Corrections, Retractions & Expressions of Concern Policy

ARCITECH: Journal of Computer Science and Artificial Intelligence is committed to maintaining an accurate, transparent, discoverable and reliable scholarly record. This policy establishes the journal's procedures for corrections, retractions, expressions of concern, clarifications and other post-publication updates affecting published articles, supplementary files and associated research records.

Errors and credible concerns may arise from honest mistakes, methodological weaknesses, editorial or production failures, or research and publication misconduct. ARCITECH evaluates each case proportionately on the basis of evidence, potential impact, procedural fairness and public interest. Maintaining the integrity of the literature—not punishing authors—is the purpose of these actions.

ARCITECH distinguishes the following actions according to the nature, reliability and significance of the reported concern:

These are editorial categories, not automatic penalties. The journal may consult qualified independent experts or institutions before determining the appropriate action.

A correction is appropriate when a published error affects interpretation, transparency, a meaningful element of the record, or the ability to reproduce findings, but the corrected article remains substantially reliable.

• Author-related errors (corrigenda): significant mistakes in a figure label, table, dataset description, software setting, equation, numerical value, affiliation, acknowledgment or disclosed competing interest.
• Journal/production errors (errata): incorrect typesetting, omitted content, misordered figures, incorrect article metadata or production-stage errors introduced by the journal.
• Minor typographical edits that do not affect scholarly meaning or bibliographic metadata may be corrected without a separate notice, provided the journal keeps appropriate version or production records.

A substantive correction notice should identify the exact article, describe the original error and corrected information, state whether any conclusions are affected, and link to the affected publication. Corrections must not be used to conceal unreliable findings that require retraction.

ARCITECH may publish a clarification when readers require an authoritative explanation of ambiguous reporting, or an addendum when additional information directly associated with the published work should become part of the record without misrepresenting it as newly obtained results of the original study.

Examples include a previously omitted funding disclosure, a documented change to a public code repository, supplemental experimental details or an explanation of a data-access restriction. An addendum is not a mechanism for disguising a fundamentally different study or retroactively replacing an invalid experiment.

Meaningful article updates must carry a date, explanation of the change and persistent relationship to the original work.

An Expression of Concern (EoC) may be issued when there is a credible and significant unresolved concern about the reliability or integrity of a published article, but available evidence does not yet support a definitive correction or retraction. It may also be considered when a sufficiently serious investigation is delayed or authors/institutions cannot provide conclusive information in a reasonable period.

• An EoC should identify the paper, describe the specific concern as accurately as possible and avoid presenting unproven allegations as established misconduct.
• The notice should indicate the general status of inquiry and be linked visibly to the article.
• An EoC is not a substitute for a justified retraction and is not a lesser form of punishment.
• When sufficient evidence emerges, the journal will publish an appropriate subsequent notice: correction, retraction, updated EoC, or notice of resolution, as warranted.

A retraction may be warranted when the editorial assessment establishes that the published work is unreliable or seriously compromises the scholarly record. Circumstances may include:

• Major errors in data collection, calculations, experimental design, statistical analysis or computer simulation that invalidate key conclusions.
• Fabricated or falsified data, manipulated scientific images, invented experiments, or serious misrepresentation of software/model performance.
• Substantial plagiarism, redundant publication without proper disclosure or attribution, or unlawful use of protected material.
• Compromised or manipulated peer review, fabricated reviewer accounts, paper-mill involvement, fraudulent authorship or other serious publication-process manipulation.
• Serious unapproved or unethical research practices, unauthorized acquisition or disclosure of sensitive records, or material undisclosed competing interests affecting reliability or interpretation.

A retraction may follow a good-faith error; misconduct is not required. Editorial judgment concerns the reliability and integrity of the published work, while institutions and relevant authorities may investigate individual responsibility.

Where a clearly separable element is fundamentally flawed, the journal may consider partial retraction only if the unaffected components can be reliably distinguished and supported by evidence. Partial retraction must not produce the misleading impression that broad conclusions survive when their foundational data, algorithms or evaluation are invalid.

The notice must specify the affected tables, figures, sections, datasets, code or conclusions, together with implications for the remaining results. If the principal claims can no longer be trusted, a full retraction is generally more appropriate.

Except where exceptional removal is necessary, retracted articles remain accessible on ARCITECH with an unmistakable indication of retraction on the article landing page and available full-text formats.

A retraction notice must:

• Be entitled clearly as a Retraction and be freely accessible.
• Identify the affected article by title, authors, publication details, DOI and URL, where available.
• State the responsible editorial authority and the principal factual reason for retraction, without inflammatory or unverified accusations.
• Be dated, linked in both directions where technically supported, and reflected in the published record and relevant metadata.
• Explain any author agreement or disagreement when relevant and feasible, without implying that author consent is a prerequisite to correcting unreliable literature.

The original article should not be silently overwritten with an unmarked replacement. A visible label such as RETRACTED and a link to the notice are essential to reduce the risk of uninformed citation.

Withdrawal before publication: a manuscript that has not been formally published may be withdrawn in accordance with the journal's submission procedures. Withdrawal of a manuscript under review is not a retraction and does not normally require a public scholarly notice.

Retraction after publication: once a Version of Record is publicly published (including online-first publication), serious subsequent reliability concerns must be handled through a documented, publicly visible post-publication process rather than treated as a quiet manuscript withdrawal.

An author request alone does not automatically entitle the authors to delete a published article or erase its public history.

Complete removal of a publicly released article or related material is exceptional and may be necessary when there is a compelling legal order, an immediate danger to individuals or systems, unlawful disclosure of highly sensitive personal information, or a serious rights violation that cannot be mitigated by access restriction or redaction.

Where lawful and practicable, ARCITECH will preserve an article's citation-level metadata and display a public notice explaining that access has been restricted or removed. The journal will avoid exposing the protected information in the explanation itself.

For cybersecurity, this may include exposed credentials, exploitable infrastructure details, unconsented logs, or unpublished vulnerability material that presents an immediate safety or privacy risk. Security controls must be proportionate; ordinary criticism or reputational concern is not a sufficient ground for deletion.

Authors, readers, reviewers, research institutions, or other persons may report credible concerns about an ARCITECH publication to the editorial office at arcitech.journal@iaincurup.ac.id.

Where possible, a report should contain the article title, DOI or URL, the specific passage, figure, data, repository, statistical result or other relevant material, and an evidence-based explanation of the concern. Reports from anonymous individuals may be reviewed when there is sufficient verifiable evidence.

Good-faith whistleblowers and individuals named in complaints should be treated respectfully; confidentiality will be protected as far as practicable while allowing a fair inquiry.

The journal will use a proportionate, documented workflow:

A response or investigation may be delayed by evidence complexity or institutional involvement. ARCITECH will act promptly and will consider an EoC if readers need to be alerted before a final outcome can be reached. Communications should avoid prejudging unresolved allegations.

The Editor-in-Chief or an appropriately designated, unconflicted senior editor bears responsibility for the final editorial determination about the journal's scholarly record. The publisher assists with investigation, production, metadata updates, archiving and legal matters without improperly substituting financial or institutional interests for independent editorial judgment.

Where the concern involves an editor, editorial-board member, reviewer or a paper authored by a journal officer, the affected person must not participate in the assessment or final determination. The publisher should facilitate independent oversight when the Editor-in-Chief is conflicted.

Investigations by universities or competent authorities may inform editorial decisions, but the journal remains responsible for timely and accurate published notices.

For computational studies, ARCITECH may request technical evidence needed to evaluate whether a published result remains credible, including:

• Original datasets, provenance, permission records, labeling protocols and separation of training, validation and test data.
• Source code, dependencies, model checkpoints, random seeds, parameter settings, system or hardware specifications and experiment logs where material.
• Independent confirmation of numerical claims, benchmarks, baseline comparisons, confidence intervals and reproducibility checks.
• Evidence concerning data leakage, fabricated measurements, duplicated samples, altered screenshots or synthetic/AI-generated content presented as real observations.
• Secure access to restricted data where public disclosure would violate privacy, licensing, intellectual property or cybersecurity safeguards.

A version update to code or data does not automatically require article retraction. Its significance depends on whether the published scientific claims remain valid and whether the change is transparently documented.

For significant post-publication amendments, ARCITECH should publish a separately identifiable notice and register or update Crossref DOI metadata and relationships as supported by its registration and editorial infrastructure. The original article's DOI should remain resolvable, and relationships between the original work and a correction, expression of concern or retraction should be explicit.

Notices warranting independent citation should receive their own DOI when supported by the journal's Crossref deposit arrangements. Links and article status should also be updated in OJS landing pages and PDF files where feasible, and communicated to indexing and preservation systems through supported channels.

The journal will preserve an auditable history of substantive changes and maintain traceable citations wherever possible. Correction and retraction notices should include dates and describe which version, component or research claim is affected.

Published work should not be silently substituted with revised text that conceals a substantial original error. Where an updated file is necessary, the revised version should make the change evident and link to an appropriate public explanation. Archival arrangements are subject to the journal's technical capacity and legal obligations.

Readers are encouraged to check the current status of an article before relying on its findings, particularly when a correction, expression of concern or retraction notice has been issued.

Affected authors will ordinarily have a reasonable opportunity to provide evidence and respond to substantive allegations. However, when the published record requires prompt correction to avoid harm, a journal may issue a justified notice even if all authors disagree or do not reply.

Authors or other materially affected parties may request reconsideration by providing new, relevant evidence or identifying a demonstrable procedural error. Where practicable, an appeal will be reviewed by an editor or publisher representative not responsible for the initial determination and not personally implicated in the concern. The appeal does not require the journal to suppress or suspend a justified public notice.

Substantive, evidence-based scholarly criticism may be considered in post-publication discussion, subject to civility, relevance, legal obligations and confidentiality. An adverse comment alone does not establish research misconduct.

This document should be read together with ARCITECH's publicly available:

• Publication Ethics & Malpractice Statement
• Peer Review Policy & Editorial Decision
• Artificial Intelligence (AI) & Generative AI Policy
• Author Guidelines and Submission Requirements

Editorial contact: arcitech.journal@iaincurup.ac.id.

The policy is informed by the following publicly accessible resources:

• COPE — Retraction Guidelines (2025 update)
• COPE — Expressions of Concern Guidelines (2025)
• COPE — Visibility of Corrections and Retractions
• Principles of Transparency and Best Practice in Scholarly Publishing
• Crossref — Registering Post-Publication Updates and Crossmark
• Crossref — Version Control, Corrections and Retractions

Reference to these standards does not constitute approval, endorsement, COPE membership, DOAJ inclusion or Crossmark certification.