Thursday, 15 January 2015

Implementing ORCID at York : a technical perspective

For York’s ORCID pilot project, this is what we said we would do:

Pure is a current research information system (a CRIS) offered as a commercial product by Elsevier. York have been using Pure for several years, and have connected Pure to various internal systems, including those for finance and HR, to provide a single source of data about research activity in York. Open access versions of research outputs can be added to Pure and are sent via a connector to White Rose Research Online (WRRO), our open access repository.

From version 4.18 onwards, Pure has included support for storing and creating ORCID identifiers against a user’s personal profile. For York’s ORCID pilot project we decided to use the built-in ORCID functionality of Pure because a) it is a system every academic member of staff (with a research contract) has to interact with during their working life at York, and b) this provided an out-of-the box solution with no development required. By asking users to create their own identifiers, we also avoided having to develop any local utilities to perform batch creation and ensured users had engaged in what was happening.

What we did

In order to create ORCID identifiers in Pure, we needed to become ORCID members. We signed up for the Creator License as Basic Members (see For the needs of this project, this license suited us perfectly although if we were to want to do more sophisticated things in future, such as connect multiple systems to ORCID or synchronise data with ORCID, there is a more costly ‘Premium’ Membership.

The Pure integration, in theory, needed only the correct ORCID member credentials and a couple of switches flipping.

ORCID configuration in Pure

In reality it was not quite so straightforward. The ORCID functionality in Pure 4.18x did not work and so we had to wait for an upgrade to version 4.19 for a fix. As Pure is a system which integrates with various business critical systems at York, an upgrade cannot just happen - a set of tests need to be carried out and work needs to be scheduled in with both the suppliers and IT Services at York.

Thanks to the support of our colleagues, we were able to push on with the upgrade speedily and the ORCID functionality was made live in November.

The process of adding an ORCID iD is relatively straightforward. The user must go into their Pure profile and select to either create a new ORCID iD, or add an existing one.

Add or create an ORCID

A check button allows the existing ORCID identifier to be verified.

Add an existing ORCID into Pure

Creating a new ORCID identifier requires three steps:
  • a single click in Pure
  • receiving an email from ORCID and following the verification link supplied
  • entering a password to claim and activate the ORCID iD

For those who wish to make more use of ORCID, creating an account allows them fill out their ORCID profile and to add publications to their ORCID record.

Verification email from ORCID

ORCID also provide access to a sandbox to allow full testing of the functionality without affecting any live systems. All screenshots above have be made in our Pure test system, talking to the ORCID sandbox.

The second step for the project was populating our institutional repository White Rose Research Online (WRRO) with ORCID identifiers in Pure. Clearly, this step could not be taken forward until there were ORCID identifiers in Pure so it was held back until the first stage of the pilot had completed.

The Pure eprints connector sends information to WRRO when there is a full-text attached to a research output and the text has undergone a copyright check by the Library. Within Pure there is a MODS XML metadata record containing information about the research output. An XSLT stylesheet transforms this MODS record in the EPrints XML format required by WRRO. We have local control of this stylesheet and so sending the ORCID identifier to EPrints is made via a simple edit to the transformation. In order to do this, though, we had to request the ORCID identifier be made available in the MODS metadata published by Pure (over which we have no control), something we had not realised needed to be requested. This caused another delay as the availability of the ORCID identifier in the MODS record was released in Pure version 4.20. This version was installed on our test Pure instance in mid December 2014 but, unfortunately, a separate problem broke the EPrints connector, requiring another fix that will be available in January 2015.

At the time of writing, we are confident that the ORCID identifer will seamlessly be added to the metadata supplied to WRRO for new outputs once the latest version is available. To retrospectively apply ORCID identifiers will require a re-synchronisation of data between the systems at a later point.

Implementation Issues

As noted above, we hit some technical issues during the testing of both the ORCID functionality and the connector to WRRO. Issues are part and parcel of running systems, but their impact on a project with a short timescale can be significant and have prevented us from delivering the full ORCID iD to WRRO integration on time.

Other issues to note include the fact that our implementation of Pure at York does not provide accounts for PhD students, and so separate communications will be needed for that target group. For the pilot, students were encouraged to register for an ORCID iD on the ORCID web site, leaving us no mechanism for tracking how many ORCID iDs had been created (something we can see in Pure for academics and research staff).

A small, but significant, issue in the usability of Pure led to a number of people not saving their profile record containing the ORCID identifier. This led to these people being assigned an ORCID iD but the identifier not being stored in Pure.

In creating an ORCID iD via Pure, the user is allowed to alter the email used (something we would not recommend) but the user has no control over the format of their name. Pure already has the ability for a user to specify a preferred format, for example to exclude a middle name or provide a commonly used short version of their forename, but this is not used for ORCID iD creation, meaning that a user would need to edit this information in ORCID, adding an extra step.

A separate issue is that Pure is an internal system whose public front-end is delivered by the York Research Database, an implementation of the Pure portal software. Having the ORCID in Pure doesn’t automatically mean it displays in the public view and needs to be separately requested. At the current time, then, ORCID iDs created in Pure are not publicly exposed.

Future Improvements

Running the pilot project has allowed us to explore the functionality offered by Pure for creating ORCID iDs. After the initial false start, this is now in place and working. Out of the work on the pilot, we are feeding back three suggestions to Elsevier on the basic functionality of the connector:

  • The ‘Save’ option in PURE is easily missed and it’s prominence should be improved.
  • Pure should use the user’s preferred name format in creating the ORCID identifier.
  • Provide a better error message when a user tries to create an ORCID when they already have one, and a link to the ORCID site.

Unhelpful error message if an ORCID already exists for that user.

Including the ORCID identifier in the York Research Database would be a quick win and should be pursued.

Better support for PhD students at York is needed, either through their being able to access Pure or by making it easier for them to create and record an ORCID iD with the University.

The potential for synchronising outputs between PURE and ORCID needs to be more fully explored. Some users have been keen on this, but it raises some questions, for example over where the canonical source of publication information is and of what ORCIDs ‘added-value’ services should be.

It remains to be seen what more we can do with ORCID iDs, but some ideas that have already been discussed include using them within identity management systems at York and providing simplified mechanisms to assist academics in moving institutions.

No comments:

Post a Comment