ENSEMBLE BASED ACTIVE ANNOTATION FOR BIOMEDICAL NAMED ENTITY RECOGNITION

Abstract

ABSTRACT An important prospect of machine learning for information extraction to deal with the problems of high cost of collecting labelled examples. Active Learning makes more efficient use of the learner’s time by asking them to label only instances that are most useful for the trainer. In random sampling approach, unlabeled data is selected for annotation at random and thus can’t yield desired result. In contrast, active learning selects the useful data from a huge pool of unlabeled data for the classifier. The strategies used often classify the corpus tokens (or, data points) under wrong classes. The classifier is confused between two categories if the token is located near the margin. We propose a novel method for solving this problem and show that it favourably results in the increased performance. Our proposed framework is based on an ensemble approach, where ID3 and C5 algorithms are used as the base classifiers. The proposed approach is applied for solving the problem of named entity recognition (NER) in the Bio-medical domain. Results show that the proposed technique indeed improves the performance of the system