Identifying Canonical Result Tree Nodes

• There is no 100% accurate general solution
• Heuristic:
  1. Select all result nodes from the target source node
  2. If some were generated in default mode, eliminate others
  3. If some were generated from an xsl:template that matched the source node, discard others
  4. If multiple nodes remain, treat it as an aggregate link
• Examples:
  • chapter processed “as itself” and as the target of an xref
  • A single database formatted as multiple reports