A Joint Model for Vietnamese Part-of-Speech Tagging Using Dual Decomposition

Abstract

Part-of-speech (POS) tagging is a fundamental task of Natural Language Processing (NLP). It provides useful information for many other NLP tasks, including word sense disambiguation, text chunking, named entity recognition, syntactic parsing, semantic role labeling, and semantic parsing. Several methods have been proposed to deal with the POS tagging task in Vietnamese. They can be divided into two types of models: word-based models and syllable-based models. While a word-based model assigns a POS tag to each word, a syllable-based model assigns a POS tag to each syllable. This chapter presents a new model for Vietnamese POS tagging using dual decomposition. The chapter shows how dual decomposition can be exploited to integrate a word-based model and a syllable-based model to yield a more powerfulmodel for tagging Vietnamese sentences. Then the chapter describes experiments on the Viet Treebank corpus, a large annotated corpus for Vietnamese POS tagging. This chapter also presents an error analysis to investigate which types of words in Vietnamese are more difficult to tag than other words. Experimental results show that the word-based model and the syllable-based model are complementary. Moreover, the proposed model using dual decomposition outperforms both the word-based and the syllable-based models.