STRUCTURAL MODELING OF LAGUAGE CLUSTERS

Abstract

In solving many practical problems, algorithms of cluster analysis are used, with the 
help of which a given set of objects is represented by a union of intersecting or non
intersecting subsets – clusters. Such problems arise, in particular, in mathematical statistics, 
control theory, pattern recognition, automatic design, operations research, taxonomy of 
language structures, in issues of studying written fiction and scientific texts. In many cases, 
the given objects have a discrete structure, and methods of discrete mathematics will be 
used to study them. An example of an increasing finite sequence of natural numbers shows 
the process of constructing clusters, the mutual connection of which is represented by a 
oriented graph and a hypergraph. With a large volume of initial information, it is necessary 
to carry out multi-stage clustering, when at each subsequent stage, generalized clusters are 
constructed, the objects of which are the clusters of the previous stage. Further issues of 
allocation clusters in language texts, which are large-volume ordered samples from a finite 
set of objects with repeating elements, are considered. The example of an increasing finite 
sequence of natural numbers shows the process of constructing clusters, the mutual connec
tion of which is represented by a directed graph and a hypergraph. The issues of clustering 
text objects have a certain specificity, which is explained by both their formal and substan
tive properties. For structural analysis and synthesis of natural language objects, intermedi
ate clusters between a sentence and a text are distinguished – texteme and macrotesteme. A 
three-stage clustering of such language objects from a sentence to a text is considered, and 
the structural model of clustering of each stage is represented by an oriented graph and a 
hypergraph. These models help, based on the semantics of the constituent linguistic units, 
to study the systemic organization of the entire text and can subsequently become a basis 
for identifying general patterns and classifying various texts.