Investigation of Viterbi Algorithm Performance on Part-of-Speech Tagger of Natural Language Processing
Yue Liu · 2017 International Conference on Computer Systems, Electronics and Control (ICCSEC) · 2017
Many tasks of NLP need to preprocess the words and sentences because it can make the future work convenient. Nowadays, the demand of POS tagging is increasing. POS tagging is an efficient method as preprocessing tagging, even meaningful in text to speech, syntactic analysis and machine translation. When it comes to the POS tagger, they need to know every word POS. It is available for human because it is easy for us to describe it. But when we increase the count of words to the million number, it is not possible to make people do POS tag. In this paper, we introduce Viterbi algorithm to help computer do the better job in tagging lexical categories. Viterbi algorithm is an algorithm that used dynamic programming to solve the POS of sentence. As we know, the word is sensitive about the position of the word. The POS of the word is related about the nearby words. We make simulations about how Viterbi Algorithms work in POS tagger and get the accuracy performance.