Evaluation of part of speech tagging on Persian text

Abstract

Part of Speech (POS) tagging is an essential part of text processing applications. A POS tagger assigns a tag to each word of its input text specifying its grammatical properties. One of the popular POS taggers is TnT tagger which was shown to have high accuracy in English and some other languages. It is always interesting to see how a method in one language performs on another language because it would give us insight into the difference and similarities of the languages. In case of statistical methods such as TnT, this will have an added practical advantages also. This paper presents creation of a POS tagged corpus and evaluation of TnT tagger on Persian text. The results of experiments on Persian text show that TnT provides overall tagging accuracy of 96.64%, specifically, 97.01% on known words and 77.77% on unknown words.