BiblioAudit: Automated Citation Integrity & Verification System

Abstract

We present BiblioAudit, an open-source framework designed to automate the verification of bibliographic references in academic manuscripts. By checking metadata across five major research databases, this system addresses the growing prevalence of hallucinated citations and metadata errors. This release serves as the official software implementation accompanying the work.

🔬 System Capabilities

5-Engine Verification Matrix

The system implements a multi-source validation pipeline that cross-references citations against a curated set of authoritative indices:

• Crossref: For universal DOI validation and metadata synchronization.

• OpenAlex: For global knowledge graph matching and disambiguation.

• PubMed: Specialized verification for biomedical and life sciences literature.

• arXiv: Targeted identification of preprints in Physics, Computer Science, and Mathematics.

• Semantic Scholar: Utilization of AI-driven citation graphs for rich metadata retrieval.

Visual Analytics & Health Metrics

The dashboard provides a real-time integrity assessment, visualizing the bibliography's temporal distribution and categorizing references into three health states: "Verified Clean," "Needs Attention," and "Not Found."

Exportable Audit Reporting

To facilitate peer review and collaboration, the system generates comprehensive CSV reports containing verification confidence scores and corrected metadata. In instances where external validation fails, the system preserves the original BibTeX data to ensure no loss of information.

⚙️ Methodology

• Precise Entity Matching: The query engine utilizes a strict Title + First Author matching algorithm to maximize precision and reduce false positives for generic paper titles.

• Smart Fallback Protocols: Entries that fail API verification are automatically routed to a generated Google Scholar search query, enabling rapid manual inspection by the researcher.

• PDF Discovery: The system integrates with Unpaywall to automatically locate legal, Open Access versions of verified references.

📄 Citation

Please cite this software as follows:

Tiwari, S. (2025). BiblioAudit: Automated Citation Integrity & Verification Tool (Version 2.1.0) [Software]. Zenodo. https://doi.org/10.5281/zenodo.18155557

BibTeX:

@software{Tiwari_BiblioAudit_2025,
  author       = {Tiwari, Satyam},
  title        = {{BiblioAudit: Automated Citation Integrity & Verification Tool}},
  month        = jan,
  year         = 2025,
  publisher    = {Zenodo},
  version      = {2.1.0},
  doi          = {10.5281/zenodo.18155557},
  url          = {https://doi.org/10.5281/zenodo.18155557}
}

Author ORCID: 0009-0006-2293-3946

📦 Installation

pip install -r requirements.txt
streamlit run app.py