Does anyone know of a program/tool that works for comparing CAD files (DWG files) against each other (compare all DWG files in the directory A with all DWG files found in Directory B) ?
The tool should provide an index of similarity (high score => drawings very similar)
The goal is to "trap" drawings that are substantially -almost completely- similar; namely with only small marginal differences, in order to analyze (manually ..) *only* files with high index (score).
Thanks in advance
Unfortunately, I don't have a good suggestion for you having recently had just such a case. The problem lies in the fact that DWG format files have multiple layers, such that deep hashing is practically useless for identifying similarities between files. If you want to delve into this, more, you can join the Open Design Alliance and get the source code for OpenDWG (now called Teigha), which will give you some understanding of the structure of DWG files.
I would recommend two things. First, look at the file metadata (you can do this with a hex editor). The metadata includes, among other things, information on the creator of the file. You may be able to find sufficient information to allow you to conclude that File B is a derivative of File A.
Second, have someone (an engineer familiar with AutoCAD) , review the drawings with you. This is a bit like detecting art forgery; there is less for the forensic investigator to do and more for the domain expert. The fact that these are digital files is almost irrelevant.
Good luck. In my case, manual review after initial screening was the only real solution.