DataSource are linked to a Measure Type

[Sacha Bruttin]

GHO DataLoader specifications states that it is possible to import a datasource linked to a point of data.

Into the GHO Database, a table DATA_SOURCES exists but is linked to the MEASURE_TYPES table.
This is contradictory with what the specification says.

Do we maintains the ability to import a datasource as mentioned in the specifications or do we stay with the current state that the datasource is defined at the measure level ?

[Jiri Dvorak (jiri-dvorak)]

The DATA_SOURCES entity is the source of any collected data. The Data Source is therefore one or more data points. Typically a Data Source represents a Time Series of values.

The MEASURE_CATEGORIES and MEASURE_TYPES linked to DATA_SOURCES will then be "grouped" by a common Data Source to be a part of the same collection (e.g. the same Time Series of data). In this sense, the values of DATA_SOURCES fields REF_SYSTEM_ORGANIZATION and REF_SYSTEM_URL point to an organization / system that "owns" (or "is responsible for") the data in the Time Series, and not necessarily the organization performing the data entry of individual records.

The Blue Infinity team needs to decide, whether this purpose of DATA_SOURCES is aligned with it usage GHO Data Loader, or whether it is not relevant to the Loader.

[Olivier Chapiteau (olivier-chapiteau)]

There's also a link from DATA_SOURCES to the MEASURE_DIMENSION_TYPES table.

[Jiri Dvorak (jiri-dvorak)]

Yes, that is correct; this is to ensure that all MEASURE_TYPES that are qualified via dimension values have their dimension definitions included in the data set's metadata.

Please note that each MEASURE_TYPES record *may* be linked to a set of child records in MEASURE_DIMENSION_USAGES; if that's the case, those child records will be linked to the same dimension definition, in the table MEASURE_DIMENSION_TYPES.

In that case, properly configured metadata will have the value of MEASURE_DIMENSION_TYPES.DATA_SOURCE_ID identical to the corresponding value of MEASURE_TYPES.DATA_SOURCE_ID. That way, if we extract all database records "owned" by a certain organization / project (e.g. "The World Health Report" or "FLUNET" within the WHO, see the FluID schema), *all* metadata - including all dimensional coding - is included in the extracted data set as well.