Named entity recognition in NLP - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the library used for named entity recognition.

NLP

import [1]

Drag options to blanks, or click blank then click option'

Asklearn

Bpandas

Cmatplotlib

Dnltk

Attempts:

3 left

💡 Hint

Common Mistakes

Importing pandas or matplotlib which are not used for NLP.

Using sklearn which is mainly for machine learning but not specifically for NER.

The nltk library is commonly used for natural language processing tasks including named entity recognition.

2fill in blank

medium

Complete the code to tokenize the sentence for named entity recognition.

NLP

from nltk import word_tokenize
sentence = "Apple is looking at buying U.K. startup for $1 billion"
tokens = [1](sentence)

Drag options to blanks, or click blank then click option'

Aword_tokenize

Bsent_tokenize

Cpos_tag

Dne_chunk

Attempts:

3 left

💡 Hint

Common Mistakes

Using sent_tokenize which splits text into sentences, not words.

Using pos_tag or ne_chunk before tokenizing.

word_tokenize splits the sentence into words, which is the first step before tagging or chunking.

3fill in blank

hard

Fix the error in the code to perform named entity recognition on tokenized text.

NLP

from nltk import pos_tag, ne_chunk

tokens = ['Apple', 'is', 'looking', 'at', 'buying', 'U.K.', 'startup', 'for', '$', '1', 'billion']
pos_tags = [1](tokens)
named_entities = ne_chunk(pos_tags)

Drag options to blanks, or click blank then click option'

Aword_tokenize

Bpos_tag

Csent_tokenize

Dne_chunk

Attempts:

3 left

💡 Hint

Common Mistakes

Using word_tokenize on already tokenized list.

Using ne_chunk before pos_tag.

pos_tag is used to assign part-of-speech tags to tokens before named entity chunking.

4fill in blank

hard

Fill both blanks to extract named entities as a list of tuples (entity, type).

NLP

entities = []
for subtree in named_entities.[1]():
    if hasattr(subtree, '[2]'):
        entity_name = ' '.join([token for token, pos in subtree.leaves()])
        entity_type = subtree.label()
        entities.append((entity_name, entity_type))

Drag options to blanks, or click blank then click option'

Asubtrees

Bchildren

Clabel

Dleaves

Attempts:

3 left

💡 Hint

Common Mistakes

Using children() which returns immediate children but not all subtrees.

Checking for leaves attribute instead of label.

We use subtrees() to iterate over chunks and check if they have a label attribute to identify named entities.

5fill in blank

hard

Fill all three blanks to train a simple NER model using spaCy.

NLP

import spacy
from spacy.training import Example

nlp = spacy.blank('en')
ner = nlp.add_pipe('[1]')

TRAIN_DATA = [
    ("Google is a tech company", {"entities": [(0, 6, '[2]')]}),
    ("I live in New York", {"entities": [(10, 18, '[3]')]})
]

optimizer = nlp.begin_training()
for itn in range(10):
    for text, annotations in TRAIN_DATA:
        doc = nlp.make_doc(text)
        example = Example.from_dict(doc, annotations)
        nlp.update([example], sgd=optimizer)

Drag options to blanks, or click blank then click option'

Aner

BORG

CGPE

Dtextcat

Attempts:

3 left

💡 Hint

Common Mistakes

Adding 'textcat' pipe which is for text classification, not NER.

Using wrong entity labels like 'LOC' instead of 'GPE'.

We add the ner pipe to the model, and label entities as ORG (organization) and GPE (geopolitical entity).