![Aggregation of cyber-attack data from multiple sources using Microsoft Power Automate Desktop UI flow [57].](https://www.researchgate.net/publication/368524442/figure/fig3/AS:11431281328532298@1743117470952/Aggregation-of-cyber-attack-data-from-multiple-sources-using-Microsoft-Power-Automate_Q320.jpg)
![Implementation of complex AI-based algorithms, such as sentiment analysis, named entity recognition, category classification, language translation using Microsoft Power Automate cloud flow [57].](https://www.researchgate.net/publication/368524442/figure/fig4/AS:11431281328428146@1743117471678/Implementation-of-complex-AI-based-algorithms-such-as-sentiment-analysis-named-entity_Q320.jpg)
Access to this full-text is provided by MDPI.
Content available from Algorithms
This content is subject to copyright.
Citation: Sufi, F. Algorithms in
Low-Code-No-Code for Research
Applications: A Practical Review.
Algorithms 2023,16, 108. https://
doi.org/10.3390/a16020108
Academic Editors: Arun
Kumar Sangaiah and Xingjuan Cai
Received: 14 November 2022
Revised: 6 February 2023
Accepted: 10 February 2023
Published: 13 February 2023
Copyright: © 2023 by the author.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
algorithms
Review
Algorithms in Low-Code-No-Code for Research Applications: A
Practical Review
Fahim Sufi
Monash University, School of Public Health and Preventive Medicine, Melbourne, VIC 3004, Australia;
research@fahimsufi.com
Abstract:
Algorithms have evolved from machine code to low-code-no-code (LCNC) in the past
20 years. Observing the growth of LCNC-based algorithm development, the CEO of GitHub men-
tioned that the future of coding is no coding at all. This paper systematically reviewed several of
the recent studies using mainstream LCNC platforms to understand the area of research, the LCNC
platforms used within these studies, and the features of LCNC used for solving individual research
questions. We identified 23 research works using LCNC platforms, such as SetXRM, the vf-OS
platform, Aure-BPM, CRISP-DM, and Microsoft Power Platform (MPP). About 61% of these existing
studies resorted to MPP as their primary choice. The critical research problems solved by these
research works were within the area of global news analysis, social media analysis, landslides, torna-
does, COVID-19, digitization of process, manufacturing, logistics, and software/app development.
The main reasons identified for solving research problems with LCNC algorithms were as follows:
(1) obtaining research data from multiple sources in complete automation; (2) generating artificial
intelligence-driven insights without having to manually code them. In the course of describing this
review, this paper also demonstrates a practical approach to implement a cyber-attack monitoring
algorithm with the most popular LCNC platform.
Keywords:
low-code-no-code; evolution of algorithm; low code application development; low code
development platform; low code application platform; low code in research; cyber-attack monitor;
cyber intelligence dashboard
1. Introduction
Since the birth of the first computer, ENIAC (electronic numerical integrator and
computer) in 1943, the way of writing algorithms has evolved from machine code towards
low-code-no-code (LCNC). As seen from Fig. 1, in 1943, machine codes were written in
binary which could be comprehended by early generation computers [
1
]. In 1949, the birth
of assembly languages simplified machine code. Then, in 1952, Autocode enabled the
programmers to write algorithms in low-level computer programming language, which
could then be translated into machine codes [
2
]. Writing algorithms became much simpler
with fewer lines of code, with the birth of high-level procedural languages (e.g., Fortran,
Algol, COBOL) in 1957 [
2
,
3
]. Then, from 1965 onwards, the way of writing algorithms
veered to be in object oriented languages with the birth of Simula and Smalltalk. In 2001,
Outsystems was born with the vision of faster delivery of digital transformation [
4
]. In
2014, Outsystems released a free version of the LCNC platform for developers. Since 2014,
there has been a trend for minimizing lines of codes and reducing upfront investment in
setup, training, and deployment for software development [
5
]. In 2017, the CEO of GitHub
mentioned that the future of coding is no coding at all [
6
]. After Forrester defined the
terminology “Low-code development platform (LCDP)” in 2014, Gartner coined the term
low-code application platform (LCAP) in 2016 [
7
]. Since then, industry giants, such as
Microsoft, Google, Siemens, and others, started releasing popular LCNC/LCDP/LCAP
platforms. For example, Microsoft released Power Platform in 2016 [
8
], Siemens published
Algorithms 2023,16, 108. https://doi.org/10.3390/a16020108 https://www.mdpi.com/journal/algorithms
Algorithms 2023,16, 108 2 of 23
Mendix [
9
] in 2018, and Google announced AppSheet [
10
], in 2020 as seen from Figure 1.
Apart from terminologies, such as LCNC, LCDP and LCAP, there are several other popular
notions, such as “citizen developer”, “robotic process automation (RPA)”, “IT-business
alignment”, and “business process management (BPM)” interchangeably being used to
represent the concept of low-code-no-code development in recent times [
11
]. The COVID-
19 pandemic accelerated the adoption of LCNC platforms as businesses sought to quickly
pivot and adapt to the changing business environment between 2019 and 2020 [
12
]. In
quarter two of 2020, an LCNC platform, Appian, reported subscription revenue rising
a little over 12% compared to the same quarter last year [
12
]. In more recent times (i.e.,
2022 to 2023), LCNC platforms continue to evolve and expand, offering more advanced
features and capabilities to users. This includes the integration of AI and machine learning,
improved user experiences, and increased integration with other tools and platforms. In
2023, global LCNC technologies are predicted to grow 20% [13].
Algorithms2023,16,xFORPEERREVIEW2of23
platforms.Forexample,MicrosoftreleasedPowerPlatformin2016[8],Siemenspublished
Mendix[9]in2018,andGoogleannouncedAppSheet[10],in2020asseenfromFigure1.
Apartfromterminologies,suchasLCNC,LCDPandLCAP,thereareseveralotherpop‐
ularnotions,suchas“citizendeveloper”,“roboticprocessautomation(RPA)”,“IT‐busi‐
nessalignment”,and“businessprocessmanagement(BPM)”interchangeablybeingused
torepresenttheconceptoflow‐code‐no‐codedevelopmentinrecenttimes[11].The
COVID‐19pandemicacceleratedtheadoptionofLCNCplatformsasbusinessessought
toquicklypivotandadapttothechangingbusinessenvironmentbetween2019and2020
[12].Inquartertwoof2020,anLCNCplatform,Appian,reportedsubscriptionrevenue
risingalittleover12%comparedtothesamequarterlastyear[12].Inmorerecenttimes
(i.e.,2022to2023),LCNCplatformscontinuetoevolveandexpand,offeringmoread‐
vancedfeaturesandcapabilitiestousers.ThisincludestheintegrationofAIandmachine
learning,improveduserexperiences,andincreasedintegrationwithothertoolsandplat‐
forms.In2023,globalLCNCtechnologiesarepredictedtogrow20%[13].
Figure1.Evolutionofalgorithmsfrommachinecodetolow‐code‐no‐code.
ModernLCNCplatformsrealizeseveralbenefitsoverpreviousgenerationsofman‐
uallywritinglengthylinesofcodeswithlow‐levelorhigh‐levellanguages.Thesebenefits
includefasterdevelopmentwithcloud‐basedenvironments,whereanyonewithoutcod‐
ingexperiencecandevelopcomplextechnologysolutionsbyintegratingtechnologycom‐
ponentsfrommultiplesources.TheseLCNCplatformsprovidesimplecloud‐basedinter‐
faceswhereanon‐programmerscientistorresearchercandrag‐and‐droprequiredalgo‐
rithmstodevelopcomplexscientificsolutions.
Hence,anyonewithouthigh‐levelcodingknowledgecanquicklydeploytheirscien‐
tificsolutionsandreportthem,asdemonstratedinReferences[14–31].Thesestudies(i.e.,
[14–31])representsthelast4yearsdevelopmentsinresearchproblemsaddressedwith
LCNCplatforms.Inthisreview,existingscientificstudiesimplementingalgorithmsusing
LCNCarecriticallyreviewedtoanswerthefollowingresearchquestions:
RQ1:WhatarethebenefitsofusingLCNCplatformsingeneral?
RQ2:WhatarethelimitationsofusingLCNCplatformsingeneral?
RQ3:WhichfeaturesofmodernLCNCplatformswereusedinexistingstudies?
RQ4:WhichLCNCplatformsweremainlyusedinsolvingresearchproblems?
RQ5:WhatresearchproblemsorwhichareaofresearchadoptedLCNCplatforms?
RQ6:HowcanaresearcheradoptmodernLCNCplatformsinsolvingcriticalre‐
searchquestions?
ThisstudyprovidesacomprehensiveliteraturereviewonLCNCplatforms.Moreo‐
ver,thisstudyguidesresearchersandscientistsinadoptingmodernLCNCplatformsfor
solvingvariousresearchproblems.Section2(i.e.,ResearchMethods)providesdetailson
thesystematicliteraturereview,coveringtheexclusionandinclusioncriteria.Sections3
Figure 1. Evolution of algorithms from machine code to low-code-no-code.
Modern LCNC platforms realize several benefits over previous generations of manu-
ally writing lengthy lines of codes with low-level or high-level languages. These benefits
include faster development with cloud-based environments, where anyone without coding
experience can develop complex technology solutions by integrating technology compo-
nents from multiple sources. These LCNC platforms provide simple cloud-based interfaces
where a non-programmer scientist or researcher can drag-and-drop required algorithms to
develop complex scientific solutions.
Hence, anyone without high-level coding knowledge can quickly deploy their sci-
entific solutions and report them, as demonstrated in References [
14
–
31
]. These studies
(i.e., [
14
–
31
]) represents the last 4 years developments in research problems addressed with
LCNC platforms. In this review, existing scientific studies implementing algorithms using
LCNC are critically reviewed to answer the following research questions:
•RQ1: What are the benefits of using LCNC platforms in general?
•RQ2: What are the limitations of using LCNC platforms in general?
•RQ3: Which features of modern LCNC platforms were used in existing studies?
•RQ4: Which LCNC platforms were mainly used in solving research problems?
•RQ5: What research problems or which area of research adopted LCNC platforms?
•
RQ6: How can a researcher adopt modern LCNC platforms in solving critical research
questions?
This study provides a comprehensive literature review on LCNC platforms. Moreover,
this study guides researchers and scientists in adopting modern LCNC platforms for solving
various research problems. Section 2(i.e., Research Methods) provides details on the
systematic literature review, covering the exclusion and inclusion criteria. Sections 3and 4
provide the advantages and disadvantages of LCNC platforms based on the systematic
Algorithms 2023,16, 108 3 of 23
literature review. Sections 3and 4answers research questions 1 and 2. Section 5is the
central part of this study answering research questions 3, 4, and 5. Section 5provides an
in-depth analysis on the usage of LCNC platforms for solving various research problems.
Section 6provides a demonstration on using LCNC platform to solve a research problem
and answers research question 6. Finally, Section 7provides a summary of achievements
with the concluding remarks on LCNC platforms.
2. Research Methods
Within this study, an extensive literature review on topics, such as low-code, no-code,
visual programming, and model-driven programming was performed. Using databases,
such as IEEE Explore, Scopus, ACM Library, and Web of Science, 76 peer-reviewed articles
were gathered. Using online search engines, such as Google and Microsoft Bing, a further
57 grey literatures (i.e., non-peer-reviewed online articles) were also gathered. Then,
133 articles were intensively reviewed and articles not contributing towards the 6 research
questions (i.e., benefits of LCNC, limitations of LCNC, features of LCNC in research, LCNC
platforms used in research, research area solved with LCNC, and demonstration of LCNC
adoption in research) were filtered out. Eventually, 47 articles were carefully selected for
answering the six research questions highlighted within the Introduction section. This
entire methodology is depicted in Figure 2. Table 1describes the common terminologies
used within the context of this paper and how the readers should attempt to interpret
these concepts. The exclusion and inclusion criteria for this study are detailed in Table 2.
General tutorial papers, tutorial videos, and online discussions were omitted (as seen from
Table 2), since the focus of this study was the usage of LCNC platforms on solving research
problems. Since generic tutorial papers, videos, and discussion did not focus on solving
research problems, the literature of this type was omitted. Short papers of less than four
pages were also excluded since they did not delve into the details of using the LCNC
platform.
Table 1. Consistent use of common terminologies.
Terminology Conceptual Usage
Studies
Within this paper, “studies” refers to existing works in the literature or existing body of knowledge
that are currently available in peer-reviewed or non-peer-reviewed (grey literature, such as websites
or portals) sources.
Research area
Research area refers to the high-level grouping or categorizations of research topics. A research area
is much broader than the scope of the research topic.
Research problem Research problems are issues or gaps in existing studies that a researcher is willing to address.
Research problems can encompass one or more research area.
Research question It is a question that a study aims to answer. Research questions essentially turns the research
problems into specific inquiries.
Algorithm Algorithm refers to set of instructions to be followed to solve a research problem or to perform
calculations on research data.
Feature
In general, the term “feature” means a distinctive attribute or aspect of something. In this paper,
“feature” with respect to LCNC has been consistently used to represent the distinctive attributes of
LCNC platforms.
Algorithms 2023,16, 108 4 of 23
Table 2. Inclusion and exclusion criteria for both peer-reviewed and grey literature.
Category Criteria
Inclusion
Peer-reviewed Literature
•
Four search keywords used: “Low Code”, “No Code”, “Visual Programming”, “Model Driven
Programming”
•Review studies, survey/questionnaire-based qualitative or quantitative studies, original
research articles
•
Paper indexed in popular peer-reviewed sources (i.e., IEEE Explore, ACM Library, Scopus, and
Web of Science)
•Papers focusing on research questions RQ 1 to RQ 6
•Studies available in English language
•Studies available in full
Grey Literature
•Websites focused on low code development platforms and their features
•Indexed in popular search engines (i.e., Google and Microsoft Bing)
•Articles authored by either by the LCNC vendor or third-party benchmarking company
•Articles available in English language
Exclusion
Peer-reviewed Literature
•Tutorial papers
•Short papers less than four pages
•Poster papers, editorials, abstract (i.e., lacking detailed information)
Grey literature
•Websites referring to the peer-reviewed literature
•LCNC platforms promoted by bloggers, consultants, or third-party companies
•Tutorial videos and discussions on LCNC
Algorithms2023,16,xFORPEERREVIEW4of23
Table2.Inclusionandexclusioncriteriaforbothpeer‐reviewedandgreyliterature.
CategoryCriteria
Inclusion
Peer‐reviewedLiterature
Foursearchkeywordsused:“LowCode”,“NoCode”,“VisualProgramming”,“Model
DrivenProgramming”
Reviewstudies,survey/questionnaire‐basedqualitativeorquantitativestudies,originalre‐
searcharticles
Paperindexedinpopularpeer‐reviewedsources(i.e.,IEEEExplore,ACMLibrary,Scopus,
andWebofScience)
PapersfocusingonresearchquestionsRQ1toRQ6
StudiesavailableinEnglishlanguage
Studiesavailableinfull
GreyLiterature
Websitesfocusedonlowcodedevelopmentplatformsandtheirfeatures
Indexedinpopularsearchengines(i.e.,GoogleandMicrosoftBing)
ArticlesauthoredbyeitherbytheLCNCvendororthird‐partybenchmarkingcompany
ArticlesavailableinEnglishlanguage
Exclusion
Peer‐reviewedLiterature
Tutorialpapers
Shortpaperslessthanfourpages
Posterpapers,editorials,abstract(i.e.,lackingdetailedinformation)
Greyliterature
Websitesreferringtothepeer‐reviewedliterature
LCNCplatformspromotedbybloggers,consultants,orthird‐partycompanies
TutorialvideosanddiscussionsonLCNC
Figure2.MethodologyforreviewingtheexistingliteratureonLCNCplatformstoanswerresearch
questions.
Figure 2.
Methodology for reviewing the existing literature on LCNC platforms to answer research
questions.
Algorithms 2023,16, 108 5 of 23
As mentioned before, the rest of the paper is organized in four different sections for
answering RQ 1 to RQ 6. Section 3deals with RQ 1. Section 4focuses on the limitations
of LCNC (i.e., RQ 2). Then, RQ 3, RQ 4, and RQ 5 are discussed within Section 5. Finally,
RQ 6 is addressed within Section 6, where a demonstration of using the LCNC platform
for obtaining global cyber intelligence is provided. This assemblage of the six research
questions into four different high-level topics (i.e., Sections 3–6) is depicted in Figure 2.
3. Benefits of LCNC Platforms
Implementing algorithms in LCNC provides several benefits over manually coding
algorithms in any low-level or high-level languages. For answering RQ 1 (i.e., benefits
of LCNC platforms), 13 crucial benefits of developing algorithms using modern LCNC
platforms are explored in this section.
3.1. Business-IT Alignment
Misalignment between business and IT is often the main reason for the failure of large-
scale digitization projects. Often the business leaders have detailed, lengthy, and complex
requirements for an organization’s digital footprint in the future. However, with poor
communication and a lack of domain knowledge of IT specialists, the business visions are
often not translated properly into IT deliverables. Low-code-no-code platforms allow the
business leaders and business analysts with in-depth domain knowledge to transform their
business vision into the IT landscape. Indeed, LCNC allows the business to realize their
vision without relying on IT experts. Hence, LCNC delivers business–IT alignment [32].
3.2. Address Resource Scarcity
Organizations often fail to recruit suitable IT personnel due to a shortage of qualified
ones. However, LCNC allows citizen developers with domain knowledge to work on
developing IT solutions, such as dashboards, applications, and databases, with ease [
30
].
This saves valuable time and resources for the organization, as the organization does not
need to recruit additional IT personnel to conduct IT digitization tasks [
30
,
32
]. Moreover,
the citizen developer can very quickly develop the required IT solutions without wasting
additional time and resources to explain their idea to third-party IT specialists [33].
3.3. Cloud Forward Approach
Modern organizations are rapidly adopting cloud-based technology with their cloud
migration strategy. Cloud-based technology inherently delivers flexibility, cost-effectiveness,
security, mobility, disaster recovery, loss prevention, sustainability, and competitive edge,
along with many other advantages compared to traditional on-premises solutions [
34
].
Since almost all the LCNC platforms are based in the cloud, adopting LCNC provides
quick cloud migration strategies for modern organizations. “No-code supports the ‘cloud-
forward’ approach, fostering faster and more convenient cloud migrations,” said Borya
Shakhnovich, CEO and co-founder of airSlate [32].
3.4. Quickly Trialling and Testing Big Ideas without Big Investment
Cost-effective LCNC solutions allow business leaders and entrepreneurs to quickly
evaluate their idea without significant investments. Citizen developers can quickly develop
prototypes and mock solutions [
35
]. Then, these fully functional IT solutions could be
trialed, tested, and evaluated for their suitability. Once they are found to be suitable and
profitable solutions, they could be further developed into full-fledged IT solutions. On the
other hand, if these mock solutions do not realize the business benefits as promised, they
would not be pursued further. According to [
36
], citizen developers said that more of their
app development effort was devoted to innovation (i.e., big ideas) instead of maintenance,
outperforming those not using low code by 5%. Thus, LCNC solutions act as a gateway
process, where only profitable and worthwhile ventures are carried forward.
Algorithms 2023,16, 108 6 of 23
3.5. Speed of Development
The LCNC platforms are inherently cloud-based. Hence, the developers do not need
to install multiple development tools and libraries (e.g., Visual Studio, Net Frameworks). A
developer can quickly create solutions, as the learning curve is not steep as in traditional
software development routes (e.g., Net, Python, C++). Using web-based drag and drop
building blocks, even a naïve user can quickly create professional IT solutions [
14
,
30
].
In fact, according to [
16
], 66% of IT professionals prefer using LCNC for accelerating
digital transformation. After the digital transformation, when it comes down to changes
or updates to the solution, modern LCNC platforms, such as Outsystems, ensure faster
change cycles [4].
3.6. Security by Design
By default, LCNC platform providers ensure international information security stan-
dards (ISO/IEC 27001, PCIDSS) [
14
]. Modern LCNC providers, such as Mendix and
Microsoft Power Platform, adhere to a principle called “Security by Design” [
37
,
38
]. The
Security by Design principle ensures overall security of the IT solution, taking a lot of
security concerns away from the citizen developer. Hence, ensuring security becomes a
crucial responsibility of the cloud-based LCNC platform [14].
3.7. Modern Patterns and Best Practices
Almost all the existing LCNC platforms support modern patterns and best practices
of software development, such as responsive design, single sign-on, authentication and
authorization, develop once deploy everywhere (e.g., iOS, Android, Windows), agile
practice, remote access, business process modelling, app statistics and reports, etc. [
17
]. As
a result, citizen developers were 15% more likely to deliver mobile applications in 4 months
or less compared to those not using LCNC [
36
]. Moreover, LCNC platform users were
20% more likely to rate their agile maturity as level 3, 4, or 5 compared to those not using
LCNC [
36
]. When citizen developers develop a solution with modern LCNC platforms,
modern patterns and best practices are automatically supported by the solution.
3.8. Integration with External Services
Modern applications rarely work in isolation. Modern applications often need to
communicate with third-party solutions through various modes, such as application pro-
gramming interfaces (API). Existing LCNC platforms support built-in connectors through
which data can be easily consumed or used [
35
]. For example, an application might read
data from Google Drive, OneDrive, or Dropbox. The LCNC platforms support connectors
to obtain data from these sources, upholding interoperability and seamless integration
from third-party systems [
39
]. Modern LCNC platforms, such as Microsoft Power Platform,
also enable researchers to consume real-time social media data from popular sources, such
as Twitter, as demonstrated in [18,19].
3.9. Application Maintenance
Existing LCNC platforms also support comprehensive application maintenance pro-
cedures with AI-driven analytics and insights. For example, Microsoft Power Automate,
used in [
19
], provided in-depth user statistics (e.g., who is using it, frequency of usage)
along with AI-driven suggestions on possible improvements. Citizen developers can easily
test the solution and deploy the solution, performing cloud-based maintenance of the
application [39].
3.10. Protection against Technology Churn
With the evolution of IT, newer technology is always replacing older technology.
Adopting mainstream LCNC platforms ensures continuous updates to the technology
platform by the vendor. This minimizes the risk of technology being completely obsolete.
Algorithms 2023,16, 108 7 of 23
3.11. Easy to Learn
While traditional low- and high-level programming languages have a steep learning
curve, anyone can easily learn any of the modern LCNC platforms within days [
35
]. This
is because modern LCNC platforms provide an interactive visual interface supporting
drag-and-drop actions. A citizen developer can obtain visual cues from the highly visual
interface and can easily develop a complex application. This “easy to learn” feature of
LCNC platforms is encouraging for non-programmer research scientists for their research
data analysis.
3.12. Disaster Recovery and Loss Prevision
Analogous to cloud-hosted solutions, LCNC platforms ensure that the system is
regularly backed-up. Moreover, in the case of disaster, automated recovery becomes the
responsibility of the vendor (i.e., the provider of the LCNC platform). While 20% of cloud
users claim disaster recovery in four hours or less, only 9% of non-cloud users could claim
the same [
40
]. Moreover, while traditional developers using their own computers are
running the risk of losing their source codes during a disaster, cloud-hosted services (e.g.,
LCNC Platforms) ensure data is always available [40].
3.13. AI, ML, & Deep Learning
One of the major benefits of using modern LCNC platforms is integration of AI,
machine learning (ML), and deep learning (DL) technologies. Citizen developers can quite
easily integrate AI-, ML-, and DL-based services within their solutions with drag and
drop-based interfaces [
41
]. Social media analysts and researchers can quite easily obtain
their required data from multiple sources, such as Tweeter, online news agencies, websites,
and portals, using API or web scraping technologies, as demonstrated in [
20
–
22
]. Once
they obtain their required data, AI and natural language processing (NLP) technologies
would allow them process their data with language detection and translation, sentiment
analysis, named entity recognition (NER), and category classification, etc., as demonstrated
in [18,19,23,24].
Citizen developers, as well as researchers, can also use other ML algorithms, such as
clustering, linear regression, logistic regression, and root cause analysis without writing a
single line of codes, as demonstrated in [
15
,
18
–
22
,
25
–
27
]. Moreover, complex deep learning
algorithms, such as the convolutional neural network (CNN), are also being applied using
LCNC platforms to solve complex research problems [
15
,
18
,
20
,
25
,
27
]. For example, manual
or hand-coding-based application of AI/ML technique on landslide data (as demonstrated
in [
42
–
45
]) could be replaced by easier LCNC-based visual programming, as shown in [
25
].
4. Limitations of LCNC
There are a few limitations of existing LCNC platforms. Hence, not all algorithms are
suitable for LCNC-based development [35]. Some of these limitations are explored in this
section for answering RQ 2.
4.1. Creation of Shadow IT
The LCNC platforms reduce dependency towards centralized IT authority of any orga-
nization. Since each of the departments can develop their own solution without depending
on the centralized IT authority, LCNC platforms create multiple shadow ITs within an
organization. Multiple solutions developed by diverse groups of citizen developers within
their own silos could challenge an organization with hostile competitive behavior and low
morale. Citizen developer-driven solutions using LCNC platforms could become a threat
to centralized governance and control of IT solutions by the designated IT authority.
4.2. Vendor Lock-In
Once citizen developers select a LCNC platform (e.g., Mendix, Outsystems, Microsoft
Power Platform, etc.) and start developing their solution on the selected platform, they
Algorithms 2023,16, 108 8 of 23
become locked-in to that platform [
35
]. If they decide at a later stage to move their solution
to another LCNS platform, then they have to redesign and redevelop the entire solution
on the new platform. This is due to the fact that all the existing LCNS platforms do not
provide seamless integration among them (i.e., a Mendix solution could not be executed in
Microsoft Power Platform) and each of these vendor specific platforms follows their own
ecosystems of application development.
4.3. Lack of Flexibility
Even though the citizen developers can quickly become experts in their chosen LCNC
Platforms, they are bound within the framework offered by the LCNC platform [
35
]. Using
traditional programming languages and platforms (e.g., C# in Net), a professional developer
can develop a wide variety of programs targeting desktops, mobile, and even embedded
platforms. Hence, traditional programming languages and platforms offer unrestricted
creativity for the professional developers. However, none of the existing LCNC platforms
allow citizen developers to develop embedded software or native mobile applications.
Hence, citizen developers are restricted in terms of viable options when developing their
solutions under LCNC platforms.
4.4. Lack of On-Premises Support
Almost all current LCNC platforms provide a cloud forward or cloud-first approach,
where the citizen developers implement their algorithms in cloud-based interfaces. So-
lutions that could not be maintained in the cloud (e.g., defense solutions, intelligence
solutions, secret government applications, etc.) are not suitable for LCNC platforms. Even
many private organizations are hosting data that could not be hosted in the cloud for
privacy compliance (e.g., privacy compliant personal health information of patients). Tech-
nology solutions that deal with these types of secret or privacy compliant data are required
to be offline or on-premises by design.
4.5. Unsuitable Mission Critical Systems
Mission critical systems are not suitable for most LCNC platforms. Existing LCNC
platforms provide a cloud-based shared environment for thousands of developers. Even
though most of the existing LCNC platforms support dedicated capacity, if the developer
wants to attain higher performance, it is often not very cost effective. For example, while
a Power BI per user license costs USD 10 per month, a Premium Per Capacity license can
cost USD 5000 per month [
46
]. Mission critical systems or real-time systems work best with
low-level languages, such as Assembly, or high-level languages, such as C/C++.
4.6. Ongoing Cost Commitment
Cloud-based LCNC requires ongoing operational expenses. Even though the sub-
scription cost is meagre (e.g., USD 10 per month [
46
]) compared to the benefit offered by
LCNC platforms, it is often not feasible for many organizations or universities located in
low-income countries. Moreover, the payments are required to be made via credit card,
which might not be suitable for many scientists or universities not possessing international
credit cards.
5. LCNC Used in Existing Research
As seen from Section 3, there are several advantages offered by LCNC that could be
useful in solving problems in all areas of research. Researchers in all areas perform critical
analysis in their domain of expertise. To conduct critical analysis, scientific studies often
need to apply a range of AI-based algorithms. The LCNC platforms enable scientists and
researchers in all areas to easily apply AI-based algorithms on their data to find critical
insight. To apply AI-based algorithms, the researchers and scientists do not need to be
a programmer or data scientist. Using the drag-and-drop features of LCNC platforms,
citizen developers can execute complex AI-based analysis that includes regression, clus-
Algorithms 2023,16, 108 9 of 23
tering, CNN, decomposition analysis, and others. As seen from Table 3, existing studies
in [
15
,
18
–
22
,
25
–
27
,
31
] have recently used these analytical features of LCNC platforms to
solve research problems in multiple disciplines (i.e., answers RQ 3).
Table 3. Features of LCNC platforms used in solving research problems.
References
AI/ML Algorithms
Automated Data Acquisition
Data Processing & Modelling
Interactive Data Visualization
NLP Algorithms
Mobile & Tablet Deployment
Linear Regression
Logistic Regression
K-Means Clustering
Deep Learning/CNN
Decomposition Analysis
Others
Sentiment Analysis
Named Entity Recognition
Category Classification
Language Detection & Translation
[15]• • • • • • • •
[18]•• • ••••••• •
[19]• • • • • • • •
[20]• • • • • • • •
[21]• • • • • • • • •
[22]• • • • • • • •
[23]• • • • • • •
[24]• • • • • • •
[25]• • • • • • •
[26]• • • • •
[27]•••• •• •
[31]• •
Similarly, researchers analyzing human behavior or perceptions of society might
want to execute NLP algorithms, such as sentiment analysis, NER, category classification,
language detection, and translations. Since sociologists, behavioral scientists, and political
scientists are often not comfortable in hand-coding NLP algorithms, LCNC platforms allow
them a quick and easy approach to analyze their data. For example, using the LCNC
platform, a social scientist may analyze anti-vaccine sentiments without hand-coding NLP
algorithms [23]. Likewise, a political scientist or researcher may obtain critical insights on
the multidimensional geopolitical impact arising from COVID-19. as shown in [15,24].
Since data-driven multidisciplinary research has gained popularity in recent times,
the studies in [
20
–
22
] automatically obtained data from several thousand sources (e.g.,
Twitter, CNN, the BBC, the New York Times, etc.) using AI-based data acquisition features
of LCNC platforms.
Social scientists and researchers working with strategic decision-makers are often
unable to create mobile apps for obtaining AI-based insights. As demonstrated in [
25
–
27
],
social scientists easily created mobile apps running in iOS, Android, and Windows for
evidence-based policy making using the LCNC platform. Without LCNC platforms, the
researchers would need to hire expert coders for developing mobile apps or AI-based
algorithms. Table 3summarizes different features of LCNC platforms currently used by
researchers, explaining RQ 3. In terms of answering RQ 4 (i.e., LCNC platforms used in
research), Microsoft Power Platform is the most popular, followed by Mendix, SetXRM,
the vf-OS platform, Aurea BPM, CRISP_DM, Primary AI, and others, as shown in Table 4.
According to Table 4, LCNC platforms supported multidisciplinary research in the area
of manufacturing, supply chain management, software development, business process
automation, education, global news analysis, COVID-19, social media analysis, and even
Algorithms 2023,16, 108 10 of 23
disaster management (e.g., landslides, tornadoes, etc.). Niche research topics in the area of
smart cities can also benefit from adopting LCNC platforms [
47
]. This answers RQ 5 for
the various areas of research supported by existing LCNC platforms.
Table 4. Different area of research supported by existing LCNC platforms.
Area of Research Reference LCNC Platforms
App creation or software development [9,48] Mendix
Software and application development [14] SetXRM
Manufacturing and logistics industry [16] vf-OS platform
Business process in manufacturing [17] Aurea BPM
Digitization of process [28] CRISP-DM
Landslides [25] Microsoft Power Platform
Tornadoes [26,27] Microsoft Power Platform
Social media analysis [15,18,19] Microsoft Power Platform
COVID-19 [15,23,24] Microsoft Power Platform
Global news analysis [20–22] Microsoft Power Platform
Industrial engineering education [29] Unspecified/Questionaire
Supply chain management [30] Unspecified/Questionaire
AI education for students (grades 3–5) [31] PrimaryAI
It becomes evident from Table 3that there are five crucial aspects of LCNC platforms
that appeal the most towards multidisciplinary researchers (i.e., the answer to RQ3), as
follows:
1. Applying AI/ML algorithms;
2. Applying NLP algorithms;
3. AI-based data collection and pre-processing;
4. Interactive data visualization;
5. Mobile and tablet deployment.
It should be mentioned that the above five features were used in [
14
–
31
] for solving
research problems. These critical features of LCNC platforms answer the research question
“RQ3: Which features of modern LCNC platforms were used in existing studies”. These
features briefly described below.
5.1. Applying AI/ML Algorithms
The AI/ML algorithms supported by modern LCNC platforms include regression,
clustering, deep learning decomposition analysis, and others.
5.1.1. Regression Algorithm
Using LCNC platforms, existing studies in disaster management (e.g., landslides,
tornadoes, bushfires, floods, earthquakes, etc.) and major event analysis have used both
linear and logistic regression algorithms [
18
,
21
,
25
–
27
]. These researchers used drag-and-
drop features of LCNC platforms (e.g., key influencer visualization of Microsoft Power
BI [
49
]), where they did not have to code the regression algorithms. Implementation of
a regression analysis automatically ranks the factors that matter, contrasts the relative
importance of these factors, and displays them as key influencers for both categorical
and numeric metrics. For numerical features, linear regression was performed using
SDCA regression implementation from Microsoft’s ML.NET [
50
]. On the other hand, for
categorical features, logistic regression was performed using L-BFGS logistic regression
from ML.NET [51,52].
5.1.2. K-Means Clustering Algorithm
Existing studies in [
19
] used visual features of Microsoft Power Platform (i.e., key
influencer visualization of Microsoft Power BI [
49
]) for implementation of k-means clus-
tering. The k-means algorithm in [
19
] used data pertaining to Australian tropical cyclone
Algorithms 2023,16, 108 11 of 23
events to analyze almost 80 parameters. It should be mentioned that the implementation
in Microsoft Power BI obviated the requirements for manually selecting the number of
clusters. Moreover, AI-based automated clustering by Microsoft Power BI eliminated the
requirement of manual pre-processing (e.g., normalization, data transformation, etc.) of
input data [19].
5.1.3. CNN-Based Deep Learning Algorithm
The anomaly detector enhances line charts by automatically detecting anomalies
within time series data. It also provides explanations for the anomalies to facilitate
root cause analysis. Researchers in [
15
,
18
,
20
,
25
,
27
] used CNN-based deep learning us-
ing the anomaly detection feature provided by Microsoft Power BI’s line chart visualization.
Within [
18
], CNN algorithm was implemented with the LCNC platform to detect anomalies
from live social media feeds on disaster events. For implementing complex CNN algorithms
in [
18
], social media analysts and researchers did not have to manually write algorithms
using any programming languages. Similarly, media analysts and political scientists can
find out abnormalities in global events using a CNN algorithm by using interactive visual-
ization or tools, as demonstrated in [
20
]. Geologists, who are unable to program complex
CNN algorithms, can now implement CNN-based anomaly detection to automatically
identify and analyze abnormalities in landslides [
25
] and tornadoes [
27
]. Moreover, the
LCNC-based solution in [
15
] used CNN for evidence-based strategic decision-making on
the COVID-19 crisis. The implementation of CNN within [
15
,
18
,
20
,
25
,
27
] used a saliency
reduction (SR) model [53,54] and NLP-based explanations [55].
5.1.4. Decomposition Analysis Algorithm
Microsoft Power BI’s decomposition tree visual is a valuable tool for ad hoc exploration
and conducting root cause analysis, while allowing the user to visualize the data across
multiple filter attributes [
56
]. Using the visual drag-and-drop interface, a non-programmer
researcher or scientist can easily perform root cause analysis with a decomposition tree
visual in Microsoft Power BI [
49
]. In [
22
], the researchers created threat-maps based on
global events by using an interactive decomposition tree visual. On the other hand, in [
25
],
decomposition analysis allowed the visualization of landslide casualty data over a range of
landslide feature attributes, such as triggers, categories, settings, sizes, and countries.
5.2. Applying NLP Algorithms
There are several NLP algorithms supported by modern LCNC platforms, such as
category classification, sentiment analysis, NER, language detection, and translation. These
algorithms have been implemented by non-programmer scientists and researchers using
the drag-and-drop features of modern LCNC platforms (as demonstrated in [15,18–24]).
5.2.1. Category Classification Algorithm
Without manually programming the category classification algorithm with C# or
other programming languages, social media researchers in [
18
] used the visual interfaces of
Microsoft Power Automate [
57
] to implement category classification. Category classification
is applied to classify text inputs into categories that are suitable for a certain business cases.
There are pre-built models of category classification (e.g., customer feedback) available
via Microsoft’s AI-builder [
58
] that can categorize a text input to any of the following
categories:
•Issues;
•Compliment;
•Customer service;
•Documentation;
•Price and billing;
•Staff.
Algorithms 2023,16, 108 12 of 23
Other than assessing customer feedback, category classification has also been used to
discover the interest or inquisitiveness of an online social media user [59].
5.2.2. Sentiment Analysis Algorithm
Research on sentiment analysis commenced early in 2002 with the publication of
[60,61]
.
Research in [
60
] represented a supervised learning corpus-based machine classifier, and [
61
]
exhibited an unsupervised classifier based on linguistic analysis. Previously, the promi-
nence of sentiment analysis was on movie and product reviews. It spread across other
domains with the advent of social media users [
60
–
73
]. Recent studies on sentiment analysis
have been used for assessing customer feedback towards comprehending the political sen-
timent of people, specifically to predict election results [
74
]. Even though several existing
studies used hand-coding skills of experienced data scientists, studies in [
15
,
18
–
24
] were
implemented using visual interfaces of LCNC platforms. Studies in [
15
,
18
–
24
] invoked sen-
timent analysis algorithms through MS Power Automate [
57
]. Microsoft Power Automate
uses drag-and-drop feature to invoke sentiment analysis with Microsoft Cognitive Services
Text Analytics API [75].
5.2.3. NER Algorithm
Here, NER is an NLP-based information extraction task that seeks to locate and clas-
sify named entities mentioned in unstructured text into pre-defined categories, such as
locations, person names, organizations, date/time expressions, monetary values, quantities,
numerical values, and percentages. Indeed, NER has been applied in almost all domains
of research for extracting crucial information from unstructured texts [
69
,
76
]. Previous re-
search has extracted three different categories of entities (i.e., “Disease or Syndrome”, “Sign
or Symptom”, and “Pharmacologic Substance”) from health-related tweet messages [
76
]
for discovering public health information and developing just-in-time disease outbreak
prediction systems and drug interactions systems. A study in [
69
] applied basic NLP-
based methodologies to extract entities and relationships, as well as to identify sentiment.
The keywords investigated within [
69
] were drug abuse—cannabinoids, buprenorphine,
opioids, sedatives, and stimulants. Research in [
70
] qualitatively evaluated posts about
methylphenidate from five French patient web-forums including an analysis of information
about misuse or abuse. Data were accumulated from French social networks that cited
methylphenidate keywords. Text mining methods, such as NER and topic modeling, were
used to analyze the chatter, including the identification of adverse reactions. Previous
research in [
69
,
70
,
76
], did not use NER as a pre-processor for AI-based algorithms. The NER
could be invoked with C# using the Microsoft Cognitive Services Text Analytics application
programming interface (API) [
75
]. The NER algorithms could also be implemented with the
drag-and-drop features of Microsoft Power Automate [57], as demonstrated in [15,18–24].
5.2.4. Language Detection and Translation Algorithm
Microsoft Cognitive Services Text Analytics API [
75
] enables social media analysis
and researchers to implement language detection and translation using the visual interface
of Microsoft Power Automate [
57
]. A non-programmer scientist can easily drag-and-
drop appropriate language detection and translation components and perform dynamic
translations, as demonstrated in [
15
,
18
–
24
]. These research studies demonstrated that
live tweets in 110 different languages can easily be comprehended and analyzed without
writing a single line of code (i.e., harnessing the power of modern LCNC platforms).
5.3. AI-Based Data Collection and Pre-Processing
Modern LCNC platforms allow researchers to obtain their research data automatically
from multiple sources, such as csv files, xlsx files, pdf files, databases, social media, or even
websites (with web scraping technologies). For example, research data was automatically
collected using Twitter API through the visual interface of Microsoft Power Automate
within [
15
,
18
–
24
]. On the other hand, research works in [
25
–
27
], automatically obtained
Algorithms 2023,16, 108 13 of 23
data from csv, xlsx, pdf, and even NASA’s global landslide databases. As demonstrated
in [
25
], LCNC platforms also allow non-programmer researchers to perform a range of data
modelling, transformation, and cleansing tasks for obtaining better AI-driven insights from
their data.
5.4. Interactive Data Visualization
Current LCNC platforms, such as Outsystems, Mendix, Microsoft Power Platform,
and others, provide interactive data visualization capabilities to non-programmer re-
searchers [
35
]. Hence, studies in [
14
–
28
,
31
] could easily portray the research findings.
The interactive nature of current LCNC platforms allows even younger children and
teenagers (e.g., grades 3–5) to seamlessly perform AI programming using simple visual
interfaces (as shown in [31]).
5.5. Mobile and Tablet Deployment
As demonstrated in Table 3, existing studies using LCNC platforms have used mobile
deployment features in [
15
,
18
–
27
]. Without writing a single line of code, the dashboards
and reports generated in LCNC platforms can be quickly deployed with the click of a
button in all major mobile platforms, such as iOS, Android, and Windows.
6. Demonstration of LCNC Adoption in Modern Research
As highlighted in the Research Methods section (i.e., Section 2), this section will
provide a practical insight into using modern LCNC platforms for answering modern
research problems. With the rising threat of global cyber-attack, a researcher might want
to develop an innovative algorithm that automatically produces cyber intelligence for
strategic decision-makers. The algorithm in Alg1 provides necessary pseudocodes for this
cyber intelligence solution. In an effort to answer RQ 6, this section will demonstrate how
to implement Alg1 with Microsoft Power Platform [8], a leading LCNC platform.
As seen from Figure 3, a non-programmer scientist or researcher can obtain cyber-
attack data from multiple sources, using Microsoft Desktop UI flow or Microsoft Power
Automate, which is part of Microsoft Power Platform [
8
]. There were two different types of
data obtained for this demonstration, as follows:
•
Real-time cyber-attack data collected from anti-virus vendors (i.e., https://statistics.
securelist.com/ (accessed on 3 January 2023));
•
Real-time cyber-related Twitter feeds obtained using Twitter API (i.e., https://developer.
twitter.com/en/portal/dashboard (accessed on 3 January 2023)).
Algorithms 2023,16, 108 14 of 23
Algorithms2023,16,xFORPEERREVIEW14of23
Real‐timecyber‐relatedTwitterfeedsobtainedusingTwitterAPI(i.e.,https://devel‐
oper.twitter.com/en/portal/dashboard(accessedon3January2023)).
Figure3.Aggregationofcyber‐attackdatafrommultiplesourcesusingMicrosoftPowerAutomate
DesktopUIflow[57].
Figure3implementsline1ofAlgorithm1;thedetailedprocessofobtainingdata
fromsocialmedia,websites,andotheronlineavenueshasbeendemonstratedin[15,18–
24].
Afteraggregatingthecyber‐attackdatafrommultiplesources,thenon‐programmer
scientistorresearchercanapplycomplexartificialintelligence(AI)‐basedalgorithms,
suchassentimentanalysis,NER,categoryclassification,languagetranslation,andothers
asseenfromFigure4.Figure4implementsline2,3,4,5,6,and7ofAlgorithm1.
Algorithm1Cyber‐attackIntelligence
1:Obtaincyber‐attackstatisticsfrommultiplesourcesincludingsocialmedia
2:foreachofthesemessagesm
1
,m
2
,m
3
,…,m
n
,do
3:s
i
=AI_Translate(m
i
)
4:t
i
=AI_AnalyseSentimentScore(s
i
)
5:{e
j
,e
k
}=AI_ClassifyEntity(s
i
)
6:c
i
=AI_ClassifyCategory(s
i
)
7:endfor
8:PerformCNN‐basedAnomalyDetectionforallmessages{s
i
,t
i
,{e
j
,e
k
},c
i
}
Figure 3.
Aggregation of cyber-attack data from multiple sources using Microsoft Power Automate
Desktop UI flow [57].
Figure 3implements line 1 of Algorithm 1; the detailed process of obtaining data from
social media, websites, and other online avenues has been demonstrated in [15,18–24].
After aggregating the cyber-attack data from multiple sources, the non-programmer
scientist or researcher can apply complex artificial intelligence (AI)-based algorithms, such
as sentiment analysis, NER, category classification, language translation, and others as seen
from Figure 4. Figure 4implements line 2, 3, 4, 5, 6, and 7 of Algorithm 1.
Algorithm 1. Cyber-attack Intelligence.
1: Obtain cyber-attack statistics from multiple sources including social media
2: for each of these messages m1, m2, m3,. . . , mn,do
3: si= AI_Translate(mi)
4: ti= AI_AnalyseSentimentScore(si)
5: {ej, ek} = AI_ClassifyEntity(si)
6: ci= AI_ClassifyCategory(si)
7: end for
8: Perform CNN- based Anomaly Detection for all messages {si, ti, {ej, ek}, ci}
Algorithms 2023,16, 108 15 of 23
Algorithms2023,16,xFORPEERREVIEW15of23
Figure4.ImplementationofcomplexAI‐basedalgorithms,suchassentimentanalysis,namedentity
recognition,categoryclassification,languagetranslationusingMicrosoftPowerAutomatecloud
flow[57].
AsseenfromFigures3and4,LCNCplatforms,suchasMicrosoftPowerAutomate
[57],allowsascientistorresearchertoeasilyobtaindatafortheirresearchfrommultiple
sourcesandthenperformAI‐basedanalysisontheirresearchdata,withoutwritingasin‐
glelineofcode.Figure5showsCNN‐basedanomalydetectiondeployedusingaline‐
chartvisualofMicrosoftPowerBI[49].Withoutwritingasinglelineofcode,twoanom‐
alies(for17October2022and25October2022)withinthecyber‐attackspectrumofthe
RepublicofPalauweredetected,asseenfromFigure5.Figure5implementsline8of
Algorithm1.
Figure5.CNN‐baseddeeplearningAlgorithmusedforidentifyinganomaliesincyber‐attacksfor
theRepublicofPalau.
TableA1inAppendixAdemonstratesthecompleteresultofimplementingthealgo‐
rithm(i.e.,Alg1)oncyberdatausingtheLCNCplatformwithoutwritinganylow‐level
orhigh‐levelcodes.Datawithinthistablerepresentcyber‐relatedpostscollectedfromlive
Tweetsfrom13October2022to13November2022.Duringthis30daysperiod,6697
Figure 4.
Implementation of complex AI-based algorithms, such as sentiment analysis, named entity
recognition, category classification, language translation using Microsoft Power Automate cloud
flow [57].
As seen from Figures 3and 4, LCNC platforms, such as Microsoft Power Automate [
57
],
allows a scientist or researcher to easily obtain data for their research from multiple sources
and then perform AI-based analysis on their research data, without writing a single line of
code. Figure 5shows CNN-based anomaly detection deployed using a line-chart visual
of Microsoft Power BI [
49
]. Without writing a single line of code, two anomalies (for
17 October 2022 and 25 October 2022) within the cyber-attack spectrum of the Republic of
Palau were detected, as seen from Figure 5. Figure 5implements line 8 of Algorithm 1.
Algorithms 2023, 16, x FOR PEER REVIEW 15 of 23
Figure 4. Implementation of complex AI-based algorithms, such as sentiment analysis, named entity
recognition, category classification, language translation using Microsoft Power Automate cloud
flow [57].
As seen from Figures 3 and 4, LCNC platforms, such as Microsoft Power Automate
[57], allows a scientist or researcher to easily obtain data for their research from multiple
sources and then perform AI-based analysis on their research data, without writing a sin-
gle line of code. Figure 5 shows CNN-based anomaly detection deployed using a line-
chart visual of Microsoft Power BI [49]. Without writing a single line of code, two anom-
alies (for 17 October 2022 and 25 October 2022) within the cyber-attack spectrum of the
Republic of Palau were detected, as seen from Figure 5. Figure 5 implements line 8 of
Algorithm 1.
Figure 5. CNN-based deep learning Algorithm used for identifying anomalies in cyber-attacks for
the Republic of Palau.
Table A1 in Appendix A demonstrates the complete result of implementing the algo-
rithm (i.e., Alg1) on cyber data using the LCNC platform without writing any low-level
or high-level codes. Data within this table represent cyber-related posts collected from live
Tweets from 13 October 2022 to 13 November 2022. During this 30 days period, 6697
Figure 5.
CNN-based deep learning Algorithm used for identifying anomalies in cyber-attacks for
the Republic of Palau.
Table A1 in Appendix Ademonstrates the complete result of implementing the algo-
rithm (i.e., Alg1) on cyber data using the LCNC platform without writing any low-level
or high-level codes. Data within this table represent cyber-related posts collected from
live Tweets from 13 October 2022 to 13 November 2022. During this 30 days period, 6697
Tweets in 42 different languages from 5984 unique users were analyzed. These Tweets
Algorithms 2023,16, 108 16 of 23
were translated into English, and sentiment analysis was performed to ascertain whether
these posts were negative, positive, or neutral. Moreover, locations mentioned on these
posts were extracted using NER. As previously mentioned in Section 5, Microsoft Cognitive
Services Text Analytics API [
75
] were used for translation, sentiment analysis, and NER on
these Tweets. As seen from Figures 3and 4, only drag-and-drop features provided by the
cloud-based LCNC platforms were used to collect and analyze the data. After collecting
the data, they were analyzed using complex AI-based algorithms, such as CNN (as shown
in Figure 5). Finally, in Figure 6, we can see the interactive dashboard visualizing the entire
dataset from 11 October 2022 to 13 November 2022. The dashboard in Figure 6shows
different types of cyber-attacks, such as ransomware, exploit, spam, and others, and a user
can click any of these types and the dashboard will filter itself automatically based on the
selected type. Moreover, a user can click on any country and the entire dashboard will be
filtered to represent data pertaining to the selected country. The global cyber-attack data
collection as shown in Figure 3and Table A1 (within Appendix A) was deployed in mobile
and tablet contexts, as shown in Figures 7and 8, respectively. In Figure 7, the solution was
deployed on a Samsung Galaxy Note 10 Lite mobile. On the other hand, Figure 8shows
the deployment on an Apple iPad, 15th generation. None of this deployment required any
high-level or low-level coding skills from the researchers.
Algorithms2023,16,xFORPEERREVIEW16of23
Tweetsin42differentlanguagesfrom5984uniqueuserswereanalyzed.TheseTweets
weretranslatedintoEnglish,andsentimentanalysiswasperformedtoascertainwhether
thesepostswerenegative,positive,orneutral.Moreover,locationsmentionedonthese
postswereextractedusingNER.AspreviouslymentionedinSection5,MicrosoftCogni‐
tiveServicesTextAnalyticsAPI[75]wereusedfortranslation,sentimentanalysis,and
NERontheseTweets.AsseenfromFigures3and4,onlydrag‐and‐dropfeaturesprovided
bythecloud‐basedLCNCplatformswereusedtocollectandanalyzethedata.Aftercollect‐
ingthedata,theywereanalyzedusingcomplexAI‐basedalgorithms,suchasCNN(as
showninFigure5).Finally,inFigure6,wecanseetheinteractivedashboardvisualizingthe
entiredatasetfrom11October2022to13November2022.ThedashboardinFigure6shows
differenttypesofcyber‐attacks,suchasransomware,exploit,spam,andothers,andauser
canclickanyofthesetypesandthedashboardwillfilteritselfautomaticallybasedonthe
selectedtype.Moreover,ausercanclickonanycountryandtheentiredashboardwillbe
filteredtorepresentdatapertainingtotheselectedcountry.Theglobalcyber‐attackdata
collectionasshowninFigure3andTableA1(withinAppendixA)wasdeployedinmobile
andtabletcontexts,asshowninFigures7and8,respectively.InFigure7,thesolutionwas
deployedonaSamsungGalaxyNote10Litemobile.Ontheotherhand,Figure8showsthe
deploymentonanAppleiPad,15thgeneration.Noneofthisdeploymentrequiredanyhigh‐
levelorlow‐levelcodingskillsfromtheresearchers.
Figure6.Visualizationofcyber‐attackdataonMicrosoftPowerBI[49]withoutwritingasinglelineof
code.
Asmentionedearlier,thecyberintelligencesolutionwasdevelopedusingMicrosoft
PowerPlatform.Allthesourcefiles(includingthepbixMicrosoftPowerBIsolution,
cyber‐relatedTweets,etc.)arepubliclyavailableathttps://github.com/DrSufi/COVID_In‐
dex_Anomaly(accessedon03Jan2023)forthesakeofresearchreproducibility.
Figure 6.
Visualization of cyber-attack data on Microsoft Power BI [
49
] without writing a single line
of code.
As mentioned earlier, the cyber intelligence solution was developed using Microsoft
Power Platform. All the source files (including the pbix Microsoft Power BI solution, cyber-
related Tweets, etc.) are publicly available at https://github.com/DrSufi/COVID_Index_
Anomaly (accessed on 3 January 2023) for the sake of research reproducibility.
Algorithms 2023,16, 108 17 of 23
Algorithms2023,16,xFORPEERREVIEW17of23
Figure7.Globalcyber‐attackdataofdifferenttypesaredemonstratedonadeployedAndroidapp
onaSamsungGalaxyNote10LiteMobile.
Figure8.Globalcyber‐attackdataalongwithcyber‐relatedsocialmediadatabeingshownona
deployediOSApponanAppleiPad,ninthgeneration.
Asseeninthissection,usingtheLCNCplatform.aresearchercaneasilyobtaindata
frommultiplesources(e.g.,socialmedia,onlinenewssites,onlinedatabases),andapply
AIalgorithmstodeploynotonlyacyberintelligencesolution,butalsopoliticalthreat
intelligence,COVID‐19intelligence,socialcohesionintelligence,militaryintelligence,and
manyotherinnovativesolutions.
7.Conclusions
Withinthispaper,severalexistingstudiesthatusedmodernLCNCplatformswere
reviewedandanalyzed.SinceadoptionofLCNCplatformsforsolvingresearchquestions
isstillatthelevelofinfancy,only47studiesofscientificsignificancewerefound.Among
therestudies,3wereLCNCreviewarticles,21wereLCNCfeature‐relatedarticles,and
only23wereonmultidisciplinaryresearchsolvingresearchquestionswithLCNCplat‐
forms.Withthese23LCNCplatformrelatedstudies,researchersused6differentLCNC
Figure 7.
Global cyber-attack data of different types are demonstrated on a deployed Android app
on a Samsung Galaxy Note 10 Lite Mobile.
Algorithms2023,16,xFORPEERREVIEW17of23
Figure7.Globalcyber‐attackdataofdifferenttypesaredemonstratedonadeployedAndroidapp
onaSamsungGalaxyNote10LiteMobile.
Figure8.Globalcyber‐attackdataalongwithcyber‐relatedsocialmediadatabeingshownona
deployediOSApponanAppleiPad,ninthgeneration.
Asseeninthissection,usingtheLCNCplatform.aresearchercaneasilyobtaindata
frommultiplesources(e.g.,socialmedia,onlinenewssites,onlinedatabases),andapply
AIalgorithmstodeploynotonlyacyberintelligencesolution,butalsopoliticalthreat
intelligence,COVID‐19intelligence,socialcohesionintelligence,militaryintelligence,and
manyotherinnovativesolutions.
7.Conclusions
Withinthispaper,severalexistingstudiesthatusedmodernLCNCplatformswere
reviewedandanalyzed.SinceadoptionofLCNCplatformsforsolvingresearchquestions
isstillatthelevelofinfancy,only47studiesofscientificsignificancewerefound.Among
therestudies,3wereLCNCreviewarticles,21wereLCNCfeature‐relatedarticles,and
only23wereonmultidisciplinaryresearchsolvingresearchquestionswithLCNCplat‐
forms.Withthese23LCNCplatformrelatedstudies,researchersused6differentLCNC
Figure 8.
Global cyber-attack data along with cyber-related social media data being shown on a
deployed iOS App on an Apple iPad, ninth generation.
As seen in this section, using the LCNC platform. a researcher can easily obtain data
from multiple sources (e.g., social media, online news sites, online databases), and apply
AI algorithms to deploy not only a cyber intelligence solution, but also political threat
intelligence, COVID-19 intelligence, social cohesion intelligence, military intelligence, and
many other innovative solutions.
7. Conclusions
Within this paper, several existing studies that used modern LCNC platforms were
reviewed and analyzed. Since adoption of LCNC platforms for solving research questions
is still at the level of infancy, only 47 studies of scientific significance were found. Among
there studies, 3 were LCNC review articles, 21 were LCNC feature-related articles, and only
23 were on multidisciplinary research solving research questions with LCNC platforms.
With these 23 LCNC platform related studies, researchers used 6 different LCNC platforms
Algorithms 2023,16, 108 18 of 23
(e.g., SetXRM, the vf-OS platform, Aurea BPM, CRISP-DM, Primary AI, Microsoft Power
Platform). A total of 61% (i.e., 14 out of 23) of these existing studies resorted to Microsoft
Power Platform as their chosen LCNC platform.
These studies predominantly used 14 features of modern LCNC platforms as depicted
in Table 3to solve different research problems. These research problems were within
the area of global news analysis, social media analysis, landslides, tornadoes, COVID-19,
digitization of processes, manufacturing, logistics, supply chain management, AI education,
industrial engineering education, and software/app development (as shown in Table 4). As
shown in Table 3, the main reasons for solving research problems with LCNC were to obtain
research data and generate AI-driven insights with complex algorithms (e.g., regression,
clustering, deep learning, classification, sentiment analysis, entity recognition, etc.) without
having to manually code them. Modern LCNC platforms have allowed non-programmer
scientists and researchers to quickly obtain their research data from multiple sources and
analyze their data with AI algorithms. During the course of this review study, it was
also practically demonstrated how to obtain cyber-attack data (Figure 3), analyze the data
with AI algorithms (Figure 4), perform CNN-based deep learning (Figure 5), visualize the
information (Figure 6), and deploy the solution on mobile platforms (Figures 7and 8) using
LCNC platforms without manually writing codes in any programming languages.
In summary, this study found the following core benefits of using LCNC platform to
solve research problems:
•
AI-powered tools: AI is increasingly being used to enhance low-code and no-code
platforms, making it easier to build applications and automate tasks. Indeed, AI can
be used to generate code, suggest best practices, and improve the overall efficiency of
the development process.
•
Improved user experience: There has been a focus on improving the user experience
of low-code and no-code platforms, making them more accessible and intuitive for
users. This includes improvements in drag-and-drop interfaces, visual representations
of data and workflow, and other features that simplify the development process.
•
Integration with other tools: Low-code and no-code platforms are now integrating
with a variety of other tools and platforms, including cloud platforms, databases, and
third-party APIs. This allows users to build more complex applications and connect
them with existing systems.
•
Increased Adoption: Low-code and no-code technology is becoming more widely
adopted, particularly among businesses. This is due in part to the ease of use and
rapid development times that these platforms offer, as well as the ability to build
applications that meet specific business needs.
Overall, low-code and no-code technology is continuing to evolve and expand, of-
fering more advanced features and capabilities to users. This technology is likely to
become increasingly popular and widely adopted in the coming years. Modern LCNC plat-
forms provide competitive advantages in solving critical research questions predominantly
through their AI-based data analysis and information processing capabilities. Widespread
adoption of LCNC platforms within the research community will be seen once critical
concerns, such as ongoing cost commitments and vendor lock-ins are addressed. However,
it is unlikely that LCNC platforms will completely remove the requirement of hand-coding
in future, since it is not a revolutionary technology [77].
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Data will be made available upon request.
Acknowledgments:
The author would like to thank Kaspersky for allowing autonomous down-
load of cyber-attack statistics from their website (i.e., https://statistics.securelist.com/ (accessed on
Algorithms 2023,16, 108 19 of 23
3 January 2023)). Because of this anonymous access to Kaspersky site, it was possible to practically
demonstrate the development of a cyber-intelligence solution with Microsoft Power Platform.
Conflicts of Interest: The author declares no conflict of interest.
Appendix A
Table A1.
Result of applying NLP algorithms on live Twitter data on a cyber-attack using the LCNC
platform.
Date
Count of
Twitter
IDs
Count of
User ID
Count of
Location
Count of
Tweet
Language
Sum of
Retweet
Count
Average of
Negative
Sentiment
Average of
Neutral
Sentiment
Average of
Positive
Sentiment
Count of
Translated
Text
13 October 2022
52 51 29 7 80,707 0.40 0.42 0.18 12
14 October 2022
211 189 122 15 77,089 0.40 0.40 0.20 67
15 October 2022
219 208 116 18 408,635 0.29 0.46 0.25 74
16 October 2022
208 205 111 18 428,407 0.31 0.44 0.25 67
17 October 2022
221 208 122 14 188,791 0.30 0.46 0.24 60
18 October 2022
186 180 101 18 49,255 0.31 0.49 0.19 56
19 October 2022
226 219 133 18 132,222 0.35 0.42 0.22 55
20 October 2022
216 215 123 17 231,915 0.32 0.47 0.21 51
21 October 2022
206 204 129 17 533,082 0.43 0.41 0.15 37
22 October 2022
219 209 118 14 134,067 0.41 0.40 0.19 46
23 October 2022
223 207 116 18 34,249 0.33 0.47 0.20 69
24 October 2022
226 218 128 16 88,944 0.44 0.35 0.21 59
25 October 2022
227 219 118 20 200,700 0.43 0.40 0.16 46
26 October 2022
219 205 113 13 30,097 0.37 0.41 0.21 48
27 October 2022
222 219 121 14 175,143 0.34 0.42 0.24 47
28 October 2022
218 212 124 14 287,112 0.39 0.38 0.23 48
29 October 2022
224 215 126 14 176,450 0.41 0.36 0.23 41
30 October 2022
222 215 114 12 217,949 0.35 0.45 0.20 48
31 October 2022
209 205 113 18 252,942 0.32 0.48 0.20 55
1 November
2022 227 223 133 14 175,690 0.31 0.44 0.25 48
2 November
2022 225 216 120 19 158,510 0.37 0.45 0.18 54
3 November
2022 219 213 126 15 435,121 0.46 0.38 0.15 47
4 November
2022 227 214 114 17 178,945 0.34 0.41 0.24 48
5 November
2022 219 208 123 12 65,565 0.45 0.36 0.19 53
6 November
2022 212 205 105 16 469,544 0.36 0.41 0.23 47
7 November
2022 221 210 107 13 89,628 0.38 0.42 0.20 43
8 November
2022 226 221 117 14 115,866 0.48 0.35 0.17 47
9 November
2022 213 205 117 19 73,431 0.43 0.38 0.19 49
Algorithms 2023,16, 108 20 of 23
Table A1. Cont.
Date
Count of
Twitter
IDs
Count of
User ID
Count of
Location
Count of
Tweet
Language
Sum of
Retweet
Count
Average of
Negative
Sentiment
Average of
Neutral
Sentiment
Average of
Positive
Sentiment
Count of
Translated
Text
10 November
2022 212 207 124 15 90,221 0.36 0.41 0.23 42
11 November
2022 216 213 109 12 110,456 0.33 0.43 0.23 40
12 November
2022 217 213 121 14 104,071 0.46 0.36 0.17 41
13 November
2022 109 105 56 14 49,361 0.52 0.35 0.13 4
Total 6697 5984 2482 42 5,844,165 0.38 0.42 0.21 1466.00
References
1.
Computer History Museum, Birth of Computer—ENIAC, CHM. [Online]. Available online: https://www.computerhistory.org/
revolution/birth-of-the-computer/4/78 (accessed on 12 November 2022).
2.
HP, Computer History: A Timeline of Computer Programming Languages. 15 October 2018. [Online]. Available online:
https://www.hp.com/us-en/shop/tech-takes/computer-history-programming-languages (accessed on 12 November 2022).
3.
Rizwan, O. A Snapshot of Programming Language History, Increment. April 2018. [Online]. Available online: https://increment.
com/programming-languages/language-history/ (accessed on 12 November 2022).
4.
Outsystems. It Began with a Vision. 2001. [Online]. Available online: https://www.outsystems.com/evaluation-guide/it-began-
with-a-vision/ (accessed on 9 November 2022).
5.
Clay, R.; Rymer, J.R.; Mines, C.; Cullen, A.; Whittaker, D. New Development Platforms Emerge for Customer-Facing Applications;
Forrester: Cambridge, MA, USA, 2014; Volume 15.
6.
Swinhoe. GitHub CEO: The Future of Coding is No Coding at All. 13 October 2017. [Online]. Available online: https:
//www.idgconnect.com/article/3578431/github-ceo-the-future-of-coding-is-no-coding- at-all.html (accessed on 3 November
2022).
7.
Vincent, P.; Iijima, K.; Driver, M.; Jason, W.; Natis, Y. Magic Quadrant for Enterprise Low-Code Application Platforms. Gart. Rep.
2016.
8.
Microsoft. Microsoft Power Platform. 2016. [Online]. Available online: https://powerplatform.microsoft.com/en-us/ (accessed
on 12 November 2022).
9.
Haan, J.D. Introducing AI-Assisted Development to Elevate Low-Code Platforms to the Next Level, Mendix. 19 June 2018. [Online].
Available online: https://www.mendix.com/blog/introducing-ai-assisted- development-to-elevate-low-code-platforms-to-the-
next-level/ (accessed on 3 November 2022).
10.
Zavery, A. Google Acquires AppSheet to Help Businesses Create and Extend Applications—Without Coding. 15 January
2020. [Online]. Available online: https://workspace.google.com/blog/product-announcements/helping- businesses-create-
and-extend-applications-without- coding (accessed on 3 November 2022).
11.
Frank, U.; Maier, P.; Bock, A. Low code platforms: Promises, concepts and prospects. A comparative study of ten systems. ICB
Res. Rep. 2021. [CrossRef]
12.
Wilhelm, A. COVID-19 is Driving Demand for Low-Code Apps, Techcrunch. 27 August 2020. [Online]. Available online:
https://techcrunch.com/2020/08/27/covid-19-is-driving- demand-for-low-code-apps/ (accessed on 6 February 2023).
13.
Gartner. Gartner Forecasts Worldwide Low-Code Development Technologies Market to Grow 20% in 2023, Gartner. 13 December
2022. [Online]. Available online: https://www.gartner.com/en/newsroom/press-releases/2022-12- 13-gartner-forecasts-
worldwide-low-code-development- technologies-market-to-grow-20-percent-in- 2023#:~{}:text=Business%20Technologists%
2C%20Hyperautomation%20and%20Composability, latest%20forecast%20f (accessed on 6 February 2023).
14.
Sahinaslan, E.; Sahinaslan, O.; Sabancıoglu, M. Low-code application platform in meeting increasing software demands quickly:
SetXRM. In Fourth International Conference of Mathematical Sciences (ICMS 2020), AIP Conference Proceedings; AIP Publishing LLC.:
Istanbul, Turkey, 2021; Volume 2334.
15.
Sufi, F.K. Automatic identification and explanation of root causes on COVID-19 index anomalies. MethodsX
2023
,10, 101960.
[CrossRef] [PubMed]
16.
Sanchis, R.; García-Perales, Ó.; Fraile, F.; Poler, R. Low-Code as Enabler of Digital Transformation in Manufacturing Industry.
Appl. Sci. 2020,10, 12. [CrossRef]
17.
Waszkowsk, R. Low-code platform for automating business processes in manufacturing. IFAC-PapersOnLine
2019
,52, 376–381.
[CrossRef]
18.
Sufi, F.; Khalil, I. Automated Disaster Monitoring from Social Media Posts using AI based Location Intelligence and Sentiment
Analysis. IEEE Trans. Comput. Soc. Syst. 2022, 1–11, in press. [CrossRef]
Algorithms 2023,16, 108 21 of 23
19.
Sufi, F. A decision support system for extracting artificial intelligence-driven insights from live twitter feeds on natural disasters.
Decis. Anal. J. 2022,5, 100130. [CrossRef]
20.
Sufi, F.K.; Alsulami, M. Automated Multidimensional Analysis of Global Events With Entity Detection, Sentiment Analysis and
Anomaly Detection. IEEE Access 2021,9, 152449–152460. [CrossRef]
21.
Sufi, F.K. Identifying the drivers of negative news with sentiment, entity and regression analysis. Int. J. Inf. Manag. Data Insights
2022,2, 100074. [CrossRef]
22.
Sufi, F.K.; Alsulami, M.; Gutub, A. Automating Global Threat-Maps Generation via Advancements of News Sensors and AI. Arab.
J. Sci. Eng. 2022,48, 2455–2472. [CrossRef]
23.
Sufi, F.; Razzak, I.; Khalil, I. Tracking Anti-Vax Social Movement Using AI based Social Media Monitoring. IEEE Trans. Technol.
Soc. 2022,3, 290–299. [CrossRef]
24.
Sufi, F.; Alsulami, M. AI-based Automated Extraction of Location-Oriented COVID-19 Sentiments. Comput. Mater. Contin. (CMC)
2022,72, 3631–3649.
25.
Sufi, F.K.; Alsulami, M. Knowledge Discovery of Global Landslides Using Automated Machine Learning Algorithms. IEEE Access
2021,9, 131400–131419. [CrossRef]
26.
Sufi, F.; Alam, E.; Alsulami, M. A new interactive system for analyzing historical records of tornedoes in Bangladesh. Sustainability
2022,14, 6303. [CrossRef]
27.
Sufi, F.; Alam, E.; Alsulami, M. Automated Analysis of Australian Tropical Cyclones with Regression, Clustering and Convolu-
tional Neural Network. Sustainability 2022,14, 9830. [CrossRef]
28.
Chhor, J.; Fischer, V.; Kröppel, F.; Schmitt, R.H. Rule-based Decision Support for No-Code Digitalized Processes. Procedia CIRP
2022,107, 258–263. [CrossRef]
29.
Adrian, B.; Hinrichsen, S.; Nikolenko, A. App Development via Low-Code Programming as Part of Modern Industrial Engineering
Education. In Advances in Human Factors and Systems Interaction; Springer International Publishing: Berlin, Germany, 2020;
pp. 45–51.
30.
Bhattacharyya, S.S.; Kumar, S. Study of deployment of “low code no code” applications toward improving digitization of supply
chain management. J. Sci. Technol. Policy Manag. 2021,in press. 1–17. [CrossRef]
31.
Park, K.; Mott, B.; Lee, S.; Gupta, A.; Jantaraweragul, K.; Glazewski, K.; Scribner, J.A.; Ottenbreit-Leftwich, A.; Hmelo-Silver, C.E.;
Lester, J. Investigating a visual interface for elementary students to formulate AI planning tasks. J. Comput. Lang.
2022
,73, 1–13.
[CrossRef]
32.
McKendrick, J. Low-Code No-Code Market Keeps Growing, and that Means Sifts in Technology Roles. ZDNET. 3 November
2022. [Online]. Available online: https://www.zdnet.com/article/low-code-no-code-market-keeps-growing-portending-shifts-
in-technology-roles/ (accessed on 6 November 2022).
33.
McLean, A. Software development trends 2021. Software Column. Can. J. Nurs. Inform.
2021
,16, 1. Available online:
https://cjni.net/journal/?p=8563 (accessed on 9 February 2023).
34.
Sangaiah, A.K.; Javadpour, A.; Ja’fari, F.; Pinto, P.; Zhang, W.; Balasubramanian, S. A hybrid heuristics artificial intelligence
feature selection for intrusion detection classifiers in cloud of things. Clust. Comput. 2022,in press. 1–14. [CrossRef]
35.
Luo, Y.; Liang, P.; Wang, C.; Shahin, M.; Zhan, J. Characteristics and Challenges of Low-Code Development: The Practitioners’ Per-
spective. In Proceedings of the 15th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement
(ESEM), New Orleans, LA, USA, 26–27 October 2021. [CrossRef]
36.
Outsystems. The State of Application Development. 2019. [Online]. Available online: https://www.outsystems.com/local- gov/
-/media/053D5BCC32364C2993C8D0BAFA880DB1.ashx (accessed on 9 November 2022).
37.
Baalbergen, F. Low-Code Security by Design with Mendix’s Low-Code Platform, Mendix. 30 August 2021. [Online]. Available
online: https://www.mendix.com/blog/security-by-design-with-mendixs-low-code- platform/ (accessed on 3 November 2022).
38.
Tandy, N. Microsoft Power Platform and Dataverse: Skill up on security by design, Microsoft. 10 October 2022. [Online].
Available online: https://techcommunity.microsoft.com/t5/microsoft-learn-blog/microsoft-power-platform-and- dataverse-
skill-up-on-security- by/ba-p/3250818 (accessed on 3 November 2022).
39.
Ruscio, D.D.; Kolovos, D.; Lara, J.; Pierantonio, A.; Tisi, M.; Wimmer, M. Low-code development and model-driven engineering:
Two sides of the same coin? Softw. Syst. Model. 2022,21, 437–446. [CrossRef]
40.
Salesforce. 12 Benefits of Cloud Computing. 2022. [Online]. Available online: https://www.salesforce.com/products/platform/
best-practices/benefits-of-cloud- computing/ (accessed on 9 November 2022).
41.
Alsaadi, H.A.; Radain, D.T.; Alzahrani, M.M.; Alshammari, W.F.; Alahmadi, D.; Fakieh, B. Factors that affect the utilization of
low-code development platforms: Survey study. Rom. J. Inf. Technol. Autom. Control. 2021,31, 123–140. [CrossRef]
42.
Wang, H.; Zhang, L.; Yin, K.; Luo, H.; Li, J. Landslide identification using machine learning. Geosci. Front.
2021
,12, 351–364.
[CrossRef]
43.
Zhou, C.; Yin, K.; Cao, Y.; Intrieri, E.; Ahmed, B.; Catani, F. Displacement prediction of step-like landslide by applying a novel
kernel extreme learning machine method. Landslides 2018,15, 1015–1027. [CrossRef]
44.
Thirugnanam, H.; Ramesh, M.V.; Rangan, V.P. Enhancing the reliability of landslide early warning systems by machine learning.
Landslides 2020,17, 2231–2246. [CrossRef]
45.
Ma, Z.; Mei, G.; Piccialli, F. Machine learning for landslides prevention: A survey. Neural Comput. Appl.
2020
,33, 10881–10907.
[CrossRef]
Algorithms 2023,16, 108 22 of 23
46.
Microsoft. Power BI Pricing. 2022. [Online]. Available online: https://powerbi.microsoft.com/en-us/pricing/ (accessed on 12
November 2022).
47.
Sangaiah, A.K.; Javadpour, A.; Ja’fari, F.; Pinto, P.; Ahmadi, H.; Zhang, W. CL-MLSP: The design of a detection mechanism for
sinkhole attacks in smart cities. Microprocess. Microsyst. 2022,90, 104504. [CrossRef]
48. Woo, M. The Rise of No/Low Code Software Development-No Experience Needed? Engineering 2020,6, 960–961. [CrossRef]
49.
Microsoft. Microsoft Power BI Documentation. 2022. [Online]. Available online: https://docs.microsoft.com/en-us/power-bi/
(accessed on 21 March 2022).
50.
Yu, H.-F.; Hsieh, C.-J.; Chang, K.-W.; Lin, C.-J. Large Linear Classification When Data Cannot Fit in Memory. ACM Trans. Knowl.
Discov. Data (TKDD) 2010,5, 1–23.
51.
Matthies, H.; Strang, G. The solution of non linear finite element equations. Int. J. Numer. Methods Eng.
1979
,14, 1613–1626.
[CrossRef]
52. Nocedal, J. Updating Quasi-Newton Matrices with Limited Storage. Math. Comput. 1980,35, 773–782. [CrossRef]
53.
Ren, H.; Xu, B.; Wang, Y.; Yi, C.; Huang, C.; Kou, X.; Xing, T.; Yang, M.; Tong, J.; Zhang, Q. Time-Series Anomaly Detection Service
at Microsoft. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, New
York, NY, USA, 25 July 2019.
54.
Zhao, R.; Ouyang, W.; Li, H.; Wang, X. Saliency detection by multi-context deep learning. In Proceedings of the IEEE Conference
on Computer Vision and Pattern Recognition, Boston, MA, USA, 7–12 June 2015.
55.
Microsoft Documentation, Choosing a Natural Language Processing Technology in Azure. [Online]. Available online: https:
//docs.microsoft.com/en-us/azure/architecture/data-guide/technology-choices/natural-language-processing. (accessed on 25
February 2020).
56.
Microsoft Documentation, Decomposition Tree Visual: AI Split. 1 October 2020. [Online]. Available online: https://docs.
microsoft.com/en-us/power-bi/visuals/power-bi-visualization- decomposition-tree#ai-splits (accessed on 10 June 2021).
57.
Microsoft Documentation, Microsoft Power Automate. 2021. [Online]. Available online: https://docs.microsoft.com/en-us/
power-automate/ (accessed on 29 August 2021).
58.
Documentation, M. Category Classification Model. 2021. [Online]. Available online: https://docs.microsoft.com/en-us/ai-
builder/prebuilt-category-classification. (accessed on 1 October 2021).
59.
Hong, T.; Choi, J.-A.; Lim, K.; Kim, P. Enhancing Personalized Ads Using Interest Category Classification of SNS Users Based on
Deep Neural Networks. Sensors 2021,21, 199. [CrossRef] [PubMed]
60.
Pang, B.; Lee, L.; Vaithyanathan, S. Thumbs up?: Sentiment classification using machine learning techniques. arXiv
2002
,
arXiv:cs/0205070.
61. Turney, P.D. Thumbs up or thumbs down?: Semantic orientation applied. arXiv 2002, arXiv:cs/0212032.
62.
Boon-Itt, S.; Skunkan, Y. Public Perception of the COVID-19 Pandemic on Twitter: Sentiment Analysis and Topic Modeling Study.
JMIR Public Health Surveill 2020,6, 4. [CrossRef] [PubMed]
63. Nemes, L.; Kiss, A. Social media sentiment analysis based on COVID-19. J. Inf. Telecommun. 2020,5, 1–15. [CrossRef]
64.
Manguri, K.H.; Ramadhan, R.N.; Amin, P.R.M. Twitter Sentiment Analysis on Worldwide COVID-19 Outbreaks. Kurd. J. Appl.
Res. (KJAR) 2020,5, 54–63. [CrossRef]
65.
Sitaula, C.; Basnet, A.; Mainali, A.; Shahi, T.B. Deep Learning-Based Methods for Sentiment Analysis on Nepali COVID-19-Related
Tweets. Comput. Intell. Neurosci. 2021,2021, 2158184. [CrossRef]
66.
Chan, B.; Lopez, A.; Sarkar, U. The Canary in the Coal Mine Tweets: Social Media Reveals Public Perceptions of Non-Medical Use
of Opioids. PLoS ONE 2015,10, e0135072. [CrossRef]
67.
Naseem, U.; Razzak, I.; Khushi, M.; Eklund, P.W.; Kim, J. COVIDSenti: A large-scale benchmark Twitter data set for COVID-19
sentiment analysis. IEEE Trans. Comput. Soc. Syst. 2020,8, 1003–1015. [CrossRef]
68.
Li, L.; Zhang, Q.; Wang, X.; Zhang, J. Characterizing the Propagation of Situational Information in Social Media During COVID-19
Epidemic: A Case Study on Weibo. IEEE Trans. Comput. Soc. Syst. 2020,7, 556–562. [CrossRef]
69.
Cameron, D.; Smith, G.A.; Daniulaityte, R.; Sheth, A.P.; Dave, D.; Chen, L.; Anand, G.; Carlson, R.; Watkins, K.Z.; Falck, R.
PREDOSE: A Semantic Web Platform for Drug Abuse Epidemiology using Social Media. J. Biomed. Inform.
2013
,46, 6. [CrossRef]
70.
Chen, X.; Faviez, C.; Schuck, S.; Lillo-Le-Louët, A.; Texier, N.; Dahamna, B.; Huot, C.; Foulquié, P.; Pereira, S.; Leroux, V.; et al.
Mining Patients’ Narratives in Social Media for Pharmacovigilance: Adverse Effects and Misuse of Methylphenidate. Front.
Pharmacol. 2018,9, 541. [CrossRef]
71.
McNaughton, E.C.; Black, R.A.; Zulueta, M.G.; Budman, S.H.; Butler, S.F. Measuring online endorsement of prescription opioids
abuse: An integrative methodology. Pharmacoepidemiol. Drug Saf. 2012,21, 1081–1092. [CrossRef] [PubMed]
72.
Al-Twairesh, N.; Al-Negheimish, H. Surface and Deep Features Ensemble for Sentiment Analysis of Arabic Tweets. IEEE Access
2019,7, 84122–84131. [CrossRef]
73. Vashisht, G.; Sinha, Y.N. Sentimental study of CAA by location-based tweets. Int. J. Inf. Technol. 2021,13, 1555–1567. [CrossRef]
74.
Ebrahimi, M.; Yazdavar, A.H.; Sheth, A. Challenges of Sentiment Analysis for Dynamic Events. IEEE Intell. Syst.
2017
,32, 70–75.
[CrossRef]
75.
Microsoft Documentation, Text Analytics API Documentation. 2021. [Online]. Available online: https://docs.microsoft.com/en-
us/azure/cognitive-services/text-analytics/ (accessed on 3 August 2021).
Algorithms 2023,16, 108 23 of 23
76.
Batbaatar, E.; Ryu, K.H. Ontology-Based Healthcare Named Entity Recognition from Twitter Messages Using a Recurrent Neural
Network Approach. Int. J. Environ. Res. Public Health 2019,16, 3628. [CrossRef]
77. Bock, A.C.; Frank, U. Low-Code Platform. Bus. Inf. Syst. Eng. 2021,63, 733–740. [CrossRef]
Disclaimer/Publisher’s Note:
The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
Available via license: CC BY 4.0
Content may be subject to copyright.
