Advances in the Witchcraft Workbench Project

Alexander Schmitt1,  Nada Sharaf2,  Wolfgang Minker1
1Ulm University, Germany, 2German University in Cairo, Egypt


The Workbench for Intelligent exploraTion of Human ComputeR conversaTions is a new platform-independent open-source workbench designed for the analysis, mining and management of large spoken dialogue system corpora. What makes Witchcraft unique is its ability to visualize the effect of classification and prediction models on ongoing system-user interactions. Witchcraft is now able to handle predictions from binary and multi-class discriminative classifiers as well as regression models. The new XML interface allows a visualization of predictions stemming from any kind of Machine Learning (ML) framework. To a special degree the workbench targets Interactive Voice Response systems and the management of large corpora from telephone applications, but is sufficiently general to cover all kinds of SDS-based corpora. We adapted the wide-spread CMU Let’s Go corpus to demonstrate Witchcraft.