Evaluating supervised semantic parsing methods on application-independent data

Abstract

While supervised statistical semantic parsing methods have received a good amount of attention in recent years, this research has largely been done on small and specialized data sets. This paper introduces a work-in-progress with the objective of examining the applicability of supervised statistical semantic parsing to application-independent data with linguistically motivated meaning representations. The approach discussed in this paper has three key aspects: The circumvention of data scarcity using automatic annotation, experimentation with different types of meaning representations, and the design of a suitable graded evaluation measure. © 2014 Springer-Verlag Berlin Heidelberg.