At ICSLP'00, we introduced the concept of data-driven semantic inference, an approach to command and control which in principle allows for any word constructs in command/query formulation. Unconstrained word strings are mapped onto the relevant action through an automated classification relying on latent semantic analysis: as a result, it is no longer necessary for users to memorize the exact syntax of every command. The objective of this paper is to further characterize the behavior of semantic inference, using a desktop command and control task involving 113 different actions. We consider various training scenarios of increasing scope to assess the influence of coverage on performance. Under realistic usage conditions, good classification results can be obtained at a level of coverage as low as 70%.
Cite as: Bellegarda, J.R., Silverman, K.E.A. (2001) Data-driven semantic inference for unconstrained desktop command and control. Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001), 455-458, doi: 10.21437/Eurospeech.2001-121
@inproceedings{bellegarda01_eurospeech,
author={Jerome R. Bellegarda and Kim E. A. Silverman},
title={{Data-driven semantic inference for unconstrained desktop command and control}},
year=2001,
booktitle={Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001)},
pages={455--458},
doi={10.21437/Eurospeech.2001-121}
}