Prevention better than cure? The case for Plagiarism Detection

1
Prevention better than cure? The case for
Plagiarism Detection

• H.C. Davis
• IAM Learning Technologies
• Electronics and Computer ScienceUniversity of
  Southampton

2
Structure of this Talk

• (Largely concerned with Source Code Plagiarism)
• What is plagiarism?
• The causes of plagiarism?
• Approaches to preventing plagiarism
• Detecting Source Code Plagiarism
• Case Study using JPlag

3
What is Plagiarism

• Obviously direct wholesale copying of sourcebut
• How much source should be the same in well
  written solutions to a small problem?
• What about buddying and team working?
• How are we going to catch sub-contractors?

4
Some Thoughts

• It is very difficult to prove copying
• It is even more difficult to prove
  sub-contracting
• This talk assumes that our intention is to
  prevent plagiarism rather than to get involved in
  post-hoc legislatation

5
The causes of plagiarism?

• Over assessment
• Unrealistic expectations
• Strategic/cynical time management
• Cultural expectation
• Team working
• Fear of failure
• Laziness

6
Approaches to preventing plagiarism (1)

• Clear university/departmental policy
• Needs to define
• What is plagiarism?
• What will be done?
• The borderline between acceptable collaboration
  and unacceptable plagiarism

7
Approaches to preventing plagiarism (2)

• Signed statements from students
• Give them the opportunity to admit to what they
  needed to borrow in order to complete the
  assignment

8
(No Transcript)
9
Approaches to preventing plagiarism (3)

• Logbooks
• and
• Time-stamped code development
• It is hard to reverse engineer this stuff
  (especially without learning!)

10
Approaches to preventing plagiarism (4)

• Reduce assessment load
• Diversify type of assessments
• (peer marking and review, multiple choice
  questions, self assessed team assignments)
• Establishing a culture of learning (rather than
  mark grubbing)

11
Approaches to preventing plagiarism (5)

• Detection
• Naming and shaming

12
Detecting Source Code Plagiarism

• Must have all code on-line (floppies or hard copy
  not helpful)
• Simple searching for suspicious strings
• Scripts as above
• Simple byte code comparisons
• Specialised tools (JPLAG, MOSS, Coursemaster)

13
Case Study Using JPLAG

• Getting Started
• First get yourself registered with JPlag at
  http//www.jplag.de/
• (email jplag_at_ira.uka.de)
• Download the JPlag applet

14
2. Collect Student Assignments

• JPLAG expects one directory per student. That
  coursework can have subdirectories and JPLAG
  doesnt care what the files are called.
• At Soton we have a handin machine which supplies
  a ZIP file of exactly such a directory file with
  the students username as the directory name, so
  all you have to do is unpack it where you want
  it.

15
(No Transcript)
16
Run the Applet

• gtc\jdk1.3\bin\java jar C\jdk1.3\jplagapplet.jar
  

After a few minutes you will be able to see the
results
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)