Plagiarism Screening & Prevention Policy

Plagiarism Screening &
Prevention Policy

Transparent similarity screening, responsible attribution, fair investigation and protection of original scholarship in Computer Science and Artificial Intelligence.

Purpose Definitions Prior Work 20% Reference Similarity Review Code & Data Generative AI Editorial Screening Author Response After Publication Independence Appeals Contact & Standards

Purpose, Scope & Principles

ARCITECH: Journal of Computer Science and Artificial Intelligence is committed to original scholarship, appropriate attribution, fair editorial assessment and the reliability of the scholarly record. This policy governs plagiarism prevention and similarity screening for submitted manuscripts, published articles, supplementary materials, code, datasets, images, figures and other research outputs.

Authors retain responsibility for the provenance and originality of submitted material. Editors must use evidence and expert judgment, not software scores alone, when evaluating potential overlap. The policy applies regardless of an author's seniority, affiliation, reputation or Article Processing Charge (APC) status.

Definitions: Similarity, Plagiarism & Attribution

Similarity is a measurable correspondence between submitted text and sources available to a comparison tool. Plagiarism involves presenting another person's words, ideas, methods, results, images, code or other scholarly contributions as one's own without adequate acknowledgment or authorization where required.

• Verbatim plagiarism: copying substantial wording without appropriate quotation and attribution.
• Paraphrase or translation plagiarism: rewording or translating another source while obscuring its origin.
• Conceptual or methodological misappropriation: using distinctive ideas, designs or research results without appropriate credit.
• Source concealment: inaccurate or fabricated references, misrepresented provenance, or citations that do not support the borrowed material.

Not every textual match constitutes plagiarism. Standard terminology, properly identified quotations, bibliography entries and conventional methods text must be evaluated in context.

Responsibilities of Authors

All authors must submit original work, properly acknowledge intellectual contributions, disclose closely related outputs, and verify every citation, DOI, quotation, data source and reused research artifact.

• Use quotation, paraphrasing and citations appropriately; do not conceal close copying through superficial word changes.
• Clearly credit third-party diagrams, images, tables, datasets, code repositories, libraries and model weights.
• Keep source material, permissions, earlier drafts and supporting records available for reasonable editorial clarification.
• Disclose preprints, thesis-derived work, conference versions and overlapping publications at submission.
• Ensure coauthors have reviewed the final work and agree with the originality and disclosure statements.

Originality, Preprints & Previous Dissemination

A publicly disclosed preprint, institutional thesis, dissertation chapter, technical report or conference contribution is not automatically disqualifying. Authors must identify the earlier version, link to it where possible, describe substantive additions, and comply with third-party copyright or reuse conditions.

Substantially duplicating previously published findings as a new contribution without appropriate disclosure is unacceptable. Fragmenting one study into misleadingly repetitive articles, recycling substantial material without acknowledgment, and concurrent submission of the same manuscript are prohibited.

When conference papers or preprints overlap with a submission, editors assess novelty, transparency and publishing rights case by case; they do not decide solely on the resulting similarity percentage.

Turnitin Screening and the 20% Reference

ARCITECH publicly identifies Turnitin as its text-similarity screening service and has published a 20% similarity reference. The journal uses this reference to prioritize manual assessment of potentially problematic textual overlap. It is an editorial screening benchmark, not a definition of plagiarism or an automatic acceptance/rejection rule.

• A result above 20% requires closer editorial evaluation and may lead to a request for clarification, correction or substantive revision, depending on source and context.
• A result at or below 20% does not establish originality or exempt a manuscript from further scrutiny.
• Significant unattributed copying can be problematic even when the overall score is low; legitimately cited content can contribute to a higher score.
• Editors may review appropriate exclusions such as references, quotations or trivial matches in line with the screening service's configuration; exclusions must not be used to conceal unacceptable copying.

How Similarity Reports Are Interpreted

A qualified editor examines the matched passages and original sources, the amount and concentration of reuse, placement within the paper, citation accuracy, author contributions, originality of research findings, and permitted or customary reuse.

• Evaluate large contiguous matches differently from numerous short, generic technical phrases.
• Review reused descriptions of standard methods without overlooking inappropriate duplication of results or interpretation.
• Consider properly cited quotations, established definitions and bibliographies separately from unattributed copying.
• Review matches with author preprints, institutional repositories or earlier drafts to avoid mistaking a legitimate self-match for another person's work.
• Document material concerns and the basis for an editorial conclusion.

Turnitin similarity reports are supporting evidence. They do not independently determine intent, research originality, or legal rights.

Self-Plagiarism, Redundant & Overlapping Publication

Authors must transparently disclose substantial reuse of their own previously disseminated text, tables, images, methods, datasets or results. Proper disclosure, attribution and permission where necessary are essential even when all works share authors.

Permissible methodological repetition is distinguished from misleading redundant publication. The journal may request comparison copies and an explicit statement describing what is new. Reusing previously published findings as if they are independent results, concealing overlap, or disguising duplicate publications is unacceptable.

Software, Code, Datasets & Figures

For Computer Science, Software Engineering, Artificial Intelligence and Data Science research, originality extends beyond prose to algorithms, code, pseudocode, benchmark results, datasets, model architectures, training assets, diagrams and figures.

• Cite the original paper, documentation, repository, dataset, license and version when reusing third-party research artifacts.
• Clearly distinguish independently developed algorithms from adaptations or implementations of prior work.
• Comply with software and dataset licenses; source-code licensing and article licensing are not automatically identical.
• Do not present reused screenshots, graphs, experimental outputs or synthetic samples as original observations without appropriate disclosure.
• Describe permitted public benchmarks and baselines accurately, including provenance and limitations.

Similarity screening of text alone may fail to identify code, image, dataset or idea misappropriation; concerns about these materials require domain-specific inspection.

Generative AI, Translation & Paraphrasing Tools

Using generative AI or automated rewriting tools to obscure copied material does not remove the obligation to credit its original source. Authors remain responsible for all text, references, code, images and findings submitted under their names.

Material generative AI assistance must be declared under ARCITECH's Artificial Intelligence (AI) & Generative AI Policy. Ordinary grammar correction is not by itself plagiarism; generated or translated scholarly content must nevertheless be checked for accuracy, attribution and potential intellectual-property conflicts.

AI-text detectors, like similarity tools, can make errors and must not alone serve as grounds for allegations or sanctions.

Editorial Screening & Peer Review

The editorial office may screen a manuscript during desk assessment, after substantial revision or when credible concerns arise. A qualified editor considers significant similarity matches before deciding whether the manuscript proceeds to independent double-blind review.

Reviewers may identify unattributed reuse, questionable citations or duplicated research material and report evidence privately to the handling editor. Reviewers must not contact authors or alleged source owners directly, nor upload confidential manuscripts to unauthorized comparison or AI services.

Reviewers offer scientific assessments and recommendations; a conflict-free editor remains responsible for editorial decisions, following the Peer Review Policy & Editorial Decision.

Pre-Publication Handling & Author Response

Where the editor identifies credible overlap concerns, the journal may request an explanation, a corrected citation, a revised passage, a comparison with a related work, or supporting permissions. Authors should receive sufficiently specific information to understand and respond to a material concern.

• No material concern: continue the normal editorial assessment.
• Correctable attribution or wording problems: request a proportionate revision and reassess.
• Serious, documented unattributed appropriation or concealed redundant publication: consider rejection after fair review and author response; refer a serious case to an appropriate institution when warranted.

The editorial record should distinguish suspected overlap, verified attribution issues and any determination of misconduct. A request for revision is not itself a finding of dishonesty.

Post-Publication Concerns

ARCITECH reviews credible reports of plagiarism or redundant publication even after an article has been published. Readers, authors, rights holders and institutions may submit a documented concern to the editorial office.

Editors assess source evidence, seek the authors' response and determine an appropriate editorial action. Depending on the significance and effect on the record, the outcome may be no action, clarification, correction, expression of concern, retraction, or institutional referral. A retraction is not automatic for every citation error or isolated overlap.

Public notices, when required, follow ARCITECH's Corrections, Retractions & Expressions of Concern Policy and remain linked to the original scholarly record.

Fairness, Evidence & Editorial Independence

A manuscript or article must not be judged solely by a similarity percentage, a competitor's accusation, the author's status or the presence of a critical review. The journal evaluates documentary evidence, relevant circumstances and the author's explanation proportionately.

Editors and reviewers must declare conflicts of interest. When a case concerns an editor's own manuscript, collaborator, institution or personal interest, the individual must be recused and a suitably independent editor assigned to handle the matter. Where institutional fact-finding is necessary, ARCITECH may refer the allegation to the relevant research institution.

Confidentiality & Screening Data Protection

Unpublished submissions, source comparisons, reviewer identities, similarity reports and investigation records are confidential editorial materials. Access should be limited to people with a legitimate role in reviewing the matter.

Editors must use authorized comparison services and consider applicable data-retention, copyright, security and privacy requirements. Sensitive source code, credentials, personal data and restricted datasets must not be transmitted to unauthorized external tools. Record-sharing with relevant institutions should be proportionate to the investigation and lawful.

Appeals, Corrections & Complaints

Authors may request reconsideration when they can identify a material factual error, overlooked attribution, procedural problem or other substantive evidence relating to an editorial decision. Disagreement with an evidence-based decision alone does not guarantee re-review.

Appeals and complaints follow the journal's Complaints & Appeals Policy. The appeal should be handled by a person without a conflict of interest, where practicable. Post-publication changes follow the published corrections and retractions policy.

Preventive Guidance for Authors

Before submitting, authors should verify that every direct quotation is properly identified, paraphrases accurately credit the source, earlier related publications are disclosed, figures/data/code are licensed and attributed, and references are valid and relevant.

• Use the official ARCITECH manuscript template and citation style.
• Clearly describe the original contribution relative to related work and preprints.
• Do not rely on a low similarity score as proof of originality.
• Check referenced DOIs, quotations, dataset and source-code attributions.
• Retain material necessary to substantiate originality and permissions if clarification is required.

Consult the Author Guidelines and Publication Ethics & Malpractice Statement for related obligations.

Recordkeeping, Periodic Review & Improvement

The editorial office should maintain proportionate records of screening, substantiated concerns, requested revisions, author explanations, decisions, conflicts of interest, and any post-publication action, with access controls appropriate for confidential editorial material.

ARCITECH will periodically review this policy and its screening practice in light of new scholarly publishing guidance, evolving research technologies and practical editorial experience. Any material change to the 20% screening reference must be reflected consistently in About the Journal and applicable author instructions.

Reporting, Related Policies & Standards

Authors, reviewers, readers or other stakeholders with credible plagiarism concerns may contact ARCITECH Editorial Office at arcitech.journal@iaincurup.ac.id. Identify the manuscript or published article and provide source links, passages, repository identifiers or other verifiable evidence where available. Allegations should be submitted in good faith and without unnecessary disclosure of confidential information.

Related ARCITECH policies: Publication Ethics · AI Policy · Peer Review · Corrections & Retractions · Complaints & Appeals · Author Guidelines.

Guiding sources: DOAJ Guide to Applying; COPE Core Practices; COPE: Suspected Plagiarism in a Published Article; Turnitin: Understanding the Similarity Score.

This policy describes ARCITECH's editorial expectations. Application follows the journal's formally approved procedures and applicable legal and institutional requirements.

ARCITECH: Journal of Computer Science and Artificial Intelligence  •  Ethical scholarship, transparent assessment and reliable research communication.