Visual Assessment of Alleged Plagiarism Cases

Abstract

We developed a visual analysis tool to support the verification, assessment, and presentation of alleged cases of plagiarism. The analysis of a suspicious document typically results in a compilation of categorized "finding spots". The categorization reveals the way in which the suspicious text fragment was created from the source, e.g. by obfuscation, translation, or by shake and paste. We provide a three-level approach for exploring the finding spots in context. The overview shows the relationship of the entire suspicious document to the set of source documents. A glyph-based view reveals the structural and textual differences and similarities of a set of finding spots and their corresponding source text fragments. For further analysis and editing of the finding spot's assessment, the actual text fragments can be embedded side-by-side in the diffline view. The different views are tied together by versatile navigation and selection operations. Our expert reviewers confirm that our tool provides a significant improvement over existing static visualizations for assessing plagiarism cases.

Visual Text Analysis in Digital Humanities

“…Sometimes, even the visualization entirely depends on manually collected data through crowdsourcing (e.g. [WMN*14, RPSF15, HFM16, JFS16]).…”

“…[JRS*09, Mur11]) and aligned text segments (e.g. [ZNMS15, RPSF15]) in parallel texts, or when exploring a text of interest , for example, to highlight the automated or manual classification of words or phrases (e.g. [KJW*14, WMN*14]), or to visualize various sound patterns in poems (e.g.…”

“…Heat maps are also applied to visualize similarities or differences among text editions [JG15, PMMR15], or to highlight re‐used passages between the texts of a corpus [JGBS14, RARC*15, ZNMS15]. For the analysis of potentially plagiarized texts, socalled Difflines reveal structural differences between several suspicious text fragments and their alleged originals in a Focus+Context view [RPSF15], an example is shown in Figure . A further heat map variant are fingerprinting techniques as introduced in [KO07] in order to visualize characteristic textual features of literary works.…”

“…Heat map highlighting potential plagiarized text passages in a PhD thesis (figure reproduced with permission from Riemann et al[RPSF15]). …”

TransVis: Integrated Distant and Close Reading of Othello Translations

“…Plagiarism detection is an application of visual text comparison as in White and Joy [87]. Also, Riehmann et al [65] combine an overview slope graph and glyph-based detailed representations to explore given text against multiple sources.…”