AN EVALUATION FRAMEWORK FOR ADAPTIVE USER INTERFACES
Enrique Noriega Atala · 2014
With the rise of powerful mobile devices and the broad availability of computing power, Automatic Speech Recognition is becoming ubiquitous. Despite the rapid advance and active research in the field, a flawless ASR system is still far from existence. Because of this, interactive applications that make use of ASR technology not always recognize speech perfectly and when not, the user must be engaged to repair the mistakes in the transcriptions. When the user is engaged, he can repair these errors with different strategies. These strategies work better under different contexts. In this thesis, we explore a rational user interface that makes use of machine learning models to make its best effort in presenting the best repair strategy available to reduce the time in spent the interaction between the user and the system as much as possible. A study is conducted to determine how different candidate policies perform and conclusions are drawn from the analysis of the collected results. Once the analysis is performed, the methodology that was used is generalized in terms of a decision theoretical framework that can be used to evaluate the performance of other rational user interfaces that try to optimize an expected cost or utility. After the generalization, the framework is placed in the context of the relevant literature and some potential future work is outlined.