Kernel-based learning for biomedical relation extraction

  • Authors:

  • Jiexun Li;Zhu Zhang;Xin Li;Hsinchun Chen

  • Affiliations:

  • College of Information Science and Technology, Drexel University, Philadelphia, PA 19104;Department of Management Information Systems, University of Arizona, Tucson AZ, 85721;Department of Management Information Systems, University of Arizona, Tucson AZ, 85721;Department of Management Information Systems, University of Arizona, Tucson AZ, 85721

  • Venue:
  • Journal of the American Society for Information Science and Technology
  • Year:
  • 2008

Quantified Score

Hi-index 0.01

Visualization

Abstract

Relation extraction is the process of scanning text forrelationships between named entities. Recently, significant studieshave focused on automatically extracting relations from biomedicalcorpora. Most existing biomedical relation extractors requiremanual creation of biomedical lexicons or parsing templates basedon domain knowledge. In this study, we propose to use kernel-basedlearning methods to automatically extract biomedical relations fromliterature text. We develop a framework of kernel-based learningfor biomedical relation extraction. In particular, we modified thestandard tree kernel function by incorporating a trace kernel tocapture richer contextual information. In our experiments on abiomedical corpus, we compare different kernel functions forbiomedical relation detection and classification. The experimentalresults show that a tree kernel outperforms word and sequencekernels for relation detection, our trace-tree kernel outperformsthe standard tree kernel, and a composite kernel outperformsindividual kernels for relation extraction. © 2008 WileyPeriodicals, Inc.