running on a method level

[SehamAAlharbi]

Hi,

I am using JPlag API and was wondering if I could let it run on a method level rather than a file level. how to do we adjust the JPlagOptions to make that?

thanks

[SehamAAlharbi]

Hi again,

just another related question, is there a way for ignoring comments and white spaces? so they do not interfere with the similarity percentage? or this is a cleaning pre-process that should be done manually to get better results?

thanks in advance.

[tsaglam]

• Comments and whitespaces should not interfere with the similarity calculations, as this calculation is done on a token level. Whitespace should not affect plagiarism detection at all.
• To improve the results, you may adjust the minimum token match parameter, use basecode if you give your students a prescribed framework, and adjust the clustering if desired. There is no need for pre-processing, except maybe ensuring file format (should ideally be UTF-8).

Hope that helps!

[SehamAAlharbi]

Thanks, does this mean that JPlag will not identify commented-out code as a clone of active code?

[tsaglam]

No, comments are ignored; they are not extracted as tokens. See, for example, the token types of the Java language module. Note that the token types vary between language modules.

[SehamAAlharbi]

Thank you, that's really helpful. How about setting the threshold value? is there a recommended range of values?

[tsaglam]

If you mean -m, then I would say ignore it for now. The first parameter you should start to vary if your results are inconclusive is the min token match -t. Increase it if you want less sensitivity, and decrease it if you want more sensitivity. Start from the default values, e.g., 9 for Java.

[tsaglam]

Regarding your first question:

I am using JPlag API and was wondering if I could let it run on a method level rather than a file level.

What do you mean by method level? If you give your students a class with pre-defined empty methods, then you can use the basecode functionality (-bc). If you mean something different, please let me know.