Feedback from ISTAT

[pafrance]

Dear colleagues, ISTAT contribution is ready and available in the attachment. We're available for further discussion. Thank you. coos ontology_IstatComments.docx

[InKyungChoi]

For simplicity, I copied comments here:

Proposal for revision of COOS Ontology by Istat – 15/11/2021 (A.Bianco, P. Francescangeli, M. Riccio, M.Scannapieco)

Activity definition revision

Figure 1. “Current” Coos model with GAMSO, GSIM and GSBPM concepts

Coos model identifies similar concepts in several models and tries to connect them in a coherent model. In the current formalism there are no other roles than that of “subclassOf” and “skos:broader” (see Figure 1).

There are some aggregation/composition relations between some of the concepts included in Coos that could be introduced. This is the case between

1. Phase and Sub Process
2. Statistical Program and Statistical Program Cycle

Furthermore concerning GSIM concepts, the notion of Statistical Program should be connected Statistical Production Activity , not including also Overarching Activities, as it can be read below:

GSIM definition quote The Statistical Program is one of a family of objects that provide the environmental context in which activities to produce statistics within a statistical organization are conducted. Statistical Program is the top level object that describes the purpose and objectives of a set of activities. Statistical Program will usually correspond to an ongoing activity such as a survey or output series. Some examples of Statistical Program are: Labour Force Survey - Multipurpose Household Survey - National Accounts - Demography - Overseas Arrivals and Departures This does not include statistical support functions such as metadata management, data management (and other overarching GSBPM processes) and design functions. These activities are conducted as part of Statistical Support Programs.

As stated in the definition above, there is an additional notion of GSIM Statistical Program Support that actually realizes Overarching Activities.

Furthermore, in GSIM Business Function could be used instead of the generic coos task definition, as reported below.

GSIM definition quote A Business Function may be defined directly with descriptive text and/or through reference to an existing catalogue of Business Functions. The phases and sub processes defined within GSBPM can be used as an internationally agreed basis for cataloguing high level Business Functions. A catalogue might also include Business Functions defined at a lower level than "sub process". For example, "Identify and address outliers" might be catalogued as a lower level Business Function with the "Review, validate and edit" function (5.3) defined within GSBPM

So, the revised model proposed looks like this:

Figure 2. Revised Coos model with more contribution from GSIM

Summary:

1. Roles between Phase and subprocess should be replaced with a composition role instead of skos:broader, because these are not specializations.
2. Overarching activity is “overarching with respect to what? If it is overarching with respect to phase, we should insert a role for this: It needs to define a role for overarching activities with respect to phases.
3. An aggregation role called wave (or edition) was created to relate Statistical Program cycle to its corresponding Statistical Program.
4. Statistical Program is realized by GSBPM:phases instead of being simply derived from the top most activity class.
5. GSIM Business Function concept is proposed instead of the generic Task concept

INFORMATION OBJECT Revision There are several concepts related data structures, but there’s a general lack of connection between them.

The general term Information Object should be revised at least in name, because the term seems to refer to an object instead of a class and could be deceiving. Furthermore, it is used as root class for any GSIM concept, when there actually are concepts with that use in GSIM, so why not to use those?

Summary:

6. “Information Object” name should be revised because using the word “object” in a concept related to a class could be deceiving.
7. A GSIM concept like data resource could be included to replace information object
8. If we need a general concept to include both data and referential metadata, maybe information resource concept from GSIM could be used instead.
9. The relation between Statistical Program and Statistical Program Cycle to Information Object is not clear

[FlavioRizzolo]

The information object revision is one we should address early on, and it's related to an issue identified also in the GSIM task team (see https://github.com/UNECE/GSIM/issues/1)

Statistical Program and related classes have also been revised in GSIM, we should look at the Activity definition revision in the context of the ongoing changes.

Should we coordinate both revisions with the GSIM task team? There is a large overlap between both development teams, but enough non-overlapping members to perhaps justify a closer coordination for both ISTAT proposed revisions.

[InKyungChoi]

Regarding comment #5 "GSIM Business Function concept is proposed instead of the generic Task concept"

GSIM Business Function should be able to refer to any level of GSBPM activities while "Task" is meant to describe more granular level activity than sub-process.

[InKyungChoi]

Regarding comment 6 “Information Object” name should be revised because using the word “object” in a concept related to a class could be deceiving"

GSIM teams had quite a many discussions on this particular topic and concluded last month to use "class" instead of "information object". So unless anyone in COOS disagrees, we can accept this conclusion?

[pafrance]

There were many interesting comments in last call. I'd like to have them summarized here for reference if possible. Thank you everybody.

Shall we open another issue for introducing and linking gsim concepts maybe?

[FranckCo]

4th July meeting decided to close until matter clarified by GSIM revision team. @pafrance will reopen in due time if needed.