Maxenttagger for Malays Jawi POS-tags

• Main Authors: Abu Bakar, Juhaida, Omar, Khairuddin, Nasrudin, Mohammad Faidzul, Murah, Mohd Zamri
• Format: Conference or Workshop Item
• Language: English
• Published: 2017
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://repo.uum.edu.my/24498/1/SICONSEM%202017%2031%2033.pdf; http://repo.uum.edu.my/24498/
• Institution: Universiti Utara Malaysia

Purpose - Malay is a major language of the Austronesian family spoken in many countries. Malay Jawi is lacking in annotated resources and tools. In addition, Part-of-speech (POS) ambiguity in Natural Language Processing (NLP) is a vague important phenomenon that needs to be solved immediately. Since POS is an important feature of the word, and is the link between the words and syntax, POS tagging (POST) needs to provide intermediate results showing superior performance to the next NLP tasks. POS ambiguity is a main problem in increasing POST performance. POST performance is often measured with accuracy and precision of a tag and it was considered critical to NLP application. Some of the standard package POS tagging provided in Natural Language ToolKit (NLTK) are Brill tagger, HMM tagger, and CRF Tagger. In this paper, POST Malay Jawi implemented NLP tools, NLTK for the state-of-the-art methods tagger; maximum entropy models. NLTK is used as the implementation tool for Jawi tagging, as syntax and semantics of the language is transparent, and it has the good functionality of NLP-operator. The tool also uses Python as the implementation language.