Multiobjective Optimization Approach for Named Entity Recognition

In this paper, we propose a multiobjective optimization (MOO) based technique to determine the appropriate weight of voting for each class in each classifier for Named Entity Recognition (NER). Our underlying assumption is that reliability of predictions of each classifier differs among the various named entity (NE) classes. Thus, it is necessary to quantify the amount of voting for each class in a particular classifier. We use Maximum Entropy (ME) as the base to generate a number of classifiers depending upon the various feature representations. The proposed algorithm is evaluated for a resource-constrained language like Bengali that yield the overall recall, precision and F-measure values of 79.98%, 82.24% and 81.10%, respectively. Experiments also show that the classifier ensemble identified by the proposed multiobjective based technique outperforms all the individual classifiers, three different conventional baseline ensembles and an existing single objective optimization based approach.