DL Project Suggestion: Similar ETDs / Plagiarism Detection

• Number of people: 2-4
• Goal: To develop a solution for detecting plagiarism among ETDs, or for helping researchers find similar work. One approach was described in the Proceedings of ACM DL'2000. Another, older effort led to the SCAM package from Stanford. This will allow students to look for closely related work, and for university staff to make sure that a new ETD is not a (partial) copy of one already in NDLTD.
• Contact information: The instructor or PhD student Paul Mather (paul@csgrad.cs.vt.edu).
• Required background: familiarity with library, digital library and/or information retrieval system operation
• Description: This project involves work on the MatchDetectReveal system and/or extending / enhancing / customising SCAM from Stanford for application to ETDs. See pp. 226-227 in Proc. ACM DL'2000: Document Overlap Detection System for Distributed Digital Libraries - copy available from instructor. See relevant Stanford PostScript papers in particular
  • dSCAM: Finding Document Copies across Multiple Databases
  • The SCAM Approach To Copy Detection in Digital Libraries
  • SCAM: A Copy Detection Mechanism for Digital Documents
  • Building a Scalable and Accurate Copy Detection Mechanism
  See also the project on ResearchIndex, which will support plagiarism detection.
• Students involved: The following students are working on this project: