<!DOCTYPE videoTopic SYSTEM "videoTopic.dtd">
<videoTopic num="062"  interactive="Y-I" automatic="Y-A" knownItems="Y-K">
  <!-- Answer of question detection -->
  <textDescription text="Search the clip which can answer the
following question: 'How long has Malcolm Baldrige National Quality
Award existed?'"/>
  <!-- The programmer of an application can suppose the case what the
user would ask the system to do, but they cannot only solve the
problems they have predict. The user maybe an common one, and they
would be interested in some information that the video clip can
provide. These information are probably embedded in the videotext,
shown in the pictures, hidden in the audio. So the application can
analyze the question and find the possible answer from the clip.-->
  <!-- Retrieval should be based on both the image and the audio
contents.Some rules should be applied to associate the supposed answer
with the features of image and audio. -->
  <knownItem src="ure1.mpg" start="01m28s" stop="01m30s"/>
  <knownItem src="ure1.mpg" start="01m01s" stop="01m10s"/>
  <knownItem src="dbe1.mpg" start="01m58s" stop="02m10s"/>
</videoTopic>