▽TOP▽活動実績▽２００１年▽

(27)　A Robust Method for Automatic and Dynamic Acquisition of Japanese Syntactic Rules
　　　　　　from Raw Text Using Inductive Learning
　　　　　　　　　　　　Proceedings of the IASTED International Conference
　　　　　　　　　　　　　　　　ARTIFICIAL INTELLIGENCE AND SOFT COMPUTING,pp.301-306,2001-5

　We propose a robust method for automatic and dynamic acquision of Japanese syntactic rules from raw text using inductive learning. The previous methods have not had the robustness, that is, the capability of unsupervised learning, a dynamic adaptation to a current domain, and an acquisition from the condition of no initial rules. We describe the framework for dynamic unsupervised learning and the learning algorithm. Then, We demonstrate its robustness in the experiments.