Most computer marking systems evaluate the results of the answers reached by learners without looking into the process by which the answers are produced, which will be insufficient to ascertain learners’ understanding level because correct answers may well include lucky hunches, namely accidentally correct but not confident answers. In order to differentiate these lucky answers from confident c...

We present a method to generate effective confirmation and guidance using concept-level confidence measures (CM) derived from speech recognizer output in order to handle speech recognition errors. We define two conceptlevel CM, which are on content-words and on semanticattributes, using 10-best outputs of the speech recognizer and parsing with phrase-level grammars. Content-word CM is useful fo...