The Oracle Endeca Server allows you to switch refinement ordering on and off on a per-query basis.
A use case for this refinement ordering would be an application that renders refinements as a tag cloud. Such an application may adjust the size of the tag cloud at query time, depending on user preferences or the page from which the query originates.
You set the refinement ordering at the refinement value level that you want to control. For managed attributes, ordering is applied to that managed attribute value and all its children. For example, assume that you have a managed attribute named WineType that has three child attribute values (named Red, White, and Sparkling), which in turn have two child attribute values each. The attribute's hierarchy would look like this:
- If you set the ordering on the root attribute value (which has the same name and ID as the managed attribute itself), the refinements in the Red, White, and Sparkling attribute values will be returned.
- If there are multiple child attribute values, you can set an order on only one sibling. In this case, the refinements from the other siblings will not be exposed. For example, if you set an order on the Red attribute value, only the refinements of the Merlot and Chianti attribute values will be returned. The refinements from the White and Sparkling attribute values will be not be shown, even if you explicitly set orders for them.
The settings of the per query ordering of refinements are not persistent. That is, each query must have its own configuration, because it is not carried over from the previous query.
