pushpendrapratap / Build-a-Language-Model-and-Part-of-Speech-POS-tagger

In this, I have created, trained and evaluated the language model using Brown corpus and implemented POS tagger using Viterbi algorithm.

Build-a-Language-Model-and-Part-of-Speech-POS-tagger

In this, we have created, trained and evaluated the language model using Brown corpus and implemented POS tagger using Viterbi algorithm.

performance/accuracy of the model

perplexity measurement for diff. Unigram, Bigram and Trigram model. https://cloud.githubusercontent.com/assets/9404205/20698652/956adefa-b628-11e6-8350-af3f44fc64ea.PNG

Accuracy of Part of Speech(POS) tagger, evaluated on Brown corpus. https://cloud.githubusercontent.com/assets/9404205/20698754/35db8902-b629-11e6-8ecc-80c6d78c66cf.PNG

This is the result I got with my implementation of the Viterbi algorithm:

Percent correct tags: 91.6848750171

This is the accuracy that I got with NLTK’s tagger accuracy (implemented using back off):

Percent correct tags: 86.7918775773