The benefits of authority management in an IR; more than name disambiguation

Abstract

The many problems of identity management are further exacerbated when dissimilar Chinese (漢字 Hanzi) names are transliterated into Roman homonyms and homographs. The IR team of The University of Hong Kong Libraries (HKUL) in 2009 extended the data model of DSpace with an authority management system that now enables, among other things, name disambiguation, and machine attribution of incoming articles to proper author profile. We created a structure of synonymy that handles authorized heading and non-preferred headings in Roman, Hanzi, and other scripts, similar to MARC authority tags 100 and 400. Unlike MARC authority, it will show preferred heading in Hanzi parallel to the Roman preferred heading (tag 700 is sometimes used for this in authority management of Greek, Russian, Hebrew or Hanzi). It will then display a third preferred form of abbreviated name used in scholarly publications.

The HKU Scholars Hub (“the Hub”), the IR of HKU, is hosted in DSpace. Publication records live in DSpace relational tables, and now with customization, additional tables show other objects, and attributes of these objects, not traditionally shown in an IR. One interface makes calls on any of these tables, and serves up an integrated display to the user. Non-publication objects now in the Hub include patents, grants (or projects), and author profiles. Attributes of these objects include, co-authors, co-investigators, co-winners of prizes, co-committee members, co-inventors, co-advisers of graduate students, professional qualifications, languages spoken, research interests, etc. The name attribute in all of these objects has authority control. All of these attributes further disambiguate identity, and contextualize HKU research and publications.

We work with several sources internal to HKU and external, to populate the Hub, control authority entries, and manage impact. We create ResearcherID profiles for each of our professoriate staff. We regularly rectify HKU data in Scopus, allowing us to capture the clean and full Scopus metadata on HKU publications and authors, and re-display in the Hub. APIs from these sources and others then allow us to show citation count “on-the-fly” against corresponding Hub publications, and paper count and citation count cumulated to each of our authors. Other sources include Microsoft Academic Search, Google Scholar Citations, ACM Digital Library, SSRN, RePEc, CiteULike, PubMed, BioMed Experts, MathSciNet, etc.

This work has allowed us to provide new functionality. We show visualizations of networks of co-authors, co-investigators, co-inventors, etc., with the user controlling the degree of separation. We show impact tables for each author, giving download & view counts, and paper & citation counts, according to time, source silo, and geography, allowing the author or his reader to drill down and find which publication became “hot” in which month.

This work prepares us for ORCID implementation. Auto population of ORCID now works from Scopus, and will be available for other sources as well. Auto-population could be disastrous. Our work with Scopus corrected 10,000s of egregious Scopus errors on HKU data, again exacerbated by the use of Romanization for Hanzi names. These errors include one author with two or more Scopus AU-ID profiles, one author’s papers distributed amongst two or more profiles of orthographically dissimilar people, two or more homonymous individuals erroneously shown as one Scopus profile, erroneous affiliations, and more.

All of this new functionality of DSpace at HKU has been funded by the HKU Office of Knowledge Exchange. Desiring to realize mutual benefits for HKU with society, they have now further funded a proposal to donate this work in open source to the DSpace community. With our partner, CINECA of Italy, we are preparing the first release of this add-on to DSpace, “DSpace-CRIS”. The reasons for moving from an IR to a Current Research Information System include, a) research is described and contextualized by more than publications, b) the benefits of authority work extend beyond publication author name disambiguation, c) libraries can do more to support their institutions than process publications. E-science and e-research beckon.