Data Cube Implementations

Name: Bill Roberts
Affiliation: Swirrl IT Limited

1) Open Data Communities, UK Department for Communities and Local Government
Already listed on http://www.w3.org/2011/gld/wiki/Data_Cube_Implementations

Additional information:

Contains over 100 different data cube datasets on housing, planning, deprivation, departmental business metrics.

Different datasets may have different results against the Integrity Constraints. We haven't run all datasets against the Integrity Constraints.
Where we have tested against IC, the constraints are satisfied except those relating to Data Structure Definitions as we have not yet added DSDs to this data.  However, we agree with the inclusion of DSDs in the Data Cube specification and intend to add these to the existing and future datasets on this site.

Use of vocabulary: measure dimension
There is one dataset on this site (http://opendatacommunities.org/data/housing-market) where we use the 'measure dimension' approach and the qb:measureType property.  This dataset relates to housing market metrics and includes a variety of measure types - the value of an indicator, the percentage change of that indicator since some previous time, the maximum and minimum values over time.  We use qb:measureType to distinguish between these different values.

Use of vocabulary; units
We make frequent use of http://purl.org/linked-data/sdmx/2009/attribute#unitMeasure to relate an observation to the units it is measured in.

Use of vocabulary: hierarchical code lists
We have a clear use case for the qb:HierarchicalCodeList class, for representing relationships between geographical areas.  So far our implementation of this is experimental and not yet available online, but we have plans to add this approach to the published datasets.  In particular, we typically use the statistical geography URIs from the UK Office for National Statistics (http://statistics.data.gov.uk).  Many of the DCLG datasets include data points for England, as well as for local authority districts in England.  Similarly datasets may have observations for a county (the upper tier of local government) as well as for districts within that county (the lower tier of local government).  We would like to express these relationships and the Hierarchical Code Lists feature of the Data Cube meets that need.  As in Example 16 of the Data Cube spec, we would use the OS Spatial Relations 'contains' property as the qb:parentChildProperty.

There are many examples on the DCLG site of datasets that include this kind of hierarchical geographical code list requirement, including http://opendatacommunities.org/data/households/projections/totalhouseolds http://opendatacommunities.org/data/housing-market/supply/additions/net-additions 

2) Hampshire County Council
http://linkeddata.hants.gov.uk/datasets/net-additional-dwellings

A dataset of predicted numbers of houses to be built at various locations in Hampshire, used to be transparent about their building planning process.

Makes use of http://purl.org/linked-data/sdmx/2009/metadata#obsStatus to register that the values are forecast, not 'measured'.


3) Scottish Government
http://data.opendatascotland.org - not yet live but due for release in December 2013.  Includes approximately 12 separate data cube datasets on deprivation and education.


EU OpenCube project.

Swirrl is a partner in an EU FP7 funded project called 'OpenCube' which has recently started and is scheduled to run for 2 years.  (Project website http://www.opencube-project.eu - due to be live by end November 2013).  This focuses on creating tools to support both the creation and consumption of RDF Data Cube datasets.  The fact that this project has been set up, funded by the EU and attracted a number of unfunded 'use case partners' from public and private sector indicates a widespread need for and recognition of the value of the RDF Data Cube vocabulary.