Paper: A Re-examination of Dependency Path Kernels for Relation Extraction

ACL ID I08-2119
Title A Re-examination of Dependency Path Kernels for Relation Extraction
Venue International Joint Conference on Natural Language Processing
Session Main Conference
Year 2008
Authors

Extracting semantic relations between enti- ties from natural language text is an impor- tant step towards automatic knowledge ex- traction from large text collections and the Web. The state-of-the-art approach to rela- tion extraction employs Support Vector Ma- chines (SVM) and kernel methods for classi- fication. Despite the diversity of kernels and the near exhaustive trial-and-error on ker- nel combination, there lacks a clear under- standing of how these kernels relate to each other and why some are superior than oth- ers. In this paper, we provide an analysis of the relative strength and weakness of several kernels through systematic experimentation. We show that relation extraction can bene- fit from increasing the feature space through convolution kernel and introducing bias to- wa...