Morfawk.ja: A Japanese Token Level Pattern Matching and Processing Language with Dependency Analysis

Abstract

The quality of development documents written in natural language is basically ensured by developer's review with quite a few man-hours. Natural language processing (NLP) tools can partially automate the time-consuming review. The authors have developed morpheme level pattern matching and processing tools morfgrep and morfawk for Japanese texts for this purpose; moreover, in this paper, extends them to perform pattern matching on dependency among phrases. Example applications of the tools, proofreading of development documents and domain-specific term extraction, are also presented.