Anti-Plagiarism utility

[Allan Brown]

I had a request from a group of instructors to suggest a way to test 2 Mastercam files for evidence of file sharing.

 

Ideally, a directory of files could be checked for identical database elements, or creation time.

 

For those with a finger on the pulse of the database structure, perhaps you could tell me if anything like that would even be possible.

[bryan314]

Dave,

 

I don't see why not. It should be easy enough to go through a directory of files and pull the entity number and creation times and look for and count the matches.

 

If the student is half way smart though (but if they're copying they can't be too smart) they could reset the numbers and time by importing/exporting the geometry or merging it into a file with existing geometry.

 

It should also be possible to compare entities looking for full or partial matches based on properties such as color, level, radius, length, etc and spit out a percentage of overlap. A certain percentage would be expected if the students are working on the same part but a higher number might indicate something else.

 

Bryan

[bryan314]

I had another thought if they are dealing with straight geometry convert it to an ascii format like iges and they can run it through normal plagiarism software.

 

If they're dealing with geometry and toolpaths it should be easy enough to get a chook or vbscript written to convert the significant bits to text and do the same. level, color, time, entity number, radius, length, toolpath comment, feeds and speeds, etc.

 

Bryan

[Allan Brown]

The concern on comparing entity database numbers, colours, and levels is that in a cookbook geometry creation tutorial, we'd expect them to be the same.

 

Is entity creation date/time part of the database?

 

Yes, if something was written to dump relevant database info out into an ASCII file, we could simply file compare 2 ASCII files.

 

Thanks for your thoughts Bryan.

[Allan Brown]

Cross-posted on the Instructor Forum:

 

http://www.emastercam.com/cgi-bin/ultimate...ic;f=7;t=000321

[Pete Rimkus from CNC Software Inc.]

"...Is entity creation date/time part of the database?..."

 

yes - every entity has a modification time stored with it.

[Jon Griffin CNC Software]

Expanding on what Pete said, every piece of geometry has a unique id number associated with it too. Going through the database you could check for matches that not only had the same color, level etc., but also had identical id's and modification times. A high percentage of hits here, while not conclusive, would be indicitive.

[Zerren Gedeon]

This very crude (CHook) utility does the following....

You select 2 (MCX) files and it scans each one looking at the ‘geometric’ data in each file’s database. It checks the ‘unique id number’ that every piece of geometry has and the modification (creation) time stamp of each entity. It then reports its findings.

 

I said it was crude !

 

This reports that T.MCX contains '7' geometric entities and that T1.MCX has '8' and that '7' of the entities in each file "match".

 

It does not check for matches on the same color, level, etc. It certainly would be easy to do so, but I think that would be redundant after looking for identical id's and modification times.

 

Obviously this utility would have to be enhanced and expanded to really be useful.

It could compare one file against each file in a folder of MCXs.

And the reporting needs some work! (I see that I misspelled “comparison” and entity! )

Did I mention that this was a quick 'proof of concept' job ?

 

Questions...

1> Is this what you are looking for?

2> What file<=>file, or file<=>folder of files comparisons do you need?

3> What type (and format) report needs to be generated?

4> How detailed does the report need to be?

Just an overall "matched" count? A "matched" count for each type of entity? (e.g. a line match count, arc match count, etc)?

[Allan Brown]

Thanks Zerren.

 

To be honest, that's almost perfect. I could bounce that off the instructors right now for feedback.

 

I suppose ideally you'd have a directory select option, and you'd compare each file against every

other. For highly correlated files, the report could be an ASCII text file dump of the filename, each entity #, type, creation time/date stamp. Just as proof of the matching data to present to a student.

[Eeyore-nonhuman forum member]

I can overpass all your checks every time EASY !

i CAN PROVE IT PRIVATLY ANY TIME .

aND IF i CAN ,ANY YOUNGSTER WITH good brains can do

it too.

And if he has good brains ,what does it matters ?

Don`t waste your time !