How to train a unigram tagger in nltk

Recipe Objective

This recipe explains how to train a unigram tagger.

Step 1: Importing library

Let us first import the necessary libraries. We'll import UnigramTagger and treebank from nltk.tag and nltk.corpus respectively.

from nltk.tag import UnigramTagger
from nltk.corpus import treebank

Step 2: Unigram Tagger

In simple words, Unigram Tagger is a context-based tagger whose context is a single word. To determine the POS(Part of Speech) tag Unigram tagger only uses a single word as its context. In the below example, we'll be training using the first 3000 tagged sentences of the treebank corpus as data. These tagged sentences are used by unigram to build a model. Since unigram is inherited from ContextTagger context() method is implemented instead of choose_tag() method.
UnigramTagger -> NgramTagger -> ContextTagger -> SequentialBackoffTagger.

train = treebank.tagged_sents()[:3000]
tag = UnigramTagger(train)
treebank.sents()[1]
tag = UnigramTagger(model ={'Elsevier': 'NN'})
tag.tag(treebank.sents()[1])