A Concise Review of Named Entity Recognition System: Methods and Features

Abstract

Named Entity Recognition (NER) is an elementary tool for all application areas in Natural Language Processing (NLP) such as Automatic Summarization, Information Extraction, Information Retrieval, Text Mining, Machine Translation, Question Answering, and Genetics. NER is a task to discover and categorises the named entities (‘atomic elements’) in the text into predefined classes such as the names of persons, organizations, locations, terminologies of time, quantity and etc. Different languages may have different morphologies and thus involve dissimilar NER procedures. For example, an Arabic NER system cannot be practically used in processing Malay texts due to the different morphological features. The morphological features of every language are rich and complex and donates to the difficulties of implementing an actual method to develop the accurate NER system. In this paper, we review on three main techniques that commonly used to develop an NER system well-known as Rule-Based, Machine Learning, and Hybrid approach. This paper also highlights the features of each technique.