What should the scope of a MELODIES demonstrator portal be?

[jonblower]

In MELODIES we wil have various "public faces":

1. Our public website
2. Datasets that are published through various means (see other discussion How to publish Open Data from MELODIES #3 !)
3. The web portals and applications that are developed by the individual service work packages.
4. A "technology demonstrator" portal, whose primary purpose is to demonstrate geospatial Linked Data techniques, in a way that may not be obvious from the above websites.

The question is, what should the scope of the "technology demonstrator" portal be? This discussion is to collect "use cases" that could be enabled by the portal, with a view to producing a specification and development plan. The portal should be developed in such a way that its components could be used by the individual workpackages. Additionally the portal should interact with the endpoints through which we publish MELODIES data.

[jonblower]

Just a note so I don't forget - we should try to demonstrate annotations of datasets (and at the sub-dataset level) using CHARMe.

[letmaik]

Being a tech demo portal it's probably a good idea to have some stronger integration than just showing off some raster or vector map layers that are produced. So something where you can see the linked data aspect clearly. And something which is interactive and invites people to explore and play around. Any ideas anyone?

[BethanPerkins]

I have an idea - how about creating tutorials on how to use the portal (I'm assuming we're thinking about Sextant here..) which use our datasets as a foundation? That way, learning about the existence of the data, the outputs of the services and the complexity of the technology behind them happens on a less conscious level but - crucially - still happens. We could construct eight or more "lessons" each drawing on data/techniques etc. used by each of the services.

[jonblower]

Hi Bethan - I think that’s a nice idea, thanks! I guess ultimately it will depend on how much time we have to generate this kind of material, but we should definitely consider this. I wasn’t just thinking about Sextant by the way - Maik and I are working on some other technology that could bring in some new kinds of data. Sextant could be part of the story though. - Jon

[jonblower]

Maik - yes, agreed, the tech demo portal should be more than just “web-GIS” and should show linked data clearly. Ideas could include: (1) bringing in “commentary” information from the CHARMe system, (2) showing links to other sources of information (inc. other datasets) by discovering them at run time (not sure how yet).

[debbiejkite]

WP3 has come up with two initial ideas for linked data portals based on exploring land cover data:

1. Comparison of classes in different LC products and aggregation of LC data to a smaller number of classes - eg from Corine (40+ classes) and LCM (30ish) to the six UNFCCC classes. Allow the user to see the impact on %age of land in output classes based on different input choices, eg handling of mixed pixels, different definitions of grassland/bare ground/wetland. Could use e.g. a traffic light system comparing how different datasets match on the same item. We can use the vocabulary and UML structures from the EAGLE project to compare classes in different classification systems and support intelligent default mapping to specified output classes (this would be the "linked data" part).
   http://sia.eionet.europa.eu/EAGLE
2. Subsetting of large raster files (eg annual LC) by a variety of shapefiles such as counties/local authorities/Water Framework Directive (WFD) catchments, and calculate LC statistics on-the-fly. Potential to bring in linked data shapefiles from external sources (eg WFD API). Bring in datasets within a user session, not to be made available more widely (as some of the necessary datasets cannot be made public)
   Possible additional data sources:
   MAGIC (magic.gov.uk) - Natural England, EA, etc.
   NERC Soils portal
   Local Authority boundaries (possibly from EDINA? data.gov.uk?)

[mattfry-ceh]

The "subsetting" described under (2) could be a good way to show soil moisture grids compared to points (in the WP3 soil moisture demonstrator). It would require aggregation / time series extraction rather than summarising classifications.

[jonblower]

Hi Matt - can you go into a little more detail about how you would compare the soil moisture grids with points? What kinds of plots might you end up with?

[jonblower]

EMODNet "checkpoints" could be a source of inspiration - gathers data to asses specific user needs: e.g. http://www.emodnet.eu/northsea. Could use semantic information (and CHARMe?) to provide data recommendations in response to a user need (e.g. datasets that could support implementation or monitoring of a policy)?

[jonblower]

SmartOpenData has a dataset of tourist points of interest, available through SPARQL: http://ha.isaf2014.info:8890/sparql. I have more details in my email. Should be possible to throw up a visualization of this using Sextant, perhaps (like this: http://portal.sdi4apps.eu/tourist-data).