cuboideum
commented
7 years ago There are two principal ways to approach the problem:
- Model the upper and lower boundaries of a value range as separate scalar value specifications.
- Introduce a new type of value specification for value ranges.
I would advocate the latter option for several reasons:
- The data structure is less complicated, as no extra instances representing the boundaries need to be defined.
- As the boundaries have the same measurement unit, there is no need to define it more often than once.
This is my proposition for the definition of value range specifications, exemplified for an age-at-death estimate:
Figure: Draft for modelling value range specifications. In the example, 'age estimation individual A' is an osteological age-at-death estimate, based on an individual skeleton (individual A). Instance 'measurement of age of individual A' represents the true age at death, e.g. in a collection of skeletons for which this information is known (cf. key to symbols in network graphs)).
The OBI uses the concept of value specifications (obo:OBI_0001933) to model data estimations, simulations, targets and settings. Currently, there are classes representing specifications for categorical and scalar measurement data.
In osteological anthropology, it is common to estimate biological qualities of dead people from their skeletal remains. Results are often expressed as a value range with a lower and an upper boundary, e.g. a minimum and a maximum age at death. The real value (which no longer can be measured directly) most likely falls within these limits.
The RDFBones core ontology needs to provide a standard definition how to model such value ranges.