Learn how to use the Veracity platform for citation checking.
This guide is intended for advanced users—editors, peer reviewers, and other “super users”—to understand the deeper philosophies behind Veracity’s citation checking tools and to interpret the results effectively. While Veracity offers powerful automation to streamline scholarly review, human judgment remains vital. This documentation will equip you with the context and best practices to derive maximum value from our system.
In academic and professional fields such as science and law, citations are the bedrock of trust and traceability. Yet, research consistently shows that up to 25% of citations in the general scientific literature are inaccurate (Peoples & Østbye, BMJ, 2023). With the increased use of AI in text generation—where hallucination rates for citations can climb to 55% (Walters & Wilder, 2023)—the potential for further degradation in citation quality looms large.
Veracity addresses this growing concern by automating the process of checking and validating citations. Our aim is to raise the bar on quality and integrity, ensuring that authors, editors, and peer reviewers can confidently assess the veracity of referenced works.
Integrity and Trust
Veracity’s chief goal is to uphold scholarly integrity. We offer tools to ensure citations are both accurate and contextually appropriate, reinforcing the academic and professional trust chain.
Augmenting Human Judgment
Automation can reduce manual workload and highlight potential issues, but final decisions about the accuracy and relevance of a citation ultimately rest with human experts.
Transparency and Accountability
By flagging inconsistencies or errors, Veracity promotes a transparent environment where issues can be corrected early in the publication cycle, thereby elevating overall quality.
Veracity’s process consists of several distinct but interrelated steps, each designed to detect and flag different types of citation issues.
Identification of In-Text Citations
We scan the document to locate every citation within the main text. This includes numeric, author-year, and other citation styles.
Bibliographic References
We parse the references section (or bibliography) to capture titles, authors, DOIs, and other metadata.
Normalizing Citation Data
Our system converts all extracted elements into a standardized format, making subsequent comparisons more reliable and easier to interpret.
Internal Consistency Checks
Outcome: A structured dataset linking each in-text citation to its corresponding reference, ready for further analysis.
Source Matching
Veracity queries third-party indices (e.g., Crossref, Semantic Scholar) and search engines (e.g., Google, Brave) to locate and confirm each cited source.
Metadata Validation
Crosschecks the author-provided metadata (title, authors, publication year, DOI) with what is found in external databases, flagging discrepancies.
Retraction and Editorial Notices
Checks the retraction status of cited works using services like Retraction Watch. Any editorial notices (e.g., corrections, expressions of concern) are also highlighted.
Content Gathering
Where possible, Veracity retrieves the full text or abstract of the cited source to enable deeper relevance and support analysis.
Outcome: Each cited reference is verified for existence, metadata accuracy, and editorial status. Any retractions or major inconsistencies are flagged for further review.
Using both the extracted content and any additional data gathered, Veracity evaluates the relationship between each in-text statement and the cited source.
Custom AI Models
Proprietary machine learning models trained on curated citation-data sets to determine how well a reference supports or relates to the in-text claim.
Scoring System
Examples
Context Limitations
If full-text access is restricted, Veracity relies on title, abstract, and metadata to make a preliminary determination. Human reviewers can then investigate further.
Outcome: A structured relevance score for each citation, plus recommendations for further action or clarity where needed.
Below is a concise way to rename and organize your checks section under the heading “Summary of Checks.” This approach is clear to readers while preserving the idea of grouping checks by what they apply to (the entire document, individual citations, references, etc.).
To help you quickly see all the checks Veracity performs, we’ve consolidated them here in one place, grouped by level of analysis. Some checks require the document to be fully parsed before running (e.g., whether all references match in-text citations); others can be run earlier in the workflow. Refer back to each step of the process (Extraction, Source Resolution, Relevance Check) to see where these checks typically happen.
| Check Name | Description |
|---|---|
| All Sources Found | Ensures that every reference has been matched to a real, identifiable source. |
| No Retracted References | Flags any references linked to retracted articles or editorial notices. |
| All References Match Metadata | Verifies that cited reference details (e.g., title, authors, year, DOI) align with the retrieved source metadata. |
| All References Cited in Text | Confirms that every bibliographic entry is cited at least once in the text. |
| No High-Frequency Author Citations | Identifies potential over-representation by authors cited above a set frequency threshold. |
| No Self-Citation Above Threshold | Checks whether self-citations exceed a predefined threshold. |
| No Ambiguous Matching References | Ensures that each in-text citation uniquely links to a single reference. |
| All In-text Citations Relevant | Provides an aggregate relevance assessment, confirming in-text citations point to pertinent sources. |
Use these checks to get a high-level sense of the document’s overall integrity. For example, an abundance of self-citations can signal potential bias, while retracted references may indicate outdated or invalid research.
| Check Name | Description |
|---|---|
| In-text Citation Has Matching Reference | Confirms that each citation in the text points to a unique entry in the bibliography. |
| In-text Citation Not Ambiguous | Verifies that the citation unambiguously refers to exactly one source. |
| In-text Citation Relevant | Assesses whether the cited source is contextually relevant to the statement being made. |
| In-text Citation Supports Statement | Determines whether the source contains specific evidence or arguments that back the claim. |
These checks happen during or after Extraction and are refined by the Relevance Check. If Veracity flags a citation for irrelevance, consider whether the source truly supports the text’s claims or if a different citation is needed.
| Check Name | Description |
|---|---|
| Source Found | Confirms that a referenced source can be accessed via external databases or search engines. |
| No Editorial Notices | Verifies that the source is not marked as retracted or under editorial concern. |
| Matches Source Metadata | Compares the metadata provided (title, authors, publication details) to external databases. |
| Reference Cited in Text | Ensures that each bibliographic entry is used at least once in an in-text citation. |
These checks typically finalize after Source Resolution, when Veracity attempts to locate and verify each reference. If metadata discrepancies arise, the user may need to confirm the correct citation details (e.g., updated DOIs, correct spelling of authors).
| Check Name | Description |
|---|---|
| Author Representation Check | Analyzes patterns in the reference list to detect if certain authors are cited an unusually high number of times. |
| Author Previous Retractions Check | Identifies whether any cited authors have a history of retractions or corrections in their publication record. |
These checks rely on fully identifying authors across all references. They help detect potential conflicts of interest or bias, such as a paper heavily self-citing or repeatedly citing authors with a track record of retractions.
Veracity offers two modes of operation to suit varying resource constraints and use cases:
Quick Checks
Deep Checks
Outcome: Organizations can balance cost and efficiency against the depth of citation scrutiny desired.
When Veracity completes its analysis, it provides a structured report summarizing:
Internal Consistency
Source Verification
Relevance Scoring
Suggested Actions
Common Error Types
Q: What if Veracity doesn’t find a reference?
A: It may be an obscure publication or incorrectly formatted reference. Try manual checks or confirm the source with the author.
Q: Will Veracity fix my references automatically?
A: Veracity flags potential issues, but authors remain responsible for making corrections.
Q: Does Veracity store or share my manuscript content?
A: We respect privacy and confidentiality. See our Privacy Policy for details.