The once & future repository, HKU's Scholars Hub

Abstract

The HKU Scholars Hub (the Hub) began service as a traditional institutional repository of The University of Hong Kong (HKU). However this format was not compelling to HKU researchers. Fortunately a subsequent reformation of the HKU statement on university mission and vision infused new life and purpose into the project. Over the next five years, in partnership with the Italian University Consortium, Cineca, the HKU Libraries transformed the Hub from an IR to a Current Research Information System. We expect that future development will see the Hub further transformed into a research information management system supporting both internal decision support and external public discovery. We will present new work developed recently to further these goals.

IRs collect, manage and display publications, and their metadata. However, an institution’s research, expertise and capacity is described by more than publications. The Hub, hosted in DSpace, began as the IR of HKU in 2005. Asking for voluntary deposit of publications from HKU academics, it received little notice, and more importantly, little support from University senior management. In 2009 a new HKU initiative, Knowledge Exchange (KE), adopted the Hub as a key vehicle to share knowledge and skill with the community outside HKU. Upon winning grant support from the office of KE, the HKU Libraries chose Cineca as a development partner. Together we designed specifications to extend the data model of DSpace. We architected solutions to support non-publication objects, including people, grants, and patents. These entities are managed in new database tables with a flexible structure that is able to hold indexed and interlinked attributes, such as co-investigators, co-inventors, co-prize winners, research interests, languages spoken, supervision of postgraduate theses, etc. The structure has been designed to provide native support (through a backend UI) to the data model extensions. This will allow local operators to easily add new entities and new attributes, interlinkable to any internal or external corresponding record, without the need to write new code.

Beginning with local data in several HKU silos, scripts will search for corresponding or augmented records in external sources, harvest and merge with Hub data. These sources are publication databases (Scopus, WoS, PubMed, etc.), funders (Hong Kong Research Grants Council, NIH, etc), patents (USPTO, Espacenet, Japan Patent Office, etc), and bibliometrics (Scopus, Google Scholar Citations, SSRN, etc).

The DSpace user interface now delivers an integrated search and display on all objects and attributes, as well as on ones newly derived, such as a) authority work on name disambiguation and synonymy in Roman and Hanzi (漢字), b) visualizations on networks of co-authors, co-investigators, etc, c) metrics extracted from external sources, and d) internal alt-metrics of view and download counts, and more.

In order to increase utility and the interchange of information, we have augmented DSpace’s OAI-PMH with several other protocols. 1) DOI. We register DOI numbers with CrossRef for all our e-theses (23,000 in 2015), and will begin sending separate XML delineated bibliographies to CrossRef to allow us to harvest citation counts from CrossRef. 2) ORCID. As an early adopter of ORCID, we have created ORCID accounts and populated them with data from the Hub for all our professoriate staff. During the OR2015 Meeting we will present the new developments made in collaboration with Cineca that have brought DSpace integration with ORCID to a new level: using the extended data model for DSpace described above, we are developing scripts to automatically query the scholar when new information comes into the Hub. Upon this scholar’s confirming one click, the script will automatically update the ORCID account. A similar future script, upon the scholar’s one click, will create and populate an ORCID account with information harvested from the Hub. 3) CERIF. We will soon implement the Common European Research Information Format (CERIF), for data exchange with other research entities. The Catalan Consortium and CSIC (mentioned below) have now used CERIF in this context to build central repositories.

Beyond the open scholarship functions of an IR, the Hub now performs as a system for reputation management, impact management, and research networking and profiling -- all of which are concepts included in the broad term, “Current Research Information System” (CRIS). These new objects and attributes curated from several trusted sources, and integrated into the present mashup, contextualize and highlight HKU research, and attract more hits, than an IR with only publications. A CRIS can provide discovery on internal resources to external users, and decision support to internal administration. Linked CRISes across several organizations or countries, using a common data interchange format such as CERIF, on top of providing federated discovery, can provide the sociological and organizational functions needed in e-science, e-research and cyberinfrastructure projects.

The Hub has repurposed data once held in dark archives for HKU in-house administration purposes only. This data, after augmentation by the Libraries, provides discovery on HKU research and expertise. This data is then further harvested from the Hub for a multitude of purposes. HKU faculties are slowly abandoning their own data collection and web page creation methods in favour of using the web service of the Hub, to extract any Hub displayed element into XML. We re-package our data to ship to SSRN, so that the HKU Law Faculty can provide a monthly journal in SSRN’s LSN, etc. The Hub also enables the repurposing of Library staff. With the demise of print cataloguing, several staff found themselves without a job. Verifying, augmenting, and curating data incoming to the Hub has provided these staff with continuing and richer employment. And, the Hub project has newly purposed the HKU Libraries with another means of supporting their underlying institution. This is much welcomed in this era of lessened importance of the onsite physical collection.

Using grant funding from the HKU Office of Knowledge Exchange, Cineca and HKU Libraries modularized this development work and placed it in open source on GitHub, with the name, “DSpace-CRIS”. Any DSpace site can now download and install as an optional add-on to DSpace. The Catalan Consortium of Research Universities has used it to build a regional portal that aggregates research information from all Universities in their region. EuroCRIS has used it for publication management and researcher profiles on their member organizations. Sites of other uptake include the Hong Kong Polytechnic University and the CSIC Consejo Superior de Investigaciones Científicas (Spanish National Research Council). In recognition of DSpace-CRIS’ great flexibility, potential, and cost effectiveness, Cineca adopted DSpace-CRIS as the basis for their new “IRIS”, an enterprise CRIS platform which they are now installing in over 60 universities in Italy and abroad. A separate proposal to OR 2015 will give further detail on this.

The new version of DSpace-CRIS that we will present at OR2015, besides the important update relating to ORCID integration above described, includes a reworked search interface. We are introducing a new global search across all entities managed by the system to give surgical precision in search and discovery. These searches will return snippets of text with search terms highlighted in context. The DSpace-CRIS repository eco-system is now very rich. As an example, the researcher’s profile is composed of more than ten different sections covering general contact information, publications, projects, bibliometrics, theses, etc. Clicking on the returned highlighted search terms in context will take the user to the exact section where the desired information is displayed, rather than dropping the searcher confusingly at the top of a long complex record. This new global search will also enable operators to easily make new reports showing correspondence between data elements in disparate Hub entities, such as an interactive “Faculty Used Journals for Publications Report”, sortable by publication frequency, impact factor, year, author, underlying grant scheme, patents generated, etc.

In summary, all involved with DSpace-CRIS wish to move this project further, to increase the utility of DSpace-CRIS, and its adoption in the open source community. We hope to find new ways to work with DuraSpace. This could mean providing a road map to the future for libraries stalled in IR development. It could mean providing an open source solution for IR / CRIS, as an alternative to commercial products. We would very much like to work with the community to find ways of convergence, DSpace-CRIS with mainstream DSpace. IRs and CRISes are changing, and their users and administrators are asking for new types of information and new ways to use them. We wish to fully collaborate with DuraSpace, to define a vision of the future, and make it happen for everyone that wishes this collaboration.