• Query a database of (Calvin and Hobbes, Asterix, Dilbert, …) comic strips by content.
• We want to approach the system with queries like:
  • Find all strips featuring Calvin.
  • Find all panels with Calvin screaming at Hobbes.
  • List all discussions between Calvin and his mom.

• Although fairly simplistic, the previous stage clearly constitutes an improvement.
• XML comes with a number of additional constructs which allow us to convey even more useful information, e.g.,
  1. Attributes may be used to qualify tags to reduce the number of different tag names (tag soup):
     • Instead of

       <question>If you could wish ... ?</question>

       <scream>A STUPID FIELD?! ...</scream>

     • we could use

       <bubble tone="question">If you could wish ...?</bubble>

       <bubble tone="scream" mood="excited">A STUPID FIELD?! ...?</bubble>

  2. References establish links internal to an XML document:
     • Establish link target:

       <character id="calvin">Calvin, a six-year-old boy, named after the 16th-century theologian John Calvin.</character>

       <character id="hobbes">Hobbes, stuffed and anthropomorphic Bengal Tiger.</character>

     • Reference the target:

       <bubble speaker="hobbes" to="calvin" tone="answer" mood="relaxed">A big sunny field to be in.</bubble>

Calvin-4.xml

<strip copyright="Universal Press Syndicate" year="1986">
  <prolog>
    <series url="http://www.gocomics.com/calvinandhobbes/">Calvin and Hobbes</series>
    <author>Bill Watterson</author>
    <genres>
      <genre>humor</genre>
      <genre>family life</genre>
      <genre>politics</genre>
      <genre>satire</genre>
    </genres>
    <characters>
      <character id="calvin">Calvin, a precocious, mischievous, and adventurous six-year-old boy.</character>
      <character id="hobbes">Hobbes, sardonic, stuffed and anthropomorphic Bengal Tiger.</character>
      <character id="wormwood">Miss Wormwood, Calvin's world-weary teacher.</character>
    </characters>
  </prolog>
  <panels length="4">
    <panel no="1">
      <scene visible="calvin hobbes">Calvin and Hobbes are walking down the alley.</scene>
      <bubbles>
        <bubble speaker="calvin" to="hobbes" tone="question">If you could wish for anything, what would it be?</bubble>
      </bubbles>
    </panel>
    <panel no="2">
      <scene visible="calvin hobbes">Hobbes looking happy with a smile. Calvin sitting next to Hobbes, listening to him.</scene>
      <bubbles>
        <bubble speaker="hobbes" to="calvin" tone="answer" mood="relaxed">A big sunny field to be in.</bubble>
      </bubbles>
    </panel>
    <panel no="3">
      <scene visible="calvin">Calvin, very excited, is screaming.</scene>
      <bubbles>
        <bubble speaker="calvin" to="hobbes" tone="screaming" mood="excited">A STUPID FIELD? YOU'VE GOT THAT NOW! THINK BIG! RICHES! POWER! PRETEND YOU COULD HAVE ANYTHING!</bubble>
      </bubbles>
    </panel>
    <panel no="4">
      <scene visible="calvin hobbes">Calvin, thoughtful, looking at Hobbes. Hobbes, absolutely satisfied, lying in the field, sleeping.</scene>
      <bubbles>
        <bubble speaker="calvin" to="calvin" tone="whispering" mood="thoughtful">Actually, it's hard to argue with someone who looks so happy.</bubble>
        <bubble speaker="hobbes" mood="relaxed">zzZZZzzz</bubble>
      </bubbles>
    </panel>
  </panels>
</strip>

XML was designed to satisfy the needs of two worlds:

1. document-centric applications (e.g. Information Retrieval)
   • Information with little structure (e.g. text documents with chapters, sections, ... )
2. data-centric applications ("traditional" database applications)
   • Very regular data; sometimes, however, we might want some flexibility (e.g. persons without a phone number, or with even two, ... )

While in data-centric application domains we expect high-volume data, you probably imagine how an XML-based approach to represent document-centric data can ultimately lead to a high-volume data processing challenge as well:

<comic-strips>
  <calvin-strips>
    <strip date="1985-11-18"> ... </strip>
        ...
        <strip date="1995-12-31"> ... </strip>
  </calvin-strips>
  <dilbert-strips>
    <strip date="1988-07-21"> ... </strip>
        ...
        <strip date="2005-10-12"> ... </strip>
  </dilbert-strips>
</comic-strips>

<strip>
  <prolog> ... </prolog>
  <drawing encoding="base64" mime-type="application/pdf">
  JVBERi0xLjMKJcTl8uXrp/Og0MTGCjIgMCBvYmoKPDwgL0xlbmd0aCA0IDA... bGF0ZURlY29kZSA+PgpzdHJlYW0KeNorVAhUKFTQD0gtSk4tKClNzFEoygQ... ...
  </drawing>
  <panels> ... </panels>
</strip>

• Also see XOP ( XML-binary Optimized Packaging: http://www.w3.org/TR/xop10/)

• In the course of this lecture we will have a look at database-supported XML processors able to process input XML documents of 1 GB size and beyond (see next assignment).
• XPath and XQuery will be used to filter, transform, join XML documents much like tables in the relational data model.