Smart Grammar: A Dynamic Spoken Language Understanding Grammar for Inflective Languages

Abstract

Spoken language understanding (SLU) is a key requirement of spoken dialogue systems (SDS). The role of SLU parser is to robustly interpret the meanings of users' utterance using a hand-crafted grammar that is expensive to build. This task becomes even harder when the developer is creating a SLU grammar for inflectional languages due to the different conjugations and declensions. This causes long grammar definition files that are hard to structure and also to manage. In this paper, we propose a new and alternative method, called Smart Grammar to facilitate the development of speech enabled applications. This uses a morphological analyzer, in addition to the semantic parser, in order to convert each user utterance in the canonical form.