Course, Chapter 3, entities in 7) Complex components

[oxidian24]

Hi,
I just wanted to ask you what's the point of creating the entities in the chapter 3, point 7. I mean, why this example uses this approach, why not just to list the span directly? Does this lesson mean that I can define any data I want to be collected in the entities?

@Language.component("animal_component")
def animal_component_function(doc):
    matches = matcher(doc)
    spans = [Span(doc, start, end, label="ANIMAL") for match_id, start, end in matches]
    doc.ents = spans
    return doc

I just dont understand why this needs to be collected in the span and with the ANIMAL label, why not just using the doc[x:y] approach?

[danieldk]

Does this lesson mean that I can define any data I want to be collected in the entities? [...] Does this lesson mean that I can define any data I want to be collected in the entities?

You can extract entities however you would like and assign them to doc.ents, however the list or tuple that you assign to doc.ents must consist of spans.

I just dont understand why this needs to be collected in the span and with the ANIMAL label, why not just using the doc[x:y] approach?

doc[start:end] would create a span without a label and spans without labels are ignored when assigning to doc.ents. So a solution with doc slicing would look something like this:

@Language.component("animal_component")
def animal_component_function(doc):
    matches = matcher(doc)
    spans = []
    for match_id, start, end in matches:
        span = doc[start:end]
        span.label_ = "ANIMAL"
        spans.append(span)
    doc.ents = spans
    return doc

which is less compact than the solution in the course.