search

EnvironmentOntology / envo

A community-driven ontology for the representation of environments
http://www.environmentontology.org
Creative Commons Zero v1.0 Universal
136 stars 53 forks source link

Rural #272

Open mark-jensen opened 8 years ago

mark-jensen commented 8 years ago

There is need in SDGIO for a term that refers to rural geographical regions and populations.

Presumably this would be somewhere under anthropogenic terrestrial biome, perhaps as a superclass to cropland biome, village biome, and rangeland biome.

Another option might be a composed class, such as: 'rural biome' equivalent to 'anthropogenic terrestrial biome' and not 'dense settlement biome' Would that be too inclusive?

mark-jensen commented 8 years ago

Any progress here?

mark-jensen commented 8 years ago

This UN doc may be of help.

pbuttigieg commented 8 years ago

'rural biome' equivalent to 'anthropogenic terrestrial biome' and not 'dense settlement biome' Would that be too inclusive?

I'd be more inclined to go this route, as this seems more like an aggregation based on some quality. Is there no clear definition offered by a UN-linked resource?

From the doc you linked to:

There is more than one “correct” definition of rural and what is most appropriate will depend on the policy problem being considered. National definitions are continuously under debate and are in fact adjusted from time to time, reflecting, for example, changes in socio-economic and administrative structures or in mobility and communication. Similarly the level at which classifications are applied (that is, the size of the territorial units and the level of geographical hierarchy) will depend on the analytical purpose or on the policy problems that have to be solved. Within Member Countries, the OECD scheme distinguishes two hierarchical levels of geographic detail: local community level (small, though not necessarily the smallest possible, basic administrative or statistical units) and regional level. This Handbook recognizes that a hierarchical system (not necessarily just two-level) is good practice Chapter VII).

This suggests that this is going to get quite involved.

pbuttigieg commented 8 years ago

I think we should tag all ENVO classes that could be "rural" with an SDG-linked synonym. We can't 'solve' the semantic problem here, just better represent it.

Alternatively, we can create an SDGIO class for rural, import all ENVO classes that are relevant, create new ones where needed, and axiomatise the SDGIO class definition to include all of them. Subclasses in SDGIO can be used to define different flavors or rurality.

pbuttigieg commented 8 years ago

The OECD definition is very PCO-ish @rlwalls2008:

... defines a rural area as a local community with a population density below 150 habitants/km2 (500 in the case of Japan) (OECD, 1994).

The wording is odd, but the idea of population density will be key for this and the other anthropogenic biomes in ENVO. Are there any semantics for population density present?

mark-jensen commented 8 years ago

Drafted a class for rural in SDGIO

I put it under anthropogenic terrestrial biome, but wonder if it rather belongs under populated place. I'm not sure I understand the difference between dense settlement and dense settlement biome.

A second question is if it should be widened in scope to include non-anthropogenic areas/biomes, such as forests. Some criteria for rural may include unpopulated areas. @pbuttigieg @cmungall Thoughts?

definition: A anthropogenic terrestrial biome that isn't a dense settlement biome, with a population density below some established threshold. subclass of: anthropogenic terrestrial biome comment: What qualifies as rural varies, usually on the national level. Population density that is below some established threshold is the standard criteria. Use of 'rural' can also apply to populations, as in "rural workers", but this is an extension of the geographic one, which is primary. A rural population is one that lives in a rural area. editor note: This class may belong under ENVO 'populated place', awaiting resolution. Will most likely be a defined class, composed of any non-dense settlement or biome.

mark-jensen commented 8 years ago

I edited the comment above to reflect a fix in sdgio-edit.owl because of an error I made.

pbuttigieg commented 8 years ago

I put it under anthropogenic terrestrial biome, but wonder if it rather belongs under populated place. I'm not sure I understand the difference between dense settlement and dense settlement biome.

The features refer to the actual entities themselves (buildings, streets, etc) while the biome classes refer to the ecosystem surrounding them, defined by community composition. I think this class is more aligned to the former.

Also, "rural" does not belong to either - as we discussed, the primary label should be as clear as possible - we're talking about "rural settlements" or similar. Best to add an annotation property like 'UNEP preferred label' (documented on a new wiki page for user-group-specific features) and add "rural" there. @cmungall and I can cook up an export.

A second question is if it should be widened in scope to include non-anthropogenic areas/biomes, such as forests. Some criteria for rural may include unpopulated areas. @pbuttigieg @cmungall Thoughts?

We should probably create subclasses here, or new classes altogether as this would not be a subclass of "populated place" . This SDGIO:'rural area' (?) class would then aggregate all these possibilities (which may be scattered among the ENVO hierarchies for the semantics to make sense).

I anticipate we'll see a lot more of these due to the national and sub-national definitional heterogeneity, so we should also document our strategies on a wiki page ("Handling definitional variation" or similar). @mark-jensen, could you create it and transfer the gist of these discussions? I'm currently in favour of the model where we create SDGIO classes that aggregate and map to all the ENVO classes needed. This way, we show that it's a definitional issue, rather than an issue with the logic of representing bona fide entities.

mark-jensen commented 8 years ago

The wording is odd, but the idea of population density will be key for this and the other anthropogenic biomes in ENVO. Are there any semantics for population density present?

@pbuttigieg I see there is a population quality in PCO. That may help, perhaps a subclass can be added for density.

mark-jensen commented 8 years ago

'rural' usage in SD goals, targets, indicators:

We have

GEMET and UN Thesaurus show a few other uses, such as 'rural development' or 'rural education'.

However, I take it that all uses hinge on geographical area, which is primary, and typically determined (for statistical purposes at least) by considering population density.

I believe my original definition above is mostly on track, but needs to be altered to reflect the the fact is it a geographic region or area that has a low density population.

Since ENVO populated place is defined as "Place or area with clustered or scattered buildings and a permanent human population", I think it is a sensible parent for this class, now relabeled rural settlement.

rural settlement = A populated place with a population density below some established threshold.

I will then add a term for rural population as a subclass of PCO population, as well as a broad organizational class rural area for grouping any biome, settlement, feature, etc. that is rural. This can axiomatized to infer membership. A question is how best to do that. One idea is to use RO 'location of'. Another is to build a disjucntive equivalency.

We will also need to have a way in SDGIO to link the word 'rural' to all these uses. That will be discussed on a different thread on the SDGIO site.

@pbuttigieg @cmungall @rlwalls2008 A lot to parse here, but your feedback is appreciated.

pbuttigieg commented 8 years ago

rural infrastructure

I think ENVO can handle this one too, but I could also see this in SDGIO.

Since ENVO populated place is defined as "Place or area with clustered or scattered buildings and a permanent human population", I think it is a sensible parent for this class, now relabeled rural settlement.

rural settlement = A populated place with a population density below some established threshold.

This is workable, I'll add it and release ENVO again later today.

We will also need to have a way in SDGIO to link the word 'rural' to all these uses. That will be discussed on a different thread on the SDGIO site.

Wouldn't this just be a question of importing/creating all acceptable definitions and using ORs in a aggregating axiom for rural?

mark-jensen commented 8 years ago

Wouldn't this just be a question of importing/creating all acceptable definitions and using ORs in a aggregating axiom for rural?

Yes, I think that makes sense. We had also discussed using an annotation property, but I don't think that is going to satisfy UNEPs needs. Better to have an organizational class with some loose textual definition or elucidation. Asserting rural in the hierarchy isn't easy, it may be best to simply leave it outside the continuant/occurrent split. Perhaps group it together with other similar terms as 'primitives'.

pbuttigieg commented 8 years ago

Asserting rural in the hierarchy isn't easy, it may be best to simply leave it outside the continuant/occurrent split.

I definitely think it will be a continuant, but its dependence or independence is less clear.

mark-jensen commented 8 years ago

This is workable, I'll add it and release ENVO again later today.

If we follow this pattern, would it make sense then to have informal settlement in ENVO as well? See #265

pbuttigieg commented 8 years ago

If we follow this pattern, would it make sense then to have informal settlement in ENVO as well? See #265

I have the feeling that this should be in SDGIO as the "formalities" are somewhat specific. On the other hand, we can treat these the same way we treat population thresholds - the document that formalises the settlement must be some instance of a yet-to-be created information artifact class. For speed, let's create it in ENVO and then debate this in a new issue if needed.

cmungall commented 8 years ago

There is a paper by Stefan Schulz on applying this sort of pattern for disease ontologies. The idea is there is a set of core ontologies that conform to rigid upper ontology distinctions ("reality"), and a view ontology, that corresponds to loose concepts, defined by unions of the former, and automatically classified.

Will try and look it out later...

On 2 Feb 2016, at 6:30, Mark Jensen wrote:

Wouldn't this just be a question of importing/creating all acceptable definitions and using ORs in a aggregating axiom for rural?

Yes, I think that makes sense. We had also discussed using an annotation property, but I don't think that is going to satisfy UNEPs needs. Better to have an organizational class with some loose textual definition or elucidation. Asserting rural in the hierarchy isn't easy, it may be best to simply leave it outside the continuant/occurrent split. Perhaps group it together with other similar terms as 'primitives'.

Reply to this email directly or view it on GitHub: https://github.com/EnvironmentOntology/envo/issues/272#issuecomment-178598907

mark-jensen commented 8 years ago

@cmungall I seem to recall that as well. When you find, please send along. Thanks!

pbuttigieg commented 8 years ago

@mark-jensen I think you can also pull in [village,rangeland, cropland] biome. More description here

pbuttigieg commented 8 years ago

@cmungall

There is a paper by Stefan Schulz on applying this sort of pattern for disease ontologies. The idea is there is a set of core ontologies that conform to rigid upper ontology distinctions ("reality"), and a view ontology, that corresponds to loose concepts, defined by unions of the former, and automatically classified.

This sounds great. Useful for writing up too.