A context-based grammar generation in mixed initiative dialogue system for visually impaired

Abstract

In this paper, we present some principles of designing mixed-initiative dialogue systems suited to the needs of visually impaired users related with dialogue based web page generation. These principles have been implemented in the online BrowserWebGen system that allows users to generate their web pages. The primary user interface is implemented as a dialogue system with implicit confirmation. The users are able to enter one piece of semantic information at each step of a dialogue interaction. Second part of the application allows the user to activate mixed initiative dialogue interface in each dialogue state. This interface is called Dialogue Prompt Window and it provides an additional support to the users as well as control over the application in natural language. © 2010 Springer-Verlag Berlin Heidelberg.