A Classification of Grammar-Infused Templates for Ontology and Model Verbalisation

Abstract

Involving domain-experts in the development, maintenance, and use of knowledge organisation systems can be made easier through the introduction of easy-to-use interfaces that are based on natural language. Well resourced languages make use of natural language generation techniques to provide such interfaces. In particular, they often make use of templates combined with computational grammar rules to generate grammatically complex text. However, there is no model of pairing templates and computational grammar rules to ensure suitability for less-resourced languages. These languages require a modular design that ensures grammar detachability so as to allow grammar re-use across domains and applications. In this paper, we present a model and classification scheme for grammar-infused templates suited for less-resourced languages and classify existing systems that make use of them. We have found that of the 15 systems that pair templates and grammar rules, and their 11 distinct template types, 13 have support for detachable grammars.