NeuroCollective Internal Schema and Linguistic Definitions

The NeuroCollective is a knowledge graph database, basically following the entity/relationship and triple store ideas. It’s differences to other triple stores is that it is, out of the box, designed to support language independence.

Language independence is constructed in the following way:

A ‘script’ is a piece of text, in any alphabet or another coded format, that represents a single language element of a language (e.g. a word). This script can be a search item, a component of a phrase, code of a coding system, chemical formula, etc.. The script links to one or more lexemes. Scripts are case sensitive.

A lexeme is a collection of morphological variants of a particular gender, in a single language and single syntactic role, that represents the same word stem. A lexeme contains the morphological variants of the various grammatical cases, variants indicating different gender and plurality.

The same script may link to multiple lexemes of:

  • different syntactic role – eng: the book, eng: to book
  • different language – por: cavar, spa: cavar
  • different genders – deu: der Marsch, deu: die Marsh

An interlingual is a node that combines lexemes of the same syntactic role but different languages. In a lexicographic sense, an interlingual is representing a semantic sense of a dictionary entry.
Interlinguals are the primary nodes in the construction of semantic networks.

The networks are constructed supporting:

  • head/modifier/modifier relations
  • type-of/part-of
  • meta properties
  • properties

The head/modifier concepts are links to other interlinguals. This is the primary means to construct composite concepts. the head relation functions as the default type-of relation, which can be overwritten through a specific type-of relation.

deu: Originalversion→ eng:original (modifier, adj)  + eng:version (head, n)

The modifier relation concept itself is an interlingual and allows thus a dynamic extension of the relations between head and modifier components.

deu: Dampfschiff→ eng:steam(modifier, n) + eng:ship(head, n) + eng:“with an engine propelled by”(modifier relation)

Depending on the specific presentation template the deu: Dampfschiff would be translated as eng: ship with an engine propelled by steam. The modifier relation, in this case, is a complex semantic network. there are simpler variants of modifier relations, the simplest and default relation is “of”.

Interlinguals allow the composition of entire phrases, up to complex ideas, that can be presented in any language known to the system. Interlingual networks represent translation memory.
The possibility of circumscription of terms (the explanation of a term by utilization of the underlying semantic network), the automatic circumscription of a term in a translation, when the direct translation is not possible (e.g. the concept does not exist in a language), the categorization of modifier relations are subject for another article in the future.

The type-of (ISA) relation is a single relation to an interlingual that represents the hierarchically more abstract type of a concept.

eng: coffee table → ISA eng: table

The part-of relation can be relation(s) to multiple container concepts defined as interlinguals.

lat: Aedes → ISA eng: genus, PART-OF lat: Culicidae

An interlingual may have meta-properties that define if the object is seen as a type, what properties should be inherited by the instances, either by the value or by property.

An interlingual may have a collection of properties.

You may also like...

Popular Posts