Abstract:

This paper introduces a terminology extraction method based on dependency parsing. An empirical study was conducted by using an English maritime text to compare this method with two traditional methods, namely the n-gram pattern method and the keyness analysis method. Our aim is to explore the most useful term extraction method or combination thereof for non-expert term extractors. The results indicate that the three methods complement each other, with the dependency parsing method showing the best performance. For non-expert term extractors, a combination of dependency parsing and keyness analysis achieves the optimal balance between efficiency, scope, and accuracy.

Key words: dependency parsing, terminology extraction, English maritime texts