PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

The paper presents a new kind of Software Architecture, based on Semantic Network- Semantic Network Based Architecture (SNBA). This is a realization of an old idea briefly presented in [1]. The basic features of the Semantic Network construction for SNBA are described as another insight into the understanding of semantics. Some comparisons with existing architectures are made. A new SNBA-approach for IS building and the platform for this are described. The future SNBA-work is presented as a framework of intentions.
1
Semantic Network Based Architecture
Lyubomir Blagoev, INATO Ltd,
lyuboblagoev@gmail.com
Tihomir Blagoev, INATO Ltd
tihomir.blagoev87@gmail.com
Abstract
The paper presents a new kind of Software Architecture, based on Semantic
Network- Semantic Network Based Architecture (SNBA). This is a realization of an
old idea briefly presented in [1]. The basic features of the Semantic Network
construction for SNBA are described as another insight into the understanding of
semantics. Some comparisons with existing architectures are made. A new SNBA-
approach for IS building and the platform for this are described. The future SNBA-
work is presented as a framework of intentions.
Keywords: Semantic Network, Software Architecture, Semantic Interoperability, AI, NLP.
Construction of Semantic Network for Semantic Network Based Architecture
The nodes and arcs in the Semantic Network for Semantic Network Based Architecture (SNBA)
are treated as objects. The object in the SNBA is a set of data, which can be identified, created
and deleted as a whole item.
The SNBA-object can be a single or composite object. The SNBA-arcs are single objects but
SNBA-nodes are composite objects, consisting of at least one arc and one node.
Every SNBA-object is described by his model. The SNBA-model consists of data, which
describes the model instances or can be used to prove that given object is an instance of the
corresponding model. The correlation between instance and his model is pointed out in SNBA by
the notation Instance[Model].
The SNBA can be understood as a Semantic Network consisting of models and their instances
presented by objects, which are nodes and arcs in the Semantic Network graph. In the SNBA
exists one root object- Model[Model], which is a model and instance of itself. Every object in
SNBA is an instance of the corresponding model and in turn can be a model for other objects,
which will be his instances.
The semantic presentation in SNBA-environment is a little bit different from other ways of
semantic presentation. Let us take an example from [2]:
Figure 1: The connection between „Adam“ and „Ben“ with familiar
presentation as a part from Semantic Network
2
The same semantic will have the SNBA-presentation as follows:
Figure 2: „Adam“ and „Ben“ with SNBA- presentation of their
semantic connection
Somewhere in the SNBA-environment are defined Man and Dog as a „living being“ (Man[Living
being]; Dog[Living being]), and Living being is defined (let us say at the end“) as a „model“-
Living being[Model]. It is necessary of course similar definition to be provided for Arc as a
„model“- Arc[Model].
The model Man[Living being] describes that the Objects[Man] have properties for given name,
family name and so on. The model Dog[Living being] describes that the Objects[Dog] have
properties for name, breed and so on. All of those properties have their model definitions with
model chains ending into Model[Model].
Objects identification in SNBA-environment
The identification of an object made by humans is supported by two properties- name and
explanation. The name is an expression which presents in short the semantic, which is
embedded in the object. The names have to be short and easy for human communication.
Because of that they can’t express the whole semantic of the object. This is the reason why it is
necessary additional information supported by explanation.
The human identification of the SNBA-arcs doesn’t need the usage of the pointed above
properties. This identification is provided as a composition from the human identifications of the
nodes, which are connected by the corresponding arcs. The composition is a text like “..the arc
connecting node A and node B…”. The sequence of so pointed nodes presents the direction of
the arc. The SNBA-arcs as objects consist of two obligatory properties- the object identifiers of
the source and of the target node.
The SNBA-objects can have names and explanations written in different human languages.
The SNBA-objects used as models have a property for machine identification- model identifier.
SNBA as an environment with embedded Semantic Interoperability
The coexistence of the model and their instances in the same environment ensures the semantic
presentation of every SNBA-object. This guarantees the semantic equivalence of the objects
which are instances of the common model.
The same equivalence can be provided and for different SNBA-environments, if they share
common models of objects. This means that Information systems with SNBA which share
common models of objects will be semantically interoperable.
The process of model sharing needs the corresponding tools:
A. Semantic Repository
The Semantic Repository (SR) itself is a specialized SNBA Information system which
provides:
Models storage
Public access to stored models
Exchange of models with external consumers or suppliers of models
Semantic Clearing Process
Adam[Man]
Ben[Dog]
Owns[Arc]
3
B. Semantic Clearing Process
The purpose of the Semantic Clearing Process is to ensure:
Uniqueness of identification of models
The necessity of the model existence/usage and the sufficiency of description of a
model
The activities in the Semantic Clearing Process can be and will be better to be supported
by the Information system of the Semantic Repository.
The embedding of Semantic Interoperability is easy to be defined but is very difficult to be done.
This article is not the place where this has to be explained. The achievement of Semantic
Interoperability in IoT by transfer from e-Governance maybe is the solution of the problem as it is
pointed out in [3]. The environment of e-Governance is really the most suitable place where the
Semantic Interoperability to be born- [4] and [5].
Another insight into the understanding of semantics
As it was pointed above the semantic embedded into models supports their identification and
presents the meaning of their instances as a notion. This isn’t enough when the instance is a
data construction. In this case the model has to consist of more properties (that is to say more
semantic), as follows:
Description of data construction
Description of data content presentation
Description of data content visualization
There exist enough tools for data construction description. But the existing tools for description
of data content presentation can’t cover the whole amount of possible cases. Because of that it
is better to use the text form of description with Natural Language. The same is the problem with
Description of data content visualization. The usage of Natural Language is the only possible
choice.
Those two descriptions in Natural Language will be used by programmers when they develop a
software support of data content creation and data content visualization. The programmers write
program code which will do the things, presented with the two descriptions. That program code
will be interpreted by program code processor.
But in case there existed a Natural Language processor it would be possible to interpret textual
descriptions to lead to the same results as program code interpretation.
The semantic equivalence of those two kinds of presentation ensures a way to define pieces of
program code, based on semantic definitions, embedded in SNBA-models.
SNBA decentralizes program code configuration
The base problem in program code decentralization is how to define the pieces of the frittered
program code. In SNBA environment this problem has a natural and easily managed decision,
based on naturally defined pieces of semantic, embedded in SNBA-models.
Embedding program code into SNBA-models transforms their instances from passive into active
system architecture components. Such active components can collaborate among and this
becomes a powerful tool to create information system components with collective behavior. An
example for such collaboration is given in [6].
Of course the SNBA-models can be equipped with different program code realizations, written in
different languages.
This decentralization is a good basis for API definitions which will ensure external access to
SNBA-system functionality known as „Microservices“, or a variant of the service-oriented
architecture (SOA).
4
SNBA and Model Driven Architecture
The developed by Object Management Group (OMG) Model Driven Architecture (MDA) is based
on three steps of information systems development [7]. Those steps are the creation of two
models- Platform-Independent Model (PIM) and Platform-Specific Model (PSM) and eventually
the Generation of the Application.
The MDA-approach is academically clear- the first level (PIM) represents the semantic
supported by the developed Information system in extremely pure form. But after this step, the
beginning PIM-model is transformed two times. Every transformation leads to more detailed
semantic presentation. Usually in these processes are discovered some errors, discrepancies
and so on in the initial PIM-model.
The two models and the generated application/system are supported by different environments
and in different presentations. The correction of the discovered errors/discrepancies needs
efforts in the reverse direction- from application/system to the PIM-model. The real practice
points out that this process of correction propagation is not made as is necessary. So at the end
of application/system development usually there is an essential difference between the three
levels of semantic description.
In SNBA exist in explicit/ apparent form first of all only PIM of the application/system and the
generated program code for it. It is possible to be recognized and some kind of PSM-
presentation, but for SNBA this is not important. The essential difference is that the program
code is a part from semantic presentation in the SNBA-models. It means that if some corrections
have to be done in program code those corrections have to be described into corresponding
parts of model, or models. This way SNBA stimulates developers to create well documented
applications/systems by creation of semantically completed models.
SNBA as an approach for Information systems building
The first step of the approach is the same as in MDA- models building. This process is being
made easier by using the available models in the Semantic Repository. It means that for new
Information systems have to be created only models, which don’t exist into repository. The
existing models will be used without correction. If some correction is needed it means that has to
be created a new kind of information object, irrespective of similarity with the old one. The so
created new models will be put into Semantic Repository and will be a ready „material“ for new
systems building.
This way, the building of Information systems is based on of suitable models selection from
Semantic Repository. This means that when for the new system is necessary to change the
language for text presentation or language for program codding this will be done only for the
selected models, if this has not already been done.
Unified Platform for Innovations is a SNBA-platform
INATO Ltd created the Unified Platform for Innovations (UPI) which is SNBA-environment for
new SNBA-systems building. UPI contents Semantic Repository which INATO Ltd is ready to
share with other developers.
Till now INATO Ltd developed a specialized information system for control and monitoring of
processes in the discrete production. An interesting feature of the system is that the working
places in production are controlled as simple CNC-machines, but are monitored as systems with
autonomous behaviour. This is achieved by usage of models of some (deterministic) Finite
automata. The creation of those models is an easy task in SNBA-environment.
At the moment it is in progress a project for development of Administrative Information System
for e-Governance.
The so accumulated experience shows that the creation of UPI-based Information system
decreases with 20-30% the time and efforts for creation, even without usage of the existing
models in UPI Semantic Repository.
5
SNBA, UPI and Artificial Intelligence
What is Deep Learning without models, presenting the knowledge
What is Natural Language Processing without models of the meaning of words and
expressions…
It is possible to set more and more questions concerning Artificial Intelligence and all of them will
include the usage of models. Not usage in some abstract form, but usage in some (SNBA)
environment where models and their instances collaborate in accordance with built in them
semantic. UPI is the platform where this can happen.
And what about the programing language for AI in SNBA-environment- is it necessary a new
one?
We remembered the old strange language FORTH [8]. The language paradigm is based on the
meaning of the words which are the language primitives. And the meaning, the interpretation
and so on can be modeled in SNBA-environment.
Maybe FORTH, or maybe not- we are curious to try.
But let us experience the pleasure from research and development and after that we will share it!
Bibliography
[1]
Lyubo Blagoev, "Data organization in Information Systems as a Semantic network," CIO,
2013.
[2]
Pshtiwan Qader Rashid, "Semantic Network and Frame Knowledge Representation
Formalisms in Artificial Intelligence," Eastern Mediterranean University, 2015.
[3]
Lyubo Blagoev, Kamen Spassov, "The Role of e-Governance in IoT Semantic
Interoperability," 1st American University in the Emirates International Research Conference
2017, At Dubai, 2017.
[4]
Plamen Vatchkov, Kamen Spassov, Roumen Trifonov, Slavcho Manolov, Radoslav
Yoshinov, Lyubomir Blagoev, "Interoperability in Electronic Government Applications,"
Avangard Prima Publisher 2015, ISBN: 978-619-160-456-2, 2015.
[5]
Lyubo Blagoev, Kamen Spassov, "National Model of Data and Processes in Administration
(NMDPA) -Part of the Semantic Network of the Administration," International Conference
InfoTech, 2013.
[6]
Lyubo Blagoev, Kamen Spassov, "Smart home as a Digital Environment," Sofia University,
Spring Scientific Session of Faculty of Mathematics and Informatics, 2015.
[7]
Object Management Group, "Developing In OMG's Model-Driven Architecture," White paper,
2001.
[8]
Moore, Charles H. and Leach, G.C. , "FORTH A Language for Interactive Computing,"
Amsterdam, NY: Mohasco Industries Inc. (internal pub.), 1970.
ResearchGate has not been able to resolve any citations for this publication.
Book
Full-text available
The book is intended for the students at Sofia University and Technical University Sofia - undergraduate, graduate and PhD students in mathematics, informatics, computer science, computer systems and technologies, computer engineering, and software engineering who follow courses in e-governance, e-government and public administration. It is also addressed to employees in state and local administrations, ICT professionals, computer and software engineers working in the area of e-Governance.
Conference Paper
Full-text available
The article presents a new concept of Smart Home. It is based on the approach toward the contemporary home as an environment in which people and autonomous devices coexist. The paradigm of conventional integration of systems and devices is replaced with content exchange within a semantic network environment. The control of home related devices is replaced with ad hoc communication. The Smart Home notion is presented as a part of the Internet of things space. Important common issues between development of Smart Home and Internet of Things are discussed.
Conference Paper
Full-text available
The article presents the National Model of Data and Processes in Administration (NMDPA) as a part of the integrated system of models constructing the semantic network of the administration. Relationships between external models or configurations of models and models of NMDPA are discussed. Semantic characteristics of such configurations are described. It is given a general description of the semantic network of the administration within the national specifics of the semantic network of the country.
Data organization in Information Systems as a Semantic network
  • Lyubo Blagoev
Lyubo Blagoev, "Data organization in Information Systems as a Semantic network," CIO, 2013.
Semantic Network and Frame Knowledge Representation Formalisms in Artificial Intelligence
  • Pshtiwan Qader Rashid
Pshtiwan Qader Rashid, "Semantic Network and Frame Knowledge Representation Formalisms in Artificial Intelligence," Eastern Mediterranean University, 2015.
The Role of e-Governance in IoT Semantic Interoperability
  • Lyubo Blagoev
  • Kamen Spassov
Lyubo Blagoev, Kamen Spassov, "The Role of e-Governance in IoT Semantic Interoperability," 1st American University in the Emirates International Research Conference 2017, At Dubai, 2017.
Developing In OMG's Model-Driven Architecture
  • Object Management Group
Object Management Group, "Developing In OMG's Model-Driven Architecture," White paper, 2001.