Stanislav Taktaev personal web site
ѕерсональный сайт —танислава “актаева.








‘орум / Discussion ||

Dimensions of semantic space co-ordinate hub. Reference point.

Dimensions of semantic space co-ordinate hub. Reference point.

Constructing of notion space co-ordinate system is untrivial task. This field is weakly researched (fragmented) in order to the fact, that this area slips the interests of traditional УconceptualФ sciences Ц linguistic, semantic, hermeutic. ThatТs why to suppose unequivocal theoretical decision without advance studying would be irresponsible.

In different periods author reviewed several theoretical approaches to this problem. This approaches presented lower in chronological order.

The 1st approach. Working hypothesis is that co-ordinate hubs have to be co-directed to mostly frequent vectors (actions, that is to say verbs). It follows from: number of notion in developed (Russian) language is about 200000. Number of verbs Ц about 10000, moreover widely used are 500. They could be grouped to about 150 groups. Thus, estimated number of dimensions in notion space is from 150 to 500.

Main hubs should reflect such linguistic notions like Уto increase Ц to decreaseФ, Уto complicate Ц to simplifyФ, Уto move toФ Ц changing of condition, changing of sense, changing of position.

The 2nd approach. Hubs should correspond to qualitative (wooden, cell, plated) and quantitative characteristics (features) of notions (cold Ц warm Ц hot). For example, hubs should correspond to adjectives and adverbs in usual language, but not only to them. As ranged features should be used qualitative characteristics of notions, such as size, time, color (like length of wave), novelty (adjectives with comparative degrees), part Ц unit.

The 3rd approach. Constructing of hubs by semantic differentials. Semantic differential suppose existence of two polar semantic notions (antonym in language). It means that semantic distance between them is extreme contrary values of certain co-ordinate hub. Imagine that moving of point along this hub is obviously, each position of this point will correspond to measure of existence of certain feature (or several features), determined by semantic differential (beauty Ц ugliness, immobility Ц movement). Consequently, semantic differentials could be guides for co-ordinate hubs to semantic space, hence they are dimensions of semantic space.

Then if imagine certain hierarchical catalog, where would be enumerated all met notions, then the most upper, root level of catalog (root domain) will correspond to notion (describing object) УworldФ and will have all co-ordinates. That is to say it will be comprehensive and will have all possible features. It means also, that, dialectically, notion УworldФ doesnТt have concrete co-ordinates in notion space, because it itself is space of notion.

Accordingly, next catalog level will have appointed way of dividing objects (and notion, describing them) to domains Ц that is to say certain character of grouping and ranging, which will be the first dimension (co-ordinate) of semantic space. Further, each next hierarchical level of this catalog УWorldФ will add new dimensions, corresponded to certain features of notion (domains) groups, describing objects and their groups.

That is to say, notion (group of notions), describing concrete object, will have the most numerous co-ordinates, describing its position in semantic space, hence will take the smallest area in semantic space as regards all upper members of hierarchy (domains).

Then each dimension will be number of features, ranged by sole criterion and will contain certain amount of those criteria.

For example: <feature: Weight><Notions: weightless, light, weighable, heavy>.

Conditions of right calculation of semantic distances in notion space are: semantic orthogonality of co-ordinate hubs, isotope and linear dimension and right choice of reference point.

Problem of semantic orthogonality solves by choosing those variants which donТt correlated between themselves form all offered variants of semantic space dimensions and is only computing task.

Problem of isotope of semantic space and linear of dimensions straightly corresponded to choosing the unit of measurement and measure of distance in semantic space (look further). At present time this problem doesnТt have unequivocal decision.

Concerning reference point also possible several variants. Here is more philosophic, than mathematical choice. Practical approach is also important for simplifying practical using of the theory. Possible variants:

Human. Because all notions created by humans, they are anthropomorphic and it is clear that counting should be held from the human.

World (God, Universe, Absolute). Because all notions are part of this notion and automatically included in this domain, than using decrement hubs each notion could be estimated as part of the World. Then moving along the hubs will corresponded to decreasing of information and increasing of entropy (moving to Chaos).

У, УZeroФ Ц this notion is reverse to the World, symbolizing absolute nothing, Chaos. Here, on the contrary to the World, using increment hubs each notion could be estimated by its difference from Chaos. Then moving along the hubs will correspond to increasing of information in each concrete notion and decreasing of entropy (moving to Absolute).

It is clear, that УZeroФ and УAbsoluteФ are semantic differential, polar points of notion space. It is obvious, that Human lies between this poles and not worse as reference point.

If conduct simply analysis it is clear that co-ordinate model is simplifying of fractal model. Simplifying reached by decreasing of number of dimensions, because for fractal model describing it will necessary to abide a condition of equal number of notions and dimensions. Total arity of graph, describing notion space, will consist of about 1.5 million. ItТll make calculations of each concrete notion co-ordinates very chargeable by resources. So, additional researching and calculating experiments are necessary to adjust which approach is more attractive and are there any additional variants.

You can ask author:

Pls, carefully type your email!

Your E-mail:
Your Question:

Website from "Summatech"
I'll ready to answer to you: Stanislav Taktaev