In data modeling and knowledge organization we often find the need to connect item as subclasses or instances of each other. These two types of relationships are often confused, but why?
Where does subclass and instance come from? Object oriented modeling? Artificial Intelligence? Knowledge Organization? Plato? ...
Some influental sources should be identified.
See http://marciazeng.slis.kent.edu/Z3919/43hierarchy.htm on hierarchical relationships. Traditional thesauri don't differentiate Generic Relationships and Instance Relationships but just support Hierarchical Relationships.
Teorey et al (1986) extend the Entity-Relationship Model with hierachies to support generalization but the idea can be trace back further
Object Oriented Modeling was very influental starting in the 1990s.
See https://en.wikipedia.org/wiki/Class_%28knowledge_representation%29.
Where has the confusion been studied before?
There is no objective rule when to use subclass and when to use instance relationship because these categories are artificial. It all depends on...
Description logic does not work to describe reality.
Probably it has already been said by Kent (1978).
I believe the basic reference for the notions of subclass and instance is ancient or Medieval logic.
There are proposals published in past decades to better specify the NT relationship into NT-subclass, NT-part and NT-instance, but they have not been used frequently. Some papers by Fulvio Mazzocchi and others by Douglas Tudhope mention this and provide references.
Further refinement are categories of faceted classifications, such as: Kind, Part, Property, Material, Process, Operation, Agent, Space, Time. E.g., Bicycles havePart Wheels
Part of the problem appears to be related to whether members of a set are all identical (like all oxygen atoms are) or variously different. A single oxygen atom is an individual, that is, an instance of oxygen atoms. A single person still is an individual, although it is partly different from other persons. However, Felids and Canids are usually considered not as individual carnivores, but as subclasses of carnivores. Still some biologists believe that taxa, especially species, should be considered as individuals. After all, in the history of Earth, Canids only appeared once, in specific conditions and with specific characters... The same can be said about a language or a dialect taken as a whole (taking for granted that a single fox or a single sentence in some language are individuals).