MediumNLPnlp-108

Span to BIO Tags

Problem

Convert entity spans back to BIO tags for a sequence of n tokens. Each span is [start, end, label] with an exclusive end.

Emit 'B-label' at start, 'I-label' for the remaining positions up to end, and 'O' everywhere else.

Hint

Slice the span against n so an over-long end cannot overflow the tag list.

case 1

in: [4,[[0,2,"PER"],[3,4,"LOC"]]]

expected: ["B-PER","I-PER","O","B-LOC"]

case 2

in: [2,[]]

expected: ["O","O"]

case 3

in: [0,[]]

expected: []

case 4

in: [3,[[1,3,"MISC"]]]

expected: ["O","B-MISC","I-MISC"]

Starter codePython

def span_to_bio(n, spans):
    # Your code here
    pass

4 test cases · run your code against them in the editor