Exploring Features for Myanmar Named Entity Recognition

Abstract

Named Entity Recognition (NER) is the task of classifying or labeling atomic elements in the text into predefined sets of named entity categories such as Person, Location, Organization or Number. NER is also a crucial piece of Information Extraction System. Robust handling of proper names is essential for many applications in natural language processing and key knowledge acquisition infrastructure for the Semantic Web. For Myanmar Language, exploring features for NER with machine learning approach is a still challenging task because of the complex nature of the language. This paper describes our effort on feature exploring using local and external information for Myanmar NER that applied Conditional Random Fields (CRFs). The experimental results show that the best result is obtained by combining the local feature, such as neighboring words, and the external information such as clue words and name lists.