MediumNLPnlp-107

BIO Tagging Decode

Problem

Decode a BIO tag sequence into entity spans. Return a list of [start, end, label] triples where end is exclusive.

A B- tag starts a new entity, an I- tag continues the current entity when the label matches and otherwise starts a new one, and O closes the current entity.

Hint

Keep the current entity start and label, flushing at every O or label change.

Test cases

case 1

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

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

case 2

in: [["I-PER","O"]]

expected: [[0,1,"PER"]]

case 3

in: [["O","O"]]

expected: []

case 4

in: [[]]

expected: []

case 5

in: [["B-PER","I-LOC","I-LOC"]]

expected: [[0,1,"PER"],[1,3,"LOC"]]

Starter codePython

def bio_decode(tags):
    # Your code here
    pass

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