User:ArthurPSmith/Draft:Elements, Nuclides, Chemicals Ontology
Current status as of mid-2017
edit
How things are right now in wikidata (note indicates statements that would be removed, statements that would be retained, and ???? statements I'm not presently sure about, under the "new ontology" proposed below).
classes relating to elements and isotopes: edit
- (for example)
- and
- ????? (of the periodic table - in turn a subclass of row, class) - similarly for the groups
- and
- (for example) but also
- and
classes relating to chemicals and substances: edit
- and
- ????
- (for example)
- - also and ????
classes relating to molecules, atoms and nuclei: edit
- and which lead via subclasses through particle (Q1621273) to physical object (Q223557).
- similarly and
- ???? and while ????
- and then via subclasses to physical object (Q223557).
- ,
Critique of current status edit
- Meaning of "instance of": saying that entity A is an "instance of" class B should imply that the sentence "A is a B" makes sense.
- "A is a chemical element" makes sense. So does "A is an isotope", "A is a nuclide", even "A is an isotope of neon".
- But "A is a neon" or "A is a neon-24" does not make sense.
- "A is a lithium atom" is fine. So is "A is a diamond", etc.
- "A is a period 2" or "A si a group 18" also makes no sense - "A is a period 2 element" however would be fine.
- Meaning of "subclass of": if A is a "subclass of" B, then every instance X of A should also be an instance of B. That is, if "X is a A", then it is also true that "X is a B".
- if it doesn't make sense for anything to be an "instance of" neon, then having neon be a "subclass of" anything, or having anything be a "subclass of" neon is also nonsense.
- "nuclide" subclass of "chemical element" (or the other way around, or with "isotope") doesn't make sense - "isotope" and "element" are different kinds of things. "neon-24" is not an element, "neon" is not an isotope or a nuclide.
- the relation between "isotope" and "nuclide" may be somewhat more nuanced - "nuclide" is the better term in most cases but "isotope" is the older one. Usage is not entirely clear, but if different nuclear energy states of a given nucleus are treated as distinct nuclides but the same isotope, then just as "isotope" and "element" are distinct, "nuclide" and "isotope" would be distinct and shouldn't have subclass relations between them. If they are not distinct then they perhaps should have a "said to be the same as" relation.
- Transitivity: "subclass of" is transitive, "instance of" is not. That is, if A is a subclass of B and B is a subclass of C, then A is a subclass of C. The same does not follow for P31 relationships.
- Inconsistency: something shouldn't be both an "instance of" X and a "subclass of" X. There are numerous inconsistencies in the current status.
Proposed new ontology edit
- elements like neon or lithium are NOT classes; they are abstract concepts that are not instantiable. There is nothing that "is a" "neon", and the elements should not have subclass relations.
- isotopes like neon-24 are similarly NOT classes, but non-instantiable abstract concepts.
- groupings of isotopes, like "isotope of neon", ARE classes, whose instances are the specific isotopes.
- similarly the concepts "chemical element", "isotope" are classes - and are normal classes (not metaclasses) whose instances are the elements and isotopes respectively - these happen to be abstract and non-instantiable
- groupings of elements like "period 2 elements", "group 18 elements" can also be classes, subclasses of "chemical element"
- instantiable abstract concepts like "lithium atom" or "diamond" or a specific chemical, enzyme, etc ARE classes, whose instances are specific atoms or other physical objects.
Examples of P31/P279 statements under new ontology:
- each of the elements is an instance of "chemical element", for example (or possibly instead an instance of a subclass of chemical element, for example "period 2 elements")
- any wikidata item for a type of atom is a subclass of "atom", for example
- specific groups of isotopes are subclasses of "isotope": (or given current usage, subclass of "nuclide" may be more correct here)
- specific isotopes are instances of "isotope" (or "nuclide") or a subclass:
Questions still to be addressed:
- how should the relationship between neon-24 (Q14662251) (and/or isotope of neon (Q470437)) and neon (Q654) be expressed? (?)
- how should the relationship between lithium atom (Q18201918) and lithium (Q568) be expressed? (facet of (P1269)?)
- how should the relationship between diamond (Q5283) (and/or allotrope of carbon (Q622460)) and carbon (Q623) be expressed? (?)
- what shall we do with period 2 (Q207712), noble gases (Q19609) etc.?
- Maybe not needing to be addressed right now, but allotrope (Q21198401) seems to me more a metaclass-level concept - it refers to the different "types" of a simple substance. I think and are the right relationships here.
- Are items about chemical compounds (like difluorine (Q1963030)) about the molecule, or about the substance, or are they also abstract non-classes (similar to elements), and only items specifically about the molecule ("difluorine molecule") or the substance ("fluorine gas") would be considered actual classes that could have instances?
Alternate (consistent) ontology edit
User:TomT0m has proposed what I believe is the following:
- elements like neon or lithium ARE classes - subclasses of "atom", for which the instances are individual atoms of those elements
- isotopes like neon-24 similarly would be classes and "neon-24" would be a subclass of "neon", representing those neon atoms that happen to have 14 neutrons in the nucleus
- "chemical element" and "isotope" would stay as first-order metaclasses
- groupings of isotopes, like "isotope of neon", would similarly be first-order metaclasses: "neon-24" would be an instance of, but NOT a subclass of, "isotope of neon". And "isotope of neon" would NOT be a subclass of "neon".
- "lithium atom" would be "said to be the same as" "lithium"
The following would be the retained and removed statements from current wikidata under this ontology, for elements and isotopes:
Current relationships: edit
- (for example)
- ???? and ????
- ????? (of the periodic table - in turn a subclass of row, class) - similarly for the groups
- and
- (for example) but also
- and
New relationships: edit
elements and groups edit
Under either system, the current items for groups and periods such as period 2 (Q207712) and noble gases (Q19609) don't make much sense, particularly given the existing subclass of (P279) etc. relations. I propose the following:
- Relabel these as "group N element", "period N element" (these are already aliases for example for group 3 (Q108307)
- Under TomT0m's approach make all of them subclass of (P279) atom (Q9121)
- Ensure the individual elements subclass of (P279) their respective groups and periods (this is already in place for most of the elements - see Current relationships above.) Then the "New relationship" for element subclass of (P279) atom (Q9121) is redundant in TomT0m's approach.