Northwest England DCC roadshow at the University of Salford: a report

  Although not directly related to current Sonex work on analysing requirements for dataset transfer via Sword, the international workgroup was interested in attending a DCC roadshow for gathering a view on -and providing its own input to- RDM-related training initiatives that complement direct institutional experience in RDM acquired through JISC MRD projects. So when a chance showed up to attend the Northwest England DCC roadshow at the University of Salford, we were happy to engage with the UK Digital Curation Centre for being there at the University Library on March 20th and 21th.

Training initiatives regarding research data management were also thoroughly discussed at the 'Research Data Management: Activities and Challenges' workshop organised by the Knowledge Exchange Primary Data Workgroup in Bonn last November, which Sonex also attended (and provided a report for), so there were opportunities in Salford for identifying synergies between national and international RDM initiatives in this regard. This was also the right occasion for highlighting an example of best practice in RDM-related training activities based on local network building and promoting extensive debate on where to start and how to carry on with the work, while disseminating the appropriate tools to do it along the way.

The DCC roadshow proved to be a very effective complement indeed to JISC MRD programme and other RDM-related initiatives for reaching the 'common university' - i.e. those ones where preliminary efforts -be it at researcher survey level- are taking place to build some kind of RDM infrastructure but with no particular 'official' support outside -sometimes even inside- their institutions. The event in Salford gathered representatives from many NW universities -Salford, Manchester, Liverpool, Sheffield, Leeds- and debates along the roadshow were very much enriched by the mixture of institutional profiles attending it, from librarians to research office managers to researchers to ethics committee members. The experienced DCC roadshow team -Martin Donnelly, Andrew McHugh and Patrick McCann- were also very efficient in promoting dialogue and passing on guidelines and expertise along the event.

A shorter schedule was applied to this NW England roadshow, that took just two days instead of three: a first day devoted to presentations on data management initiatives taking place in the region and a second day for group discussions on research data management needs and how to use tools provided by DCC to identify them (such as DAF, Cardio or DMPOnline). The DCC Data management roadshow is also an evolving creature and there are slight variations in content among different editions thereof, this meaning that the event focus can be adapted to different levels of regional RDM implementation: emphasis can be for instance made on advanced RDM tools such as DMPOnline where RDM initiatives are well under way while mainly focusing on the Data Assessment Framework initiative in regions where RDM lies yet at a preliminary implementation stage.

Highlights from the first day included an estimulating 'Towards Open Worlds' keynote speech by Professor Martin Hall, a Vice Chancellor showing an unusually high commitment to Open Access and keen to debate related issues with the audience. An inspiring presentation of the two-stage MaDAM/MiSS JISC MRD project at U of Manchester was also delivered by Meik Poschen, providing some guiding light for preliminary initiatives in RDM currently being carried out at other institutions. Finally, Day I sessions were closed with a four expert panel discussion, in which presenters at the event were asked to stress a specific issue in RDM they considered worth deeper examination. The answers were: Cost model (Meik Poschen, UoM), Limits to researcher time availability (Rachel Kane, U Sheffield), Who shoud lead RDM tasks - is the Library able to? (Julie Berry, Salford U) and Research motivation as a decisive argument (Graham Pryor, DCC).

Along subsequent discussions Sonex became aware of three relevant points:

- Benefits can arise regarding these issues from a deeper analysis of international RDM initiatives -including ongoing and forthcoming European projects- connected to the institutional activity in the area,

- Besides disseminating specific funder mandates, finding a way for estimating the institutional costs derived from universities not managing their research data could be a potentially very effective argument for engaging universities with RDM activity.

- There is much emphasis in discussions on how to train researchers, but not so much on how to set up and train a team of dedicated data librarians - this strongly depending on Library staff figures and on whether or not librarians see themselves as fit for the task.


On the roadshow Day II the EPSRC policy framework on RDM and its implications for RDM strategy implemention at universities and research centres were discussed, and several joint RDM planning activities were carried out by different groups using DCC tools for examining aspects such as benefits to be obtained from RDM, where each institution stands in terms of RDM implementation strategy or how to deeper engage research groups and university management into RDM.

From a Sonex point of view, attending the roadshow proved very useful for identifying successful models of RDM training and dissemination, and we would humbly recommend to provide this RDM training initiative an international profile once complete so that similar efforts may be applied to a broader context. It would also be useful that participants in the DCC roadshows could provide feedback on the impact of their taking part in the initiative on their institution's work on RDM implementation a few months afterwards. Any future reporting from the DCC roadshow team on their initiative will be a very interesting read indeed and we shall be following dissemination initiatives outside the UK -such as the talk on DMPOnline at the Future Perfect 2012 Conference in Wellington this week- and hoping they'll soon arrive to continental Europe, where their work on Data Management Plans may be particularly valuable in the near future.

Thematic parallel session on metadata - actions to be taken

  On Day II of the JISC MRD Programme 2011-13 launch event in Nottingham, last Dec 2nd, specific subject-based discussion sessions were held among the different JISCMRD02 Projects for research data management in order to promote synergies and joint work on common issues. This is a brief report on the outcomes of such discussions at the parallel session on metadata - some other were simultaneously held for Institutional, Life Sciences, Engineering or Archaeology MRD projects, whose discussions have been reported elsewhere (and there are also other posts summarizing talks for this one too).

It was really hard for some of us to pick a single of those groups, since many projects actually belonged to several strands (some lucky ones had also two representatives at the event, it should be noted). The session on metadata was attended, among others, by:

- Anna Clements (U St Andrews)
- Simon Kerridge (U Sunderland)
- Kevin Ginty (U Sunderland)
- Charlotte Pascoe (British Atmospheric Data Centre)
- Pablo de Castro (SONEX Workgroup)
- Simon Hodson (JISC MRD Programme manager)
- David Shotton (U Oxford)
- Louise Corti (UK Data Archive)
- Marco Fabiani (Queen Mary U London)
...


Discussion

Metadata standards were repeatedly discussed along the session - there was a joint (and unsuccessful) attempt to recall whether anyone knew about a metadata standard registry available for different disciplines. Representatives from CERIF4Datasets Project, University of Sunderland, mentioned they were using the MEDIN metadata standard for their work in marine sciences data management. The Core Scientific Metadata Model (CSMD) standard, developed at STFC for the I2S2 Project was also mentioned as an interesting approach to multi-disciplinary metadata standard for structural sciences such as Chemistry, Materials Sciences, Earth Sciences or Biochemistry. Finally, the PIMMS Project (BADC/U Reading), mentioned Metafor as a Climate Science metadata standard and their goal of using PIMMS software tool to generate CIM-based content.

At some point the idea catched up that metadata standards should perhaps be mandated by publishers in order to harmonise discipline-specific data description procedures. Publishers are actually involved in several very successful international RDM projects, such as Dryad, but -save for REWARD- are significantly missing in JISCMRD02 projects.

Having previously developed the Semantic Publishing and Referencing (SPAR) Ontologies, David Shotton said he was now working on their extension to CERIF-based metadata description of datasets, which is closely linked to dataset CERIFication work being carried out at the CERIF4Datasets Project.


Actions

The following actions were proposed for improving the chances of metadata standard harmonisation - hence enhancing dataset discoverability:

• Trying to locate (or otherwise collect) an already existing registry of metadata standards for different disciplines, in order to offer researchers from a given discipline an already tested metadata schema they can re-use,


• Mapping metadata standards to each other aiming to produce a minimum-sufficient-information metadata set that may be widely applicable accross disciplines,


• Taking steps towards organising a workshop in order to have metadata issues discussed among relevant stakeholders. ANDS Metadata Workshop in 2010 might be a potential source of inspiration for this with all those discipline-based approaches to metadata standards. Proposed dates for this Metadata WS were spring-summer 2012.

Finally, there was a wrap-up by different subject-based project groups which showed strong possibilites for a more stable cooperation among them (Biomedical/Healthcare projects even discussed the possibiity of building a common wiki). Some cooperation frameworks (googlegroups, mailing lists) might be set for promoting this disciplinar trans-project collaboration. Regarding the metadata strand, it should be noted it was also an issue in discussions held at most subject-specific workgroups, so it would potentially allow contributions from all of them.

The dawn of a new JISC MRD programme - Day I

  After a successful first stage of the JISC Managing Research Data (MRD) Programme (2009-2011), a second phase of JISC MRD was launched yesterday at the NCSL Conference Centre in Nottingham, along a 2-day event that will continue today. JISC MRD02 Programme includes 27 projects classified in three different strands:

Strand A. Research Data Management Infrastructure: 17 projects, to be completed from Mar to Jul 2013, comprising Institutional Pilot projects, Institutional Embedding and Transition to Service projects, Disciplinary projects for creative arts and archaeology, and a Metadata project,

Strand B. RDM Planning: 8 projects running until Mar 2012, aiming to design and implement data management plans and supporting services for researchers,

Strand C. Enhancing DMPOnline projects: 2 projects, aiming to customize and enhance the DCC DMPOnline Tool to improve its interaction with institutional/ disciplinary information systems).

It is worth noting that a number of funded RDM projects along this 2nd programme stage are building upon previous pilot work (projects carried out along JISC MRD programme 2007-2011) in order to for instance extend and embed data management services accross the whole institution.

On describing the research data management programme, Simon Hodson, JISC MRD programme manager mentioned there will be two further JISC MRD calls as early as Jan 2012, dealing with:

- Research data publications, aiming to build partnerships among involved stakeholders and encouraging data citation and publication,

- RDM Train, aiming to design and implement data management training strategies for specific disciplines and support roles (including librarians), to be performed by linking to professional bodies.

Emphasis will also be made along this 2nd JISC MRD programme stage on evidence gahering for project benefits and impact. A session devoted to these issues will be held on Dec 2nd, with practical work with both the Benefits Framework Tool and the Value Chain Impacts Tool. Developing metrics for measuring project impact is a specific programme goal along this 2nd implementation stage.

Project blogging

Another JISCMRD02 main objective -and closely related to impact measurement- is promotion of project dissemination and interaction among themselves and with the broader community via blogging. A specific presentation on 'blogging practices to support project work' was delivered for the purpose by Brian Kelly, UKOLN. The presentation highlighted the relevance of publishing project blogposts as an alternative means of expression to writing research papers or code, and engaged the audience in finding shared views regarding potential benefits blogging may bring to RDM projects, also providing some useful technical advice along the way.

Subsequent discussion focused on pros and cons of blogging as a communication technique (both from regular bloggers' and researchers' viewpoint), as well as on potential advantages of JISCMRD project blog aggregation, with a common RSS feed embedded back into the JISC site.

Parallel sessions and poster-session networking

Two parallel sessions came afterwards, dealing with two principal RDM issues: a first one on DCC Tools, introducing Data Asset Framework (DAF), DMPOnline and CARDIO, and summarized by Paul Stainthorp, U Lincoln, on his JISCMRD02 Day I blogpost.

The 2nd parallel session dealt with UMF Tools and related RDM projects. This 2nd session featured presentations by John Milner on JANET Brokerage and Andy Powell on Eduserv Cloud Pilot, along which the strategy for Academic Cloud service implementation was described - based on the "work with the willing" driving line. The Dynamic Purchasing System (DPS) -originally developed for utilities such as water or light- will be re-used as purchasing framework for cloud-related services. Regarding Eduserv, a 2-month 'introductory tier' will be available (just for institutions) along the service gradual implementation (storage being currently single-site, with no backups at this pilot stage, though there are plans for offering tape backup for part of the stored infrastructure).

After an interesting Q&A time, in which backup was suggested to be an absolute requirement for the success of the initiative and there were questions on various Eduserv use mode details (such as the possibility of using departmental orders/purchase order instead of credit cards for academic use), five projects from the UMF strand were briefly presented which are already working either based on a SaaS approach or in the cloud, or both: these were BRISSkit (Jonathan Tedds, U Leicester), DataFlow (David Shotton, U Oxford), Smart Research Framework (or ELB software as a service, Tim Parkinson, U Southampton), VIDaaS and YouShare Projects. Slides for these presentations will shortly be available and will be linked from here.

Finally, Day I official programme ended with a poster session and networking event, which meant a really good opportunity for RDM projects to interact with each other and with 'fellow travellers'. Synergies among projects became quite evident when having all them displayed together on a set of panels, and having their representatives available and willing to discuss each project aims, challenges and similarities to others offered a very good chance to get the general picture along with the details, as well as for establishing inter-project liasons that went well over closure time.

Progress on Researcher ID initiatives: IRISC 2011 Helsinki

  
The problem with names...

Prof. Carlos Martínez-Alonso is a renowned Spanish senior biochemist. He was actually President of the Spanish National Research Council (CSIC) when the Berlin Declaration was signed by the institution in January 2006. Prof. Martínez-Alonso has published hundreds of papers in high impact factor journals. However, when retrieving a complete list of his publications from PubMed database, you find out it is not possible unless several parallel author queries are carried out: there is a Martinez-A C entry under which most of his publications get listed [222]. But then there's also Martinez-Alonso C [21] and even Alonso CM [1].

It might be argued it's all about funny Spanish names with two surnames in them. That's a problem alright. Not just for Spanish names though: it's quite the same for Portuguese/Brazilian authors as well. Not to mention transliteration of Asian author names (see "Which Wei Wang?" Phys Rev 2007 editorial). PubMed is presently running its Author ID project in order to tackle this problem, which is by no means exclusive of theirs: around 2/3 of the over 6 million authors in MEDLINE share a last name and first initial with at least one other author, and an ambiguous name refers to 8 persons on average (Torvik and Smalheiser, "Author name disambiguation in MEDLINE").

Name disambiguation and proper attribution is a well-known problem in the scholarly publishing ecosystem. There have been and there are lots of initiatives trying to tackle this complex issue at subject, institutional or even national level - with remarkable success in the case of the Dutch Digital Author Identifier (DAI).

However, this is not an issue to be tackled at national nor subject level, but globally. Commercial stakeholders such as ThomsonReuters or Elsevier-Scopus are then in a privileged position to implement some international author unique identification schema. From a knowledge discovery viewpoint there are however some problems in this commercial-stakeholder approach: the ResearcherID, ThomsonReuter's author identifier, will provide seamless integration with ISI Web of Knowledge and show all author publications registered in that database, but will otherwise leave out most of the research output.

Some joint effort between public institutions and private stakeholders (remarkably publishers) must therefore be attempted to unify the multiple author identification standards and devise a single, comprehensive one at a global level. And that's where ORCID comes in.

  
... and strategies to tackle it: IRISC 2011 workshop

The Open Researcher & Contributor ID (ORCID) initiative started in Dec 2009 as a non-profit organisation. Currently over 240 participants have joined the project for developing the one research identifier which is not limited to discipline, institution or geographical area. Many other projects are working in this issue at the same time (such as abovementioned discipline-based PubMed Author ID and Cornell University initially institutional then grown to national VIVO initiative).

ORCID and VIVO were two of the main topics of the IRISC 2011 Workshop on Identity in Research Infrastructure and Scientific Communication held this week (Sep 12-13) in Helsinki - see the event programme with attached presentations. Gudmundur "Mummi" Thorisson, Research Associate at University of Leicester and member of ORCID Technical Working Group, was IRISC 2011 main organizer.

There were two major IRISC 2011 strands: identity regarding knowledge discovery and identity for security & access control (focusing mainly on identity federation). A third big cross-issue along the Helsinki event was research data management, from three different perspectives:

i) dealing with a rapidly increasing amount of biomedical research data (Andrew Lyall, EMBL, ELIXIR Project)

ii) dealing with clinical research sensitive data (see Tony Brookes GEN2PHEN Project presentation)

iii) benefits the ORCID implementation might bring to research data attribution and management (mentioned in most ORCID-related presentations and discussions along the workshop)


There were several presentations dealing both with ORCID and closely resembling VIVO initiatives. Martin Fenner, Hannover Medical School and member of ORCID Board of Directors announced the ORCID registration service will start operating in spring 2012. ORCID will be open: researchers will be able to manage & maintain their profiles, filed data will be openly available, ORCID-related software will be released as open source, and researchers will control their privacy settings (with a chance too to share with particular members). Finally, for ORCID identity definition purposes, self-claim as well as external claiming sources will be used.

Brian Lowe, University of Cornell, presented the already running NIH-funded, institutionally-managed VIVO initiative. VIVO is aiming for an extensible semantic model-based more comprehensive approach than ORCID. However, links have already been established between both initiatives and ORCID is hoping to build upon VIVO success in the US.


Breakout sessions were held on IRISC Day 2 on the workshop's two main strands: "Unique identifiers and the Digital Scholar" (lead by Cameron Neylon and Jason Priem) and "What do researchers need from the authentication and authorisation infrastructure (AAI)?" (chaired by Michael Linden, CSC). Breakout session #1 was devoted to discussing potential tools and services to researchers ORCID could provide in the short term (6 months from adoption). Several groups were set up for the purpose and proposed ideas were later voted and discussed for selecting three main future worklines for ORCID to deal with. The proposed and selected use cases were the following:

-> data submission to repositories (multiple task attribution)

service to enable attribution or comment

pre-populate ORCID data

-> manuscript/grant tracking system

ORCID app gallery

-> automatic CV maintenance (potentially including data citations in CVs)

connecting different author research & social network profiles

Selected ORCID use cases were later introduced by Cameron Naylon along his talk 'ORCID and researchers' at the second annual ORCID Outreach Meeting held at CERN on Sep 16th, 2011.

Research data management in crystallography at the XXII IUCr Congress

  On Aug 29th a session on research data management will be held at the XXII Congress of the International Union of Crystallography (IUCr2011). The session will feature talks by Brian McMahon (IuCr), Brian Matthews (I2S2 Project), Peter Murray-Rust (CrystalEye), John Westbrook (wwPDB) and Nick Spadaccini (DDLm). Peter Murray-Rust will deliver a talk along the session on Open Crystallography.

STM research data management and the Quixote Project

  A one-day seminar was held yesterday Thu Aug 25th at the Zaragoza Scientific Center for Advanced Modeling (ZCAM) on research data management and the Quixote Project for data management in Computational Chemistry. The session, entitled “Research data management: The experience of the Quixote project for Quantum Chemistry data. Can it be extended into a collection of research data management repositories?”, was attended by a rather diverse group of researchers (both computational chemists and from other disciplines) and repository managers, aiming to learn about research data management initiatives and specifically about the progress of the Quixote Project, in which two researchers from the University of Zaragoza and the CSIC Institute of Physical Chemistry "Rocasolano" are involved.

The Quixote Project (see paper "The Quixote project: Collaborative and Open Quantum Chemistry data management in the Internet age", in press with the J Chem Inf) is developing the infrastructure required to convert output from a number of different molecular quantum chemistry (QC) packages -such as NWChem or Gaussian- to a common semantically rich, machine-readable format and to build repositories of QC data results.

The session started with an introduction to "STM Research data management initiatives in Spain and abroad" delivered by SONEX member Pablo de Castro, in which different national approaches to RDM were presented based mainly on the information collected at the JISC MRD Programme International Workshop held last March in Birmingham.

Different approaches to data management taken from the JISC and SURF Foundation were discussed at Q&A time: for the JISC, datasets are assets per se, regardless of where they are attached to a research paper as supplementary material, whereas the 'Enhanced publication' approach from the SURF Foundation in the Netherlands, regards datasets mainly as digital objects connected to research publications. Some emphasis was made on the fact that the upcoming OpenAIREPlus European project shares the SURF approach.

Two presentations on the Quixote Project followed, "From Databases in QC 2010, ZCAM, Sep 2010 onwards: a brief history of Quixote" by Jorge Estrada and "The Quixote Project: a pioneering work in managing Computational Chemistry research data" by Pablo Echenique. Both Quixote project members explained the results, the challenges and the cooperation opportunities of this non-specifically-funded RDM project, engaging in a fruitful dialogue with the attending researchers and repository managers on how the QC data assets could be best managed.

Finally Peter Murray-Rust closed the morning interventions with some reflections on the subject "Entering a new era in data management" - see his blogpost for a summary of his ideas.


In the afternoon there were joint debates on how to improve implementation of research data management initiatives. Researcher motivation for dataset sharing was extensively debated: this motivation should ideally not just arise from a given funding agency actually requiring those data to be made available, but from the sheer advantages (as summarized by Peter Murray-Rust) that doing so would bring to the research practice and communication ("improving methodology").

An independent debate session was held for discussing how to start developing some kind of research data management infrastructure in those countries where work in this area is presently beginning. These are some recommendations that were put together by the participants in the debate:

- Some workgroup of (not just library-based) IT professionals should be put together for analysing the current infrastructure and the opportunities for launching new initiatives upon potentially reusable pre-existing ones,

- It would be advisable to analyze the researcher behaviour and needs in terms of storing their datasets into international platforms for data sharing (in case they are available for their specific discipline),

- It would be interesting to examine the motivation for data sharing from research groups in different research areas, so that initial efforts to develop data management infrastructures can start working with those areas more willing to share their data (Earth Sciences recurrently showing up when analysing the international perspective),

- Pioneering initiatives for providing services to STM researchers regarding data handling and storing from given Institutional Repositories (such as eSpacio UNED and Digital.CSIC) should be highlighted as a role model to be spread,

- The OpenAIREPlus/SURF Enhanced papers approach could be a good starting point for Institutional Repositories to work at, by finding out which of their presently filed papers have supplementary data attached at the journal site and trying to independently manage those ones,

- A need was detected along the session talks with researchers for a dataset management system at research centres for basic internal organisation purposes. Datasets filed in this internal storage system may or may not be aimed for publication,

- Production and publication of potentially citable datasets should be acknowledged as a relevant scientific contribution for research assessment purposes,

- There are big differences in needs, procedures and required infrastructure regarding data management between Big Science and long-tail science (the greater part actually being groups of three researchers in a lab with specific needs of their own),

- The Library is a potential supplier of know-how on data processing and storing for researchers, and that role should be promoted within the institutions,

- The Spanish e-Research National Network, mostly dealing with Grid and supercomputing initiatives, might be a good workgroup infrastructure for pioneering data management initiatives in Spain,

- There are real collaboration opportunities between the Quixote Project and the research information management infrastruture at the University of Zaragoza (two IRs being currently available, Zaguan at the University and Digital.CSIC at the Spanish Nacional Research Council, CSIC),

- Research staff (mainly PhD students) getting involved in the management and operation of the dataset information management systems (such as Chempound data repository at the University of Cambridge) seems a prerequisite for the success of the data management initiatives

- Due to the specific data features for various research areas, the incipient data management infrastructure available is more developed for the Social Sciences and Humanities than for STM research areas.

A first analysis of data management

As previously mentioned in this blog, the Sonex workgroup is now try to extend its use case scenario analysis on 'Deposit opportunities into repositories' to the realm of research data. A first meeting held at EDINA on Mar 30th served the purpose of drawing a general picture of the data management landscape.

Stress should be put on the fact that the way of handling SSH and STM data may substantially differ. Given the strong IASSIST-attachment of some Sonex members, the workgroup initial approach to data management may therefore be a bit biased towards procedures in the area of Social Science and Humanities. However, attention will be paid as well to specific ways of dealing with STM datasets as the analysis gets fine-tuned.

Moving along the same lines as we did for research articles, we first try to tackle the ACTIONS scope. Data deposit is certainly an issue, but there's more to data-related processes than just deposit. It's also about Access to data and also about Data Notification/Register.

Next we get on to the WHAT and the WHO. Answer to WHAT? is a data set. Previous analysis by Peter Burnhill shows -at least- three different types of research data (see image below).


Dealing mainly with the data file itself, this data type classification is somewhat narrow for the general picture of data management, so Sonex would rather set a new and more generic data classification for answering the question WHAT is there to deposit:

• Metadata record


• Codebook or user guide, where all necessary information is provided to allow for data re-use*


• Raw data or dataset file(s)

* See a DCMI-based description at: Inter-university Consortium for Political and Social Research (ICPSR). (2009). Guide to Social Science Data Preparation and Archiving: Best Practice Throughout the Data Life Cycle (4th ed.). Ann Arbor, MI. Section 'Important documentation elements', p. 22

These three elements should ideally be supplied as a single package.

As to the question of WHO performs each data-related operation (Notification-Deposit-Grant access), a handful of running projects within the JISC MRD (phase I) programme should serve to test the different use cases resulting from a double-entry 'Action/What' table as featured below.


Next step as we proceed to further development of this preliminary analysis should be a survey for gathering information on procedures for data handling as carried out in specific JISC MRD projects.

Could external cooperation improve collection of specific JISC MRD project-related information?

  In forthcoming days SONEX will be publishing some posts on the JISC MRD Programme International Workshop held last March 28-29th at Aston Business School Conference Centre, Birmingham. Certain aspects debated at this comprehensive meeting were very useful for establishing an approach for dealing with research data management from a SONEX viewpoint, as debated in a SONEX meeting at EDINA on Mar 30th whose outcome will also be shortly blogged.

See IUCr Brian McMahon's report for a general review on the JISC MRD workshop.

One of the most visible disciplinary approaches to data management presented at the JISC MRD event -which featured all kinds of institutional and subject-based initiatives in the area- was the one coming from meteorology, palaeoclimatology and climate-related sciences: there was a presentation of the PEG-BOARD Project (U of Bristol) at the Subject-Oriented Approaches session on Monday, followed by ACRID (U of East Anglia & STFC) and Metafor (BADC & STFC) Project presentations on Tuesday afternoon.

One of the most relevant features of these climate-related projects is interdisciplinarity. PEG-BOARD Project in particular aims to serve the archaeology research community by supplying them their paleoclimate data.


A few specific aspects about PEG-BOARD were discussed after the project presentation. Interesting thing about them is they were not mentioned along the talk, nor are they reported at the project site:

- Due to the project interdisciplinarity, there are two clearly different user groups for palaeoclimatology data produced: climatologists, who will understand the nature of involved datasets, as they're central to their discipline, and archaeologists, who don't and need not know much about the data format but need the information contained in it for their own purposes - thus functioning as regular non-technical users to the project instead of researchers. However, as they are indeed researchers, the feedback they may provide on the project outcome could be so much more valuable.

- What archaeologists care about in the end is the data plottings, and Data Centres will not provide such processing. So what PEG did was implement specific software capabilities that will address the needs of non-technical data users (i.e. archaeologists), as to allow them to search for the plots or false-colour graphics they need. This piece of middleware is a conceptual key feature of the project in terms of deliverables.

- Climate data is usually archived in binary format, so it's often not easy to process. UK Met Office provided lots of info, often incomplete or in old formats. The adaption process of raw data to the project needs was very interesting and worth disseminating.

- Climate models were written in FORTRAN. When re-written or translated into C++, the results would vary for the same data arrays due to specific treatment by the code. That poses a quite amazing challenge in terms of model interpretation.

- When asked on whether researchers provided enriched metadata for their data, the answer was there's usually an input in terms of past experiments, i.e. "this is the data outcome of such and such experiment when changing initial conditions in such a way". Such-and-such experiment would be described the same way until one was reached that wasn't described at all.

The fact that none of these project aspects is recorded or discussed at the project blog poses a question on whether an external approach to data management projects might collect and disseminate very interesting information that researchers may not consider relevant enough to discuss from project blogs. Such an external approach to running projects might be carried out by data librarians in order to
share these specific project details with the data management community.

For whatever it may be worth, Sonex would be keen to do this kind of job for the MRD community.

I2S2 Project workshop at RAL-STFC

  Along a busy week in terms of research data management events (due to be shortly reported from this blog), last Friday Apr 1st Sonex had the opportunity -thanks to Simon Hodson, JISC MRD programme manager- to attend the I2S2 Project workshop at the Rutherford-Appleton Laboratory (RAL) at STFC in Didcot. I2S2 -standing for 'Infrastructure for Integration in Structural Sciences' is a JISC MRD project ending in Mar 2011 aiming to "identify requirements for a data-driven research infrastructure in "Structural Science", focusing on the domain of Chemistry, but with a view towards inter-disciplinary application".


Several presentations were delivered along the meeting: Brian Matthews on the I2S2 project achievements, ICAT architecture and CSMD metadata standard, Brian McMahon, International Union of Crystallography (IUCr) on 'Information Management and Publication in Crystallography', Tom Griffin on TopCAT GUI for management of data coming out of STFC ISIS and DIAMOND facilities, Steve Androulakis on the TARDIS ANDS-supported project at Monash University, Mark Borkum on OreCHEM files, Chris Morris on on PiMS (Protein Information Management System) and Juan Bicarregui on the EU PANData project.

Along the IUCr presentation the need was identified for filing & preserving different data categories such as raw measurements, processed numerical data, derived info and the paremeters. The convenience of providing access to raw diffraction images was also stressed along the talk, these files being a few GB in size, and thus not large enough for Data Centres but too big for sites such as CCDC. A review on Crystallographic Information Framework (CIF) file formats was provided, with imgCIF being used for raw data storing out of the experiment, .fcf for including structure factors after data reduction and a final stage of structure solution and refinement being performed in the lab before the author starts formatting those into a IUCr paper, which would translate CIF into SGML for producing final fcf, cif, pdf and html versions.

Raw data was mentioned to be kept for 183 days at SFTC and 3 months at Australian Synchrotron (in which TARDIS is involved), and a discussion followed on the fact that some agreement shoud be reached on the kind of data that ought to be stored and preserved. The process of attachment of DOIs to datasets was also discussed, IUCr being presently involved in projects such as XYZ or Open Bibliography in order to promote this objective.


A TopCAT demo was provided by Tom Griffin. This open source GUI (see image above) is being used for storing raw data from STFC facilities such as ISIS and DIAMOND. TopCAT provides access to its contents through an open registration system, thus operating as a sort of STFC institutional data repository, and would be potentially applicable to other institutions, facilities and disciplines.

TARDIS presentation by Steve Androulakis, Monash Univ, Australia, mentioned their using of XML/METS metadata standards for research data description at the federated institutional repository-platform initially meant to store X-ray diffraction images, later evolving into a much larger initiative with application into microscopy (MicroTARDIS), particle physics and gene processing through the Squirrel software.

Finally, extra presentations were delivered on PiMS (Protein Information Management System) by Chris Morris, STFC and on the European PANData project by Juan Bicarregui, STFC e-Science. PANData aims to build Photon and Neutron Data Infrastructure through a consortium of European synchrotron facilities and neutron sources.


A final summary was made on the whole set of presented I2S2-related features (imgCIF, CIF, IuCr/XML/RDF BIBLIO, PDBML, CML, ICAT, TopCAT, ICAT Lite/CSMD, TARDIS, PiMS, PANData, NeXuS) by mapping them on the I2S2 Idealized Scientific Research Activity Lifecycle Model (see image above - may click on it for an updated version). References were also made to other initiatives not represented at the meeting such as Quixote Project for Computational Chemistry CML data management or Protein Production and Crystallization.

Strategies for research data deposit in ongoing data management projects

  Prior to start performing pattern analysis for research data deposit into (institutional or subject-based) data repositories –whether or not open access– first step by Sonex is to scope ongoing projects dealing with that kind of deposit, as well as already closed projects which supplied relevant guidelines on the subject. A list of projects working on data management follows, with their specific approach on how to deal with actual data deposit as taken from project blogs:


TARDIS (Monash University–Australian National Data Service).
“There is a pressing need for the archival and curation of raw X-ray diffraction data. However, the relatively large size of these datasets has presented challenges for storage in a single worldwide repository. This problem can be avoided by using a federated approach, where each institution or university utilizes its institutional repository”.


ADMIRAL: A JISC-funded data management infrastructure for research across the life sciences.
"The purpose of the ADMIRAL Project is to create a two-tier federated data management infrastructure for use by life science researchers, that will provide services (a) to meet their local data management needs for the collection, digital organization, metadata annotation and controlled sharing of biological datasets; and (b) to provide an easy and secure route for archiving annotated datasets to an institutional repository, The Oxford University Data Store, for long-term preservation and access, complete with assigned Digital Object Identifiers and Creative Commons open access licences".
(See Oxford University Library Services' Databank)


XYZ Project. “The XYZ Project will create a demonstrator of a new workflow for publishing data in support of full-text. The author prepares data for publication (if possible with validation) in a third-party trusted repository before the paper is submitted to a publisher. Our software will manage the deposition, release to reviewers, dis-embargo and for conventional publication or as a data journal. Two Open Access publishers (International Union of Crystallography and BioMed Central) are engaged with the project and will test the new workflow”.
Anticipated Outputs and Outcomes: A demonstrator repository hosted by the IUCr.


FISHnet: Freshwater information sharing network. “This project will allow researchers in multiple academic, governmental and voluntary-sector institutions to share their data. Data will be held securely in a sustainable subject repository which preserves and disseminates multiple datasets as part of the FreshwaterLife.org information portal. Data creators will be able to manage access rights to their content, from Open Access to sharing with trusted colleagues”.


DMBI: Data Management in Bio-Imaging. “The quantity of data generated by modern high-throughput bio-imaging systems presents a significant challenge in both data management and processing. Furthermore, there is no explicit system/way to record the processing algorithms and parameters that are used to produce results. Thus there is no strong link between images, software and results. This projects aims to address these issues”.
Anticipated Outputs and Outcomes: Build a prototype DMBI system around OMERO.


CaiRO: Curating Artistic Research Output. “No prominent subject-based repository exists to act as the custodians of arts practice-as-research data. Where institution provision for data management is in place (for instance, an institutional repository service) the arts researcher-practitioner cannot always rely on an understanding of the special nature of arts research data. More commonly, data is retained in departmental collections, built and maintained by small teams which often include researchers themselves”.


BRIL: Biophysical Repositories in the Lab. “The BRIL project aims to enhance the repository facilities at the Randall Division of Cell and Molecular Biophysics at King’s College London. This will involve:
» Embedding the repository within the researchers’ day-to-day research and experimental practices;
» Integrating the repository into the wider King’s infrastructure”.
Example of KCL “internal” repository: Mutation Testing Repository.


ADS+: Enhancing and Sustaining the Archaeology Data Service digital repository. The project aims to “Increase the sustainability of the ADS, by implementing Fedora (Flexible Extensible Digital Object Repository Architecture). This is a world-leading open source digital repository application which will allow the automation of many ADS curatorial functions, according to the Open Archival Information System (OAIS) Reference Model (ISO 14721:2003). This will help ensure the long term preservation of all ADS digital archives, as well as making the ADS archival procedures more cost-effective”.


IDMB (Institutional Data Management Blueprint) Project, U. Southampton.
The project’s aims are to provide the University of Southampton with a ten-year roadmap for delivery of a comprehensive data management infrastructure.

[IDMB Recommendations] The data management audit and gap analysis indicates where improvements can be made in the short, medium and long-term to improve data management practices and capabilities at the University. The following preliminary recommendations are put forward for short (one year), medium (one to three years), long (more than three years) term action.
[Short Term (1 year)] Crucial to supporting researchers is the consolidation of data management into a coherent framework that is easy to understand, use, and has a sustainable business model behind it. A number of major recommendations are put forward here for the short-term:
• Create an institutional data repository
• Develop a scalable business model
• One-stop shop for data management advice and guidance


MaDAM: Pilot data management infrastructure for biomedical researchers at University of Manchester.
A pilot infrastructure for Biomedical Researchers at the University of Manchester, which covers data capture, data storage and data curation. This infrastructure comprises procedural support, hardware and software.
[18/03/2010] The development team have built a prototype data management front end which fits a generic set of needs amongst our Life Sciences researchers. It is aimed at being flexible enough to allow researchers themselves to assign attributes (i.e. metadata) to their experiments and datasets for them to be usefully categorised and tagged. The prototype is also entirely dispensable and intended as a catalyst for feedback from our use cases on their specific functionality requirements.


DISC-UK DataShare Project. The DISC-UK DataShare project, led by EDINA National Data Centre and the Edinburgh University Data Library, with partners at the Universities of Southampton and Oxford, has advanced the current provision of repository services for accommodating datasets in the UK.
Key conclusions: 1) Data management motivation is a better bottom-up driver for researchers than data sharing but is not sufficient to create culture change, 2) Data librarians, data managers and data scientists can help bridge communication between repository managers & researchers, 3) Institutional repositories can improve impact of sharing data over the internet.

Repository take-up and embedding: the future of repositories

  Being already in Birmingham for the JISC Deposit Project Meeting on Mar 1st, Sonex stayed in town for attending the JISC Repositories Take-Up and Embedding Meeting as well. Start up meeting for this new JISC programme aimed to outline the future of repositories, dealing with specific issues such as (automated) deposit, shared services like RoMEO or OpenDOAR, repository integration into general software infrastructures for research information managament and promoting national (via RSP) and international (via KE, COAR and OpenAIRE) collaboration.

Six projects were presented along this programme start up meeting:

- Bringing a Buzz to NECTAR (Miggie Pickton, University of Northampton)
- Hydrangea: letting the repository flower (Richard Green, University of Hull)
- MIRAGE 2011: Repository Enrichment from Archiving to Creation (Xiaohong Gao, Middlesex University)
- Enhanced interface design for supporting take-up and embedding of the Glasgow School of Art research repository, including visual
engagement with practice led and applied outputs (Robin Burgess, Glasgow School of Art)
- eNova (Marie-Therese Gramstadt, VADS)
- EXPLORER: Embedding eXisting & Propriatary Learning in an Open-source Repository to Evolve new Resources (Alan Cope, De Montfort University)

An extra postprandial presentation on repository consolidation within a university research information management environment and the way it was done at University of Glasgow Enlighten IR was delivered by Willian Nixon. Statements like "Silos are the past, embedding repositories -through the use of tools like Sword or LDAP- is the future" made the point on how repositories should evolve in the future. According to William, repositories are to exploit new opportunities for data mining, business, intelligence, KPIs, analytics, 'stickiness' and visibility (some of these issues being thoroughly dealt with at Enlighten repository blog).

There was a remarkable presence of image-related projects among the presentations, Glasgow School of Arts, eNova and MIRAGE 2011 dealing with archiving of images into repositories one way or another. This is great news for momentum-gaining development of new information infrastructures in the area (also traceable at the JISC Deposit Programme meeting the day before), which will no doubt benefit from these projects outcomes.

After watching project presentations from a Sonex point of view, it seems they could particularly benefit from interacting with JISC Deposit projects in terms of implementing resulting strategies for automated content ingest into repositories. A handful of the take-up and embedding projects would thus be the soundest candidates for initial "customer implementation" of the various resulting methods for quick population of repositories with institutional research output (the take-up bit, prior to embedding) coming from the Deposit strand. As these projects will run
until the end of 2011 and the ones from Deposit strand should deliver around July, interaction among them could probably be easily achieved.

There was one particular project among those presented that captured Sonex's attention: MIRAGE 2011, Middlesex Medical Image Repository with a Content-Based Image Retrieval Systems Archiving Environment. MIRAGE is both an image-related repository project (as it deals with medical images) and a research data project, and it's this latter feature what gets it fully within scope of Sonex activity with regard to research data management. Ongoing data management projects (either JISC-funded or otherwise) usually deal with either numerical or textual data, but projects dealing with the deposit of graphical research data are rare (save for Data Management in Bio-Imaging - DMBI project run at The John Innes Centre, BBSRC, Norwich).

A couple of references were shared with MIRAGE project manager Dr. Xiaohong Gao, 'Feeding Neuroimaging Repositories' poster presented at OR2010 Madrid last July by a team of Universitat Autònoma de Barcelona (UAB)-Hospital de la Santa Creu i Sant Pau researchers in Barcelona, and the MIDAS/National Alliance for Medical Image Computing (NAMIC) medical image repository as to promote synergies among different projects on the same area.

The meeting presentations will shortly be available.