Plagiarism Detection Software: How to Read Your Reports Well!

1
Plagiarism Detection Software How to Read Your
Reports Well!

• Dr. Gene Kleppinger
• Eastern Kentucky University
• Friday 1045-1145, Auditorium
• Cross-Conferenced to BbSEUG, Savannah
• via Elluminate! Live

2
Plagiarism Detection

• SafeAssignments and Turnitin reveal how students
  texts match source materials.
• Text matches can be honest citations, innocuous
  coincidences or outright plagiarism
• Wise interpretation of the programs reports is
  crucial.

3
About SafeAssignment

• Provides results similar to Turnitin
• Implemented at EKU in 2004 because they had the
  first Bb Building Block, and we continue using it
  because it works.
• Data from MyDropBox, the parent company
• Client institutions 368 (including 74
  installations of SA for Blackboard and over 100
  institutions using our services through our
  partner publishers) and individual users from
  over 500 institutions
• Papers processed peaks up to 6,000/day now,
  projected peaks of over 10,000/day by the end of
  this year

4
The SafeAssignment database . . .

• Almost 8 billion Internet documents
• All major publicly accessible digital databases,
  such as PubMed, Project Guttenberg and other
• The FindArticles database by LookSmart (over
  5.5 million articles from over 900 periodical
  publications)
• The ProQuest ABI/Inform databases (over 1100
  publication titles, about 2.7 million articles)
• MyDropBox proprietary database of over 300,000
  documents that are known to be offered for sale
  by paper-mill Web sites
• Proprietary institutional archives and databases
  provided by clients
• ZIP archives and password-protected areas
  available on the Internet indexed on demand.

5
Accessing SafeAssignment . . .

• Through Blackboard integrated to work like Bbs
  Assignment file-exchange feature, available in
  every course (no limits).
• Students submit files through a View/Complete
  link, placed in any content area
• Instructors create links just as for Assignments,
  and obtain reports in the Control Panel and
  Gradebook
• Instructors also can submit any file at any time
• Through partnerships with textbook publishers
  Web sites (individual and institutional
  accounts), primarily in the writing area.
• Pearson Education, McGraw-Hill, Houghton Mifflin,
  Bedford/St. Martins, Stoas, etc.
• Not necessarily connected directly with Bb

6
Open the list of Safe Assignments for this class
7
Open the SA report window for Essay 2
8
(No Transcript)
9
Text-matching score
Link to students file
Link to SA report
The Gradebook holds links to the same
information, listed by student.
10
(No Transcript)
11
Basic information
Sites with matched text
Students text, with color highlights for
passages detected as matching, keyed to the
list of sites.
12
(No Transcript)
13
Email this report anywhere
14
(No Transcript)
15
Click to see the contextual matchingtexts
16
(No Transcript)
17
Evidence shows that the students text matches
another students paper, at the level of quoting
similarly from the textbook, with citation.
(GOOD WORK!)
18
(No Transcript)
19
43 figure is attention-grabbing, but you
mustinvestigate further
20
(No Transcript)
21
The evidence shows that this studenthas copied
verbatim from a Webdocument without
citation,representing this work as her own.
22
Matching texts can be

• honest citations
• Praise good documentation!
• innocuous coincidences
• Seeing multicolored sentences here and there is
  far less alarming than single-color paragraphs,
  or the same color again and again.
• Always give clear directions about quoting from
  the textbook or your questions.
• outright plagiarism
• Punishment must be heavy (not -5 points).

23
Weighing the evidence . . .

• Text-matching scores of 0-25 usually represent
  low probabilities of plagiarism. 21 is the
  average SA score.
• Scores over 40 almost always indicate
  significant plagiarism or too much quoted content.

• Scores of 95-100 indicate duplicate submissions,
  which may or may not represent plagiarism.
• A source matched by the software may not be the
  actual source used by the student.

24
Uncovering Unusual (?) Plagiarism

• CD liner notes students expositions of
  musical performances, etc., contained passages
  copied from blurbs on music store Web sites.
• Quoting the citation students texts appear
  to contain appropriate citations for quotations,
  but the SA report reveals that the citation
  itself may have been plagiarized.Compare these
  stylized SA report fragmentsHe says that blah
  blah (Miller, 1954a, 63).He says that blah
  blah (Miller, 1954a, 63).

25
Is SA Foolproof? No.

• Foolish instructors might misunderstand the
  text-matching score to be a direct plagiarism
  indicator, or they might overlook other clues,
  e.g., citing sources that arent available, or
  sudden changes in writing style

• Foolish students might submit irrelevant texts,
  or attempt to trick the text-matching algorithm
  (such as by deliberately misspelling words or
  packing white spaces)

26
Attempting to trick the text-matcher

• I took two sentences from a famous speech and fed
  them to SafeAssignment. It reported (only) 97
  matching, apparently because the source it
  located first was actually an abridged version.
• I doctored the text with some misspellings and
  some wording changes . . . and submitted that.

27
My doctored text (earning 0 from SA)

• I have a dream that one day soon this
  nation will raise up and live out the true
  meaning of this creed "We hold these truths to
  be self-evident that all people are created
  equally." I have a dreem that some day on the
  hills of Georgia and Alabama the sons of former
  slaves and the daughters of former slaveholders
  will be able to sit down at a table of brotherly
  love.

28
Special SA design features available to
instructors on setup

• Draft option each students file is scanned
  and the results are preserved, but the scanned
  text is not added to the database. So, when the
  final version is submitted (through an additional
  SA link), it will be scanned like new.
• Resubmittable option each students file is
  scanned and the results are available as usual,
  but a student can elect to submit a new file, and
  the previously submitted text is deleted from the
  database.

29
Strategies for Course Design

• Prepare a plagiarism policy statement that
  explains plagiarism detection.
• Design essay assignments to emphasize
  originality. Avoid asking for
• standard sets of facts,
• canned interpretations, or
• traditional position statements.
• Monitor student-chosen research topics
  closely!

30
Strategies for Course Design (cont.)

1. Design assignments to require personal
   reflection/evaluation, or connections to current
   events. Change assignments every time you teach
   the course.
2. Wherever possible, arrange semester-long projects
   in sequential stages. SAs draft option allows
   the stages to be treated separately from the
   final product.
3. Use SAs Quick Submit method to get files from
   earlier semesters scanned.

31
SA limitations/challenges

• Current file formats accepted Microsoft Word,
  RTF, text, PDF, HTML. Cannot process native
  formats from WordPerfect, Microsoft Works, etc.
• Current file types analyzed word processor
  documents. Cannot process spreadsheets,
  presentations, databases, or tables and charts
  loaded in other documents.

32
Discussion-and-Demonstration Time

• EKU Blackboard
• Plagiarism Policy
• SafeAssignment
• gene.kleppinger_at_eku.edu