This next phase sees us widening the community, which will now include HPC specialists, clusters, large scale data storage, middleware and applications programmers, across 8 universities and CRIs, namely Auckland, Victoria, Otago, Massey, Canterbury, Lincoln, Landcare, GNS Science, and the NZ Supercomputing Centre / InterGrid.
Our primary deliverables focus on building infrastructure for specific groups in the Bio and Geo sciences, while maturing our infrastructure, offering training, and extending the reach and depth of the community. I'll post detailed information on the project based on the successful proposal to MoRST onto the BeSTGRID wiki in the next couple of weeks.
Richard Li from the Centre for Software Innovation will be joining us from Monday as project manager for this project, which starts next week and runs through to end June 2010.
We'll be happy to discuss with interested groups once we're up and running.
This survey is the most publicly visible sign yet of a process that has been ongoing for at least 3 years, and is aiming to establish a national eResearch and HPC capability over an initial 10 year period supported by the MoRST established Research Infrastructure Advisory Group (RIAG) process.
For those unaware of this group, RIAG was established after funding was secured for the national research and education network KAREN and the New Zealand investment in the Australian Synchrotron facility. The mandate given to RIAG was to identify and prescreen similar significant research infrastructure investments, through receiving, screening, and advising MoRST on the science merits of any significant proposal.
Once constituted, RIAG carried out a national survey for research infrastructure needs, and reported back to MoRST on their findings in their "Scan of New Zealand's Large Scale Research Infrastructure Needs 2007-2012 report". On reading this report it becomes clear that a significant proportion of the investments identified have an eResearch or HPC component or dependence.
Meanwhile other initiatives have laid the foundations on which a national eResearch and HPC capabiliity can be built. The national research and education network KAREN, the initially TEC supported and now MoRST funded BeSTGRID national grid computing initiative, the significant investments in HPC capabilities at BlueFern, NIWA (arguably the first HPC in NZ), and the New Zealand Supercomputing Centre and InterGrid, have all contributed to a vibrant and growing community with a broad range of interests. There is a strong collaborative and collegial spirit within and between these groups, which is a pleasure to experience and is crucial to foster as we work towards developing a national capability.
The proposal being developed is being informed by a broad section of the community most active in eResearch and HPC. MoRST have run workshops to assess need, and are supporting the process directly. This survey is the latest initiative from MoRST to reach out to the community and ensure any such proposal represents the needs of the wider community.
Hopefully there is a strong response to this survey, as it will provide crucial evidence of the communities needs and preferences to MoRST, Treasury, and others involved in and consulted on this proposal. It is likely to be the most direct opportunity for adding your opinion to the conversation currently underway.
Thanks to Kathleen Logan and Julie Watson for pulling the survey together, and for opening up the discussion to receive the opinions of the wider community.
High Performance Computing survey
The speakers at the conference were excellent, with conference chair Patricia McMillan noting several comments suggesting there wasn't a single timeslot that didn't have compelling presentations. I'd wholly agree. The highlights for me were as usual in the exemplars of research being done in novel and innovative ways. There were many talks addressing the data capture and semantic discoverability issues in research, with several approaches worth further investigation.
John Wilbanks spoke of his work with neurocommons.org, which seeks to make scientific work as accessible and useful as possible through using the latest semantic web technologies to bring together multiple databases and link their contents to a very fine grained level of detail, across taxonomies.
Michael Fulford spoke engagingly about his work leading the Silchester Roman Town archaeology project, with the recent addition of the JISC funded VERA project blog. The data they're capturing from the site excavation is being rigorously archived, and represented along with evidence based artistic interpretations within interactive data based town reconstructions, with impressive depth of detail and rendering. Using this technology Michael was able to trace through the collections database and give specific justification to all details, right down to the goats depicted in various town life scenes.
During dinner I caught up with Mark Birkin who is PI on the MoSeS node, one of 2 geographic simulation nodes, part of the research programme run by the UK's National Centre for e-Social Science. We discussed potential overlap with simulation research the Social Statistics Network are carrying out on Primary Care in an Aging Society.
We ended the afternoon on Thursday with a session focused on building trans Tasman eResearch collaborations, which followed from a similar Birds of a Feather session held at APAN 26 in Queenstown. I'll blog the outcomes from Thursdays session once I've put them up on the wiki.
The week was full of opportunities to discuss leading edge research techniques and technologies, and was topped off by a visit to the Australian Synchrotron, by the perfect guide for such an occassion, Richard Farnsworth.
Mozilla recently announced http://labs.mozilla.com/2008/08/introducing-the-concept-series-call-for-participation/
their Concept series http://labs.mozilla.com/projects/concept-series/
which had some interesting ideas around browser use cases, with particular
relevance to the ideas of concept mapping, navigation, and semantic mashups on
the web. Some interesting videos on the Concept Series site worth reviewing. The work demo'd is open for use and contribution - check the site for details.
Last week, Mark Gahegan presented to the Software Engineering Research Group (SERG) a seminar reviewing his academic work. Mark provided insights into his work within the GeoVista Center, Penn State University. There's some obvious cross overs between Mark's work with Improvise and the evolution of concept mapping and related visualisations. When coupled with the Aurora concept in Mozilla there's obvious opportunities for extending these rich visualisations to the crowds.
I'm curious as to the possibilities coming from inferring tight and loose associations of concepts, which are visualised in the spatial view in Aurora, and in the Improvise examples (TGraph, Music), along with some of the concept mapping developments Mark was talking to. (I'll see if I can get Mark's slides to share for those that might be interested).
I can see valuable information management techniques and tools arising from the confluence of these developments, which could allow us to start to confront the info glut that is the web, while also providing easy access to mashups of data services and visualisations (as shown in the Aurora video).
For example the browser could automatically associate commonly related browsing patterns, and reinforce these maps of associated websites through accessing the browsing habits of the crowd through social networks. These visual maps would be nicely displayed in the spatial views shown in Aurora, while their associations could be used to infer opportunities for semantic data exchange between websites and services. The end user tools shown in the Aurora demo would then allow for Through The Browser authoring of connectivity between semantic services and further visualisations a la mashups.
I should note that I'm most excited by the interaction design of these approaches to some very complex information management problems. The discovery and composition of semantic services is facilitated through a social network mechanism inside an intuitive zooming user interface which could run in a browser and be directly manipulated by most, shared, reused, refuted, etc. This is the promise of the semantic web, which I'd love to see as reality.
