=Paper=
{{Paper
|id=Vol-2362/paper6
|storemode=property
|title=Development of Information System for Textual Content Categorizing Based on Ontology
|pdfUrl=https://ceur-ws.org/Vol-2362/paper6.pdf
|volume=Vol-2362
|authors=Victoria Vysotska,Vasyl Lytvyn,Yevhen Burov,Pavlo Berezin,Michael Emmerich,Vitor Basto Fernandes
|dblpUrl=https://dblp.org/rec/conf/colins/VysotskaLBBEF19
}}
==Development of Information System for Textual Content Categorizing Based on Ontology==
https://ceur-ws.org/Vol-2362/paper6.pdf
Development of Information System for Textual Content
Categorizing Based on Ontology
Victoria Vysotska[0000-0001-6417-3689], Vasyl Lytvyn[0000-0002-9676-0180],
Yevhen Burov[0000-0001-6124-3995], Pavlo Berezin[0000-0003-1869-5050],
Michael Emmerich[0000-0002-7342-2090], Vitor Basto Fernandes[0000-0003-4269-5114]
1-4Lviv Polytechnic National University, Lviv, Ukraine
2Silesian University of Technology, Gliwice, Poland
5Leiden Institute of Advanced Computer Science, Leiden University, The Netherlands
6University Institute of Lisbon, Portugal
Victoria.A.Vysotska@lpnu.ua1, Vasyl.V.Lytvyn@lpnu.ua2,
yevhen.v.burov@lpnu.ua3, pavlo.berezin@gmail.com4,
emmerix@gmail.com5,vitor.basto.fernandes@gmail.com6
Abstract. The methods and means of using ontologies within systems for the
categorization of textual content were created. Also, a method for optimizing
the definition of which rubrics best relate to a certain text content was devel-
oped. The intellectual system that will use the methods developed earlier, as
well as other research results was implemented. The results will allow users to
easily filter their text content. The system developed has an intuitive user inter-
face.
Keywords: ontology, content, text categorization.
1 Introduction
The rapid development of the World Wide Web, in particular the increase in the
amount of content, raises the difficult task of analyzing such amount of content [1].
First of all, it defines the information that a person will ultimately get. Precisely be-
cause of this, the task of automating the rubricating of text content is very important
[2]. The main problem of manual rubricating is that it takes a lot of time and effort on
a person who does it. Also, with manual heading, there is no unification of headings
between people who read text. Automatic rubricating solves this problem, because it
will allow [3]:
To simplify the search for required information.
Discard the need for manual rubrication.
Unify rubrics.
Improve understanding of the content itself.
The aim of the work is development of a system for categorizing text content. To
address this goal, the following research tasks structure was developed:
1. Research of various methods of constructing a system of heading text content.
2. Exploration of various languages describing ontologies.
3. Investigation of the finished decisions of the heading of text content.
4. Search for ready ontologies.
5. Construction of the method of allocating the most relevant rubrics and categories
among the system received.
The object of research is the process of creating intelligent systems for categorizing
text content for convenient categorization of text records of a large size.
The subject of study is means and methods of creating intelligent systems for ru-
bricating of text content based on the ontological approach for analysis and rubricat-
ing the text data arrays. This system should free the text’s authors from categorizing
their text, as well as simplify the search for content for a specific topic for users.
2 Analytical review of literature
2.1 The analysis of key concepts
Analyzing the basic methods of text content rubricating, one can conclude that meth-
ods using ontologies are among the best, even though these systems do not use all the
features and benefits of ontologies [1-9]. Compared with other methods of construct-
ing knowledge bases, the advantages of using ontologies are obvious, since it is on-
tologies that are the standard of knowledge engineering and have proven themselves
as the best method for representing objective knowledge[10-16] . In particular, the
following tasks are not solved: criteria for filling and optimizing ontologies; modeling
processes for processing information resources and the emergence of new knowledge
based ontologies; assessment of the novelty of ontology knowledge. It is the solution
of these tasks that will allow to build effective application information systems for the
processing of text content, namely, its rubricating [17-21].
Under the formal model of ontology O, three of the following form are understood:
O=, where C is finite set of concepts (concepts, terms) of software [1-9],
which sets the ontology O; R:C->C is finite set of relationships between concepts
(concepts, terms) of a given software; F is finite set of functions interpretation (axio-
matization, restrictions) defined on the ontology concepts or relations O. It should be
noted that the set C is limited to finiteness and non-emptiness, and F and R must be
also finite. One of the most common ways to represent an ontology is the graph (con-
ceptual graph). In this graph, the vertices are software concepts, arcs are the relation
between concepts. The vertices can be interpreted or not interpreted. Interpretation
depends on whether the axioms of concepts are defined [22-26].
The arcs are divided into vertical and horizontal. Vertical relationships are given
by taxonomy of software concepts. To set a horizontal relationship, you need to de-
fine the set of values and the area of determination of the relation [27-31].
The structure of ontology, in its general form, is a set of four categories of ele-
ments: concept, relation, axioms and individual instances [32-36].
Concepts (classes) are general categories that are hierarchically arranged. Classes
can be regarded as the conceptualization of all representatives of a certain entity. Each
class describes a group of instances that have certain common properties. Concepts
are combined with each other through different relationships. The most common,
among systems using ontologies, the relationship type is categorization, that is, the
allocation of a certain class under the appropriate category [37-42].
Instances - are other elements of ontology, which are representatives of a certain
class, that is, the elements that belong to a particular category. Elements of ontology
have a specific hierarchy. Instances, like specific representatives, occupy the lower
level of the hierarchy. Above them are categories (concepts). Above categories there
are relations between them. The rules and axioms combine all these levels of the
hierarchy. To construct an ontological model, first of all, it is necessary to determine
the hierarchy of concepts (set C). When constructing an infological model in the form
of an ontology, experts should be involved in subject areas, which as a rule will use
abstraction and combination. The main method for constructing ontologies is the clas-
sification. Classification is a method for organizing knowledge. In object-oriented
analysis, the simplicity of the architecture of the infological model of the system is
achieved by identifying the general properties of objects, through which represent the
key abstractions and mechanisms. However, there are no strict classification methods
and well-organized rules now on how to classify objects and their classes. Therefore,
there are no such concepts as "ideal class hierarchy", "proper classification of ob-
jects", since the choice of classes is a compromise solution. It is through the classifi-
cation that we combine different objects into one group, according to their structure or
behavior. It should be noted that no system of rubrication of text content can be
achieved without parsing and identifying keywords. Keyword is a word, or a constant
expression of a natural language, used to express some aspect of a document's content
(or query); a word that has a significant semantic load. It can serve as a key when
searching for information. Parsing (parsing) is a process of analyzing the input se-
quence of characters, in order to parse the grammatical structure according to the
given formal grammar. During parsing, the text is transformed in the data structure,
usually in a tree that corresponds to the syntactic structure of the input sequence, and
is well suited for further processing. Stemming is the process of reducing the word to
the base by rejecting auxiliary parts, such as a suffix. The results of the stemming are
sometimes very similar to the definition of the root of the word, but its algorithms are
based on other principles. Therefore, the word after processing by the stemming algo-
rithm may differ from the morphological root of the word. Stemming is used in lin-
guistic morphology and in the information search. There are two types of errors in the
stemming algorithms: overstemming; understemming. Overstemming is when, during
this process, the two words are reduced to one basis, although this should not happen.
Understemming is the opposite mistake when words get different bases, where they
should have one common. Stemming algorithms try to minimize similar errors, but
reducing one type of errors can lead to an increase in the errors of another.
2.2 Analysis of recent research and publications
In [1], the task of developing methods for processing information resources of intel-
lectual Web-systems that were scientifically sound was considered and solved. Based
on these methods, technical tools that allowed the creation, distribution and develop-
ment of e-business systems were developed. To accomplish this task, a study was
conducted, the main purpose of which was to determine the peculiarities in the pro-
cess of processing information resources and finding patterns and dependencies in
such processes. After analyzing these factors, we can conclude that there is a contra-
diction between the active development and the spread of applied information systems
and IT in general on the one hand, and a small amount of research on this topic on the
other [20, 28-33]. It is precisely because of this contradiction that we can see the ex-
isting problem of stalling the innovation development of this sector, which develops
much slower than the whole IT sphere, which negatively affects both the growth rates
of applied research on this topic. The concept of adaptive ontology, which is used to
form mathematical models of the functioning of intelligent systems based on ontolo-
gies, was introduced in [4-5]. After analyzing works [4-8, 21-24] in general, one can
conclude that the field of development and use of ontologies in the creation of applied
systems and research on this topic are actively developing. However, fundamental
research on the use of ontologies to make optimal decisions based on the development
of information resources in a particular subject area is still absent. Such decisions are
based on ontology training methods. Scientific researches that study and develop
information processing systems and use the ontologies in the development and opera-
tion of information systems in general began at the end of the twentieth century. Peo-
ple and works that have developed the basic theoretical foundations of mathematical
models of ontologies include [25-36]:
J. Salton, T. Gruber, who were the first to suggest using a three-dimensional tuple
as a way of representing an ontology;
N. Guarino, M. Shambard, who in their works identified the ways of develop-
ment of ontologies, as well as methods of their construction;
J. Owl, who first introduced such a concept as a conceptual graph;
M. Montes-Gomez, who used a conceptual graph to represent ontologies.
In the works of R. Knapp, E. Kaufman, I.P. Norenkova, E. Meena, A. Kalli, M. Yu.
Uvarova, we can find a description of exactly how to use ontologies in applied solu-
tions. And T. Andreasen, T. Berners-Lee, O. Lazsil, O.V. Palagin, AV Anisimov,
considered the problem of building systems that would be based on ontologies, in-
cluding the problems of working out the Ukrainian language in information resources.
After analyzing the research, carried out by foreign specialists in the field of pro-
cessing on-line information resources, we can conclude that the main aspects of learn-
ing ontologies are [37-45]:
Assess of the quality of the ontology.
Building of ontologies, using knowledge derived from heterogeneous sources.
Integrations of different ontologies within a common subject area.
The main areas of research that are associated with the use of intellectual resources
for the development of information resources and ontology training are [46-51]:
Learning of ontologies using the analysis of natural language texts. [2, 11-13].
Building of decision support systems that would be based on ontologies [14].
Development of software tools that allow the development of ontologies in man-
ual and automated mode (OntoEdit, Ontosaurus, Protégé) [15-17].
Implementation of applications based on ontology and queries to the knowledge
base [2, 18].
Development of ontology description languages (XML, OWL, RDF, DAML
+OIL) [19].
2.3 Analysis of the available software products
After looking for available software products, one can come to conclusion that the
market for text rubrication is pretty empty. Most solutions are proprietary, or they are
designed purely for scientific research. In this case, the quantity of products that
would be available to a large mass of people is very small. All found products have a
number of common disadvantages: only English is available, bad user interface and
lack of storage of the result. One of such products is «uClassify». The lack of lan-
guages, other than English, can negatively affect the popularity of the product, since a
very large part of the potential audience is not attracted by such a solution.In «uClas-
sify» there is one more significant minus - a low number of rubrics. The small number
of rubrics is also a big disadvantage, since the presentation of the rubrics is very gen-
eral and the rubricating within such a limited number of topics is very superficial. The
superficiality of rubrication is bad, because the authors of the content usually write all
their articles in one of the global rubric (sports, art, science, etc.) and they are inter-
ested in a deeper rubrication of the text, usually performed manually and being very
time consuming. The problem with this user interface (Fig. 1) is that it is completely
unsuitable for writing large-size articles. That is, the user will have to write his article
in a text editor and then copy it in the appropriate field of input on the site. This nega-
tively affects user experience, which significantly reduces the chance that the user
will continue to use this software product. Another product is «twinword» company
service. This service also supports only English and has a non-convenient user inter-
face (Fig. 2). As in the previous product, the text input field is completely unsuitable
for writing the large amount of textual content. It is worth highlighting that even
though this site is rubricating better than the previous one, the way to present the re-
sult of the rubrication is counter-intuitive and many users may not find the results
they need. It is worth noting another common minus both for "twinword" and "uClas-
sify" is the lack of the ability to save results. For people who generate a large amount
of text content, it is very important to save this content, as well as the rubrics to which
this or that article relates. Consequently, the available systems have flaws regarding
the number of languages, the usability, as well as in the ability to store the results.
Fig. 1. The user interface of the service «uClassify»
Fig. 2. The user interface of the service «twinword»
3 The object system analysis in the research
The main purpose of the system being developed is the textual content rubrication.
We break this goal into smaller goals, which in turn are divided into sub-goals:
1. Interaction with users: authorization, request for rubrics, viewing of the sectioned
articles and filtering of articles by categories.
2. Content rubrication: processing of request for rubric, search for entities in the arti-
cle and analysis of found entities.
3. Defining of rubrics for articles.
Fig. 3. Tree of objectives for information systems categorizing the textual content
In the developed system, two main external entities can be distinguished: User and
Dandelion API. The user has access to his personal profile on the system, and can add
his own articles to the system, which will be grouped by the system. Dandelion API is
a data source for the system. With this external resource, you can get well structured
data from DBpedia, namely the essentials contained in the article being analyzed.
DBpedia is a large crowdsourcing project whose main goal is to structure all the
data that is located in Wikipedia. DBpedia itself can be considered the largest ontolo-
gy in the world with open access. General context diagram of the information system
is shown on the Fig. 4. For a better understanding of the work of the system, a de-
composition has been made which will help to depict how our system works internal-
ly and how data is transmitted between components of the system.
Fig. 4. General contextual diagram
On Fig. 5.rubrics are shown. In the process of elaboration three main functions were
selected: user authentication, processing user queries, and analyzing the text. The user
authorization feature allows guests to log on to the system. What will allow them to
get into their personal cabinet, and attribute the article to the rubric? The function of
processing user queries is responsible for routing all queries between the user and the
internal functions of the system. The text analysis function corresponds to both the
processing of the text and the connection with the external data source, from where
we get the essentials needed for the rubrication of the text.
Fig. 5. The detailed diagram
In the text analysis process we can distinguish three main functions: preparing a re-
quest, connection with external resources, analysis of entities (Fig. 6).
Fig. 6. Detailed text analysis function chart
The request preparation implies sending an inquiry to Dandelion API. The better re-
quest is formed, the better results we obtain from the external resource. This function
directly affects the quality of our system. This function also directly influences the
performance of the system, the better it will work, the more requests the system can
handle per unit of time. Next is the entity, which is responsible for the analysis of data
received from the external resource and defining categories. This function is the cor-
nerstone of the system. In the process of specification of the query preparation, we get
3 main functions: language definition, text splitting, and installation of additional
parameters (Fig. 7). The language definition function is responsible for correctly de-
termining the language of the article, which is very important for the correct analysis
of the article.
Fig. 7. The chart for request preparation
The text splitting function is responsible for the preliminary preparation of the text
before being sent to the external resource. Since the DandelionAPI capabilities are
limited, we must adapt the text for the requirements of this resource.
The function of setting additional parameters is responsible for setting additional
parameters that the DandelionAPI takes, for example, or when looking for entities to
focus more on the context (which is good for a large article) or on the keywords
themselves. In the process of specification of the analysis of entities functions, we get
three main functions: the definition of the most relevant entities, getting the catego-
ries, the definition of the most relevant categories (Fig. 8).
Fig. 8. The chart of entities analysis
The function determining the most relevant entities accepts the DandelionAPI entities
and processes them. The function of obtaining categories is responsible for receiving
categories of entities, as well as the proper structuring of these categories. The func-
tion of determining the most relevant categories takes categories and analyzes them.
In the process of analysis is determined, which categories will be the most appropriate
for article entered by the user. In the process of specification, we obtain two basic
functions: the rejection of entities with low confidence and the rejection of entities
that are an alternative name (Fig. 9).
Fig. 9. The chart of the most relevant entities determination
The function of dropping entities with low confidence accepts the DandelionAPI enti-
ties and processes them. During the process, the acceptable level of confidence in the
relevance to the essence is determined, and all the entities that have the result below
this level are rejected. The function of discarding entities is responsible for finding
duplicates of the same entities (e.g. auto and car). In the process of elaboration of the
function of determining the most rewarding categories, we obtain two main functions:
the same categories weighting and the categories rejection with low weight (Fig. 10).
Fig. 10. The chart of entities weighting and selection
The function of weighting the same categories is responsible for the combination of
weight categories that were repeated in several entities at the same time.
The category weight function defines the most optimal rated category, and all cate-
gories that have the result below are rejected. Figure 11 depicts the hierarchy of tasks
of the informational system of textual content rubrication.
Fig. 11. The hierarchy of objectives for information system categorizing textual content
The root is the main purpose - the rubricated article. To accomplish this goal, we need
information about the author of the article, the text of the article and the topics to
which it relates. For a proper analysis of the article text, we need to know the lan-
guage in which it is written and what entities are present in this text. For the proper
use of entities we need to know its relevancy ratio for this article. On the other side of
the tree, we can see the rubrics that are made of categories. Categories in their turn
include entities, and also have their weight.
4 The choice and justification for software platform
The system which categorizes the textual content is best conceived as a web-
application. Therefore, the development of this system requires reliable and fast plat-
form that would support Web. Since most modern systems are web-oriented choice is
very large.
In order to build an optimal system we need to consider several platforms and se-
lect one which fits into our goals. For our system are the following requirements:
High throughput for client requests.
Low time serialization data.
High performance.
The ability to use different types of servers.
Easy maintenance.
Among the above platforms were selected Node.js, because the requirements of a
particular system it is best suited. To implement the entire system was chosen MERN
(MongoDB, Express.js, React.js, Node.js) stack. In fig. 12 can see the overall struc-
ture of the system.
Fig. 12. The scheme of software based on a system MERN-stack
So, on the client side is used React.js, in conjunction with ordinary HTML / CSS. in
order to use the latter we utilize the method Babel. Babel - is a compiler that translates
any dialect of JavaScript, including CoffeeScript, TypeScript and other add-on lan-
guage JavaScript ES5, which is supported by almost all browsers including IE8. The
power of Babel is in its modularity and extensibility through plug-ins.
To speed up the download of our resources at the request of the client, we will use
a minification. This is an approach where the code removed all unnecessary gaps and
spaces are removed comments, and big names of variables and methods vary for less.
On the server side is used due Node.js + Express. Node allows the postponed oper-
ation records into the database, while continuing to work in this mode, as if these
recordings successful. To process client requests for server we use Express.js.
Express is a minimalist and flexible framework for Web applications built on
Node.js, providing a wide range of functionality. Having to his disposal many HTTP-
assisted methods and intermediate handlers, the robust API can be easily and quickly
created. For the database was selected MongoDB. This is document-oriented database
(RDBMS). MongoDB occupies a niche between fast and scalable systems that oper-
ate on data in the format of a key / value, and relational DBMS, functional and con-
venient in forming queries. MongoDB has developed a new approach to building
database with no tables, charts, queries, SQL, foreign keys and many other things that
are inherent in object-relational databases. So the basic architectural idea of the sys-
tem is the use of isomorphic JavaScript. This means that the whole application is writ-
ten only in that language. This approach was chosen because JavaScript is ideally
suited for this system, since it is asynchronous and non-blocking I / O to allow this
system to be used by a large number of users while maintaining high system perfor-
mance. Also, this approach will allow us to quickly develop a core part of the func-
tionality that will be fully operational, after which developers will eventually be able
to easily add new features to the system, in the form of new modules.
5 The description of text categorization software
Figure 13 depicts an "essence-to-link" diagram for the information system of text-
based content rubricating. The user writes articles that are read from Essences and are
described by Category. In this, the Essences are categorized.
Fig. 13. Chart "entity-relationship" information system for categorization of text content.
To operate the system requires a constant connection to the server. The server is re-
quired to connect to the Internet because the system uses external data. The whole
system was written using language JavaScript, and is based on technology stack
MERN. The user interface is presented in a web page where users may access the
system. The purpose of the program is automating the rubricating of text content. This
problem is quite important in today's realities, since handwriting is a long process, and
correctly-selected rubrics improve both search optimization and the overall experi-
ence of using the system. The system being developed also uses the open Dandelion
API. The Dandelion API processes receive text, and find the basic essence, and then
receives data from DBpedia about the received entity and returns this data to our sys-
tem. This API is actively developed in DBpedia, and stably supports 7 languages, and
more than 50 are in the beta stage. This will allow our system to be used by a large
number of people, since it has the ability to articulate text in many languages.
It's worth noting that DBpedia is a structured version of knowledge presented in
Wikipedia, which means that we can process articles on a variety of topics. Most
systems that deal with text content rubrication have a common problem in that they
can categorize articles in just one area of knowledge, but DBpedia has no such prob-
lem. The most important part of the whole process that occurs on the server is pro-
cessing the received entities. For large text Dandelion API returns a large number of
entities and should identify the category that would fit simultaneously to a large num-
ber of entities. The task involves the following steps:
1. Determining the language of the article.
2. Breaking the text as DandelionAPI is limited to the number of characters that are
sent for processing.
3. Extra options are provided such as entities or when searching for more focus on the
context (which is good for large articles) or the keywords themselves. Also, addi-
tional parameters may be represented by the maximum number of entities, confi-
dence limit for each entity and country of the author.
4. Sending request to Dandelion API, including entities from the article.
5. Discarding entities with low confidence. During the process of determining the
most optimal level of confidence relevant to the essence, and all the entities that
have the result below are rejected.
6. Finding and removing entities, which are the alternative name of another entity.
7. Getting the categories of entities.
8. Combining weights for categories, which are repeated in a number of entities sim-
ultaneously.
9. Determining the threshold for weight, and all categories that are below the desired
result are discarded.
10.Use categories that have reached this step, as a category for the article.
There are many popular sites for blogs / articles, which don’t provide an automatic
categorization, and the category must be manually put by the authors, or site admin-
istration. Using this system, the category will be assigned automatically. In addition to
simplifying the life of the author of articles, it will simplify the lives of people who
read this article, because in this case, the reader can immediately understand the
theme of the article. It can also help in filtering articles by topics. The web – applica-
tion categorizing the textual content can be used not only in browsers PCs or laptops,
but also on mobile browsers. After authentication, the user opens the home page.
Above, in the so-called "headers" site posted menu. The menu consists of three but-
tons: Home; Create - a page for writing; Profile – the private office of user. The sys-
tem automatically determines the categories of article and assigns them to this article.
The user can see both the home page and his private office. After authorization, the
user has the opportunity to do the following: view revised articles; filter articles by
categories; writing an article for its categorization. See article user to the home page
where he can see the cards with basic information about the article, namely, author,
subject, category, date of writing (Error! Reference source not found. 14). To the
left is a filter card where the user can enter the rubrics that interest him, and the sys-
tem will filter records for these categories (Error! Reference source not found. 15).
Fig. 14. The home page of application
Fig. 15. Simple filter application
To write an article, the user can go to the "Create" tab. This tab has a form with two
input fields. In the first field, the author can enter the name of his work. In the second
field, the author introduces the text of his work. After that, the author clicks on the
"Submit" button and this article is shown by the system and falls into the list of all
articles. We consider four very popular language among the authors of the text con-
tent: English, German, Italian and French. Each author will work the same way. First,
each author must authorize. After that, the author can go to the "Create" page, where
he can enter the name and the body of his article. After the system has broken down
his article, the author can review its text, sections, and can delete it. First, consider the
example of the system in English, German, Italian and French (Fig. 16-19).
Fig. 16. Viewing results of English text categorization
Fig. 17. Viewing results of German text categorization
Fig. 18. Viewing results of Italian text categorization
Fig. 19. Viewing results of French text categorization
6 Conclusion
The implementation of the task of automating textual content rubrication allows pro-
cessing the large amounts of textual data and filter it before manual analysis by ex-
perts. In process of research the key concepts of the topic were studied, and the analy-
sis of recent research and sources was carried out. Also, an analysis and comparison
of the available software was conducted, and common deficiencies were found. The
system analysis and design of the information system provided the basis for system
development. In the course of this analysis, a tree of goals was developed, functional
diagrams were also created reflect the data flows inside the system. At the end of the
system analysis, a task hierarchy was constructed, which shows what tasks should be
executed in order to achieve the final result. As a platform for textual content catego-
rization software, the MERN-stack technology stack was selected. The description of
the finished software product described with examples of categorization of texts com-
ing from different languages is provided.
References
1. Alipanah, N., Parveen, P., Khan, L., Thuraisingham, B.: Ontology-driven query expansion
using map/reduce framework to facilitate federated queries. In: Proc. of the International
Conference on Web Services (ICWS), 712-713. (2011).
2. Euzenat, J., Shvaiko P.: Ontology Matching. In: Springer, Heidelberg, Germany, (2007).
3. Maedche, A., Staab, S.: Measuring Similarity between Ontologies. In: Knowledge Engi-
neering and Knowledge Management, 251-263. (2002).
4. Xue, X., Wang, Y., Hao, W.: Optimizing Ontology Alignments by using NSGA-II. In: The
International Arab Journal of Information Technology, 12(2), 176-182. (2015).
5. Martinez-Gil, J., Alba, E., Aldana-Montes, J.F.: Optimizing ontology alignments by using
genetic algorithms. In: The workshop on nature based reasoning for the semantic Web,
Karlsruhe, Germany. (2008).
6. Calvaneze, D.: Optimizing ontology-based data access. KRDB Research Centre for
Knowledge and Data. In: Free University of Bozen-Bolzano, Italy. (2013).
7. Gottlob, G., Orsi, G., Pieris, A.: Ontological queries: Rewriting and optimization. In: Data
Engineering, 2-13. (2011).
8. Li, Y., Heflin, J.: Query optimization for ontology-based information integration. In: In-
formation and knowledge management, 1369-1372. (2010).
9. Keet, C.M., Ławrynowicz, A., d’Amato, C., Hilario, M.: Modeling issues & choices in the
data mining optimization ontology. (2013).
10. Keet, C.M., Ławrynowicz, A., d’Amato, C., Kalousis, A., Nguyen, P., Palma, R., Stevens,
R., Hilario, M.: The data mining Optimization ontology. In: Web Semantics: Science, Ser-
vices and Agents on the World Wide Web, 32, 43-53. (2015).
11. Montes-y-Gómez, M., Gelbukh, A., López-López, A.: Comparison of Conceptual Graphs.
In: Artificial Intelligence, 1793. (2000).
12. Biggs, N., Lloyd, E., Wilson, R.: Graph Theory. In: Oxford UP, 1736-1936. (1986).
13. Bondy, J., Murty, U.: Graph Theory. In: Springer. (2008).
14. Lytvyn, V., Sharonova, N., Hamon, T., Vysotska, V., Grabar, N., Kowalska-Styczen, A.:
Computational linguistics and intelligent systems. In: CEUR Workshop Proceedings, Vol-
2136 (2018)
15. Vysotska, V., Fernandes, V.B., Emmerich, M.: Web content support method in electronic
business systems. In: CEUR Workshop Proceedings, Vol-2136, 20-41 (2018)
16. Kanishcheva, O., Vysotska, V., Chyrun, L., Gozhyj, A.: Method of Integration and Con-
tent Management of the Information Resources Network. In: Advances in Intelligent Sys-
tems and Computing, 689, Springer, 204-216 (2018)
17. Gozhyj, A., Vysotska, V., Yevseyeva, I., Kalinina, I., Gozhyj, V.: Web Resources Man-
agement Method Based on Intelligent Technologies, Advances in Intelligent Systems and
Computing, 871, 206-221 (2019)
18. Korobchinsky, M., Vysotska, V., Chyrun, L., Chyrun, L.: Peculiarities of Content Forming
and Analysis in Internet Newspaper Covering Music News, In: Computer Science and In-
formation Technologies, Proc. of the Int. Conf. CSIT, 52-57 (2017).
19. Naum, O., Chyrun, L., Kanishcheva, O., Vysotska, V.: Intellectual System Design for
Content Formation. In: Computer Science and Information Technologies, Proc. of the Int.
Conf. CSIT, 131-138 (2017)
20. Vysotska, V., Lytvyn, V., Burov, Y., Gozhyj, A., Makara, S.: The consolidated infor-
mation web-resource about pharmacy networks in city. In: CEUR Workshop Proceedings
(Computational linguistics and intelligent systems), 2255, 239-255. (2018).
21. Lytvyn, V., Vysotska, V., Burov, Y., Veres, O., Rishnyak, I.: The Contextual Search
Method Based on Domain Thesaurus. In: Advances in Intelligent Systems and Computing,
689, 310-319 (2018)
22. Lytvyn, V., Vysotska, V.: Designing architecture of electronic content commerce system.
In: Computer Science and Information Technologies, Proc. of the X-th Int. Conf.
CSIT’2015, 115-119 (2015)
23. Furgala,Y., Rusyn, B.: Peculiarities of melin transform application to symbol recognition.
In: Advanced Trends in Radioelectronics,Telecommunications and Computer Engineering,
251-254. (2018)
24. Kapustiy, B., Rusyn, B., Tayanov, V.: Peculiarities of application of statistical detection
criteria for problems of pattern recognition. In: Journal of Automation and Information
Science, 37(2), 30-36. (2005)
25. Lytvyn, V., Vysotska, V., Uhryn, D., Hrendus, M., Naum, O.: Analysis of statistical meth-
ods for stable combinations determination of keywords identification. In: Eastern-
European Journal of Enterprise Technologies, 2/2(92), 23-37 (2018)
26. Rusyn, B., Lytvyn, V., Vysotska, V., Emmerich, M., Pohreliuk, L.: The Virtual Library
System Design and Development, Advances in Intelligent Systems and Computing, 871,
328-349 (2019)
27. Gozhyj, A., Kalinina, I., Vysotska, V., Gozhyj, V.: The method of web-resources man-
agement under conditions of uncertainty based on fuzzy logic. In: International Scientific
and Technical Conference on Computer Sciences and Information Technologies, CSIT,
343-346 (2018)
28. Rusyn, B., Kosarevych, R., Lutsyk, O., Korniy, V.: Segmentation of atmospheric clouds
images obtained by remote sensing. In: Advanced Trends in Radioelectron-
ics,Telecommunications and Computer Engineering, 213-216. (2018)
29. Rusyn, B., Prudyus, I., Ostap, V.: Fingerprint image enhancement algorithm. In: The Ex-
perience of Designing and Application of CAD Systems in Microelectronics, CADSM,
193-194. (2001)
30. Su, J., Sachenko, A., Lytvyn, V., Vysotska, V., Dosyn, D.: Model of Touristic Information
Resources Integration According to User Needs. In: International Scientific and Technical
Conference on Computer Sciences and Information Technologies, 113-116 (2018)
31. Lytvyn, V., Peleshchak, I., Vysotska, V., Peleshchak, R.: Satellite spectral information
recognition based on the synthesis of modified dynamic neural networks and holographic
data processing techniques, 2018 IEEE 13th International Scientific and Technical Confer-
ence on Computer Sciences and Information Technologies, CSIT, 330-334 (2018)
32. Kapustiy, B., Rusyn, B., Tayanov, V.: A new approach to determination of correct recog-
nition probability of set objects. In: Upravlyaushchie Sistemy i Mashiny, 2, 8-12. (2005)
33. Varetskyy, Y., Rusyn, B., Molga, A., Ignatovych, A.: A new method of fingerprint key
protection of grid credential. In: Advances in Intelligent and Soft Computing, 84, 99-103.
(2010)
34. Gozhyj, A., Chyrun, L., Kowalska-Styczen, A., Lozynska, O.: Uniform Method of Opera-
tive Content Management in Web Systems. In: CEUR Workshop Proceedings (Computa-
tional linguistics and intelligent systems, 2136, 62-77. (2018).
35. Kravets, P.: The control agent with fuzzy logic, Perspective Technologies and Methods in
MEMS Design, MEMSTECH'2010, 40-41 (2010)
36. Davydov, M., Lozynska, O.: Information System for Translation into Ukrainian Sign Lan-
guage on Mobile Devices. In: Computer Science and Information Technologies, Proc. of
the Int. Conf. CSIT, 48-51 (2017).
37. Davydov, M., Lozynska, O.: Linguistic Models of Assistive Computer Technologies for
Cognition and Communication. In: Computer Science and Information Technologies,
Proc. of the Int. Conf. CSIT, 171-175 (2017)
38. Vysotska, V., Hasko, R., Kuchkovskiy, V.: Process analysis in electronic content com-
merce system. In: 2015 Xth International Scientific and Technical Conference Computer
Sciences and Information Technologies (CSIT), 120-123. (2015).
39. Rusyn, B., Torska, R., Kobasyar, M.: Application of the cellular automata for obtaining
pitting images during simulation process of thei growth. In: Advances in Intelligent Sys-
tems and Computing, 242, .299-306. (2014)
40. Martin, D., del Toro, R., Haber, R., Dorronsoro, J.: Optimal tuning of a networked linear
controller using a multi-objective genetic algorithm and its application to one complex
electromechanical process. In: International Journal of Innovative Computing, Information
and Control, Vol. 5/10(B), 3405-3414. (2009).
41. Precup, R.-E., David, R.-C., Petriu, E.M., Preitl, S., Rădac, M.-B.: Fuzzy logic-based
adaptive gravitational search algorithm for optimal tuning of fuzzy controlled servo sys-
tems. In: IET Control Theory & Applications, Vol. 7(1), 99-107. (2013).
42. Ramirez-Ortegon, M.A., Margner, V., Cuevas, E., Rojas, R.: An optimization for binariza-
tion methods by removing binary artifacts. In: Pattern Recognition Letters, 34(11), 1299-
1306. (2013).
43. Solos, I.P., Tassopoulos, I.X., Beligiannis, G. N.: Optimizing Shift Scheduling for Tank
Trucks Using an Effective Stochastic Variable Neighbourhood Approach. In: International
Journal of Artificial Intelligence, 14(1), 1-26. (2016).
44. Martin, D., del Toro, R., Haber, R., Dorronsoro, J.: Optimal tuning of a networked linear
controller using a multi- In: objective genetic algorithm and its application to one complex
electromechanical process. In: International Journal of Innovative Computing, Information
and Control, Vol. 5/10(B), 3405-3414. (2009).
45. Nazarkevych, M., Klyujnyk, I., Nazarkevych, H.: Investigation the Ateb-Gabor Filter in
Biometric Security Systems. In: Data Stream Mining & Processing, 580-583. (2018).
46. Nazarkevych, M., Klyujnyk, I., Maslanych, I., Havrysh, B., Nazarkevych, H.: Image filtra-
tion using the Ateb-Gabor filter in the biometric security systems. In: International Confer-
ence on Perspective Technologies and Methods in MEMS Design, 276-279. (2018).
47. Nazarkevych, M., Buriachok, V., Lotoshynska, N., Dmytryk, S.: Research of Ateb-Gabor
Filter in Biometric Protection Systems. In: International Scientific and Technical Confer-
ence on Computer Sciences and Information Technologies (CSIT), 310-313. (2018).
48. Nazarkevych, M., Oliarnyk, R., Dmytruk, S.: An images filtration using the Ateb-Gabor
method. In: Computer Sciences and Information Technologies (CSIT), 208-211. (2017).
49. Nazarkevych, M., Kynash, Y., Oliarnyk, R., Klyujnyk, I., Nazarkevych, H.: Application
perfected wave tracing algorithm. In: Ukraine Conference on Electrical and Computer En-
gineering (UKRCON), 1011-1014. (2017).
50. Nazarkevych, M., Oliarnyk, R., Troyan, O., Nazarkevych, H.: Data protection based on
encryption using Ateb-functions. In: International Scientific and Technical Conference
Computer Sciences and Information Technologies (CSIT), 30-32. (2016).
51. Babichev, S., Gozhyj, A., Kornelyuk A., Litvinenko, V.: Objective clustering inductive
technology of gene expression profiles based on SOTA clustering algorithm. In: Biopoly-
mers and Cell, 33(5), 379–392. (2017)