Pattern matching in the TEXTRACT information extraction system
Tsuyoshi Kitani, Yoshio Eriguchi, Masami Hara · 1994
In information extraction systems, pattern matchers are widely used to identify information of interest in a sentence. In this paper, pattern matching in the TEXTRACT information extraction system is described. It comprises a concept search which identifies key words representing a concept, and a template pattern search which identifies patterns of words and phrases. TEXTRACT using the matcher performed well in the TIPSTER/MUC-5 evaluation. The pattern matching architecture is also suitable for rapid system development across different domains of the same language.