POS tagging using the Viterbi algorithm and n-gram models.

               

You can easily run this code on google colab by just clicking this badge

the dataset is a train set and the test set we create the validation set from the train set for validating our model.

here is part of the train set frame :

here is part of the test set frame :

here is our formula we need

so for calculating the P(T|W) we should calculate the emission and transition model.
the P(W|T) is the emission section and the P(Ti) is the transition section calculated by ngram

here is different accuracy from different N-gram model we can compare them together :

Uni-gram :

Accuracy: 25.5772
Precision: 40.7354
Recall: 40.7354
F1-score: 40.7354

Bigram :

Accuracy: 92.4962
Precision: 96.1019
Recall: 96.1019
F1-score: 96.1019

Trigram :

Accuracy: 92.3538
Precision: 96.0249
Recall: 96.0249
F1-score: 96.0249

This project is licensed under the MIT License.