Managing Lexical-Semantic Hybrid Records of FAIR Metrics Analyses with the NPDS Cyberinfrastructure

Abstract

Current approaches to plagiarism detection often focus on finding lexical matches rather than semantic similarities in the text content that is compared. But the more important unanswered questions remain whether similar concepts expressed in related topical contexts are semantically equivalent as idea-laundering plagiarism by humans or algorithm-generated plagiarism by machines. Now publicly available and easily accessible, text-generating algorithms have automated the process of assembling a text derived from but not attributed to published content scraped from the web. The FAIR Metrics, with FAIR an acronym for Fair Attribution to Indexed Reports and Fair Acknowledgment of Information Records, measure how appropriately a document cites prior records based on whether they contain similar claims that are equivalent in meaning. We demonstrate herein a workflow with results for manual evaluation of the FAIR Metrics to quantify the extent of plagiarism in 8 articles retracted or reported for plagiarism. We also demonstrate use of the Nexus-PORTAL-DOORS-Scribe (NPDS) Cyberinfrastructure to manage semantic descriptions of the concept mappings and entity equivalence evaluations made using concepts and relationships from the PDP-DREAM Ontology.