Content uploaded by Grzegorz Nowakowski
Author content
All content in this area was uploaded by Grzegorz Nowakowski on Dec 22, 2023
Content may be subject to copyright.
The 12th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications
7-9 September, 2023, Dortmund, Germany
979-8-3503-5805-6/23/$31.00 ©2023 IEEE
Chatbots Lifecycle Support Platform
Grzegorz Nowakowski
1
, Sergii Telenyk
2
, Yevhenii Vovk
3
1,2
Faculty of Electrical and Computer Engineering, Cracow University of Technology, Cracow, Poland,
1gnowakowski@pk.edu.pl, 2 stelenyk@pk.edu.pl, https://www.pk.edu.pl
3 National Technical University of Ukraine “Igor Sikorsky Kyiv, Polytechnic Institute”, Kyiv, Ukraine,
3kafedra@acts.kpi.ua, kpi.ua/en
Abstract—This article proposes the concept of a
platform for the development, accumulation and use of
specialised applications – chatbots – that automate functions
related to providing information, placing orders and
fulfilling orders, and the implementation of multi-stage
processes using the capabilities of social networks and
messaging programs.
This platform is built on the basis of the approach,
according to which chatbots are built in several stages. For
each of the stages, there are formal models and methods
that ensure a high level of automation of work on the
creation, development and maintenance of chatbots.
Thus, the creation of a chatbot consists in a chain of
automated transformations of business processes into
transactional systems that allow describing the
corresponding state machines, and state machines into
formal means of their description in popular messengers.
Natural language processing models are used to
communicate with the bot and identify user intent.
The stages of describing the bot's state machine and the
transition to the formal definition of a chatbot in the
messenger are demonstrated on a real example.
Keywords—bots; formal systems; lifecycle support
platform; IT systems
I. INTRODUCTION
An important feature of the modern stage of
information technology (IT) development is the formation
of the IT environment and, according to many experts, the
IT society as a whole. Accordingly, people's demands for
service in establishments, institutions and organizations
designed to meet their needs are growing. Indeed, a
person's work, medical care, financial management,
education and daily needs are regulated by the template-
algorithmic processes of state institutions, banks, health
care institutions, enterprises, companies and other
organizations. For citizens, this means that it is necessary
to wait in queues, write official letters and contact the
employees of these institutions and establishments.
For example, consider the process of opening a bank
account. An experienced bank employee will quickly help
the client to carry out all the necessary steps of a certain
already-developed process for opening a bank account,
which takes into account all the peculiarities of the
institution, the client and legal aspects. However, this
requires many workers, and they must be given
appropriate pay. Similar situations occur daily in
educational institutions, the communal services of cities,
other organizations. People must be provided with
appropriate services, references and certificates. This
should be done by employees, and they carry out this
work by implementing fairly established processes that
govern the work but might introduce delays.
Hence, there exists a requirement for technologies
capable of managing the virtual interactions among
individuals, capable of performing a broad spectrum of
tasks. Typically, interaction tools comprise electronic
document management systems, educational services for
student-teacher interaction, email, specialized services for
searching individuals based on their interests, location,
and other characteristics, diverse messaging applications,
and social networks.
Given that the utilization of such tools in the
information society has disadvantages that can lead to
adverse outcomes for both the owners of these systems
and their users, suitable advisory programs, personal
dashboards, and other tools have emerged within the
information systems of banks, educational and medical
institutions, and governmental authorities. However,
creating such tools requires IT resources that sometimes
cost more than core business resources. Therefore,
effective tools for creating technologies for managing the
virtual interactions of people will be important for the
above-mentioned establishments, institutions and
organizations, which are designed to meet the needs of a
large flow of people. By investing in chatbots (a chatbot is
a computer program that can engage with the user through
auditory or textual methods [1,2]), they save money on
service workers. In order to ensure investment in the
technology of creating chatbots by banks, educational and
medical institutions and the authorities, these technologies
can be provided in the form of software as a service
(SaaS). The lessee of the system pays a comparatively
small yet fixed rental fee for utilizing the system as
outlined in a license agreement with the developer; this
fee is significantly lower than the expense associated with
developing such a system.
At the same time, the costs of acquiring and
maintaining this system, adding the necessary content,
scaling in case of an increase in the number of users will
be charged to the provider of the software services that
308
2023 IEEE 12th International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS) | 979-8-3503-5805-6/23/$31.00 ©2023 IEEE | DOI: 10.1109/IDAACS58523.2023.10348794
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
will appear on the market. It is obvious that the provider
of the software services, in turn, will pay for infrastructure
as a service (IaaS), which is currently available from other
providers.
Therefore, the provider of the software services will
need effective tools for creating technologies for
managing the virtual interactions of people. There are
tools on the market for creating technologies for managing
virtual interactions of people, which generally meet the
basic requirements of both users and service providers.
But they require improvement in the conditions of rapid
development of the IT environment and changes in the
organization of human activity in various fields [1,2].
Drawbacks for the user are linked to the utilization of
technologies for managing the virtual interactions among
individuals. Presently, users have access to online
management systems for nearly all aspects of their lives,
and each of these systems possesses its unique interface,
which doesn't always align with the user's accustomed
experience. Also, each of these systems uses own method
of authorization, login information and password (because
the use of a single password for all systems is a violation
of security rules in the web space). Hence, there exists a
necessity for a systematic decentralized review of
information encompassing each sphere of life within the
system or group of systems. Additionally, a substantial
number of passwords and the distinct characteristics of
each system need to be remembered. This demands extra
time, and on the whole, there are ample drawbacks
accompanied by adverse consequences for the user.
Regarding software service providers, the drawbacks
are linked to supporting the system's usage and
development. Generally, employing decentralized
management systems for each individual component isn't
just time-consuming for the user but also resource-
intensive for the provider. In addition, the development of
the functionality of each individual component of the user
does not create conditions for the effective work of
developers. Hence, the issue of centrally managing all
aspects of user activity on social networks emerges. The
advancement of contemporary technologies and the
knowledge gained from employing solutions built upon
these technologies indicate that the previously mentioned
problem can be partly addressed through the swift
automated generation and utilization of a broad category
of chatbots [3-6].
The outlined problem requires a solution that includes
a platform that integrates the development, accumulation
and use of chatbots. This platform amalgamates the user's
engagement across all messaging applications, social
networks, and information systems within the information
society, encompassing content, graphics, and owner
functionalities. The developed platform, which integrates
the development, accumulation and use of chatbots, will
also be useful for providers of platform services (PaaS).
Indeed, providers of this class, by renting infrastructure
services, will be able to eliminate the abovementioned
disadvantages associated with decentralization and
increase the level of quality of the user service they
provide.
This is the subject of this article and, in the near future,
a cycle of articles devoted to the creation of such a
platform, as it is a complex topic that requires research,
design and implementation processes.
II. FORMULATION OF THE PROBLEM OF THE DEVELOPMENT
AND IMPLEMENTATION OF THE TECHNOLOGY FOR CREATING A
WIDE CLASS OF APPLICATIONS SUCH AS CHATBOTS BASED ON
FORMAL MODELS
It is imperative to formulate a comprehensive
approach, mathematical models, and methodologies as the
foundation for a platform aimed at creating, accumulating,
and employing an extensive range of chatbots. Indeed, this
entails resolving conventional coordination issues for
scenarios such as 'client – bank employee,' 'resident –
utility worker,' 'employee – employee,' 'customer –
dispatcher,' 'student – employee of the dean's office,' and
'student – teacher,' through an innovative approach. This
solution is seen as a platform for the development,
accumulation and use of specialized bot applications that
automate human functions related to providing
information, placing orders and fulfilling orders by
implementing multi-stage processes. At the same time, it
should be taken into account that the individual stages of
these processes depend on various circumstances, events
and participants' features, and they use the capabilities of
social media messaging applications.
The overarching approach, delineated models, and
methodologies must be directed towards swiftly
generating applications that enact the choice and
arrangement of interactions (support), contingent upon the
kind and distinctive attributes of numerous conceivable
sub-scheme arrangement alternatives. Simultaneously, the
mentioned approach should not exhibit the shortcomings
inherent in the algorithmic approach, which are linked to
the necessity of reprogramming the bot for novel user
characteristics, modifications in the executed sub-
schemes, or the emergence of new sub-schemes.
This platform must adhere to stringent demands,
particularly concerning attributes such as flexibility,
scalability, accessibility for end users without specialized
knowledge and skills, reliability, user-friendliness,
distribution, utilization of remote resources, load
adaptation, speed, and seamless integration with
information systems.
III. REVIEW OF PREVIOUS PUBLICATIONS
Numerous sources provide valuable information about
the instant messaging systems that are extensively utilized
by individuals and communities globally [3]. Presently,
instant messaging systems have progressed from being
mere communication tools between individuals to
becoming a means of acquiring information and an
incredibly potent marketing instrument. The pivotal
309
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
characteristic of systems within this category is their
capability to facilitate communication with bots.
An analysis of the ideas, technologies and experience
related to both the development and application of instant
messaging systems allows us to conclude that they can be
used as a basis for creating technologies for managing the
virtual interactions of people. However, a more in-depth
analysis of developments in this area, scientific
publications and user reactions is needed to select the
necessary tools for developing bots. Such analysis will
also aid in establishing an integration foundation that can
be utilized to amalgamate them into a unified system.
Naturally, examples showcasing the efficacy of the
suggested solution are also required. This will help to
avoid mistakes in the creation of technology for managing
the virtual interactions of people. This holds particular
significance during the initial stages of development – the
phases encompassing the contemplation of fundamental
bot creation principles grounded in logical, linguistic, and
other models. It involves shaping a comprehensive
approach that captures, accumulates, synthesizes, and
leverages positive user experiences to guarantee utmost
efficacy in the user's social network activities.
Currently, a chatbot is defined as a computer program
responsible for engaging with the user through auditory or
textual means. Chatbots are classified by the type of
messages (commands), level of interactivity, type of
interaction with the chat server and other features [3].
Most of the available messaging applications provide the
option to choose the communication layer interface
between the chatbot and the messenger server. In the
architecture of the solution, it is expedient to establish two
modes of interaction – polling the server through the API
at an adjustable interval and subscribing to events.
This tool can be beneficial for businesses, especially
for providers of platform services that pertain to the
creation, accumulation, and utilization of specialized bots.
Bots have been in active use for several years, and their
popularity continues to grow. Consequently, a promising
space has emerged for IT developers. Regarding potential
clients, the understanding of bot commands, the focus on
the user's query's core, and the pertinence of user
responses are vital criteria for selecting solutions [4].
The need to create a large number of chatbots, and
mainly as components of information systems of
establishments, institutions, organizations and companies,
including service providers, gave rise to the problem of
developing tools for creating and supporting specialized
chatbots. Since this topic is very popular, there are many
literature sources in which can find descriptions and
applications of tools for creating and maintaining chatbots,
formalisms used, approaches to evaluating chatbots, and
other aspects. We selected and analyzed the sources that
are directly related to the formulated research problem and
influenced the obtained results [7–23].
Existing works can be divided into several categories.
Most often, there are works devoted to the features of
creating chatbots for various fields of human activity, such
as medicine, education, and others [7–11]. The problems
of creating and maintaining chatbots are considered here,
but in close connection with the implementation of
functionality. The second category consists of articles that
reveal certain aspects of creating chatbots, such as the
development of formal chatbot models, natural language
communication, implementation [12–17]. The third
category consists of works related to the evaluation of
chatbots by users, their social aspects [18–19]. Finally, the
fourth category consists of comprehensive works that
describe tools for creating and supporting chatbots,
primarily information technologies and platforms [20–26].
We have started the analysis from the first category.
This category is also interesting because, in addition to
focusing on the implementation of specialized functions, it
often contains experience using formal models and
evaluation of results by users. The article [7] demonstrates
that chatbots powered by artificial intelligence (AI) are an
exciting educational development. They allow to
implement all the necessary functionality of the field of
education (for example helping students and employees,
providing information about the institution). Models of AI
allow to train chatbots to give better and better answers to
users' questions and expand the set of questions without
increasing costs, as they implement training without a
teacher. Simultaneously, the time to receive correct
answers is reduced, chatbots are available at any time,
customer requests are documented. In addition, the article
provides interesting information about user evaluations of
these chatbots, for example, users are often reported to
feel confused and even scared.
Undoubtedly, the use of AI models is an important
factor in the improvement of chatbots.
The article [8] is devoted to the study of the use of
messenger chatbots to provide automated financial advice
to employees, for example, a chatbot helps to develop
personal savings programs. A life cycle model is used. In
addition, the factors influencing users' decisions regarding
consultations were investigated, which allowed to
consider their preferences, problems, financial literacy,
experience and confidence when working with investors.
Unfortunately, chatbots work only through Telegram. But
the accepted concept of API and chatbot development
using the Java programming language is easy to extend to
other messengers.
In the work [9], using the example of corporate
training, an approach to the development of courses as
chatbots using Telegram is proposed. The proposed
approach is based on the concept of micro-content, the
access to which through the messenger is regulated by
accepted work procedures. The use of inclusive language
is also a feature of the implementation.
The work [10] developed a system that, based on the
knowledge of the driver's behavior while driving,
performs the functions of a supervisor. The system
notifies the driver and the administration when the driver
310
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
makes a mistake and suggests the necessary actions. The
on-board diagnostics unit allows to analyze the driver's
behavior in real time. The use of Internet of Things and
messenger technologies helps to drive the car, guided by
the desired rules.
In the work [11], a chatbot was developed that
performs the functions of a virtual assistant for remote
submission of applications for enrollment in educational
organizations. The chatbot is implemented on the basis of
a standard process description template, but taking into
account the important characteristics of the user and
educational programs, educational institution.
Then we have analyzed the works of the second
category. The work [12] demonstrated the use of Petri nets
for modeling chatbots and automatic generation of
executable code for event-driven systems based on the
Telegram messenger. The use of such formalisms allows
to create chatbots in the following stages - modeling, code
generation, verification, and analysis. The stages are
performed in a formal and precise manner.
Although several methods and tools have been
developed for modeling and analyzing chatbots, this
aspect of creating chatbots requires active development.
First of all, for the automated creation of chatbots for
various fields of activity, it is necessary to expand the
class of formal models, including transactional systems
and finite automata. it is also desirable to generalize the
used methods of automatic code generation based on
developed and verified formal models for a wide class of
messengers.
The work [13] presented a methodology for creating
chatbots based on a formal representation of processes.
The methodology is based on methods of transforming a
formal description of the process into a conversational
agent that can guide the process participant through the
necessary steps in a convenient way of communication.
The developed system also includes tools for natural
language processing and generation, dialogue support.
BPMN, the markup language of artificial intelligence, was
used in the development of the system prototype. Human
empowerment methodology aims to make formal
representations of processes more accessible to the actors
involved. Unfortunately, process descriptions in the
BPMN language require additional costs and time delays.
The work [14] combined the use of traditional
components of control systems and chatbots for
monitoring the values of the controlled process. An
algorithm and software for data exchange in a real control
system have been developed. Values are accepted in the
feedback loop of the control system. The bot sends
predefined alerts about values of the controlled process for
example temperature, humidity, pressure, and gas
concentration as soon as they occur in the system and are
published. It allows to deliver instant messages with a
user-friendly presentation of data within a specified time
interval. The implementation integrates the Grafana
project, LoRa and applications.
The work [15] proposed the integration of chatbots in
the IoT environment. The well-known Long-Polling
method [9] was chosen to create bots. The method is built
on templates, which provides a certain level of
automation. The proposed approach demonstrated the high
potential of using chatbots in the IoT environment and
highlighted the advantages of a light and simple interface
provided to users. The use of more formal chatbot models
has made it possible to meet the needs of the IoT
environment in chats for practical applications of all areas.
In the work [16] a model engineering approach to the
development of chatbots is proposed. It is a pattern-based
approach that is based on a neutral meta-model and a
domain-oriented language for describing chatbots.
Forward and reverse engineering analysis as well as
model-based analysis are supported. The developed
prototype tool includes code generators and parsers for
multiple chatbot platforms. The approach is aimed at
practical users who want to develop chatbots for their
problems.
The paper [17] is devoted to the application of formal
models and natural language text processing methods for
the creation of intelligent chatbots. The conducted study
of emotional transition and dialogue structures allowed the
authors to develop the foundations and implement
chatbots for text processing and recognition of emotions
in recorded chats. The theoretical aspects of the research
are based on artificial intelligence algorithms. The
implementation is made using the Telegram API in
Python. Chatbots can be used in medical applications for
psychological assessment, clinical counseling, autism
diagnosis, and expanding cognitive models.
Thereafter we have analyzed the works of the third
category. The paper [18] analyzed the perception of
chatbots by clients using the example of chatbots to ensure
communication between policyholders and insurance
companies. Using a semi-structured survey based on the
Technology Acceptance Model (TAM), the advantages
and disadvantages of using chatbots to perform procedures
related to existing insurance contracts were evaluated.
Quantitative (structural equation modelling with partial
least squares (PLS-SEM)) and qualitative analysis
methods were used. Perceived usefulness and perceived
ease of use of chatbots were evaluated, taking into account
social influence on behavioral intention and trust.
The analysis showed a general reluctance to use
chatbots, resistance to their use. The explanation lies in
the complexity of interaction with chatbots and the
frequent need to complete communication with the help of
a human operator. Another aspect of dislike for chatbots is
the perception by users that they provide a dehumanized
service without empathy.
Thus, in order to demonstrate the advantages of
chatbots, their developers need to ensure that the
interaction with chatbots is perceived as simple, efficient
and low-cost in terms of time and price, flexible in terms
of time. The authors conclude that chatbots must have
311
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
conversational skills, the ability to provide human-like
interactions, and user perceptions that chatbots actually
add value to policyholders.
It is important to take these properties of chatbots as a
basis. But at the same time, it is necessary to provide
chatbots with the ability to understand user intentions,
which helps chatbots acquire the above-mentioned
properties.
In the work [19], a few quality and efficiency
assessments of created chatbots based on information
from various sources are proposed. It would be good to
develop methods for evaluating chatbots based on
messages in social media.
Finally, we have analyzed the works of the fourth
category. The work [20] describes a platform EREBOTS
for creating and supporting chatbots for providing medical
services to users. Chatbots are generated based on
previously developed scenarios. Actually, the scenarios
are formal models of chatbots, but they need to be
described and this requires costs and time.
The approach aimed at developing a complete
platform for creating and supporting chatbots seems to be
the most acceptable. But to provide the created bots with
the properties that users expect, the platform must be
generalized. First, the creation and support of chatbots
should cover all areas of human activity. Second, to speed
up the creation of chatbots, it is necessary to use all
existing chatbot descriptions. For example, in some
industries, formal descriptions of the business process
system have existed for a long time. Annotators can also
be used to describe the behavior of chatbots, but then it is
necessary to develop the most simplified form of
descriptions of chatbot functioning, for example in the
form of regular expressions, as well as algorithms for the
synthesis of formal models of chatbots, such as transaction
systems and finite state machines of various types, on
based on regular expressions.
In the paper [21], the surge of digital educational tools
is demonstrated by the creation of a platform of Telegram
bots in teaching and learning English. The comprehensive
nature of the platform determines a number of advantages,
including the effectiveness of language teaching and
learning, providing students with more useful practice,
and rapid acquisition of skills and knowledge. The
platform allows to provide many educational programs,
systematically develop and improve them.
The work [22] presented classifications of chatbots
based on various criteria, the general architecture of
modern chatbots and the main platforms for their creation.
Also here, a historical review of the evolution of chatbots
is made and the motives for using chatbots, their accepted
evaluations and the influence of social stereotypes on the
development of chatbots are described.
The accumulated experience of using the specified
tools for the development of specialized bots has shown
that the scope of their practical use is rapidly expanding.
Really, state institutions, communal services, banks,
health care facilities, enterprises and companies are
increasingly using these tools to automate functions
related to providing information, placing orders and
fulfilling orders to support business processes. It is natural
that these tools have begun to receive more attention from
researchers, developers, theorists and practitioners. New
views on the development of specialized bots are being
formed, and new industrial solutions are emerging.
The above analysis of the available literary sources
showed that the approach developed by the authors earlier
[23-24] to the creation and maintenance of chatbots
corresponds to modern trends. In particular, the use of
models of mathematical logic, AI and accumulated
descriptions of business processes for the formal
description of a chatbot with their subsequent
transformation into a chatbot according to the templates of
a specific messenger is inherent in all modern
methodologies for creating chatbots. The use of natural
language processing models and methods is the main
approach to ensuring the ability of chatbots to
communicate with users. Based on this approach, the
technology for creating and supporting chatbots has
proven to be useful and effective.
At the same time, the accumulated experience,
dynamics of changes and forecasts of the development of
the IT industry indicate that the previously proposed
holistic approach of the authors and the developed
technology can be improved in order to quickly create,
accumulate and effectively use chatbots.
The main directions for improving the approach and
technology of creating specialized chatbots should be
determined:
1) transition from bot creation technology to a
platform for development, accumulation and use of
specialized chatbots;
2) using formal descriptions of business processes to
structure chatbots;
3) modeling chatbots behavior using formal and well-
researched classes of models, primarily finite automata,
Petri nets, and others depending on the features of the
chatbots being created;
4) the use of methods of synthesis of formal models
of chatbots based on their behavior, expressed in a user-
friendly form of regular expressions;
5) expansion of the used linguistic models by means
of determining user intentions to improve the
effectiveness of communication with the chatbot;
6) application of user information and social networks
for evaluating chatbots, determining the needs for their
development and improvement.
IV. ADVANTAGES AND DISADVANTAGES OF THE CURRENT
SOLUTION – IMPLEMENTING INFORMATION TECHNOLOGY FOR BOT
CREATION THROUGH A LOGICAL APPROACH AND A STATE MACHINE
The individual stages of bot creation processes depend
on various circumstances. The most important and
influential circumstances are events and the features of
participants who are members of these social networks
312
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
and users of messaging applications. Differences in such
features and circumstances affect the complexity, structure
and overall composition of the whole application,
determining the entire end-to-end flow of the development
process. As a result, the creation of these applications
requires thorough planning and a coherent, thought-out
approach to performing design work on stages crucial to
the whole multistage process.
The general approach to creating bots, based on those
assumptions, was proposed by authors at the work [23]
and developed later in the cycle of articles. This approach
was based on a system of formal models and methods and
provided for the structuring of the bot creation process.
Initially, the following stages were highlighted and
supported by appropriate tools: the acceptance of
descriptions of business processes as source information
and the use of logical models to build solutions to user
problems in the space of states; presentation of received
logical conclusions by chatbot state machines as chatbot
models; transformation of chatbot state machines into
formal descriptions of state machines of well-known
messengers. The approach can be thought of as an attempt
to build diagram specifications based on the analysis of
business processes to facilitate the implementation of the
proposed bot applications.
The author's concept of automated creation of bots
based on a logical approach was used to describe the
relevant information technology. As part of the
implementation of the specified information technology, a
general scheme of a web-oriented system using a chatbots
has been developed. The general architecture of the
solution is determined by the features of the approach, and
the implementation of the components that support
individual bot creation processes are based on the models
and methods developed by the authors. The experience of
using the technology to create a bot (an example is given
below) has demonstrated the workability of the approach
and the practical value of the developed information
technology for the rapid creation of bots with software
support.
The practical use of bot creation technology built on
the basis of the proposed approach revealed both its
advantages and disadvantages.
The advantages, in addition to the practical benefit, are
the following:
• the labor productivity of developers increases;
• the time needed to create bots decreases;
• user interactions with the system are simplified;
• experience is gained and conditions are created
for the development of more effective solutions.
Regarding drawbacks, firstly, the presented solution
does not encompass the complete spectrum of tasks
associated with centrally managing all aspects of user
activity on social networks via the establishment of web-
oriented technologies. Secondly, the execution of such an
approach necessitates a solution facilitating the
construction of user interactions through chat within a
designated messaging application, relying on a target
platform for request processing. This target platform
should incorporate a comprehensive algorithm of actions
applicable to all scenarios outlined within a given chat.
However, the solution was realized using the server
architecture of the interaction level with the bot, as
outlined in [23].
Within this technology, the components serve standard
roles: the message provider functions as an intermediary
for the message source (the messaging application);
command handlers manage messages, assigning tasks to
services, updating the state container, and producing user
responses; the repository stores parameters concerning the
state of user request processing; and services execute
supplementary logic for handling user requests.
The mentioned benefits of user interactions within
messaging applications prompted the necessity to
incorporate a mechanism for receiving user requests via a
chatbot. An illustrative case involved employing a chatbot
for processes like registration, taxi booking, scheduling
the order's execution, and monitoring the plan's progress
within the framework of an international service provider.
This example showcased the viability of the approach
founded on state and transition models. One of the
conceivable approaches for specifying actions, states, and
prerequisites for the input data model within the car-
ordering process follows this sequence: 1. Selection of
country; 2. Selection of city; 3. Choosing car class; 4.
Entering the client's mobile phone number; 5. Confirming
the car delivery time; 6. Providing the car delivery address
(street, house).
Given that the states and transitions (which constitute
a loop in this context) are established, it becomes
straightforward to assign a corresponding set of attributes
to each state. In essence, we are dealing with a
conventional control theory model of states and
transitions, which has proven to be a convenient structure
for addressing the bot creation challenge (for this specific
illustration, the model is presented in [23]).
Numerous approaches exist for implementing this
model, ranging from a strictly algorithmic approach to
methods founded on problem-solving within the system's
state space to tackle subproblems. The algorithmic
approach is straightforward to enact, yet it might not be
very user-friendly for the system owner. Its drawbacks,
like a lack of interactivity and potential alterations in the
action sequence during order execution, could result in
considerable setbacks.
It is easy to verify that adding a seventh step to specify
the payment terms in the given taxi order example will not
only change the data, to account for the new state and new
transitions, but also rewrite the software module.
Therefore, in order to implement technology for creating
bots, it was necessary to abandon such a simple but
inflexible mechanism and switch to an approach that
responds to changes in states (situations) not by rewriting
the program but by expanding the space of possibilities for
313
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
achieving certain states (situations) and using a universal
solution search mechanism to solve the problem in the
changing space of these possibilities based on the
decomposition of the problem and derivation. To achieve
this, each action within the plan integrated into the system
needs to be delineated in terms of objects. Accomplishing
this objective within such systems is linked to formulating
an action plan that entails solving the problem.
Nonetheless, implementing this approach necessitated
the creation of a formal logical model for specifying
conditions and solutions, along with the corresponding
information technology. This technology must possess the
capability to leverage experience and knowledge to
autonomously generate bots rooted in logical models and
employ methods to deduce suitable action plans within the
context of messaging applications, bots, and information
systems.
To realize the information technology for constructing
bots grounded in a logical model, the formalism
introduced by the authors in [23-24] was used.
A web application component was built on the basis of
this formalism, which will guarantee the correct behavior
of the bot. In order to create bots with a behavior that
makes them useful in practice, the phases of action
planning and plan execution were emphasized. This
facilitated interaction with the user during the planning
phase, ensuring prompt and precise selection of
subsequent steps. During the plan execution phase, it
enabled monitoring the prerequisites of previous actions
and the fitting response from the environment,
reallocating actions among participants, and introducing
other necessary adjustments.
The second crucial step involved subdividing the
planning phase into two stages: constructing detailed plans
and formulating schemas. During the first stage, a plan is
established, while in the second stage, a comprehensive
action plan is composed using the system's aggregation of
simple (template) actions.
Following the logic of action planning, objects
(TAgent, TRalObject, TConception, TRelation, TValue,
TProblem), situations (states) (TSituation), events
(TEvent), and actions (TAction), along with their
corresponding operations and relationships, underwent
formalization. This was essential for establishing problem
formulations. To model solutions for problems, plans were
subjected to formalization, along with their corresponding
operations and relationships. These were outlined in the
context of states and transitions, considering the
determination of phases for constructing detailed plans
and crafting schemes.
The state (situation) of the control object is delineated
by its parameters. An implementation comprises a
sequence of states and transitions that fulfill the
prerequisites of initialization, transitionability, accuracy,
and completion.
This formalism facilitated the acquisition of responses
to queries that specify the initial and target situation
(state), enabling the quest for or formulation of a
particular action plan. The execution of this plan
guarantees the transition of the control object from the
initial situation (state) to the desired situation. The core
component of this plan is the sequence of actions.
A formal logical AP system (the construction of
detailed plans) was used to construct detailed plans, and a
formal logical PC system (the composition of ready plans)
was used for drawing up plans. These formal systems are
proposed in [23]. Both systems are typed logical systems.
The knowledge base of the AP system comprises clauses
that describe situations and their categories, objects and
their categories, the prerequisites and outcomes of actions,
as well as the event component of prerequisites. The
knowledge base of the PC system encompasses clauses
detailing action programs, problem-solving schemes, and
scheme properties.
The derivation procedures proposed in [23-24] made it
possible for requests to solve problems in the form of a
pair <initial situation, target situation> to create a basis for
the implementation of the bot within the framework of the
adopted architecture, since the program of actions of each
participant, the conditions that require control and the
reaction of the environment were deduced. The listed
derivation procedures used a number of well-known
techniques to reduce the search. The web-oriented
architecture of the information technology for
constructing bots grounded in a logical model is outlined
in [23-24].
This approach made it possible to expand the
functionality of information technology and make it more
convenient for the user. Furthermore, it granted the system
owner the advantage of modifying the order processing
sequence and incorporating/removing structural elements
from the previously established sequence of actions. We
will demonstrate the capabilities of the technology
pertaining to template utilization during the construction
of web applications for determining action sequences.
Let's modify the prerequisites for the input data model of
the taxi-ordering process as follows: 1. Selecting the
country; 2. Selecting the city; 3. Choosing the car class; 4.
Confirming the mobile phone number (if unavailable,
accessible within the messaging application); 5.
Specifying the car delivery time; 6. Providing the precise
address (street, house); 7. Selecting the payment method;
8. Completing the payment (in case of non-cash
settlement); 9. Utilizing one of the previous routes; 10.
Receiving a status update about the order.
By means of technology, the logic of states and
transitions can be presented in the form of the graph
shown in Fig. 1. The behavior described by the system
owner can be implemented using the available capabilities
of the messaging application. However, user requirements
change; for example, a car rental option may be added,
which will require a check of the user's driver's license,
the user's health. In this case, the described behavior
cannot be implemented based on the available machine
314
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
states in modern messaging applications. State and
transition logic is essential to generate an outcome by
utilizing the prerequisites and outcomes of each state. This
involves constructing a tree of transitions (method calls)
that accurately manage situations from the initial state to
the concluding state, based on user-defined prerequisites
and outcomes.
Figure 1. Modified state container
Next, we will delve into the specific actions of the
system within a precisely defined situation.
It is necessary to order a car by specifying the country,
city and class or brand of the car to be leased or used as a
taxi. Precondition: The user is authenticated and possesses
the required driving privileges. Postcondition: The user
expresses the intention to hire a car for a one-day period.
The system encompasses a collection of methods that
dispatch the contract, verify driver's licenses, validate the
city and country, assess the user's eligibility to operate a
vehicle, and engage with the payment service. The
quantity of methods is limitless, and each method is
distinguished by a precondition (which dictates if it should
be invoked) and an postcondition (which determines the
result).
Hence, the system presently functions with a series of
states demarcated by the vertices of a graph, where
transitions occur in accordance with prerequisites and
outcomes. For each distinct user request, a prerequisite
(initial state) and an intended outcome (postcondition) are
stipulated. Through this approach, to execute the
technology for automated bot generation employing the
chatbot module, the following steps must be taken:
1. Create a state construction logic template based on the
data entered by the user (build a graph, attach
conditions to its vertices).
2. Create an API endpoint to receive messages from the
messaging application.
3. Set up a set of available commands and an API
endpoint for receiving messaging application requests
when creating a chatbot in the messaging application
configuration.
4. Add a message provider for the messaging application
at the business logic level using the suitable template.
5. Obtain the message by invoking the message provider
through the API endpoint.
6. Outline the behavioral logic of the handler for each
command in the form of the outcome upon execution.
7. Specify the state repository and design a template for
populating this repository.
8. Establish a connection with an external server for text
message recognition (with plans for the subsequent
development of the system's proprietary server) and
obtain a series of commands from it.
9. Create a class interaction with a third-party API (after
step 8) using available methods and add handler
behaviour at the business logic level.
10. Enhance the handler calls with additional invocations
of the state repository if a call to the third-party
service from step 8 is necessary.
The implementation of the system of the formal
logical derivation in the existing information technology
for creating bots was carried out using the Camunda
framework [3, 4, 23] in the form of a separate service with
existing modifications for the implementation of the
template and the preservation of behavior; this
significantly simplifies the complexity associated with
creating systems of this particular class.
The logical model turned out to be effective for
variable conditions, situations and actions, especially in
those cases in which the conditions of the target and initial
states were quite complex. If we take into account the
branching of the inference tree, these conditions require
many steps in the operation of the inference mechanism.
However, even then, the method was not too convenient,
because it assumed experience in working with logic
models. In addition, certain actions are performed in such
systems not only in certain states but also based on events
that occur.
At the same time, banks, state institutions, communal
services, enterprises, companies and educational
institutions have accumulated models that describe the
order of customer service. These models are business
processes, which are quite convenient to use as a basis for
building bots. At the same time, you can combine states
and events, as well as use sequences of actions, a
hierarchy of actions, roles and actors that can be assigned
to subprocesses and operations, as well as certain
additional conditions, inputs and outputs; that is, business
processes naturally contain elements of a logical
approach.
Therefore, there was a need to enrich the previously
proposed approach for the creation of specialized bots and
to enhance the information technology developed on the
basis of this approach.
V. DEVELOPMENT OF THE SOLUTION –
IMPLEMENTATION OF INFORMATION TECHNOLOGY FOR
CREATING BOTS THROUGH BUSINESS PROCESS
DESCRIPTION
An example of employing a practical implementation
of a component for bot creation is elaborated upon to offer
a clearer illustration of the distinct characteristics and
attributes of various process stages, the realization of bot
applications, and the overarching development process.
The conception of a platform comprised of these
applications is also envisaged.
315
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
First, the basic idea of using messaging applications
within a service-oriented architecture in which the
function of processing user requests is rationally
distributed between the messaging application, command
handlers, storage and services is preserved.
Second, flexible mechanisms for transforming the
general space of system capabilities into the logic of
processing user requests that are clearly described with the
help of a state machine form the basis for the
implementation of a new solution.
Third, the improvement of the approach is related to
its orientation towards the problems that service providers
encounter in managing the virtual interactions of people.
These issues are suggested to be addressed by means of
swift automated generation and utilization of a broad
category of chatbots.
Fourth, an important new concept of the approach is to
supplement the state machine model, which is directly
used as the basis for controlling the processing of the user
request, and the logical derivation model of the request
execution plan in the possibility space (in essence, the
derivation of the state machine for the request processing)
is added to the business process model, which must be
transformed into a state machine model.
Fifth, taking into account the previous changes, a
solution is proposed in the form of a platform as a service
(PaaS), which integrates the development, accumulation
and use of chatbots. Such a solution will be useful both for
providers of software services that are used to manage the
virtual interactions of people and for providers of services
in the form of a PaaS.
The general template of the platform architecture is
shown in Fig. 2.
Figure 2. The general template of the platform architecture
The implementation of modules of the natural
language communication subsystem is determined by the
purpose of the platform aimed at the creation,
development and use of telegram bots capable of
performing a given business process.
The general idea is that, based on the input
information provided by the user, determine the
appropriate state within the state machine built for this
business process. Traditionally, solving the current
problem includes two separate sub-problems.
1. Text recognition and its analysis to determine
the current state of the user.
2. Speech recognition, its transformation into text,
and subsequently, similarly to point 1, setting
the current state of the user.
The task of establishing the user's current state is
solved by selecting key elements that are specific to one or
another vertex of the state machine graph and transition
criteria.
For our part, when creating the bot itself, we can
require the user to enter a description and keywords that
will characterize the current top.
The use of models and appropriate methods for
detecting user intentions in solving this problem will
significantly improve the speed of the chatbot.
To solve the second subtask, well-known common
means of speech recognition are used.
Figure 3. Example of a diagram defined according to the technology
specifications within the user interface
316
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
The platform implementation plan outlines the
practical realization of each component. This guarantees
the establishment of the logic for handling user actions
within the execution of the specified business process,
aligned with the diagram's specifications.
As part of the practical implementation of the
platform, a component was created to facilitate the
formulation of logic for handling user actions during the
execution of a specific business process, following the
specifications outlined in the diagram depicted in Fig. 3.
An example of the available options for calling and
processing child commands within the main command is
shown in Fig. 4.
This solution makes it possible to create any bot that
solves a task and/or implements any business process that
involves receiving and processing user information,
without requiring the user to know programming
languages and to be able to implement algorithms in any
of the languages. Such a solution can be called a 'zero-
code' application.
This approach also greatly simplifies and speeds up
the process of creating a Telegram bot (e.g., messenger)
for existing needs or according to existing business
processes, because there is no need to write code,
subscribe to events and write the processing logic of these
events.
Figure 4. Example of a diagram defined according to the technology
specifications with child commands within the user interface
To set the rules, a 'flow' diagram is used; it provides a set
of control components and connections between
components. According to the developed technology, the
software tool has the following control elements for
setting the rules for the interaction of the bot with the end
user:
• a starting point – an element denoting initial
commands ('/start' in this case);
• interaction with the user from the bot side – an
element that provides the output of a question or
provides information to the user;
• the element of processing a specific action – the
diagram presents it as an element responsible for
the logic of saving information through the
available provider of interaction with the data
source;
• an element of logical branching capable of
checking a given condition;
• an element that specifies the logic of calculations
or data processing for determining the age of the
user;
• relationships between blocks that regulate the call
sequence or available branches;
• a point marking the end of the command
processing logic.
The following control elements are available to the user:
• Save – saves the chart in the format *.bpmn;
• Load – loads an existing chart;
• Apply – sets the execution rules for the bot;
• Save as default – saves the chart as the default
chart;
• Load default – provides an option to load a chart
that has been saved as the default chart.
After pressing the 'Apply' button, the user receives a
ready-made bot; an example interaction with this bot is
shown in Fig. 5.
Figure 5. An example of an interaction with a chatbot
Appropriate technologies were used to implement this
software product. To ensure the preservation of
information, a NoSQL database, namely MongoDB, was
chosen as a data source, since this software solution can
store completely different and untyped data within the
limits of the bot design according to the business process.
The client part is implemented using the TypeScript
language, and the Python server part is implemented
using the FastAPI framework.
VI. CONCLUSIONS
Drawing upon the examination of messaging
applications, chatbots, and the challenges entailing their
conception, utilization, and evolution, the approach for
automated bot creation, grounded in business process
descriptions and logical and linguistic models, has been
enhanced. The rationale for implementing an approach
predicated on the Platform as a Service (PaaS) model is
317
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
justified. Furthermore, the architecture of the platform has
been formulated.
The basis of the subsystem for the use of chatbots is an
architecture focused on the use of a state machine. The
'State Machine Engine' component determines how to
process a user's request by interacting with the messaging
application, event and situation handlers and process
support services.
The chatbot creation subsystem ensures the
construction of a state machine for the bot based on a
logical approach. The formal logic systems previously
developed by the authors make it possible to formally
determine the achievement of goals and the conditions
that must be met by action plans to ensure the
achievement of goals, and the proposed derivation
procedure helps to form action plans for solving user
requests; these requests are formulated as pairs of initial
and ultimate states.
Within the framework of the platform, it is proposed to
implement a number of new concepts. First, the synthesis
of chatbots based on business processes is complemented
by the synthesis of formal models of chatbots in the form
of transaction machines and finite automata based on the
behavior of chatbots described by regular expressions.
Secondly, it is proposed to add a method of detecting
user intentions to the linguistic models and methods on
which the components of the platform for processing
natural language texts are based. Implementation of the
method significantly speeds up the chatbot's response to
user requests and improve the effectiveness of
communication with the chat-bot.
Thirdly, within the framework of the platform, it is
proposed to implement a chatbot evaluation component
based on user reactions and social network data in order to
determine the direction of chatbot development.
An example in which the platform is used to fulfil a
user request is given. The execution of the proposed
solution enables the rapid creation of bots for a diverse
array of tasks articulated in terms of initial and final state
pairs. This is achieved through a fusion of functions from
corresponding object classes, which are also represented
as pairs of initial and final states. Bots can be collected
and employed to address an extensive array of user tasks
associated with acquiring essential information, placing
orders, facilitating access to educational resources, and
other generally regulated tasks linked to the execution of
the described business processes.
The report demonstrates an approach to the
transformation of a formal chatbot model into a chatbot
description according to the requirements of the Telegram
messenger.
Subsequent research will encompass the advancement
of the platform, the formal methodologies underpinning
bot creation and utilization, the utilization of technologies
for efficient communication between the bot and the user
in natural language, and the integration of user activities
within prevalent messaging applications, social networks,
and information systems.
ACKNOWLEDGMENT
This research was funded by the Faculty of Electrical and Computer
Engineering, Cracow University of Technology and the Ministry of
Science and Higher Education, Republic of Poland (grant no. E-1/2023).
REFERENCES
[1] M. Eeuwen, Mobile conversational commerce – messenger
chatbots as the next interface between businesses and consumers,
http://purl.utwente.nl/essays/71706, 2017.
[2] D. Griol, J. Carbó, J. M. Molina, “An automatic dialog simulation
technique to develop and evaluate interactive conversational
agents,” Applied Artificial Intelligence, vol. 27, issue 9, pp. 759-
780, 2013.
[3] One AI for chatbots - [electronic resource] - 2023 -
https://www.oneai.com/industries/chatbot.
[4] Botkit - Building Blocks for Building Bots, - [electronic resource]
- 2023 - https://www.npmjs.com/package/botkit.
[5] Camunda, [electronic resource] - 2023 - https://camunda.com/.
[6] S. Dixon, Most popular global mobile messenger apps as of
January 2022, based on number of monthly active users,
[electronic resource] – 2018. – https://www.statista.com/ statistics/
258749/most-popular-global-mobile-messenger-apps/.
[7] M. Kurni, M. S. Mohammed, K. G. Srinivasa, “Chatbots for
education,” A Beginner's Guide to Introduce Artificial Intelligence
in Teaching and Learning. Cham: Springer International
Publishing, 2023. 173-198.
[8] V. Kobets, S. Savchenko, “Using telegram bots for personalized
financial advice for staff of manufacturing engineering
enterprises,” Design, Simulation, Manufacturing: The Innovation
Exchange. Cham: Springer International Publishing, 2022. 561-
571.
[9] R. P. Díaz Redondo, M. C. Rodríguez, S. Torres-Guijarro, I. V.
Silva, M. M. Vázquez, “A micro learning approach based on a
telegram bot: A gender-inclusive language experience,”
Proceedings of the 2021 10th IEEE Mediterranean Conference on
Embedded Computing (MECO), 2021.
[10] M. Sinev, et al., “Virtual assistant for remote submission of
applications for enrollment in educational organizations,” CEUR
Workshop Proceedings, vol. 2922, 2021.
[11] E. Husni, F. Hasibuan, “Driver supervisor system with telegram
bot platform,” Proceedings of the International Conference on
Computational Collective Intelligence Computational Collective
Intelligence ICCCI, 2018, pp 436–444.
[12] D. I. Samokhvalov, L. W. Dworzanski, “Automatic code
generation from nested Petri nets to event-based systems on the
telegram platform,” Trudy ISP RAN / Proc. RAS, vol. 28, issue 3,
2016. pp. 65-84. DOI: 10.15514/ISPRAS-2016-28(3)-5
[13] A. López, J. Sànchez-Ferreres, J. Carmona, L. Padró, “From
process models to chatbots,” Proceedings of the 31st International
Conference on Advanced Information Systems Engineering CAiSE
2019, Rome, Italy, June 3–7, 2019.
[14] S. Penkov, A. Taneva, “Telegram and discord as a part in control
system,” AIP Conference Proceedings, vol. 2449, no. 1, AIP
Publishing, 2022.
[15] Rosid, M. A., et al., “Integration telegram bot on e-complaint
applications in college,” IOP Conference Series: Materials
Science and Engineering, vol. 288, no. 1, IOP Publishing, 2018.
[16] S. Pérez-Soler, E. Guerra, J. De Lara, “Model-driven chatbot
development,” Proceedings of the International Conference on
Conceptual Modeling. Cham: Springer International Publishing,
2020.
[17] T. S. Gunawan, et al., “Development of intelligent telegram
chatbot using natural language processing,” Prceedings of the
2021 7th IEEE International Conference on Wireless and
Telematics (ICWT), 2021.
[18] J. de Andrés-Sánchez, J. Gené-Albesa, “Attitude and behavioral
intention toward chatbots in an insurance setting: A mixed method
318
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
approach,” International Journal of Human–Computer
Interaction, pp. 1-16, 2023.
[19] N. Radziwill and M. Benton, Evaluating Quality of Chatbots and
Intelligent Conversational Agents, 2017.
[20] D. Calvaresi, et al., “EREBOTS: Privacy-compliant agent-based
platform for multi-scenario personalized health-assistant
chatbots,” Electronics, vol. 10, issue 6, 666, 2021.
[21] D. Mandalika, F. Ramadhanty, D. Dasuki, M. Suseno, “Telegram
bot as a digital learning platform in English language teaching and
learning,” Proceedings English Language and Literature
International Conference (ELLiC), vol. 6, 2023.
[22] I. Maglogiannis, L. Iliadis, E. Pimenidis, “An overview of chatbot
technology,” Artificial Intelligence Applications and Innovations,
vol. 584, pp. 373–383, 2020. doi: 10.1007/978-3-030-49186-4_31
[23] S. Telenyk, G. Nowakowski, E. Zharikov, J. Vovk, “Conceptual
foundations of the use of formal models and methods for the rapid
creation of web applications,” Proceedings of the 2019 10th IEEE
International Conference on Intelligent Data Acquisition and
Advanced Computing Systems: Technology and Applications
(IDAACS), 2019, pp. 512-518, 10.1109/IDAACS.2019.8924416.
[24] S. Telenyk, G. Nowakowski, E. Zharikov, Y. Vovk, and O. P.
Tokmenko, “Action planning logic in intelligent IT systems using
the example of bots,” Proceedings of the 2020 21st IEEE
International Conference on Research and Education in
Mechatronics, REM 2020, 2020. doi:
10.1109/REM49740.2020.9313874.
[25] K. Lipianina-Honcharenko, R. Savchyshyn, A. Sachenko, A.
Chaban, I. Kit and T. Lendiuk, “Concept of the intelligent guide
with AR support,” International Journal of Computing, vol. 21,
issue 2, pp. 271-277, 2022. doi: 10.47839/ijc.21.2.2596.
[26] M. Sarosa, M.H. Wijaya, H. Tolle and A.E. Rakhmania, “Concept
of the implementation of chatbot in online classes using Google
Classroom,” International Journal of Computing, vol. 21, issue 1,
pp. 42-51, 2022. doi: 10.47839/ijc.21.1.2516.
319
Authorized licensed use limited to: Politechnika Krakowska. Downloaded on December 22,2023 at 19:36:07 UTC from IEEE Xplore. Restrictions apply.
