点击次数：

论文类型：期刊论文

发表时间：2009-04-01

发表刊物：Journal of Information and Computational Science

收录刊物：EI、Scopus

卷号：6

期号：2

页面范围：1067-1075

ISSN号：15487741

摘要：Along with the rapid development of the biomedicine, automated protein-protein interaction (PPI) extraction from scientific literature is a task of significant interest in the BioNLP field. Addressing this problem, in this paper we propose an ensemble kernel combing word feature based kernel and path-kernel which is defined in this paper and based on paths of the parse trees, and this ensemble kernel achieves much better performance than word feature based kernel and path-kernel. At the same time the ensemble kernel performs better than the graph kernel in recall and F-score on the IEPA corpus. The best F-score of our method on IEPA corpus is 75.9%, the recall and precision is respectively 84.3% and 69.1%. Copyright ?2009 Binary Information Press.