|
|
Rank: Member
Joined: 9/9/2009
Posts: 15
Location: Lebanon
|
Can someone please tell us what algorithm is being used for the comparison of XML files in DiffDOG? Is it a published work?
thanks
|
|
Rank: Advanced Member
Joined: 12/13/2005
Posts: 2,856
Location: Mauritius
|
Implementation is Altova internal and cannot be shared with public
|
|
Rank: Member
Joined: 9/9/2009
Posts: 15
Location: Lebanon
|
ok thanks for the reply
|
|
Rank: Member
Joined: 9/9/2009
Posts: 15
Location: Lebanon
|
k i have been testing diffdog lately and I am not very impressed.
Tree Edit Distance is NP-Hard, so at first I was under the impression that DiffDog was doing Constrained Tree Edit Distance (which belongs to P). So I was impressed by the performance since the best algorithm for the problem takes more than cubic time.
Now I see that DiffDog is doing just another heuristic :( and that explains the relatively fast performance.
Feed diffdog the following two files:
<a>
<b>
<c>aa<\c>
<\b>
<\a>
and
<b>
<c>aa<\c>
<\b>
It can't detect that the second file was obtained from the first by deleting the root element...
|
|
Rank: Member
Joined: 9/9/2009
Posts: 15
Location: Lebanon
|
btw are you using X-Diff algorithm or a modified version of it ? ^^
|
|
Rank: Advanced Member
Joined: 12/13/2005
Posts: 2,856
Location: Mauritius
|
Algorithms are trade secret of Altova and cannot be disclosed.
By the way, DiffDog is able to process your sample comparison - you only need to set comparison option "Ignore Node Depth"
|
|
|
guest |
