Using knowledge acquisition to build spoken language systems
The development of spoken language interfaces is of increasing importance due to the progress in speech recognition technology. Unfortunately, the development of comprehensive dialog system is still too labour-intensive to allow their widespread dissemination. This paper presents a new approach to the design of spoken language dialog systems. We view the design task as a knowledge acquisition and knowledge maintenance task. We developed a workbench, PIA, which allows continuous knowledge acquisition and maintenance without a knowledge engineer. It uses also the available data to produce suggestions of how to amend the current knowledge base to accommodate new cases of recorded dialog interactions. First experimental results with our workbench are promising.
Unable to display preview. Download preview PDF.
- 1.Large vocabulary continuous speech recognition using HTK. In Proceedings of ICASSP, pages 125–128, 1994.Google Scholar
- 3.P. Compton and R. Jansen. A philosophical basis for knowledge acquisition. Knowledge Acquisition, 2:241–257, 1990.Google Scholar
- 4.P. Compton, B. Kang, P. Preston, and M. Mulholland. Knowledge acquisition without knowledge analysis. In Proceedings of the European Knowledge Acquisition Workshop, pages 277–299. Springer-Verlag, 1993.Google Scholar
- 8.G. Schreiber, B. Wielinga, and J. Breuker. KADS A Principled Approach to Knowledge-Based System Development. Academic Press, 1993.Google Scholar