Abstract:Semantic analysis is one of the fundamental and key problems in the research of content-based Text Mining. Most of supervised machine learning methods led to poor performance when work on limited tagged data. This paper investigated a novel semi-supervised learning algorithm—Transductive Support Vector Machine for shallow semantic parsing. An optimization strategy of selecting training instances, based on active learning, was integrated with TSVM. The experiment result shows that the method integrating TSVM and optimization strategy for shallow semantic parsing outperforms supervised methods on small tagged data.
[1] Xavier Carreras, Lluís Mrquez. Introduction to the CoNLL-2005 Shared Task: Semantic Role Labeling [A]. In: Proceedings of CoNLL-2005 [C]. 2005. [2] T.Joachims. Transductive inference for text classification using support vector machines [A]. In: Proc. of ICML-99 [C]. 1999. 200-209. [3] Levin, Beth. English Verb Class and Alternations: A Preliminary Investigation [M]. Chicago: University of Chicago Press. 1993. [4] AP Engelbrecht et al. Incremental Learning Using Sensitivity Analysis.IJCNN’99 [A]. International Joint Conference on Neural Networks [C]. 1999(2):1350-1355. [5] CA Thompson, ME Califf, RJ Mooney. Active Learning for Natural Language Parsing and Information Extraction [A]. In: Proceedings of the Sixteenth International Machine Learning Conference [C]. Slovenia, 1999. [6] 张健沛, 徐华. 支持向量机主动学习方法研究与应用[J]. 计算机应用, 2004, 24(3): 1-3. [7] Nianwen Xue, Martha Palmer. Automatic Semantic Role Labeling for Chinese Verbs [A]. In: The Proceedings of the Nineteenth International Joint Conference on Artificial Intelligence, Edinburgh, Scotland [C]. 2005.