212
V. Babenko et al.
To date, issues related to the automatic analysis of complexly organized table
structures as sources of knowledge for the formation of the Semantic Web have not
been sufficiently developed.
A technology for automatically extracting ontological specifications of ODR
objects from HTML tables of unknown structure based on top-level ontologies to
facilitate the processes of semantic linking of document contents with models of
subject areas of information systems was proposed in [2]. The possibility of generating new knowledge based on the analysis of ontological descriptions of documents,
which means the evolution of information systems in the direction of ontological
knowledge bases, is noted.
The approach to the formation of knowledge bases based on the use of structural analysis methods involves the identification of objects based only on the analysis of the structural characteristics of SSK, which allows you to build specialized
procedures for the formation of knowledge bases focused on specific presentation
structures and SSK storage formats.
The limitations of structural analysis methods are manifested in the wide variety
of structures necessary for organizing data and attracting a person to “clean” the
extracted data.
The approach to the formation of OKB based on top-level ontologies involves
the formation of OKB in accordance with the terminology defined in the top-level
ontology. Thus, the boundaries of the representation of the components of ODR
objects, in contrast to structural analysis, are found as a possible combination of terms
defined in the ontology by calculating measures of semantic similarity. To establish
a correspondence between the terminological representations of source objects and
ontologies, the methods of Lee, Wu, and Palmer or Reznik are often used. This
approach allows you to build procedures for obtaining new knowledge, abstracting
from the method and format of storing the contents of structured sources of knowledge. The process of forming an OKB based on this approach is that for each SSK a
structural-logical data scheme is constructed, which, like an ontology, is represented
in the form of a graph of objects and relationships, objects of a structural-logical
scheme are characterized by properties. Each property corresponds to a domain,
which, in addition to the set of acceptable values, determines the recording format
adopted in a particular source of knowledge.
After constructing a structural logical diagram, an expert can modify it by deleting
part of objects or narrowing domains. This makes it possible to simplify the work with
SSK in the case when it is necessary to use only a part of the data contained in it, while
the structural-logical scheme of SSK is compared with the top ontology by searching
for matching terms of sources and ontology, as mentioned above. Considering the fact
that it is not always possible to establish a direct correspondence between the terms
of the ontology and the structural-logical scheme of SSK, additional restrictions may
be imposed on the scheme, in particular, restrictions on the values of properties. Next,
a comparison is made of the properties associated with domains by constructing a
formula for transforming source values. This formula plays the role of a translator
V. Babenko et al.
To date, issues related to the automatic analysis of complexly organized table
structures as sources of knowledge for the formation of the Semantic Web have not
been sufficiently developed.
A technology for automatically extracting ontological specifications of ODR
objects from HTML tables of unknown structure based on top-level ontologies to
facilitate the processes of semantic linking of document contents with models of
subject areas of information systems was proposed in [2]. The possibility of generating new knowledge based on the analysis of ontological descriptions of documents,
which means the evolution of information systems in the direction of ontological
knowledge bases, is noted.
The approach to the formation of knowledge bases based on the use of structural analysis methods involves the identification of objects based only on the analysis of the structural characteristics of SSK, which allows you to build specialized
procedures for the formation of knowledge bases focused on specific presentation
structures and SSK storage formats.
The limitations of structural analysis methods are manifested in the wide variety
of structures necessary for organizing data and attracting a person to “clean” the
extracted data.
The approach to the formation of OKB based on top-level ontologies involves
the formation of OKB in accordance with the terminology defined in the top-level
ontology. Thus, the boundaries of the representation of the components of ODR
objects, in contrast to structural analysis, are found as a possible combination of terms
defined in the ontology by calculating measures of semantic similarity. To establish
a correspondence between the terminological representations of source objects and
ontologies, the methods of Lee, Wu, and Palmer or Reznik are often used. This
approach allows you to build procedures for obtaining new knowledge, abstracting
from the method and format of storing the contents of structured sources of knowledge. The process of forming an OKB based on this approach is that for each SSK a
structural-logical data scheme is constructed, which, like an ontology, is represented
in the form of a graph of objects and relationships, objects of a structural-logical
scheme are characterized by properties. Each property corresponds to a domain,
which, in addition to the set of acceptable values, determines the recording format
adopted in a particular source of knowledge.
After constructing a structural logical diagram, an expert can modify it by deleting
part of objects or narrowing domains. This makes it possible to simplify the work with
SSK in the case when it is necessary to use only a part of the data contained in it, while
the structural-logical scheme of SSK is compared with the top ontology by searching
for matching terms of sources and ontology, as mentioned above. Considering the fact
that it is not always possible to establish a direct correspondence between the terms
of the ontology and the structural-logical scheme of SSK, additional restrictions may
be imposed on the scheme, in particular, restrictions on the values of properties. Next,
a comparison is made of the properties associated with domains by constructing a
formula for transforming source values. This formula plays the role of a translator