Stephan Meyer, Clearing Process Manager, and Martin Herzog, Asset Designer
The collaborative process that will result in the publication of a conform pan-European asset modelling a 'core person' has started.
The issue is discussed lively at the SEMIC.EU LinkedIn Group, and downloads of interoperability assets have increased significantly since the start of the conformance process for data on persons.
The Asset Designer, Martin Herzog (right), and Stephan Meyer, the Clearing Process Manager, explain the aims and procedures in this conformance process and compare it to Lego.
SEMIC.EU What does conformance of an interoperability asset mean?
Stephan Meyer SEMIC.EU defines semantic interoperability assets as entities which support data exchange with semantic accuracy. Examples include code lists, data models, thesauri, taxonomies, ontologies, and their relevant documentation. Conformance of such an asset represents the highest level in our three-step quality assurance process, the Clearing Process. It is the final stage of the clearing process for an asset. Conformance indicates that an asset is complete, sound and applicable from a technical/formal point of view (stage one in the clearing process), has received feedback and modifications based on community review (stage two), and ? essentially ? has been evaluated and recommended by a group of experts, has been implemented and has proven its functionality in practice (stage three).
Conform assets are strongly recommended for reuse by SEMIC.EU.
SEMIC.EU Can you describe the steps already taken and what you think the next stages will be?
Martin Herzog In this particular case of the person data model, starting the conformance process has not been requested by any asset agent (persons responsible for an asset in the repository). We, however, realised the necessity and demand for such a generally usable asset ourselves.
As it stands, none of the national person assets looks like it is going to turn into the conform asset eventually. The backgrounds and particularities are too diverse.
Thus, we?re designing an independent version of a person data model that makes use of the strengths of the individual source assets and consolidates their elements so the new model can be mapped to the original ones. What we?ve done so far is we have analysed the structure of the existing models, i.e. we compared their respective attributes. We also examined whether the concepts the models describe (?natural person?, ?person?, ?physical person?) are sufficiently congruent. The answer to this tricky question is yes and no, mainly because of the different levels of detail: the Danish model is quite concise and short, while the German and Austrian ones feature a lot more details, and the French model reflects the attributes in UN/CEFACT if not all of them.
In a nutshell, they all describe persons but stress different aspects.
SEMIC.EU So how do you decide on which aspects to choose for the new model?
Martin Herzog The goal is to have a person model to be used in as many cases as possible. Our approach is to define a ?core person? as a set of attributes suitable for the purpose of identifying a person. For specific use cases, more attributes can be added. The core person is an adjustable, amendable concept.
SEMIC.EU Based on this approach: What do you expect as attributes of such a core person after the community consultation?
Martin Herzog At the moment, I would count name, birth date, gender, birth place among them and perhaps nationality as an identifier in pan-European contexts.
Address is often named as a potential attribute of a person, but there are some caveats: Address is a very complex and volatile concept. It bears a lot of inherent interoperability challenges, especially with regard to the different concepts of addresses which European countries have. This makes it less of a candidate for identification. Just consider that it is more likely that a person does not have an address than it is that they don?t have a gender. Lastly, core address will certainly become a conform asset itself at some point.
SEMIC.EU Why do you need a use case to agree on ways to model natural persons? ?Person? appears to be a pretty straightforward concept.
Martin Herzog Yes, but there are different ways of identifying individuals. Many countries, for instance, have a personal identification number for all citizens or residents. Others don?t.
Similarly, bank data, tax numbers, social security numbers are all use case dependent. In their respective use cases, the numbers are always used, without exception. They will complement the core person we?re aiming at. There?s no contradiction to the core person idea. The latter identifies a person with no regard to their function or role. In taxation contexts you?ll use the core person and add their respective tax-related ID (usually a number) to turn the core person into a tax paying individual.
SEMIC.EU Does that mean that the use case you say is required for any data model to become a conform asset is ?no particular use case? for the core person?
Martin Herzog The core person is not a complete model. Imagine it as a box of Legos, the generic ones with four and six buttons. You can build anything you want out of them. But for a house you?ll need roof and window parts in addition. The core is the box with the simple bricks. To build a house you?ll have to bring your own roof tiles. To make a tax payer, you?ll have to bring your own tax number attribute.
SEMIC.EU Why has there never been a conformance process before?
Stephan Meyer The distribution of states in the clearing process has the shape of a pyramid: The higher the level of quality, the fewer items will have reached it.
Usually there will be one conform asset per use case in European eGovernment. Competing assets should be ruled out under the conditions of conformance whereas it?s different for the maturity level (?mature assets?). Three to four assets can realistically assumed to become mature out of an undefined number of assets originally submitted to SEMIC.EU from different national, organisational, lingual contexts. A conform asset can evolve from the consolidation of competing mature assets.
?Person? is the first item with such a basis of several competing (registered and/or mature) assets in the SEMIC.EU repository. For this first conformance candidate, we decided to document and discuss the process publicly. It?s unlikely that every process of this kind will have the same degree of public visibility, although they will all be public, of course.
Why is this the first one? The basic collection of assets for all kinds of uses is still going on. The repository needs to grow further for the pool of assets to get large enough to trigger further conformance processes of information units other than ?natural person?.
Martin Herzog Person is a good starting point because it is so widely used. You can hardly think of any eGovernment process, transaction, reporting or application that does not contain some sort of person-related data. Even if the subject of the information is not directly related to citizens or people in general (e.g. in environmental data reporting, finance data), you will virtually always have to identify a sender and/or receiver of a message or a person responsible.
SEMIC.EU Are there any other data models / assets that are candidates for conformance?
Stephan Meyer As a general rule, models that can be used in more than one context are the most likely candidates. ?Organisation?, ?address?, ?bank data?, ?education? come to mind.
SEMIC.EU Does that apply irrespective of the type of artefact we?re talking about?
Stephan Meyer Yes, of course. Every asset can potentially receive the conform label. What is true for data models or schemas must apply for, say, code lists as well. ISO code lists, as an obvious example, are already widely used and have a high degree of acceptance. SEMIC.EU would not give different recommendations here.
SEMIC.EU The process is supposed to be collaborative. What does that mean in practice? Is there any special input you need?
Stephan Meyer In general and for this particular process we expect constructive contributions from the stakeholders. And we have quite a few already. It?s important that people share their experience with the data sets we talk about. To stick with the the Lego analogy, we need suggestions for the basic bricks. Our task is to evaluate, sort and clear the suggested items and decide whether they should be included.
In other words, we need real use cases, e.g. cases that require eye colour, body height, driver?s licence number, blood type to figure out which of these should reasonably be included in the core.
SEMIC.EU What do you expect the core person to be used for?
Martin Herzog People will download it from the repository and implement it in their respective use case. The basic bricks can be used for different purposes because they allow for complementation by other attributes.
The full potential of semantic interoperability will be reaped when the core is implemented in several countries, projects or cases and when existing concepts will be mapped to the core. A conform interoperability asset functions as a mediator between different sources.
SEMIC.EU Does that imply that you?re planning standardisation as well?
Stephan Meyer Standardisation per se is not our task. But it?s well conceivable and reasonable to transfer our results to standardisation organisations, especially this conform asset and other classified as such.
Download: Interview: The 'core person' and the conformance process (PDF)
More information: Conformance