Tuesday, 11 December 2012

Software choices for Metadata Stores projects

In the October Metadata Stores Newsletter #9 there was a call for your rationales for choosing Metadata Stores software. The following rationales have been provided for sharing. If you have not already done so and would like to share yours, then please email a paragraph to be added to this (unordered) list:

University of Adelaide: The two candidate Metadata Store Solutions originating from academic institutions are very different in nature. ReDBox is an Australian developed Metadata Store focused on curating metadata within an institution. VIVO, originating from Cornell University, is much wider in scope aiming to use Semantic Web functionality to provide an overarching discovery and network building tool for researchers across institutions.

The comparative review examined the attributes of each system in terms of functionality, vendor capability, service and support, vision and cost. This document summaries that comparative review.

In general ReDBox can be considered to be a much smaller system in comparison to VIVO in terms of scope, complexity and vision. This lighter weight system is better focused on the needs of the University of Adelaide’s Research Metadata Store Project, with key advantages in the areas of complexity, workflow implementation and support. ReDBox overall represents a better fit for the needs of the Research Metadata Store Project - (Evaluation document attached).

The University of South Australia has evaluated a number of open source software solutions as part of the Research Metadata Store project. In order to ensure that UniSA has an enterprise-wide Research Metadata Store solution that not only meets our deliverables for the ANDS project but also fulfils the need for an overall metadata store solution, the Project Team has documented the complete set of ANDS requirements in addition to capturing all the UniSA requirements for a research metadata store then compared each open source software solution against the combined ANDS and UniSA requirements - (Evaluation document attached).

University of Queensland is developing custom software for its metadata store. Existing software that was available at the time did not complement other systems at the university and required non-trivial customisation to integrate. A custom approach gave the project more control over what functionality is available, how it is implemented and the timeline for delivery -- which was important for reducing the project risk, since many aspects of the project had never been done before. The approach requires more development resources, but it is expected the investment will result in a solution that is better aligned and evolves with the diverse and changing needs of the university.

University of Western Australia: UWA chose VIVO for its ANDS Metadata Store for several reasons: its potential to develop into a full ‘Research Profiles’ system, its use of Linked Data / Semantic Web technology, and the availability of Australian expertise at Melbourne, Griffith and QUT. We chose not to use an integrated repository/metadata store system like ReDBox, largely because we were already using a commercial product (DigiTool) for our publications repository. DigiTool itself was not suitable as an ANDS metadata store, and building an ANDS metadata store directly on to DigiTool was not an option.

The Australian National University’s reasoning for the adoption of Vivo is a little more complex than it first appears.

Our abstracted high level architecture is in appendix 2 of our project management plan and our choice of orca revolves around our need to have a store of some sort in our implementation. We have looked at a number of solutions

1) virtualising the store, i.e. pulling the data in real time from the various sources and building a display tool
2) using vivo as the display tool and simply pulling the data from a variety of sources and caching it in vivo
3) using orca as a backing store for data

We've come to the conclusion that a purely virtual store is not practical which means we need an interim store if only to store newly minted NLA party identifiers.
As we were already planning to use orca as a standalone collections registry populated from our data capture and seeding the commons projects it made sense to extend the use of orca rather than have another database.

Vivo we discounted because of its implicit organisational model - we did have it running with a very sparse set of data harvested from a number of sources, but we felt it was very person centric as opposed to research output centric
However our use of orca is pragmatic and designed to save us development time - there is no reason in our architecture why it could not be replaced by something else that did the job, and because we are using a very agile methodology we may yet do so should orca have performance problems.

It's our intention to then build a front end that gives a number of views and to see the core system effectively as middleware that supports a range of standard queries.
In that way the metadata store could be used as a source to populate individual research school's web pages in a standard manner to get a consistent view (and presentation) of the information across the institution.

University of New South Wales reviewed ReDBox, Vivo, MyTARDIS and the in house ResData. The evaluation was detailed and the conclusion was that the existing system ResData could be extended to include the required functionality and combined with Mint. However the other systems would require significant work to integrate with the existing systems.


Post a Comment

Note: only a member of this blog may post a comment.