Semantics

Full Title or Meme

When an Entity can determine the "significance" of words in sentences.

Context

• Based on the obsolete idea that all of language can be divided into Syntax (order) and Semantics (meaning).
• Some data elements have meaning that is independent of the places where they are used. These cases are handled well with a Data Dictionary. See that wiki page for non-context-dependent semantics.
• For data elements that are context-dependent, the meaning cannot be known without reference to the context. These cases need a richer model.

Semantic Models

• Semantics in data processing goes back to the 1980's when EDI and XML structures were being defined.
  • In the case of EDI all semantics were context-dependent. The concept context definition was introduced by Tom Jones at a meeting at MIT where a data type was defined to say that the following data values had meaning only in that document at that location.
  • In the case of XML, the contexts were established at the top of the document and the specific context tag was prepended to the element name. So if the context name was "x", an element of that context would appear as x:tag.
• Data Dictionaries appeared in the databases with the EAR or Entity Attribute Relational models used with languages like SQL. In that case the context was the table where the data meaning was described.
• The Semantic Web can be dated to a paper by Tim Berners-Lee's paper in 2001 and the publication of the OWL language in 2004. which spawned an bewildering collection of data dictionaries which are ostentatiously called Ontologies when what they are is just what their employers imagine the reality should be.
• The result is not a collection of words that users of computer understand, but what the advertisers want them to see.
• An alternate source of Ontologies is the result of an Artificial Intelligence has determine what terms the training set is using. That might be closer to what the users that created the training set might want to know. For an example see WikiData " a free and open knowledge base that can be read and edited by both humans and machines." In "2021-12-08: "Antanina Paulavičienė", the one hundred and ten millionth item, was created."
• The above article by Hitzler also notes that "shallow non-expressive schemas often used for linked data appeared to be a major obstacle to reusability and initial hopes that interlinks between datasets would somehow account for this weakness did not really seem to materialize."
• Knowledge graphs with central control (unlike WikiData) are "usually understood to be much more internally consistent, and more tightly controlled, artifacts."

General Semantics

This is a label for a school of thought originated by Alfred Korzybski in 1933. The core idea is that meanings arise from attempts to describe an event or objects in the real world. The label that is used acquires nuances of meaning as it is used by people, but in the end, it is just an attempt by people to categorize a collection of events or objects so that communications is possible. These collections can never be a precise as Plato would have you believe with his ideal forms. An example of the varying meaning of words used for human feelings was published by Batja Mesquita where the idea of a constant set of names for human emotions was shown to be false. They emotions are separately collected by different cultures into different groups with different names.

Solutions

• Controlled effort like SNOMED-CT have worked will within the healthcare field to create a standardized set of categories. Naturally as knowledge expands the list of categories needs to expand as well.