Assigning domains to speech recognition hypotheses
Klaus Rüggenmann, Iryna Gurevych · TUbilio (Technical University of Darmstadt) · 2004
We present the results of experiments aimed at assigning domains to speech recognition hypotheses (SRH). The methods rely on high-level linguistic representations of SRHs as sets of ontological concepts. We experimented with two domain models and evaluated their performance against a statistical, word-based model. Our hand-annotated and tf*idf-based models yielded a precision of 88,39% and 82,59% respectively, compared to 93,14% for the word-based baseline model. These results are explained in terms of our experimental setup.