Discussion
This section shows the results generated by the plagiarism detection module of
the system. The performance of our new algorithm is evaluated and compared
with an existing plagiarism system, JPlag.
A set of programs given in (Malpohl 2002), which had been used to evaluate
the performance of JPlag before, was used. These programs are related to some
graphical user interface (AWT) programs. The results of comparison are
summarized in Table 1.
In general, the percentages of similarity generated by JPlag and our system
are very close to each other for very ësimilarí programs (set 1 and set 2). For
other sets of programs, the two systems have larger differences because our
system uses a different detection method to check for plagiarism, as it also
checks the meaning between the two programs. As those programs are AWT
programs, the weights of the components were adjusted and compared again.
The results after adjustment of weights are also shown in Table 1. Since the
weights are adjusted according to the nature of the programs, the searching and
checking criteria are changed accordingly. The results of our system after
adjustment of weights will differ from the original one.