An IR-Based Approach Utilizing Query Expansion for Plagiarism Detection in MEDLINE

Area of Science:

• Biomedical Informatics
• Information Science
• Text Mining

Background:

• Plagiarism detection is crucial in research, especially for identifying modified texts.
• Existing methods struggle with obfuscated original sources in large databases like MEDLINE.

Purpose of the Study:

• To develop and evaluate an Information Retrieval (IR)-based method for plagiarism detection.
• To identify potential sources of plagiarism within MEDLINE, even when text is altered.
• To improve the accuracy of plagiarism detection in biomedical literature.

Main Methods:

• Utilized a scalable Information Retrieval approach for candidate document selection from MEDLINE.
• Employed ULMS Metathesaurus for query expansion to handle obfuscated original documents.
• Investigated Word Sense Disambiguation techniques to manage multiple Concept Unique Identifiers (CUIs).

Main Results:

• The proposed IR-based approach demonstrated superior performance compared to a Kullback-Leibler Distance baseline.
• Effectively identified plagiarized sources despite word or phrase replacements in the suspicious text.
• Successfully addressed challenges posed by obfuscated documents and ambiguous terms.

Conclusions:

• The IR-based method offers a robust solution for detecting modified plagiarism in MEDLINE.
• This approach enhances the reliability of identifying original sources in biomedical research.
• Further development can improve automated plagiarism detection in scientific literature.