If you are asking of a good target to point with this issue, I would recomend the following:

• * New Bookmark parsers (like Galeon's) use to load every address as a hash key. And memorize their branch possition at every tree. Then, when building the final & unique tree, they would place every address on the most popular branch and they would place links to that address at the other branches where the same address was previously collected.
• * You should see XML files just as other text files. With the advantage that you are able to find similar tags in different files. And take advantage of this.

You might want to share your final masterpiece. But consider it as a practice for future works with XML files.