
    print(model)
    nlp = spacy.load(model)
    ruler = nlp.add_pipe("entity_ruler")
    patterns = [{"label": "ORG", "pattern": "Apple"},
                {"label": "GPE", "pattern": [{"LOWER": "san"}, {"LOWER": "francisco"}]}]
    ruler.add_patterns(patterns)

    doc = nlp("Apple is opening its first big office in San Francisco.")
    print([(ent.text, ent.label_) for ent in doc.ents])

the results are:

en_core_web_sm
[('Apple', 'ORG'), ('first', 'ORDINAL'), ('San Francisco', 'GPE')]
en_core_web_md
[('Apple', 'ORG'), ('first', 'ORDINAL'), ('San Francisco', 'GPE')]
en_core_web_lg
[('Apple', 'ORG'), ('first', 'ORDINAL'), ('San Francisco', 'GPE')]

Furthermore, when I try to add ids to the patterns like in the next section the ent_id_ are simple empty strings.

Your Environment

Info about spaCy

• spaCy version: 3.4.1
• Platform: Linux-5.13.0-21-generic-x86_64-with-glibc2.34
• Python version: 3.9.7
• Pipelines: en_core_web_sm (3.4.0), es_core_news_md (3.4.0), en_core_web_md (3.4.0), pt_core_news_sm (3.4.0), pt_core_news_md (3.4.0), es_core_news_sm (3.4.0), es_core_news_lg (3.4.0), en_core_web_lg (3.4.0), pt_core_news_lg (3.4.0), fr_core_news_md (3.4.0)

[transfluxus]

Maybe to add:
When I use the default model with

    from spacy.lang.en import English
    nlp = English()

the results are as expected

[('Apple', 'ORG', 'apple'), ('San Francisco', 'GPE', 'san-francisco')]
[('Apple', 'ORG', 'apple'), ('San Fran', 'GPE', 'san-francisco')]
...

[polm]

By default, the EntityRuler doesn't overwrite entities. That means that entities detected by the pretrained pipelines will not be modified, which includes not giving them entity IDs. You can use the overwrite_ents setting to change this, see the docs.

Moving to Discussions because this is not a bug report.