Abstract: A method and apparatus for continuously improving the performance of semantic classifiers in the scope of spoken dialog systems are disclosed. Rule-based or statistical classifiers are replaced with better performing rule-based or statistical classifiers and/or certain parameters of existing classifiers are modified. The replacement classifiers or new parameters are trained and tested on a collection of transcriptions and annotations of utterances which are generated manually or in a partially automated fashion. Automated quality assurance leads to more accurate training and testing data, higher classification performance, and feedback into the design of the spoken dialog system by suggesting changes to improve system behavior.

Abstract: A single, subjective numerical rating to evaluate the performance of a telephone-based spoken dialog system. This CE rating is provided by expert human listeners who have knowledge of the design of the dialog system. Different human raters can be trained to achieve a satisfactory level of agreement. Furthermore, a classifier trained on ratings by human experts can reproduce the human ratings with the same degree of consistency. More calls can be given a CE rating than would be possible with limited human resources. More information can be provided about individual calls, e.g., to help decide between two disparate ratings by different human experts.