Validating and monitoring bibliographic and citation data in OpenCitations collections

Abstract: Purpose. The increasing emphasis on data quantity in research infrastructures has highlighted the need for equally robust mechanisms ensuring data quality, particularly in bibliographic and citation datasets. This paper addresses the challenge of maintaining high-quality open research information within OpenCitations, a community-guided Open Science Infrastructure, by introducing tools for validating and monitoring bibliographic metadata and citation data. Methods. We developed a custom validation tool tailored to the OpenCitations Data Model (OCDM), designed to detect and explain ingestion errors from heterogeneous sources, whether due to upstream data inconsistencies or internal software bugs. Additionally, a quality monitoring tool was created to track known data issues post-publication. These tools were applied in two scenarios: (1) validating metadata and citations from Matilda, a potential future source, and (2) monitoring data quality in the existing OpenCitations Meta dataset. Results. The validation tool successfully identified a variety of structural and semantic issues in the Matilda dataset, demonstrating its precision. The monitoring tool enabled the detection of recurring problems in the OpenCitations Meta collection, as well as their quantification. Together, these tools proved effective in enhancing the reliability of OpenCitations' published data. Conclusion. The presented validation and monitoring tools represent a step toward ensuring high-quality bibliographic data in open research infrastructures, though they are limited to the data model adopted by OpenCitations. Future developments are aimed at expanding to additional data sources, with particular regard to crowdsourced data.