Letter from the Editor | Gillian Griffiths on the new Author Identifier | Tips to find out who's who | Confessions of a user | Events Calendar
Gillian Griffiths on the new Author Identifier
"One of the reasons I think Scopus is so good is that we’ve really listened – and continue to listen – to what librarians and end users have to say. The product has been built that way from the ground up. "
as a matter of fact:
- Every time an Author Identifier search is requested, the entire Scopus database of 27 million records is scanned and results sorted in just milliseconds.
- The Author Identifier assigns a unique identifier number to each of the over 20 million authors who have published articles currently covered by Scopus.
- The Author Identifier has an unprecedented 95% recall rate. That means that Scopus has successfully matched 95% of an author’s documents.
- The Author Identifier can identify up to 150 of an author’s most frequent co-authors.
- Using the Citation Tracker in combination with the new Author Identifier provides a powerful tool to evaluate the impact of a single article, an author’s entire body of work or to assess the impact of a specific journal title.
- Authors can send us feedback to adjust their details using a link on the author details page.
- You can read more about the challenges of author disambiguation. We recommend “Name Authority Challenges for Indexing and Abstracting Databases” by Denise Beaubien Bennett and Priscilla Williams, University of Florida, Gainesville, published in Evidence Based Library and Information Practice, 2006. This Open Access article is available at http://ejournals.library.ualberta.ca/ index.php/EBLIP/article/view/7
Gillian is one of the Product Technology Managers in Amsterdam responsible for the ongoing development of Scopus. She has played an instrumental role in the evolution of the Author Identifier from concept to implementation. Another of Gillian’s instrumental roles is on her cello, performing in concerts throughout The Netherlands.
We first began talking about the Author Identifier almost two and a half years ago. Our goal was ambitious – to develop functionality that would generate a complete and accurate list of articles relating to a specific author and that intelligently distinguishes between authors with similar names – not just to generate a simple list of author name variants.
This was a daunting task, especially when you take into account the enormous datastream involved. In partnership with an information retrieval company, an extremely complex and elegant set of algorithms and weighting factors was developed to match Scopus records to authors. These not only use the author’s name, but also additional data elements within Scopus itself associated with the author’s articles, such as affiliation, publication history, source title, subject area and co-authors for analysis. We discovered that differently sized clusters of names actually behave differently.
After rigorous testing involving both librarians and end users, we felt confident that the Author Identifier was ready to make its debut. The results have been astounding! I’m proud to say that we have hit our targets – a 99% precision rate and a 95% recall rate.
There’s a dynamic relationship between precision (specificity) and recall (sensitivity). We decided to err on the side of precision so that there’s less chance of multiple authors being identified as a single author and articles incorrectly assigned. The Author Identifier itself is dynamic; new data enters the system all the time resulting in constant reprocessing, rematching and reassigning of crosslinks. There’s also a robust feedback process that includes verification checks. It’s our intent that Scopus will go a long way in taking the “detective work” out of author disambiguation.
More information
For a step-by-step guide to “Getting Started” and an overview
demo about the Scopus Author Identifier, just go to
www.info.scopus.com/authoridentifier
