The corpus element

The corpus element is always root element in a .corpus file. Its DTD-based definition is:

<!ELEMENT corpus (question|probe)*>
<!ATTLIST corpus
          description CDATA #IMPLIED
          image CDATA #IMPLIED
          position CDATA #IMPLIED
          size CDATA #IMPLIED
          rank CDATA #IMPLIED
          evalall CDATA #IMPLIED>

description allows you to attach arbitrary documentation to the corpus.

image is the background image to be shown when the corpus is active. These are within the /static directory

position is the position (e.g. "200px 300px") of the background image.

size is the size (e.g. "1000px 700px") of the background image.

rank determines the priority of the corpus when compared to other corpora when showing the list of corpora to select (e.g. a corpus with rank "01" will appear before a corpus with rank "05") of the background image.

evalall, if true, lets users execute all the questions in the corpus with a single button click. This defaults to false.

Within the corpus element, sub-elements represent question or probe elements.

Here is the simple corpus within the geo configuration.

<?xml version="1.0"?>
<corpus>
    <question text="give me the cities in california"/>
    <question text="what is the population of new york"/>
    <question text="give me the largest state"/>
    <question text="give me the states that border utah"/>
    <question text="smallest city in the largest state"/>
    <question text="3 tallest Mountains in alaska"/>
    <question text="3 tallest mountains on west coast"/>

    <probe text="names of states"/>
    <probe text="names of cities"/>
    <probe text="names of rivers"/>
    <probe text="names of mountains"/>
    <probe text="names of lakes"/>
    <probe text="numbers of roads"/>
    <probe text="my cities" answer="Ann Arbor,East Lansing,New York,Los Angeles,Berkeley,Austin"/>
</corpus>