Re: eliminating duplicates

If there was only one attribute, lets say "source" it would be simple:
<xsl:for-each
select="//edge[not(./@source=preceding-sibling::edge/@source)]">
   <xsl:copy-of select="."/>
</xsl:for-each>

That wouldn't really work either. That would just assure you didn't
spit out a duplicate of the preceding record.

Someone will surely followup with more comprehensive suggestions about
xsl:key. I'd research it until then. =)