Character Relation Extraction Based on the Combination of Multi-View Ensemble and Multi-Classifier Ensemble

Current research has made great progress in the information extraction, but the research achievements on the character relation extraction are scarce and preliminary. This paper proposes the approaches to extract character relations from texts based on the combination of multi-view and multi-classifier ensemble. This paper also applies over-sampling data balance technique and researches how the sequence of multi-view and multi-classifier ensemble methods affects the performance of character relation extraction. The ensemble methods have two kinds of strategies: MVC and MCV. The experimental results have shown that the performance of MCV is better than MVC. Therefore, the research done in this paper provides a theoretical basis for character relation extraction research.