restlessmedia. posts from a UK web developer and music enthusiast.

I had the task recently of creating an online decision tree application and was limited technically by the environment it was to be hosted on… i.e. no DB.

It was also a requirement that the client was able to update the decision tree and possibly add new branches - this how I hold a basic question:

<nodeCollection id=”1”>
  <node id=”1”>
    <text>
      <![CDATA[Question html]]>
    </text>
    <answers />
    <parents>
      <parent nodeCollectionId=”0” id=”1” answer=”1” />
    </parents>
  </node>
</nodeCollection>

NodeCollection is (funnily enough) a collection of questions. The node represents a question.  Text is the label.  By default, these have simple boolean (yes / no) responses but it is possible to create a answers/answer collection to override these.

Parents is how this all hangs together.  A question can have 0 or more parents (usually 1 or more but that wouldn’t be the case for the first question in the first tree) and a response/answer used to travel from that particular parent.  So in the example above, question “1” in set “1” is accessed via answering question “1” with a value of “1” from the nodeCollection with an id of “0”.

This method also has an ‘address’ system to be able to access a particular question anywhere in the decision process with an id and answer value (used for hyper-linking).

There are other small features like storing responses using the ‘address’ system and also the ability to add pass through questions which still get registered as a ‘response’ but are simply continue screens.

It’s very quick, scalable and apart from the obvious issues with updating/adding a question set when a user has already taken part (and has recorded responses).

http://public.restlessmedia.com/docs/DecisionTreeSchema.xsd

Monday Aug 2nd

• xml
• development
• project