NTR: population density

[ramonawalls]

This is a type of environmental quality, but it could also be a quality of a population.

See original request at https://github.com/EnvironmentOntology/envo/issues/259, including "We would use this neutral quality to define environmental exposures (of relevance to SDGIO type projects, public health, etc) as well as experimental treatments."

[pbuttigieg]

Does density inhere in things like a population or a collection of molecules? Sounds a bit odd.

We could even think of it as an increased disposition of members of a collection (or object aggregate) to come into contact with one another? However, density can increase with no contact at all.

I think we need a spatial (or spatiotemporal) entity like a site here (to keep aligned to the general understanding of density) and the idea that the degree of some other entity's instantiation within that window matters.

[pbuttigieg]

Regardless of the semantics, I think PCO is a good home for this class right now. Regarding usage, we have to make sure it works for different human populations as well as ecological populations

[pbuttigieg]

Perhaps this is best as a relational "quality", using spatial relations between the entities in a collection of interest. @cmungall and I are thinking about handling spatial relations a little better (or documenting how it can already be done)

[ramonawalls]

Does density inhere in things like a population or a collection of molecules? Sounds a bit odd.

I think it would be a relational quality, very similar to what we came up with for concentration. Effectively, population density is the "concentration" of organisms in a spatial area, so it is a relational quality between a collection of organisms and a spatial region (or maybe site).

[pbuttigieg]

I think it would be a relational quality, very similar to what we came up with for concentration. Effectively, population density is the "concentration" of organisms in a spatial area, so it is a relational quality between a collection of organisms and a spatial region (or maybe site).

Yes, I agree. Could you paste an axiom example here?

[zhengj2007]

OBI worked out 'molecular concentration' with logical axiom as follow: 'quality of related physical entities' ('inheres in' some material_entity) and (towards some 'scattered molecular aggregate')

If PCO would like to adopt it, the population density can be: 'quality of related physical entities' ('inheres in' some organism) and (towards some population)

[Public-Health-Bioinformatics]

A few questions, since I'm keenly interested in this with respect to epidemiology.

1) I'm curious what is this "towards" relation? I couldn't find it in OBI release. Is it like "with respect to"?

2) @pbuttigieg @cmungall I'd like to know more about your spatial domains vision. (My bias is that space and time concepts could be organized within the paradigm of dimensional analysis. Seems like time should be able to use exactly the same adjacency relations that any numbers perched on a number line use. Causal relations however would be an addition for the time dimension.)

3) Density, a prerequisite, should be defined. Since there are closely related density definitions (density as count / area; density as mass/volume), its label should distinguish its use. A general definition of "density (count per region)" could be "a (real, estimated, or hypothetical) ratio of the count of some given countable entity(/ies) that are located within some given region."

4) I guess the primary design pattern for simpler relational qualities that are ratios is that in the "forward" relation, subject acts as numerator and object acts as denominator? We could have: 'population density': 'quality of related physical entities' and density and 'is about' some 'population' and 'is about' some 'region' and 'has numerator' some 'population count' and 'has denominator' some 'region area'

Actual measurement datums of things like density depend on real-world census or experimental data which it appears the APPOLO_SV ontology is looking into: census APOLLO_SV:00000173: The count of a specific entity in a specified region at a specified time. population census APOLLO_SV:00000153: The outcome of a population survey.

[dosumis]

If you have a figure for density, why not use a data property? has_density (or has_population_density ?)

Or if you don't have exact figures but want to make a comparative statement about population density, just use an objectProperty. ('towards' has always seemed like a hack to me, introduced to find a way to deal with PATO classes that are implicitly comparative and probably should have been treated as objectProperties from the start).

[Public-Health-Bioinformatics]

I get that a data property is the quickest OWL way to associate a value of a certain type with an entity; and an object property is the quickest way to make known some association between two concepts. My own bias (which I know doesn't fit many pragmatic efforts because it involves complexity that is avoided by the property approach) is that the more we define the structure of a concept (like the ratio density) with respect to its parts, the closer we get to a future that can automatically associate the resources we need to answer a query about the concept. By defining population density as above, a search for appropriate population counts and related region areas at a given timepoint can theoretically be conducted. Defining density just as a name of a property doesn't help in that campaign. In that respect properties are shortcuts that don't add semantic value. I'm attracted to a model in which a quality of an entity at some point in time (like population density of a given population) has a "has value" data property relation to its observed value. It's a work in progress though, and admittedly more difficult to insert time in there.

[Public-Health-Bioinformatics]

p.s. Now I see the definition of "towards" in PTO object property list. "Relation binding a relational quality or disposition to the relevant type of entity." So it seeks to constrain what entity a quality pertains to. I guess above 'has numerator' and 'has denominator' have the same mission.

[dosumis]

The problem with this: "Relation binding a relational quality or disposition to the relevant type of entity" is that, AFAIK, nothing in the formal semantics of OWL-DL is available to enforce this binding. This => potentially serious co-reference issues.

[cmungall]

On 28 Mar 2017, at 20:54, David Osumi-Sutherland wrote:

If you have a figure for density, why not use a data property?
has_density (or has_population_density ?)

Or if you don't have exact figures but want to make a comparative statement about population density, just use an objectProperty.
('towards' has always seemed like a hack to me, introduced to find a way to deal with PATO classes that are implicitly comparative and probably should have been treated as objectProperties from the start).

Unfortunately this wouldn't work for any scenario where we want to say e.g. the change is due to a mutation in a gene (we could reify, but this is invisible to OWL)

RQs with towards were attempts at OWL-visible reification (with 'towards' being a syntactic predicate, like rdf:object in a reification). I agree it hasn't worked out so well. Time for a rethink.

[Public-Health-Bioinformatics]

Is this a case where the most succinct definition of a concept involves a structure of relations that OWL can't validate. One would have to resort to external tools to do that validation until the day when "OWL 3.0" can catch up. Or is there not enough agreement on the best way to define ratios in the meantime?

Population density has a flipped sibling - ecological or urban footprint, the ratio of land area/person, so the definitions need to distinguish this. Seems to me that direct reference to "numerator" and "denominator" dimensionality is necessary.