Hidden markov model and maximum entropy Markov model: A comparison in POS tagging with the AnCora corpus

Abstract

POS This paper presents the comparison between Hidden Markov Model - HMM and Maximum Entropy Markov Model - MEMM, in the POS Tagging, for the spanish language, with the AnCora linguistic corpus. The study is quantitative and descriptive. HMM achieves an accuracy of 99%. MEMM achieves an accuracy in the labeling of 98.04%, taking into account that for this model, it used features on the observations. © 2019, Associacao Iberica de Sistemas e Tecnologias de Informacao. All rights reserved.