On 18^th-19^th May, King’s College London hosted the third in a series of DReSNet workshops on policies in repositories. The workshop was attended by members of the FP7-funded project SHAMAN, colleagues from King’s and from the University of Amsterdam.  The workshop focused on understanding the derivation of processes (implemented as software) from policies.

The SHAMAN derivation procedure was used as the starting point for the discussion. The procedure takes the hypothesis that each process must be able to find a home within a policy, that is each process must be uniquely associated with a policy statement. The procedure starts with a derivation of detailed policy statements from the policy document. Each statement in the document is studied and the steps necessary to achieve the statement are documented. The dependencies on other statements are also described, as are the assumptions made in defining the steps. The steps needed to achieve the policy statement are then transformed into high-level rules (we have selected the W3C standard Rule Interchange Format, RIF, as the high-level language). Again, any assumptions made in the transformation need to be documented. The high-level rules can then be translated into low-level, implementable rules (such as iRODS rules) or processes. For many, the distinction between the high-level rule and the detailed policy is blurred, as much of the work is carried out while making the policies themselves more detailed, so that they can serve as the basis for formalisation.

The high-level rule is just a formalised form of the policy statement. However, the fact that the choice of high-level rule format (currently the RIF) may change in the future suggests that it is prudent for the time-being to keep the two steps separate. But, the procedure has been modified to take account of the optional nature of the step (see Policy Refinement Process).

Policy Refinement Process

The workshop found that that the SHAMAN procedure missed out some important information. The fact that policies or sub-policies may exist within the organisation, or that workflows may exist within the organisation were not taken into account. The impact on the procedure was to introduce the issue of conflicting rules or workflows in which a rule derived from the derivation process may conflict with an existing rule. Possible reasons for the conflict could be that the existing rule corresponds to multiple policy statements, or covers a partial statement. In this case a procedure to resolve the conflict would need to be established. The procedure to resolve the conflict would necessarily be manual as this would involve understanding the nature of the conflict and the assumptions made in creating the rules. In some cases the conflict resolution may be made more difficult due to an ‘orphaned’ existing rule where an understanding of how the rule was derived does not exist. Although conflict resolution may be manual the identification of  conflicts can be almost totally automated.

Conflicts at the policy level arise when new policies are created. An understanding of how they relate to existing policies and the impact on existing policies would need to be assessed. It can be argued the creation of the new policy would already involve an understanding of the impact on existing policies. The new policy may require changes to existing policies that would have a ‘trickle-down’ effect at all levels. The refined diagram is shown in Policy System Data Model.

Policy System Data Model

An interesting point was raised during the workshop that all policies are inherited. That is all policies are really sub-policies as they are all derived from some existing policy (for example, we ultimately operate within the law and the law is a policy that describes how society interacts). This is a refinement of the original identification of external policies that influence an organisation’s policies.

The workshop then took a couple of example policy statements from a JISC-funded project (link??????) to test the derivation procedure. The tricky part of the procedure was in understanding what assumptions had been made during the derivation. It was helpful that members of the workshop were not experts on the project and were able to ask questions on why certain choices were made in the description of the workflow. In general the derivation of rules from policies may be the result of a small group of individuals from within the organisation which may make it difficult to distinguish common, current organisational practices from standard practices. Perhaps the best rule of thumb is to justify everything to ensure longer-term sustainability. The workshop also thought it useful to review the derivation process on a yearly basis to update the assumptions to ensure they are still understandable.

The next steps for the group are to continue to take more example policies and apply the procedure to generate processes in order to
further work-harden the procedure as well as to continue to understand the conflict identification and resolution procedure.

The PI and Co-I of DReSNet – Mark Hedges and Tobias Blanke of the Centre for e-Research at King’s College London – have won best paper award at IEEE eScience 2010 in Brisbane, the prime international conference in the field of e-Science and e-Research.

The paper, ‘Humanities e-Science: From systematic investigations to institutional infrastructures’, reflects much of the joint work on Arts and Humanities e-Science at CeRch and shows how e-Science tools and methodologies can be embedded in institutional infrastructures. Much of this work grew out of collaborations and partnerships engendered by DReSNet, and demonstrates the value that this network has achieved.

DReSNet member Mark Hedges visited Australia to attend IEEE eScience 2010, and at the same time took a visit to VeRSI, the Victorian eResearch Strategic Initiative, which is directed by Dr Ann Borda, whom some of you may remember as a former JISC programme manager. Here he had a series of meetings with the team to exchange information, and in particular spoke about recent developments in the EU ESFRI project DARIAH.

After leaving Melbourne, Mark travelled to IEE eScience in Brisbane. Here he spoke at the the workshop on “Exploring the Legal and Policy Aspects of Accessing and Making Use of Scientific Knowledge and Information” about the work being carried out by DReSNet’s repository policies working group, including research at the Centre for e-Research at King’s College London, and work undertaken by the EU SHAMAN project.

He also gave a paper on “Humanities e-Science: From systematic investigations to institutional infrastructures – a status quo report“, which describes work carried out by Mark and his colleague (and DReSNet Co-I) Tobias Blanke on “scholarly primitives” and their approach to humanities eScience at KCL. The paper is a timely follow-on from one they presented at IEEE eScience 2007 in Bangalore.

DReSNet will be attending IEEE eScience 2010 in Brisbane this year. Mark Hedges will attend the workshop on “Exploring the Legal and Policy Aspects of Accessing and Making Use of Scientific Knowledge and Information“, presenting the work of the DReSNet policies working group. He will also be presenting a paper on “Humanities e-Science: From systematic investigations to institutional infrastructures – a status quo report“, which describes the approach to humanities eScience and repositories taken by the DReSNet partners at KCL.

DReSNet organised an invitation-only workshop on 10 September 2010 in Glasgow, co-located with ECDL 2010.

The working title was “From repositories to virtual research environments”, and the workshop was focused on discussion rather than presentations, with the intention of producing a “white paper”. A draft will be released shortly – watch this space!

The workshop attendees – plus others that were unable to make it – will form another DReSNet working group addressing issues around repositories and VREs.

What is citizen cyberscience?

“Citizen cyberscience is a collective term for a diverse, grass-roots movement that is enabling ordinary citizens to participate in real scientific research thanks to the Web. Practically anyone with an Internet connection can join: schoolchildren, office workers, pensioners. Using PCs, laptops and even mobile phones, volunteers can classify images of distant galaxies or track the migration patterns of endangered species, to name just two examples. Citizen cyberscience is social networking with a purpose.” – Francois Grey, CERN and Tsinghua University in Beijing.

DReSNet participated in the Citizen Cyberscience Summit held at King’s College London on 2-3 September 2010. This event was an opportunity for scientists and the public to learn about the state of the art in citizen cyberscience, and to contribute to a debate about the future of citizen cyberscience. One (forthcoming) result of the summit will be a draft citizen cyberscience manifesto.

Citizen cyberscience approaches have great potential for the development and enhancement of data repositories across disciplines, in the sciences and in the humanities. DReSNet was one of the sponsors of the event, and DReSNet participant Mark Hedges introduced the Summit, and gave a presentation on the use of “citizen cyberscience” approaches for sustaining digital archives.

More details on the Citizen Cyberscience Centre may be found here.

The Web is ushering in a new age of public participation in science. Projects such as SETI@home searching for extraterrestrial intelligence, LHC@home simulating CERN’s Large Hadron Collider, and GalaxyZoo cataloguing millions of astronomical images, have demonstrated how thousands of dedicated volunteers can make a big difference to fundamental science.

Regional initiatives involving CERN, the University of Geneva and UN organizations have exploited this trend. The short-term projects Africa@home and Asia@home trained scientists in developing countries to use internet-based volunteer support in order to address pressing health and development challenges. These initiatives highlighted the benefits of closer collaboration between scientific, humanitarian and academic partners, both North and South. To build on this success, a sustained effort is required, in the form of a Citizen Cyberscience Centre.

It was decided to set up a working group looking at policies in digital repositories, starting with (some of) the participants at the Madrid workshop. The working group will hold a series of meetings and workshops, probably co-located with international conferences. The working group will be organised by Adil Hasan and Mark Hedges. watch this space for more information!

Interoperability and Policies Workshop – 9 July 2010, Open Repositories 2010, Madrid

by Adil Hasan

Introductory slides for policies workshop, madrid

Attendees at the workshop ranged from people interested in understanding the impact of interoperability and policies when creating repositories to people actually involved in projects aimed at exploring the issues of interoperability and policies on repositories.

The workshop started with a discussion on interoperability. The discussion identified a number of different types of interoperability: the service, data and repository. Interoperation of each type is governed by policies.

The workshop then looked at a few examples of repositories and interoperability. The scientific domain example pointed out the need for the repository to be part of the scientific process as domain expertise was needed to ensure the data were interoperable (ensuring metadata information was kept up to date, terminology adequately described, etc). Which also highlighted the need for funding for ‘scientific curators’ or custodians with domain knowledge that ensure the domain knowledge relevant to the data endures.

There was also an example of event-driven interoperability in the case of an emergency response where access levels change depending on the level of emergency (for example in the case of an epedemic or natural disaster access to certain types of information may become available to users that do not normally have access to that type ot data).

An example of a repository that holds composite content in two different locations highlighted the need for policies on interoperability that do not exist within either repository. For example, if the content consists of HTML pages with images with the HTML stored in one location and the images stored in another a policy governing the migration of images from one format to another would also have to include an update of the HTML to point to the new format image. Other examples were given where sharing of data may conflict local repository policies. In some cases a resolution of this situation can be further complicated by privacy issues concerning the policy (external repositories or services may need to request special permission to view the policy to understand and resolve any conflicts).

The workshop then moved to a discussion on policies and described a high-level process on how rules/workflows can be derived from policies. The problem of mapping the policy to the rule could be reduced by iteratively refining the policy such that it maps to the model (preservation model) unambigiously. The assumptions used in the refinement should be recorded as should the updated policy. The process does not stop once a policy has been mapped to a rule. Updates to a policy would require a re-mapping to rules in order to understand the impact of the change. The process of refining the rule would hopefully mean that any change would not necessitate a full re-mapping, but only a partial re-map of the related sub-policies.

There was also a growing understanding that policies are not as explicit in existing repositories and perhaps a ‘bottom-up’ approach is needed where the services that form part of the repository are characterised to capture their function. At this point it should be possible to replace an existing service with a new service, or outsource a service. These characterisations can be used to form the policies for the repository. There was also a discussion of this approach in principle providing an ontology of services (in some respects similar to the UDDI).

A need to define a format for rules was also recognised as essential. There are currently a number of different types: Rule Interchange Format, SWRL, RuleML etc and an investigation of which ‘language’ was best to express rules is needed.

The workshop participants were very interested in a perpetuation of the group complete with mailing list and web pages and further workshops as there was a recognition of a gap that the workshop seemed to be filling.

DReSNet is sponsoring a workshop on policies in digital repositories at Open Repositories 2010 in Madrid on 9 July 2010.

http://or2010.fecyt.es/Publico/WorkShop/index.aspx#b4