Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM68652
Volume:
10 |
10 |
Oct 2026
Pages:
1-8
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Computer Engineering
Vishwabharati Academy's College of Engineering
Computer Engineering
Vishwabharati Academy's College of Engineering
Computer Engineering
Vishwabharati Academy's College of Engineering
Computer Engineering
Vishwabharati Academy's College of Engineering
Digital content, machine translation and generative AI tools make it easy to reuse an idea without reusing its wording, and this has made plagiarism harder to detect in academic and research settings. Detectors built on lexical matching were designed for verbatim copying and lose most of their signal once a text is translated, paraphrased or restructured. Knowledge graphs (KGs) and graph neural networks (GNNs) offer an alternative, because they encode relations between words and concepts instead of surface strings. This paper reviews that line of work and concentrates on the CL-UniDepWord-GNN framework of Bouaine et al. published in IEEE Access, which merges WordNet semantic relations with Universal Dependencies syntax in a single multilingual graph and learns over it with relation embeddings, a graph convolutional network (GCN) and a graph attention network (GAT). We summarise how the framework is built, which datasets it uses and what PlagDet scores it reports for four language pairs, and we read those results critically, including how the evaluation corpora were assembled and how the baselines were compared. We then collect the open problems, namely evaluation realism, dependence on linguistic resources, computational cost, explainability and AI-generated text, and outline a prototype architecture for future work.
KEYWORDS
The framework reviewed here shows that adding syntac-tic structure to WordNet-based graphs can improve cross-language plagiarism detection in three of four language pairs, with the language-dependent pattern in the ablation being its most informative result. The size of the reported gains over earlier systems should be treated with caution until they are confirmed on held-out corpora with like-for-like baselines. Semantic resources combined with graph learning remain a promising basis for next-generation detectors. Future systems should add reliable handling of AI-generated content, scalable deployment and explanation of their output, while preserving multilingual support.
[1] C. Bouaine, F. Benabbou, A. Bouaine, and C. Zaoui, “Multilingual Plagiarism Detection Using GNNs and Syntax-Semantic Knowledge Graphs,” IEEE Access, vol. 14, pp. 99027–99043, 2026. [2] O. Hourrane and E. H. Benlahmar, “Graph Transformer for Cross-Lingual Plagiarism Detection,” IAES Int. J. Artif. Intell. (IJ-AI), vol. 11, no. 3, pp. 905–915, Sep. 2022. [3] M. Potthast, B. Stein, A. Eiselt, A. Barro´n-Ceden˜o, and P. Rosso, “PAN Plagiarism Corpus 2011 (PAN-PC-11),” Zenodo, 2011. [4] P. Koehn, “Europarl: A Parallel Corpus for Statistical Machine Transla-tion,” in Proc. MT Summit X, Phuket, Thailand, 2005, pp. 79–86. [5] R. Steinberger et al., “The JRC-Acquis: A Multilingual Aligned Parallel Corpus with 20+ Languages,” in Proc. 5th Int. Conf. Lang. Resour. Eval. (LREC), 2006. [6] N. Reimers and I. Gurevych, “Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks,” in Proc. EMNLP-IJCNLP, 2019, pp. 3982–3992. [7] G. A. Miller, “WordNet: A Lexical Database for English,” Commun. ACM, vol. 38, no. 11, pp. 39–41, 1995. [8] Z. Sun, Z.-H. Deng, J.-Y. Nie, and J. Tang, “RotatE: Knowledge Graph Embedding by Relational Rotation in Complex Space,” in Proc. ICLR, 2019. [9] T. N. Kipf and M. Welling, “Semi-Supervised Classification with Graph Convolutional Networks,” in Proc. ICLR, 2017. [10] P. Velickovic et al., “Graph Attention Networks,” in Proc. ICLR, 2018.
Publication Date
10/31/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
4
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...