Evaluation of unique identifiers used as keys to match identical publications in Pure and SciVal – a case study from health science [version 2; referees: 1 approved, 2 approved with reservations]

• Main Authors: Heidi Holst Madsen, Dicte Madsen, Marianne Gauffriau
• Format: Article
• Language: English
• Published: F1000 Research Ltd 2016-09-01
• Series: F1000Research
• Subjects: Data Sharing; Publishing & Peer Review
• Online Access: http://f1000research.com/articles/5-1539/v2

Unique identifiers (UID) are seen as an effective key to match identical publications across databases or identify duplicates in a database. The objective of the present study is to investigate how well UIDs work as match keys in the integration between Pure and SciVal, based on a case with publications from the health sciences. We evaluate the matching process based on information about coverage, precision, and characteristics of publications matched versus not matched with UIDs as the match keys. We analyze this information to detect errors, if any, in the matching process. As an example we also briefly discuss how publication sets formed by using UIDs as the match keys may affect the bibliometric indicators number of publications, number of citations, and the average number of citations per publication.  The objective is addressed in a literature review and a case study. The literature review shows that only a few studies evaluate how well UIDs work as a match key. From the literature we identify four error types: Duplicate digital object identifiers (DOI), incorrect DOIs in reference lists and databases, DOIs not registered by the database where a bibliometric analysis is performed, and erroneous optical or special character recognition. The case study explores the use of UIDs in the integration between the databases Pure and SciVal. Specifically journal publications in English are matched between the two databases. We find all error types except erroneous optical or special character recognition in our publication sets. In particular the duplicate DOIs constitute a problem for the calculation of bibliometric indicators as both keeping the duplicates to improve the reliability of citation counts and deleting them to improve the reliability of publication counts will distort the calculation of average number of citations per publication. The use of UIDs as a match key in citation linking is implemented in many settings, and the availability of UIDs may become critical for the inclusion of a publication or a database in a bibliometric analysis.