Tagging of speech acts in Persian texts using a hidden Markov model

This document addresses one of the major challenges in computational linguistics known as Part-of-Speech (POS) tagging. POS tagging, which is among the most fundamental requirements for intelligent text processing, is dependent on the language of the text being processed. Therefore, developing a robust POS tagger for the Persian language became one of our primary priorities. The technique we employed to solve this problem is the Hidden Markov Model (HMM). This technique is widely used in various tagging approaches; for instance, it is utilized in the TNT tagger [2], which is one of the most powerful taggers across different languages [4, 5, 8]. According to our experiments, using this tagger enables the identification of Persian word morphological categories with an accuracy of 94.3%.