Building speech-based conversational systems involves the development of several speech specific control mechanisms such as validation, confirmation, disambiguation in addition to the actual application call-flow. We present an architecture for pluggable disambiguation mechanism for speech based conversational systems. The architecture provides a mechanism to decouple the disambiguation from the voice application. Several disambiguation strategies can be designed to disambiguate a user input. These strategies can be applied to the user input in a seamless manner. The disambiguated value from one component can be passed on to another component for further disambiguation. We implement the architecture by using it in the Reusable Dialog Component framework. Several illustrative examples are presented to highlight the effectiveness of having a pluggable disambiguation mechanism for voice applications.
Cite as: Faruquie, T., Kankar, P., Rajput, N., Verma, A. (2005) An architecture for pluggable disambiguation mechanism for RDC based voice applications. Proc. Interspeech 2005, 3409-3412, doi: 10.21437/Interspeech.2005-829
@inproceedings{faruquie05_interspeech,
author={Tanveer Faruquie and Pankaj Kankar and Nitendra Rajput and Abhishek Verma},
title={{An architecture for pluggable disambiguation mechanism for RDC based voice applications}},
year=2005,
booktitle={Proc. Interspeech 2005},
pages={3409--3412},
doi={10.21437/Interspeech.2005-829}
}