Biomedical Named Entity Recognition: A Review

Abstract

Biomedical Named Entity Recognition (BNER) is the task of identifying biomedical instances such as chemical compounds, genes, proteins, viruses, disorders, DNAs and RNAs. The key challenge behind BNER lies on the methods that would be used for extracting such entities. Most of the methods used for BNER were relying on Supervised Machine Learning (SML) techniques. In SML techniques, the features play an essential role in terms of improving the effectiveness of the recognition process. Features can be identified as a set of discriminating and distinguishing characteristics that have the ability to indicate the occurrence of an entity. In this manner, the features should be able to generalize which means to discriminate the entities correctly even on new and unseen samples. Several studies have tackled the role of feature in terms of identifying named entities. However, with the surge of biomedical researches, there is a vital demand to explore biomedical features. This paper aims to accommodate a review study on the features that could be used for BNER in which various types of features will be examined including morphological features, dictionary-based features, lexical features and distance-based features.