Contacts

Kashirin Igor Yurievich

e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.

Definition of Hierarchical Numbers

Kashirin I.

 

Hierarchical numbers are numbers of the form [s] a₀ . a₁ . a₂ . ... . a_i . ... .a_n ,

where a_i are positive integers from the set N = {0, 1, 2, 3...}. s is the symbol of the "+" or "-" sign,

a positive sign may not be specified.

 

 

For example, hierarchical numbers may look like this:

 

 

0.0.12.48.0 or -2.33.0.0.4 .

 

 

The symbol denoting the set of hierarchical numbers is H.

 

Applied definition . Hierarchical numbers are a branch of mathematical number theory that allows you to adequately operate on the main characteristics of taxonomies. In the theory of knowledge representation of artificial intelligence, hierarchical numbers allow:

− to determine the semantic similarity of concepts;

− calculate indexes of hyperonyms and hyponyms;

− receive regular knowledge structures;

− verify trees of generic and causal classifications.

 

 

The idea of hierarchical numbers was first proposed in 2020 in an article 

 I.Yu.Kashirin. Hierarchical numbers for designing artificial intelligence ICF taxonomies  / 

2020. № 71.  P.71-82 (rus)

 Let N be a set of positive integers with elements ni ϵ N, let there also be a highlighted character ".".

The set A = N ᴗ "." ᴗ L is defined as an alphabet with integers n, where "ᴗ" is the operation of combining sets, and L is an empty character.

Then the grammar is:

ĥ → L,

ĥ → h,

ĥ → - h,

h → < n >,

h → < n > . <h>

describes a set of hierarchical numbers H with elements ĥ.

 

The rule  ĥ → - h   allows for the full range of hierarchical numbers, including the middle negative elements of the number:

 

0.10.-1.-1.127 

 

 These numbers are already used in one way or another in the practice of classification or addressing, for example: a universal decimal code or the IP address of a computer on a global network. However, the introduction of

an algebraic system of hierarchical numbers makes it possible to perform operations with them similar to formal arithmetic, and to isolate binary relations for their comparison and analysis of non-trivial properties of operations and relations.

 

Consider the algebra of binary hierarchical numbers.

 

Let B be a set of numbers with elements {0, 1}, n ϵ B (n = 0 or n = 1), let there also be a highlighted character ".".

The set A = B ᴗ "." ᴗ  L is defined as an alphabet with integers from B, where "ᴗ" is an operation combinations of sets, and L is an empty character. Then the grammar is:

ĥ → L, ĥ → h, ĥ → -h,

h → < n >, h → < n > . <h>

describes a set of binary hierarchical numbers H with elements of h.

 

Examples of binary hierarchical numbers can be given: 0.1.0.0.1 or 1.0.-1.0 .

Binary hierarchical numbers are numerical indices of the vertices of two

binary trees: positive and negative with one common vertex 0.

The generation of a vertex to the left of 0 is performed by the binary operation 0+0 = 0.0, the generation of a vertex to the right is performed by the binary operation 0 + 1 = 0.1.

The generation of negative vertices is performed by the operation "-", respectively:

0-0 = -0.0, 0-1 = -0.1.

Graphically, this can be represented by a tree spreading in a positive or negative direction (Figure 1):

 

 

  

 Figure 1 – An image of algebraic operations in the form of trees

  

However, using negative elements can make the meaning of operations more complicated. For example, the generation of the "+" trace may look like this:

0.1 + 1.1 = 0.1.1.1, 0 + 0.1 = 0.0.1

However, the example 0.0.-1 + 1.1 = 0.0.-1.1.1 indicates the presence of more complex tree travel routes using not only descent but also local ascents. This application of hierarchical numbers will be discussed further with specific examples.

The reverse operation of generation, the removal of the terminal vertex "--", is unary:

0.1.1.1-- = 0.1.1, 0.1.0-- = 0.1, 0.-1.-1 -- = 0.-1.-1 .

 

In graphical interpretation, the number can be considered the absolute index of any vertex, i.e. starting from the top of the tree 0 or relative, displaying the path through the tree from one of any vertices to other vertices up and down.

 

The absolute index always starts with the character 0.

When solving practical problems, only the positive part of the algebra of binary hierarchical numbers can be considered. In this case, operations claiming to receive a negative index will have a result of 0.

One more rather popular operation can be cited, namely, the calculation of the most common vertex, which is interpreted as a search for a common ancestor of two argument vertices: 

 

 

 Figure 2 – Another example of hierarchy

 

This generalization/multiplication operation is commutative, i.e. a º b = b º a.

  Multiplying a positive number by a negative number always equals 0.

The important operation is "Ù" as the calculation of the path from the vertex given by the first argument to the vertex given by the second argument. For the previous figure, examples of such calculations could be given:

 

 0.1.0  Ù  0.1.1.1   = 0.1.0.  0.1.  0.1.1.  0.1.1.1

 

  0.1.1.1   Ù  0.1.0  = 0.1.1.1  0.1.1.  0.1  0.1.0

  

 0.1.0.  0.1.  0.1.1.  0.1.1.1    @  0.1.1.1  0.1.1.  0.1  0.1.0

 

 Here “@” relation of equality of lengths of two hierarchical numbers.

 

  However, such a calculation leads to an unnecessarily complex result.

 Note that the common ancestor for 0.1.1.1 and 0.1.0 is 0.1, from which both numbers begin.

 As a result, when calculating the operation Ù vertex to the second.

 This is necessary to get an idea of the complexity of the path from the first vertex to the second, regardless of the depth of the tree.

 Then the correct operation Ù will be like this:

 

 

0.1.0 Ù 0.1.1.1 = [0.1.]0. [0.1.] [ 0.1.]1. [ 0.1].1.1 = 0.1.1.1

 

0.1.1.1 Ù 0.1.0 = [0.1.]1.1 [0.1.]1. [0.1] [0.1.]0 = 1.1.1.0

 

0.1.1.1 @ 1.1.1.0

 

 

After considering the semantics of the above operations, we can define a universal arithmetic algebra of hierarchical 

 numbers H:

                                                 H = < H, Ω >, Ω = {+, -, --, º , Ù, ⊕},

 

where Ω is the signature of the algebra, i.e. many operations.

  All operations are binary, except for "--", which is unary. The meaning of the operation “⊕” will be discussed later, 

 using relevant examples.

 The considered algebra can be supplemented to the algebraic system 

 

H = < H, Ω, R > by introducing a set of relations R = {< , >, @ , = },

 

where the relations “a > b” and “b < a”, respectively, “the number a is more complex numbers b" and "number b is 

 shorter than number a."

 

A brief example of a taxonomy is shown in the figure.

 

 

 

 Figure 3 – Brief taxonomy example

 

More examples on this site in articles about icf taxonomies.

Continue. Theory Of Hierarchical Numbers

Educational materials for students

Materials for undergraduates are posted here. (by Kashirin I.Yu.)

 

Neural networks for beginners (rus)

4 Methods of Text Vectorization (rus)

A brief overview of vectorization techniques in NLP (rus)

Parameters of the support vector machine (SVC)

Parameters of the k nearest neighbors method (KNN)

 

More detailed materials are available on the Russian-language website.

The idea of hierarchical numbers

-

Binary hierarchical numbers
Kashirin I.Yu. ( Каширин И.Ю. )

 

The simplest description of the theory of hierarchical numbers

for knowledge representation models.

It can be used to calculate the semantic proximity of words and sentences in a natural language. 

The idea of hierarchical numbers was first proposed in 2020 in an article

 I.Yu.Kashirin. Hierarchical numbers for designing artificial intelligence ICF taxonomies/2020. № 71.  P.71-82 (rus)

 

 

They are numerical indices of the vertices of two binary trees: positive and negative with one common vertex 0.

The generation of a vertex to the left of 0 is performed by the binary operation 0+0 = 0.0, the generation of a vertex to the right is performed by the binary operation      0+1 = 0.1.

The generation of negative vertices is performed by the "-" operation, respectively:
0-0 = -0.0, 0-1 = -0.1.

Graphically, it looks like this:

 

 

  The operation of generating the "+" trace may be more complicated:

0.1 + 1.1 = 0.1.1.1
0 + 0.1 = 0.0.1

The reverse operation of generation, removal of the terminal vertex "--", is unary:

0.1.1.1-- = 0.1.1
0.1.0-- = 0.1

To consider a number as an absolute or relative index of the vertex of a binary tree is determined by a person solving an applied problem using hierarchical numbers. The absolute index always starts with the character 0.

Only the positive part of the algebraic system of binary hierarchical numbers can be considered. In this case, operations claiming to receive a negative index will have a result of 0.

Calculating the most common vertex is interpreted as a search for their common ancestor:

  

     

 

 

  0.1.1.1 º 0.1.0 = 0.1.0 º 0.1.1.1 = 0.1

This generalization/multiplication operation is commutative. The operation can also be designated as "*".

Multiplying a positive number by a negative number is always 0.

The concept of the inverse element (number) can be described as follows.

What can be said about the numbers 0.1.1.0 and 1.0.0.1 ?
They are obtained by the complete inversion of atomic elements 0 and 1 in all digits.

Multiplying such numbers will always give a zero result. Obviously, these numbers cannot be considered absolute (counted from the root of the tree).

 
However, the access paths to the terminal vertices for these numbers are completely opposite: if we consider "1" to be a positive choice

and "0" to be a negative one, then any descent in the tree one level lower for the first number will be the opposite of the choice determined by the second number.

 

 

When the first number says "yes", the second one definitely says "no" and vice versa.

 
If we always talk only about absolute numbers, then the first digit "0" can be omitted.

 

An important operation is "Ù" as the calculation of the path from the vertex specified by the first argument to the vertex specified by the second argument.

Examples of such calculations can be given for the previous figure:

 

0.1.0 Ù 0.1.1.1 = 0.1.0. 0.1. 0.1.1. 0.1.1.1
      0.1.1.1 Ù 0.1.0 = 0.1.1.1 0.1.1. 0.1 0.1.0

 

0.1.0. 0.1. 0.1.1. 0.1.1.1 @ 0.1.1.1 0.1.1. 0.1 0.1.0

 

Here " @ " is the ratio of the equality of the lengths of two hierarchical numbers.

 

 

However, the common ancestor for 0.1.1.1 and 0.1.0 is 0.1, which is where these numbers begin.

 

 

As a result, when calculating the operation Ù, these fragments are omitted for all vertices of the path from the first vertex to the second.

 

This is necessary to get an idea of the complexity of the path from the first vertex to the second, regardless of the depth of the tree.

 

Then the correct operation is like this:

 

0.1.0 Ù 0.1.1.1 = [0.1.]0. [0.1.] [ 0.1.]1. [ 0.1].1.1 = 0.1.1.1

 

0.1.1.1 Ù 0.1.0 = [0.1.]1.1 [0.1.]1. [0.1] [0.1.]0 = 1.1.1.0

 

0.1.1.1 @ 1.1.1.0

 

 

 

What is considered a right or left descendant? This is in the application of numbers in practice. First, the left descendant is written, then the right one.

Orderliness needs to be specifically specified.

 

(Static Dynamics) (Beautiful Ugly) (Good Bad) (Small Big)

(Simple Complex) (Delicious Tasteless) (Friend Enemy)

(Cause Consequence) (Yesterday Tomorrow)

 

Continue. Theory Of Hierarchical Numbers