draft v0.3
COOS governance is fully in line with the ModernStats Governance Guidance, the generic governance applicable to all ModernStats models. The COOS governance contains additional and COOS-specific elements to this generic governance. Please consult the ModernStats Governance Guidance first as the two documents together cover the full governance for COOS. [add link to the ModernStats governance document later]
Based on the common semantic versioning governance elements of the ModernStats models [link], the following COOS-specific rules apply to the three main version types:
Major: there are changes in the new version that affect the formal naming, definitions and descriptions of the existing concepts, as well as the relations between these concepts.
Minor: there are changes of concepts and/or relations of the model that do not affect the formal naming, definitions and descriptions. The changes might include adding new concepts or relations or removing old ones, but all of these changes are considered as backward-compatible with the existing model.
Patch: error corrections, textual or visual changes based on errors identified in previous versions that result in bug fixes. Changes implemented as part of the patch do not change the substance.
N° | Principle | Statement | Rationale | Implications |
---|---|---|---|---|
1 | Change control | COOS has a systematic way to design and adapt changes, i.e. the design principles of COOS apply to national adaptation of COOS and to the revisions of the model. |
The only way to provide a consistent level of quality information to stakeholders is if the whole organisation abides by the same principles. Therefore, it is recommended that national adaptations of COOS apply the design principles. The design principles also need to be used for future revisions of COOS in order to ensure consistency and comparability between different versions. |
|
2 | Use of agreed models and standards | COOS is built upon foundations set by other standards and helps the stakeholder to leverage them in combination. Therefore COOS optimally reuses the existing terms and definitions used by the involved models, standards. |
COOS provides a formal presentation of ModernStats standards covering the statistical business process, the information objects and the generic activities of statistical organisations. COOS is also linked to general vocabularies, data models and existing ontologies, such as DCAT, PROV, SKOS. By COOS providing the high-level formal presentation of the concepts used in these models, it is ultimately connecting them. As COOS is integrating information available in models, standards, in case of missing terms or contradictions between similar terms used in different models, using new definitions should be avoided. |
|
3 | Stakeholder appeal | COOS helps stakeholders to understand various standards and their connections. |
COOS helps stakeholders to easily understand the importance of different standards and their connections in order to realise what benefits can be expected from their use. |
|
4 | Driver for modernisation and future revisions | COOS is powerful in suggesting future modernisation and initiating revisions of the involved standards. |
COOS provides ideas for future modernisation activities. With COOS connecting several models and standards, it will also be used as a guidance on how to improve each and strengthen the interconnection between them to have better integration in the future. |
|
5 | Providing global names | COOS provides global names for some individuals, based on interoperability and use cases. |
The initial version of the COOS is guided by model interoperability, the future versions of the ontology would be more use-case-driven. |
|
6 | Simple presentation |
COOS objects and their relationships are presented as simply as possible. COOS is aimed to provide a generic solution to users and not specific elements for different statistical domains. |
Even though the use of ontology standards will ensure that the information described by COOS can be understood clearly by ontology experts, it also needs to be presented as simply as possible, avoiding notations and presentation styles that are known only to experts. |
|
7 | Agreed level of detail |
COOS represents objects only down to the level of agreement between key stakeholders. |
COOS does not represent information on the deep level of the standards but on high-level to highlight the benefits of using different models together. Too detailed level means the presentation will not be easy for several stakeholders to adopt/understand while too high-level imposes the risk of not realising the benefit of each standard. |
|
8 | Adaptability and extensibility |
COOS can be easily adapted and extended to meet new/changed stakeholder needs. COOS is meant for repeated use and not only for a one-time investment. |
COOS will need to be robust and stable to be adopted by statistical organizations, but will need ongoing review and adaptation to remain relevant to user needs. |
|
9 | Platform independence |
COOS does not refer to any specific IT setting or tool. |
Statistical organizations use a wide range of in-house and proprietary hardware and software platforms; this environment also changes over time. COOS must be platform independent to be relevant to all stakeholders and robust over time. COOS will provide the ontology based on ModernStats standards and common ontology models. The interpretation, presentation is not IT-specific, the formats used for the ontology do not limit its use in specific IT settings. Development of COOS requires stocktaking of standards to be involved in the work according to the user needs. Once it is developed, it needs to be put in use in order to realise its benefits in harmonisation and modernisation goals. The developed solution can be used and reused by the organisation. |
|
10 | Formal specification |
COOS provides formal naming, definitions and descriptions of concepts in a given domain, and the relationships between these concepts. |
COOS will provide a set of standardized, consistently described objects (OWL). This will help stakeholders understand significant relationships among the objects in the various standards involved in the ontology. |
|
11 | Relevance |
COOS is a solution that provides relevant information on the standards and their relationships to the stakeholders. |
The ontology in itself is an up-to-date solution but it also means that the information described by the ontology is relevant to the stakeholders and outdated information is removed, new information is added. COOS represents the latest versions of the underlying models it is integrating, e.g. GSIM, GAMSO, etc. |
|
The process of the COOS describes the full procedure for the establishment of the COOS, starting with specifying the needs. It describes how COOS is developed and maintained by the owner of the model. The process consists of 5 main process steps broken down by sub-processes (see Figure 1).
Figure 1: Process of COOS design and review of COOS
Source: https://ec.europa.eu/eurostat/cros/content/ess-standardisation (modified for COOS use).
It does not cover the process of adoption by National Statistical Offices as that activity might vary from NSO to NSO.
This process model for ontology also includes the iterative approach similar to the PDCA cycle of quality management as ensuring the quality of an ontology is of high importance to serve user needs.
In order to keep the creation and update of the COOS understandable, some main criteria are defined in order to help NSOs with adopting COOS or a national ontology.
The step of the design/review process of COOS begins with collecting the needs from different stakeholders. Main stakeholders are NSOs as a part of the international statistical community. The need might come from changing connecting models (GSBPM, GSIM etc.) or can be derived from the demand of HLG as well. This collection of needs is in line with the Capability Development activity area of GAMSO and the process phase of Specify needs of GSBPM where modification of existing statistical outputs or new needs for new statistical outputs might come into consideration. This step takes input from the continuous communication with stakeholders. This step ends with the approval of the proposal on the COOS creation or update. This approval has to be made by experts on ontologies and the stakeholders. This process is the task of the Supporting Standards Group in consultation with the HLG-MOS Executive Board.
This step covers the compilation of ontology specifications based on the approved proposal of the previous step ’Specify needs for ontology creation and update’. The compilation of the specification of the COOS means determining the methodological background of the ontology such as basic terms that have to be used when compiling statistical outputs, the planning of the structure and their detailed content needed for the outputs and an implementation plan. This implementation plan is only for the owner of the standard and not for NSO level.
This step covers the programming of the ontology prepared in the ’Design new ontology / updates to the ontology’ step. This step ends with the compilation of a draft proposal for adoption, including the programmed ontology and an implementation plan mentioned in the previous step.
In order to support the formal adoption of COOS the output of the step ‘Build Ontology’ is presented to the stakeholders and the feedback is collected in the form of an expert review and a public review. After stakeholder consultations the final proposal is submitted to HLG-MOS for adoption. In case of a positive decision, the new version of the COOS is released.
Between review cycles of the ontology, the use of the model by the community is monitored and feedback is encouraged, lessons learnt are collected from the community. In order to foster the use of the model, supporting material (learning and communication) is provided. Periodic checks are conducted from time to time to have an up-to-date overview on the use of the model. Based on this overview and feedback from stakeholders, the owner of COOS evaluates if a revision is needed and if so, an activity proposal is prepared and the revision process is launched.