Craig2024MMFMGC

Title

The Multimedia FAIR Metrics Grand Challenge
View PDF | Save PDF

Authors

Adam Craig, Carl Taswell

Affiliations

Brain Health Alliance Virtual Institute, Ladera Ranch, CA 92694 USA

Abstract

Brain Health Alliance (BHA), a US 501c3 nonprofit organization, will host the first annual Multimedia FAIR Metrics Grand Challenge on 9th September 2024. The FAIR Metrics, with acronym FAIR for the phrases Fair Attribution to Indexed Reports and Fair Acknowledgment of Information Records, quantify how well a scholarly work cites and discusses prior literature and the extent to which it remains devoid of plagiarism or misrepresentation of previously published work. Unlike lexical plagiarism detection, FAIR Metrics semantic analyses require identifying statements with equivalent meaning. Recording the comparison of documents in searchable records of FAIR Metrics analyses strengthens the integrity of scholarly publishing by providing a more transparent and systematic way to trace the origins of concepts, ideas, and creative contributions to the historical record of published literature. FAIR Metrics analysis by a human expert remains subject to debate because semantic similarity of concepts and ideas can be more difficult to quantify than lexical edit distance. This grand challenge will solicit automated tools to perform one or all stages of FAIR Metrics analysis with a focus on their use for plagiarism detection. Because FAIR Metrics analysis currently depends on human judgment with various opinions about which statements in a document are substantive claims and which are equivalent in meaning, we cannot currently declare a set of FAIR Metric values to be uniquely correct for a given document. Instead, we must establish expert consensus and will evaluate automated tools based on uniform formatting of the plagiarism analysis records and the ability to differentiate plagiarizing from non-plagiarizing documents. For the competition, there will be a total of four separate sets of published reports used in the data repository of documents for development versus evaluation purposes in the competition on plagiarizing and retracted versus non-plagiarizing and non-retracted. In this year's grand challenge, we will focus on the core media types found in most scholarly articles: prose text, figures (images), and tables. However, a growing number of digitally published works include supplementary multimedia content, such as video, audio, source code, and numerical data in a wide variety of formats and data repositories. Future iterations of this grand challenge will focus on extracting claims from different types of media and placing them in a shared semantic format that allows contrast and comparison.

Keywords

Research integrity, citational justice, publishing ethics, plagiarism detection, FAIR Metrics, multimedia data, artificial intelligence.

Citation

Brainiacs Journal 2024 Volume 5 Issue 1 Edoc G7ECAEAD9
DOI: 10.48085/G7ECAEAD9
PDP: /Nexus/Brainiacs/Craig2024MMFMGC
URL: BrainiacsJournal.org/arc/pub/Craig2024MMFMGC

Dates

Created 2024-02-03, received 2024-03-03, updated 2024-06-30, published 2024-06-30.