Examining and detecting academic misconduct in written documents using revision save identifier numbers in MS Word as exemplified by multiple scenarios

Abstract

Deliberate academic misconduct by students often relies on the use of segments of externally authored text, generated either by commercial contract authoring services or by generative Artificial intelligence language models. While revision save identifier (rsid) numbers in Microsoft Word files are associated with edit and save actions of a document, MS Word does not adhere to the ECMA specifications for the Office Open XML. Existing literature shows that digital forensics using rsid requires access to multiple document versions or the user's machine. In cases of academic misconduct allegations usually only the submitted files are available for digital forensic examination, coupled with assertions by the alleged perpetrators about the document generation and editing process This paper represents a detailed exploratory study that provides educators and digital forensic scientists with tools to examine a single document for the veracity of various commonly asserted scenarios of document generation and editing. It is based on a series of experiments that ascertained whether and how common edit and document generation actions such as copy, paste, insertion of blocks of texts from other documents, leave tell-tale traces in the rsid encoding that is embedded in all MS Word documents. While digital forensics can illuminate document generation processes, the actions that led to these may have innocuous explanations. In consequence, this paper also provides academic misconduct investigators with a set of prompts to guide the interview with alleged perpetrators to glean the information required for cross-correlation with observations based on the rsid data.