Hierarchy of opposition values

[viktor80]

Hello,

as I'm not a linguist I'm not sure where I have to post this question

I'm working on  natural language processing project, concentrated not on the semantic processing itself  but rather on supporting of bases for  such processing. So I need to create a several  semantic structures based on the meaning of the word

I encountered with issue where I need to build hierarchy  of oppositions, If you imagine it visually, In the form of a  pyramid, branching by dedicated groups, from main value to less value

I'm looking for  exist researches  or rules to differentiate  this groups,  close as possible to some most generalized linguistic model(s)

simple example, if on the top of  pyramid is good and bad opposition and each contains some general values, each of which should contain less generalized values down to a reasonable limit

bad
- hate
- - war
- - destruction
- death
- - sorrow
- - despair
- lie
- - betrayal
- - deceit

good
-  love
- -  peace
- -  creation
-  birth
- - happiness
- - hope
- true
- - faithfulness
- - honesty

[Daniel]

You might want to look into various hierarchical arrangements (also called taxonomies) of concepts for specific semantic fields.
The obvious example would be trees from biology showing the relationship between species, and similar relationships can be drawn for semantic fields like you mentioned.

Work like this is popular both in philosophy and in computer science (for applying relationships like you mention), and has actually been going on for centuries. One of the earliest examples is a project by John Wilkins who tried to capture all of the meaningful contrasts in language: https://en.wikipedia.org/wiki/An_Essay_towards_a_Real_Character_and_a_Philosophical_Language

Of course the problem is that taxonomies can be grouped in various ways and there is no real 'top' to any particular hierarchy, because different relationships could be expressed (or similar relationships expressed differently).

If you search for keywords like "taxonomy + semantics" you will probably find relevant research.

Another area of research that could interest you is using a database (such as WordNet: https://en.wikipedia.org/wiki/WordNet ) that tries to automatically (or sometimes with help from users) capture meaningful relationships between words that can be extracted for computational applications.