engleski

Modeling Natural Language Dialogue for Croatian Weather Forecast System

Language and speech resources are of crucial importance for research and development in language technology. The central module of any natural language dialogue system is the dialogue manager. This paper presents common dialogue manager architectures and discusses their possible application in Croatian weather forecast system together with dialogue initiative strategies and problems that spoken dialogue systems face concerning the input. Beside the fact that the language is ambiguous in itself, there are cases of the speech input which is not linguistically or grammatically covered by the system. Moreover, the interaction may occur in noisy environment. Most of these problems are discussed here and our concern will be how to overcome these problems and build a robust Croatian spoken dialogue system.

dialogue manager; semantic analysis; robustness; dialogue initiative strategies; interpretation subsystem

nije evidentirano