Java XML search engine for most similar XML files.

I looking for solution to search for the most similar XML (mean position x,y,w,h, and value) file from the database of XML files. So I will need to: 1) register in database multiple (not less than 500) XML files as templates and create some kind of descriptors of <text> key, keys example: <text x="304" y="86.95" width="46.609" height="14.861"> August</text> <text x="72.6" y="121.5" width="80.741" height="10.747">Google Ireland Ltd</text> <text x="72.6" y="132.75" width="63.706" height="10.726">Gordon House</text> <text x="72.85" y="144.05" width="30.887" height="10.758">Barrow</text> ** positions of the areas are more important, and values less - I would like to be able to put a penalty for locations and different value 2) template can by create by 1-10 XML files 3) use ultra fast algorithm (search not over 5 sec for 500 XML files) to find most similar XML file (sorting by similarity) -with XML files in database 4) ability to add/remove template form database Finally I need to know a list of N most similar XML files. I will provide set of XML files for testing purpose.