SDMX Users Forum

Please login or register.

Login with username, password and session length
Advanced search  


Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Erik van Ingen - FAO

Pages: [1] 2
Thank you for pointing me to StructureSet, very interesting and could be the way to go forward.

It would not solve all of my requirements though, I will try to explain, giving this table:
The table shows three code lists, having codes for 'things'. A code list covers fully or partially all 'things'.

What I would like to do is just to show all 'things' expressed in their code mappings, thus avoiding using one code list as the reference list where others refer to.


Would this be possible in SDMX and if yes, how?

What my group considers as part of reference data is the mapping of different codes. For instance the different country codelists:
  • ISO ALPHA-2 Code
  • ISO ALPHA-3 Code
  • UN Country Code
  • etc.
We map these codes in our internal system and would like to express them in SDMX. We do the same for species, commodities, etc.

My idea would be that a hierarchical codelist would be the appropriate SDMX artifact in order to express all the different codes of the same 'thing'. So one hierarchical codelist to express the mappings of all the different country codes.

Would that be correct or are there other SDMX best practices to do so?

I assume that title in this observation is an attribute. In this case I would advise to code the TITLE in the DataStructure (keyfamily), link the TITLE attribute to a SDMX concept and associate it with a codelist. In the codelist you can then have explicit descriptions for each code for the languages you need. Doing so, you avoid having language dependent stuff in the dataset. Would that make sense and answer your question?

Has anyone tried to plot a SDMX dataset on a map? If yes, how did you spatially map the data?

We are thinking of adding a spatial annotation like "Well-known text" to codes in codelists which can be geo-referenced. Doing so, the SDMX dataset can be formally geo-referenced by the spatial system through the DSD. Any feedback on this approach is welcome.

FAO Fisheries has currently this SDMX 2.1 REST API with SDMX 2.0 messages:
FAO will publish this year also:

Yes, it would be good to have this list. Using the tools registry would be good. There is however a bit of confusion because the way I interpret tools/features is that it is software to implement SDMX, not a running and published webservice. If the tools/features are used for both (SDMX software and SDMX content), it is probably the best place (until the registry of registries is implemented).

Xavier and Stratos,

These answers do definitely answer my question completely, thank you!


Nice and very interesting! Lately I was having a look at the Stax implementation Woodstox. Did you consider also Stax?

Thanks for asking! No that is not my case. The concept behind the OBS_VALUE is not represented by one of the dimensions. For instance we have a DSD with 3 dimensions, FishingCountry, FishingArea and the Species. The OBS_VALUE refers to the number of tonnes caught. So my question in this case is, where do I express in SDMX 2.0 that the OBS_VALUE refers to the catch?


The question was actually in the context of SDMX 2.0, I had forgotten to mention that. Since we will be moving to 2.1, your answer is still very useful, thank you! What would be the answer in a SDMX 2.0 context?

The observation value is a concept in the DSD: OBS_VALUE.

In the DSD the OBS_VALUE is used to express a certain measure. Where can be expressed in SDMX what the OBS_VALUE represents in a certain DSD?

And can, would it be expressed as well in the SDMX dataset or would that be strange practise?

Technical Specifications / Re: Getting started with SDMX - a few questions
« on: December 30, 2010, 01:38:20 AM »
I will try to respond to you quickly, others are welcome to add more!

1) Are Concepts defined centrally (for instance by the SDMX initiative) or can anybody define a Concept to use?
There are cross domain concepts mentioned in the SDMX CONTENT-ORIENTED GUIDELINES:
If you have more concepts (which you most likely have) you can create them for yourself.

(2) A given deployed Nesstar server can hold many different types of data. Should it use a global Structure Definition for that site or one definition per cube to export?
Preferably you would define a DSD for each type of data, so for one definition for each cube to export. In our project we have a similar situation and we started by only generating the genericData without a DSD. This can be useful as a start; however it is more in the spirit of SDMX to publish your data with a reference to a DSD.

3) How are Structure Definitions linked to the data (or: given some SDMX file, how does one find the Structure Definition that describes the data)?
In a SDMX dataset there is usually a link to the DSD. For instance a generic SDMX dataset can have this element:
The keyfamily is part of the DSD.

(4)I'm under the impression that SDMX is a "two-file minimum" - meaning you need at least a Structure Definition and the data file itself to pass on data. Is this correct? What is the most generic way to describe data using SDMX?
Yes, as described in (2), you usually start of by defining your concepts, codelists and DSD. Next based on the DSD, you generate the SDMX dataset.

(5)Any comments, links to documents, case studies etc would be greatly appreciated. I welcome any ideas and suggestions to how SDMX export should be.
For publishing SDMX data there is the choice whether to use SOAP or REST as a protocol. Metadata is published through the so called Registry, data through the so called Repository or Dissemination. You can use an existing SDMX Registry and/or Repository out there or develop one for yourself.

SDMX 2.1 Public review / attachmentLevel in SDMX 2.1
« on: December 16, 2010, 08:00:38 AM »
I am not able to find anymore the attachmentLevel in SDMX 2.1 draft, also not in the document Summary of Major Changes and New Functionalities.

Is attachmentLevel replaced by something else?

Technical Specifications / SDMX Query language and MDX
« on: December 02, 2010, 06:21:22 AM »
SDMX has its own query language. Has it ever been considered in SDMX to use MDX instead of having its own query language?

An advantage of using MDX in combination with SDMX would be that (1) SDMX would not have the burden of defining its own query language.  (2) MDX is a well known and adopted standard. (3) OLAP systems do not have to implement besides MDX (if they use it), the SDMX query language.

Is anyone using SDMX in combination with MDX?

SDMX Tools, technical implementations and case studies / Re: SDMX JAXB
« on: March 16, 2010, 02:36:38 AM »
Thanks for your suggestion. Is your implementation open source?

The thing is however that I would prefer to use JAXB. JAXB (finally) got mature and is part of the Java specs and therefore a good standard. XmlBeans, Jibx, Castor and so on where always superior but now it looks like that JAXB is equally good.

Pages: [1] 2