dgillman4909
commented
1 year ago InKyung -
As far as I can tell a Category does not have a Designation associated with it. I was one of the designers of the Concepts package, and we tried to differentiate Category Set, Code List, and Classification Scheme by what was associated with each Node (or entry) in each kind of set. I don't see any contradictions.
The definitions of Category, Category Set, and Category Item probably need some editing. Yes, a Category Set should be defined as a set of Categories. A Category Item is a kind of Node, gathering all the information needed to describe a Category. The problem is a Category Set in GSIM is a kind of Node Set. Therefore, the disjunction between Category and Category Item arises. Maybe a better approach is to define Category Set as a set of Category Items. A Category Item is a kind of Node that contains a Category and some other information. A Category is defined in GSIM by "A Concept whose role is to extensionally define and measure a characteristic". Other than to violate the rule about starting a definition with a leading article, this is probably good. In this definition, a Variable takes the role of a Characteristic. Take a variable called Marital Status. We could measure it through either of the 2 Category Sets {single, married} or {single, married, divorced, widowed}. These aren't the only possibilities. But note, single needs to be defined differently in these 2 cases. With these careful definitions, each set of categories defines Marital Status extensionally, meaning these are subordinate concepts whose totality (the whole set of categories) defines the superordinate - Marital Status.
The use of the word "sign" in the definition of Designation in GSIM is confusing, and it needs to be replaced. The word "signifier" should be used instead. There is a general model of designation that says a Sign is the representation of a Signified by a Signifier which denotes it. In the case of concepts and designations, the designation is a Sign, the concept is a Signified, and alphanumeric strings are Signifiers. Other kinds of signifiers can exist, too. A particular string, an example of a Signifier, is called a Token instantiating that Signifier.
I know, that was a mouthful.
Datum as defined by Farance and Gillman (see DDI-CDI documentation) is a designation of a value, where a value is a concept with a notion of equality defined. That is very technical and not at all transparent unless you really know ISO 704 and ISO/IEC 11404. However, we have simplified this definition in several places. One is the Metadata Glossary. The draft DDI Glossary is another. As long you think of a datum as the representation some underlying meaning (i.e., a designation) with a computational model defined (a datatype), you have the idea.
For defining Level, I like the definition based on the number of arcs from the root. It assumes levels are used and only make sense in hierarchies. I think that's right. And each level contains all the Categories the same distance from the root. There could be exceptions to this, though. Alternately, each level comprises concepts (Categories in this case) that form an extensional definition of the root concept. Industry or Economic Activity is defined by all the Sectors, or all the major industry groups, or all the minor industry groups, etc. Each is a level.
InKyungChoi
Please see this google doc updated based on the feedback from Metadata Glossary team. I would like to draw attention to:
1. Node and Node Set area:
2. Designation : Metadata Glossary task team proposed as below. But I wonder if "designation" is a "sign" itself rather than "association" (i.e., Designation is a sign denoting a Concept with which it is associated)
The name given to an object for identification.association of a Concept with a sign that denotes it.3. Classification Index Entry and Classification Index: it is hard to understand what they mean from the definition and explanatory text, at least for me..... :(
4. Datum: Metadata Glossary task team asked to come up with a new definition as "value" is a synonym. How about "value that was collected or derived" (borrowed from the explanatory text)
5. Level: Metadata Glossary task team proposed three candidate definitions: 1) position of a Category or a group of Categories within the hierarchy of a Statistical Classification; 2) identifiable position to which codes in a scheme of codes are related; 3) set of nodes in a statistical classification in which 1) each node in the set is the same number of arcs away from the root node in the hierarchy, and 2) the set is defined by a concept) - which one shall we go with?