ArticlePDF Available

Fraud detection automation through data analytics and artificial intelligence

Authors:

Abstract

This study aims to review the use of data analytics and artificial intelligence in fraud detection to support internal audits. This study employs a qualitative method with a scoping review approach. The research data comprised 24 online journal articles indexed by Scopus and Sinta, which were used as the basis for scoping reviews. The stages carried out in this study consisted of identifying research questions, using keywords, selecting literature, mapping the results of research data, and compiling a summary of research results. This study concludes that the fraud detection model based on data analytics and artificial intelligence has a high accuracy value in improving audit quality. This study indicates that the Indonesian Financial and Development Supervisory Agency needs to increase the use of technology, including data analytics and artificial intelligence, to detect fraud optimally.
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
103
https://ejournal.ibik.ac.id/index.php/riset E-ISSN : 2656-7113 P-ISSN : 2797-4057 DOI : doi.org/10.37641/riset.v4i2.166
FRAUD DETECTION AUTOMATION THROUGH DATA
ANALYTICS AND ARTIFICIAL INTELLIGENCE
Wishmy Meinawa Ikhsan1), Elzami Haqie Ednoer2), Winanda Setyaning Kridantika3),
Amrie Firmansyah4)
1, 2,3,4) Polytechnic of State Finance STAN
E-mail: 4132210045_wishmy@pknstan.ac.id 1), elzamihaqie@gmail.com 2), amrie@pknstan.ac.id 4)
ABSTRACT
This study aims to review the use of data analytics and artificial
intelligence in fraud detection to support internal audits. This study
employs a qualitative method with a scoping review approach. The
research data comprised 24 online journal articles indexed by Scopus
and Sinta, which were used as the basis for scoping reviews. The
stages carried out in this study consisted of identifying research
questions, using keywords, selecting literature, mapping the results of
research data, and compiling a summary of research results. This
study concludes that the fraud detection model based on data analytics
and artificial intelligence has a high accuracy value in improving audit
quality. This study indicates that the Indonesian Financial and
Development Supervisory Agency needs to increase the use of
technology, including data analytics and artificial intelligence, to
detect fraud optimally.
Keywords: Artificial Intelligence, Data Analytic, Fraud, Internal
Audit
INTRODUCTION
The term industrial revolution is not something new anymore because the
phenomenon has been introduced since the 18th century (Fonna, 2019). They started
from the industrial revolution 1.0 with a focus on the revival of machines until now.
Industrial revolution 4.0 is marked by the development of digital technology and the
internet, increasing rapidly (Fonna, 2019). Areas that are trending in the industrial
revolution 4.0 are the Internet of Things (IoT), Industrial Internet of Things (IioT),
cyber-physical systems (CPS), cloud computing, and artificial intelligence (AI) (Fonna,
ARTICLE INFO
FRAUD DETECTION
AUTOMATION
THROUGH DATA
ANALYTICS AND
ARTIFICIAL
INTELLIGENCE
Submitted:
08 July 2022
Revised:
23 August 2022
Accepted:
27 September 2022
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
104
2019). These things focus on combining technology and the internet as an automation
tool that supports the production process and activities carried out by humans.
The development of increasingly advanced technology certainly positively
impacts various fields, such as industry, business, and finance. First, technology can
automate business processes and encourage people to find innovations related to their
business. Second, computerized accounting can provide predictions or projections
regarding future business prospects for the benefit of stakeholders. Third, with the
existence of a computerized system, it is no longer necessary to process the preparation
of financial reports manually (Naqvi, 2020).
In addition to providing benefits, technological developments also provide new
risks, such as viruses that attack the system, cybercrime, and illegal activity on
computer networks and the internet. In addition, the risk that also needs attention is
digital fraud, which can be interpreted as the use of web-based technology to commit
fraud and misuse assets to obtain financial benefits. Fraud is generally defined as an
activity that includes various behaviors in the form of fraud to make a profit (Button et
al., 2009).
Fraud is one of the fraudulent acts that are of great concern to the international
community. Fraud is a global problem that affects all organizations around the world.
Fraud is often undetected and never reported, making it difficult to determine the full
scope of global losses (ACFE, 2020). Indonesia ranks first in the country with the most
fraud cases in Asia, with 36 fraud cases (ACFE, 2020). The number of cases decreased
according to the ACFE study in 2022 to 23 cases (ACFE, 2022). Even though it is
starting to improve in 2022, Indonesia is currently in the top 4 countries with the highest
fraud cases in the Asia Pacific region, after Australia, China, and Malaysia (ACFE,
2022). This fraud case is detrimental to government organizations, the private sector,
and the public, who have got the impact directly or indirectly. ACFE Indonesia has
previously conducted a survey which proved that in 2019, losses due to fraud cases in
Indonesia were around IDR 873 billion, of which acts of corruption caused almost 70%
of these cases (ACFE Indonesia, 2020). It shows that fraud cases must be a concern to
be mitigated by every organization.
The impact of fraud has affected performance in various fields and professions,
one of which is auditing. Fraud causes the emergence of new areas that need the
auditor's attention. In 2020, each person generated an average of 1.7 MB of data per
second, increasing exponentially yearly. Data must be analyzed to ensure that fraud
checking is done thoroughly. If the data are unchecked, IT developments can become a
new opening for fraud because there are many hiding places for fraudsters. It is the
background for the emergence of the need to use emerging technology to support
auditor performance.
The extensive data population has resulted in the use of the sampling method in
conventional audits is no longer effective. Without a thorough examination of each data,
there is a risk that many critical areas go unnoticed. In addition, using conventional
audit techniques also requires more time, resources, and costs to audit a more extensive
scope. Thus, conventional audit techniques increase the risk of fraud detection failure
and reduce audit effectiveness and efficiency. The audit sector is still slow in adopting
emerging technologies to support audit activities (Satyawan et al., 2021). It is generally
due to concerns over the high initial costs incurred to adopt it, even though this initial
investment will be proportional to the long-term benefits obtained. In addition, the
demand for the required competencies is a barrier for auditors to use new techniques.
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
105
Fraud techniques are developing rapidly, and auditors must be able to respond to
them. It is not easy for auditors to keep up with the rapid development of information
technology, especially with demands regarding specific technical capabilities in auditing
these areas. However, it is still a must because adequate professional skills are the
foundation of confidence in the opinion given by the auditor.
The demand for auditors to have high specific technical capabilities is reflected in
Kovanen (2020), who finds there is an agreement in the views of auditors that in
maintaining audit relevance, audit scope should not be determined only based on the
limitations of existing capabilities. Hakami et al. (2020) found that the detection of
fraud by auditors through conventional audit activities has a lower accuracy rate than
the fraud detection model. It indicates the need for auditors to expand competence.
Furthermore, Mui (2018) found data analysis ability as one of the skills needed by
auditors in detecting fraud.
Fortress (2020) stated that big data technology, a new achievement in the world of
information and technology, impacts various kinds of work, one of which is auditors.
The use of big data analytics positively affects work performance and improves the
audit quality carried out by auditors. Big data analytics will significantly add value to
organizations by discovering hidden patterns in data (Handoko et al., 2020).
Big data is a good strategy for providing basic information, which is a significant
advantage for its users. However, big data takes a long time, accuracy is not guaranteed,
and costs are expensive. Given the availability of huge and complex amounts of data,
big data processing will be optimal if it involves a computer system to analyze further
the data they get to draw conclusions and take action. This system is called artificial
intelligence (AI). AI has intelligence in thinking, has a broad knowledge base in a
limited domain, and uses structured reasoning in making decisions to solve problems. It
is considered the leading solution for various cases of auditor failure in detecting fraud.
AI can provide convenient automation and control and improve audit process
efficiency. Muawanah et al. (2022) further observed how auditors perceive the ease with
which AI can improve the audit process. It was concluded that the auditor considered
the AI system to facilitate the audit process, although it is undeniable that there will also
be direct threats to employees' work. In addition, in its implementation, auditors will
undoubtedly face challenges in implementing AI to improve the audit process.
Specifically, internal audit has a significant role in fraud detection. The reporting
and internal auditing media are the most effective means of detecting fraud (ACFE
Indonesia, 2020). Reporting media originating from reports of employees of the
company/organization is the tool with the most significant contribution to uncovering
fraud. Furthermore, an internal audit is also very effective in the early detection of
fraud. Internal audits can detect fraud better by utilizing data analytics and AI
techniques. It is at the same time to maintain and increase the relevance of the internal
auditor function.
This study aims to review the use of data analytics and AI in fraud detection to
support internal audits. In addition, the risks and challenges that may be faced will also
be discussed. This study differs from previous research because this study discusses
data analytics support and artificial intelligence for internal audits. The previous studies
mainly discussed data analytics and artificial intelligence separately (Albizri et al.,
2019; Craja et al., 2020; West & Bhattacharya, 2016). They discussed audit coverage,
whether internal audit, external audit, or both.
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
106
Internal audit is an independent and objective assurance and consulting activity
designed to provide added value and improve the organization's operations (The
Institute of Internal Auditors, 2022). Internal audit activities are carried out through a
systematic and regular approach to evaluating and improving risk management, control,
and governance effectiveness to achieve organizational goals (The Institute of Internal
Auditors, 2022).
There is a shift in philosophy in viewing the role and function of internal audits in
an organization in the current era. According to Zamzami et al. (2013), there has been a
change in the orientation of the internal auditor profession from its initial role. As a
watchdog around 1940 to a consultant role around 1970. With various developments,
auditors play a role as strategic business partners for management and the board of
directors (Zamzami et al., 2013).
Data analytics is the process of evaluating data to obtain conclusions to answer
questions within a business (Richardson et al., 2020). Data analytics usually involves
the technologies, systems, methodologies, databases, statistics, and applications used to
analyze data (Richardson et al., 2020). The main point is that data analytics aims to turn
raw data into knowledge to create value for its users.
Data analytics is often juxtaposed with big data, resulting in the term big data
analytics. Big data is a complex data set with another term, namely 'the three Vs,'
consisting of volume, velocity, and variety. Volume characteristics indicate the amount
of big data that continues to grow and be produced every time. Velocity is the speed of
growth of big data that cannot be controlled. So it is difficult to process it with
conventional methods; variety means that all kinds of data are available in big data
(Oracle.com, 2022). Big data analytics is the process of extracting useful information by
analyzing various types of large data sets. Big data analytics is used to find hidden
patterns, market trends, and consumer preferences to benefit a company's decision-
making (Kurniawati, 2020).
AI is not new because its development started around the 1950s. This term was
first coined by McCarthy et al. (1955) as science and engineering to make intelligent
machines. The proposed definition of AI has developed along with the development of
AI (McCarthy et al., 1955). However, artificial intelligence is developing a system
capable of doing work that generally requires human thinking intelligence, such as
visual perception, decision making, and language/communication understanding
intelligence.
Now the development of AI is accelerating and indeed increasingly resembles
human intelligence. According to Naqvi (2020), the factors that influence these
developments include the exponential development of data that is so large and, of
course, accompanied by storage capacity and data management. Furthermore, increased
processing power increases data processing capabilities. Naqvi (2020) also stated that
research to find new algorithms and approaches is increasingly widespread. It is also
supported by the existence of a global network for researchers to share their research &
accelerate the development of AI.
Data analytics is closely related to artificial intelligence. The tools used in data
analytics are also used in artificial intelligence. In short, an artificial intelligence system
can replace or perform data analysis work usually performed by data analysts to obtain
information or implied knowledge of these data. Thus, artificial intelligence not only
processes data as data analytics tools function but also analyzes the processing results.
Naqvi (2020) recommended several tools, including robotic process automation (RPA),
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
107
expert systems, process mining, and machine learning to support audit automation. Data
analytics and AI are kinds of leveraged technology in the continuous audit. They are
real-time reporting and analysis in the form of an intelligent dashboard to show any
anomalies, outliers, inconsistencies, and non-compliance earlier to provide feedback and
feed-forward.
This research is expected to improve auditors' knowledge about various fraud
detection methods that have the potential to be adopted according to organizational
conditions. In addition to convincing auditors to immediately adopt new techniques, this
study is also expected to increase the readiness of internal auditors to implement them.
The potential benefits obtained are also expected to convince organizations to
encourage and facilitate their internal auditors in implementing fraud detection systems
based on data analytics and artificial intelligence. In addition, this research is expected
to contribute to developing literature on data analytics and AI for internal audits.
METHOD
This study employs a qualitative method with a scoping review approach to
discuss the implementation of data analytics and AI in fraud detection automation.
Scoping review is a method that pays attention to the need for comprehensive literature,
where each literature has a different level of depth of discussion (Arksey & Malley,
2005). This study took secondary data from 24 journal articles in online media indexed
by Scopus and Sinta according to the theme of the discussion. The scoping review steps
in this study are to identify research questions: "How is the implementation of data
analytics and AI in fraud detection automation?". Next is to search the literature in the
Scopus and Sinta indexed databases with the keywords: fraud detection with big data
analytics and fraud detection with artificial intelligence. The next stage is to select the
literature and map the data. The final stage of the scoping review of this research is
compiling a summary of the results of data mapping. These stages used scoping review
follows Pamungkas & Firmansyah (2021).
RESULTS
Implementing Data Analytics for Fraud Detection
Singh et al. (2019) started their research by determining the common fraud
indicators in transactions. A total of 2,117 procurement transaction data at
telecommunications companies in America became the object of testing in a model that
utilizes binary logistic regression (BLR). However, BLR is strongly influenced by the
cutoff value, where a high cutoff value tends to reduce the number of transactions
marked as risky. In contrast, a low cutoff increases transactions marked as risky and
false positives. False positives are usually caused by an algorithm of a program that
states the existence of a symptom/signal/object that does not exist. The result is that the
lowest cutoff value significantly produces the highest accuracy for marking fraud risk
transactions of 68.6%. It should be noted that Singh et al. (2019) emphasized the
importance of continuously updating the fraud indicator base in line with the
development of fraud techniques.
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
108
Singh et al. (2019) also described the advantages offered by the built model. From
the agency theory point of view, this model can reduce the costs required to create and
execute contracts and monitor those contracts. It is a fundamental problem of agency
theory where information asymmetry appears when shareholders entrust wealth
management to the manager, so additional costs are needed to mitigate the risk. In
addition to reducing incentives/encouragements to commit fraud, this model can also
increase auditor efficiency and independence and improve overall audit quality.
Moreover, this model can also be used to detect fraud in real-time, considering that the
type of data used is also updated in real-time.
On a larger scale, Patil et al. (2018) performed big data analytics to detect fraud in
real-time. This study uses logistic regression algorithms, decision trees, and random
forest decision trees to process a large volume of banking data. The test results found
that the fraud detection precision from logistic regression was 76%, the decision tree
was 89%, and the random forest decision tree was 93%.
Rizki et al. (2017) examined how data processing applications detect fraud in
companies in Indonesia. The companies studied were listed on the Indonesia Stock
Exchange (IDX), which consisted of 100 non-fraud companies and 24 fraudulent
companies. Significant indicators in detecting financial fraud are profitability and
efficiency. Algorithm accuracy used Support Vector Machines (SVM) is 88.37%, while
Artificial Neural Network (ANN) produces a better accuracy of 90.97%. However, there
is no significant difference between the two algorithms, and the determination of the
algorithm used is based on the data set, operating time, and ease of implementation.
Data analytics can also be used as a complement to other fraud detection
techniques. Tarjo & Herawati (2015) found that the Beneish-M score can be combined
with logistic regression data mining techniques to detect fraud. The results of this study
of 70 companies in Indonesia show that the accuracy of fraud detection by this model is
77.1%. Tarjo & Herawati (2015) also found that depreciation, cost of sales, and
discretionary accrual accounting policies can reduce the model's ability to detect fraud.
Another breakthrough in fraud detection is the use of text mining. Dong et al.
(2018) examined financial and social media data to detect fraud. Social media is not
directly related to organizational operations, but hidden insights can still be drawn from
unstructured data on social media. The insights are extracted from signals such as
sentiments, emotions, topics, lexical or related words, and social networks. The data
obtained is, of course, very sufficient in number to be analyzed, considering that the use
of social media is increasing rapidly. In addition, processing can be carried out in real-
time, considering that the data can be recorded directly. The data is then processed with
four algorithms (SVM, NN, decision tree/DT, LR) to determine which produces the best
accuracy. The fraud detection accuracy rate among 128 companies ranged from 54.5%
to 75.5%, with the highest accuracy held by SVM. In addition, the analysis of the
language used in the management discussion and analysis section of the financial
statements produces detection accuracy between 52.78% to 70.33%, with the best
results given by the LR algorithm.
In comparison, fraud detection with financial ratios tested in this study has an
accuracy ranging from 41.17% to 56.17%. However, the best results are produced from
a combination using SVM, reaching an accuracy of 56.17% to 80%. This research also
shows how this method can help auditors better assess the risk of material
misstatements.
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
109
Goel et al. (2010) examined whether linguistic predictors detect fraud in finance.
This study uses datasets from a sample of 126 companies that committed fraud from
1993 to 2006. The method used for testing is Natural Language Processing (NLP) tools
to analyze, understand, and produce language. The Support Vector Machines (SVM)
method is a machine learning method to study the characteristics of the tested dataset.
This method can increase the prediction accuracy of the fraud detection model from the
initial baseline with an accuracy of 56.75% to 89.51%. Interestingly, there have been
systematic differences in communication and writing style of the annual report that
contained fraud where financial figures did not accurately capture the narrative.
Goel & Uzuner (2016) further examined text analysis by examining whether
sentiment affected fraud detection in the annual report. The methods used are Natural
Language Processing (NLP) and Support Vector Machines (SVM) which are considered
the most suitable for building the fraud sentiment classification model. This method
identifies sentiment features to distinguish between fraudulent and honest Management
Discussion and Analysis (MD&As) by measuring sentiment on polarity, subjectivity,
and intensity dimensions. The classification model built can identify 81.84% accurately.
This model can identify a more prominent use of negative and positive sentiment, a
higher proportion of subjective expression than objective expression, and greater use of
sentiment expression.
Data analytics can also detect fraud in other fields, not just financial statement
fraud, one of which is international shipping. The risk of fraud often faced in this field
is the manipulation of documents to avoid customs fees and smuggling. Usually,
random audits are conducted to check the completeness of the documents. However,
this technique cannot cover the entire sender, so the risk of passing the manipulated
document is still large. Triepels et al. (2018) investigated fraud detection used the
Bayesian Network to analyze trade and shipping route patterns. As a result, random
audits with conventional methods showed a very low fraud detection precision of
9.99%, while the data-driven model reached a value between 34.6% and 60.7%. This
detection model also has the potential to be applied to companies that have a similar
business nature as retail companies that manage lots of inventory movement. With this
model, auditors can reduce physical audits while increasing accuracy by involving a
larger population.
Not just a model, data analytics for fraud handling has been implemented directly
in the real world. Unmitigated, the giant company Alibaba uses big data as a basis for
fraud risk management. Chen et al. (2015) reviewed how Alibaba uses CTU as a real-
time fraud prevention monitoring system and a risk model in the form of a Readiness
and Investment Navigator (RAIN) score. Alibaba's fraud risk method leads to a new
generation of fraud risk monitoring and management based on big data processing and
computing technology, as well as a real-time fraud prevention system.
The findings of testing the accuracy of data analytics models that show their
benefits in detecting fraud seem to align with the auditor's perception. Syahputra &
Afnan, (2020) surveyed 221 auditors working at Indonesia Audit Supreme Board and
Financial and Development Supervisory Agency spread throughout Indonesia. This
study examined the effect of big data and forensic audits on fraud detection. The test
results suggest that big data and forensic audits are positively associated with fraud
detection.
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
110
Application of Artificial Intelligence in Fraud Detection
Tang et al. (2018) developed an intelligent fraud detection model based on a fraud
detection ontology. Ontology is a discussion of the most basic or deepest principles
(Tang et al., 2018). In this study, the ontology was compiled based on financial
statement data to produce financial variables that significantly affect detecting fraud
(Tang et al., 2018) by combining the ontology knowledge base with the C4.5. An
algorithm is used to serve the decision tree. This model was tested on 130 fraudulent
companies and 130 non-fraudulent companies. As a result, the detection rate of
fraudulent financial statements reached 86.67%, with a total accuracy of 80%,
indicating this model's validity. However, it should emphasize that this model relies
purely on financial reports' data without expert knowledge. Even though the auditor's
field knowledge has its value which is not always reflected directly in the financial
statements.
Financial variables from financial statements are also used in the research
(Temponeras et al., 2019). If Tang et al. (2018) obtained the main variables for
detecting fraud from the compilation of ontologies, Temponeras et al. (2019)
determined the variables based on professional judgment. The machine learning
algorithm used to detect fraud in this study is Denoising Diffusion Probabilistic Models
(DDPM). It is then compared with Deep Dense Multilayer Perceptron Decision tree
C4.5; k-NN type 3-NN; logistic regression; and minimal sequential optimization for
support vector machine (SVM). In testing 164 companies in Greece, DDPM obtained
the best results among other variables, with a fraud detection rate of 91.7% and a total
accuracy of 93.7%. With this model, sound professional judgment from the auditor is
needed to determine the proper characteristics of fraud as the basis for detection. Bao et
al. (2020) used raw data instead of financial ratios because raw data can be used more
flexibly and in more complex functional forms. With a data sample of 206,026
American companies, using the Normalized Discounted Cumulative Gain (NDCG)@k
model can relatively avoid high direct and indirect costs for investigating cases
suspected of fraud.
In addition to financial variables, non-financial variables were used together in the
study of Jan (2018) on 40 fraudulent companies and 120 non-fraudulent companies in
Taiwan. In the first stage, these essential variables are filtered by artificial neural
networks (ANN) and SVM. Ten-fold cross-validation was carried out in the second
stage to improve detection accuracy in 4 decision tree algorithms. From the eight
models built, the combination of ANN and CART decision tree resulted in the highest
fraud detection rate of 90.21%, with a total accuracy of 90.83%. This study also
emphasized the need for internal control, audit systems, and corporate governance to
avoid fraud and produce objective data for more accurate analysis. Internal auditors
certainly play a significant role in ensuring the system's realization.
Mustika et al. (2021) examined fraud detection in retail financing companies in
Indonesia. Departing from the pandemic conditions that limited audit space, remote
audit activities became a solution that was intensively encouraged. However, limited
direct interaction also increases the risk of document manipulation and deletion of
relevant information. This study employs 46,536 transactions as objects and employs
customer and finance profiles as predictors of fraud. Based on the test, the random
forest algorithm became the method with the highest accuracy of 74.54%. When
compared, the previously discussed studies have a total accuracy rate of more than 80%
and fraud detection accuracy above 85%. Mustika et al. (2021) suggested further
research to add appropriate predictor variables to improve accuracy. Thus, the
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
111
implication of conducting an audit is that auditors need to understand the characteristics
of fraud in depth. Professional judgment in determining these variables will affect the
accuracy of fraud detection.
AI is certainly not limited to automating the audit process itself but can be used in
all operational activities of the organization. However, not a direct part of the internal
audit process described above, AI in operational activities can also contribute indirectly
to the audit process.
Several studies have used artificial intelligence to detect fraud used by
management. Raj & Portia (2011) conducted a survey and evaluated AI-based fraud
detection techniques used the Bayesian Belief Networks (BNN) and Artificial neural
network (ANN) models. The results of this study are that the ANN and BNN models
have a medium accuracy and are pretty expensive. Fraud detection rates for BNN and
ANN are 74% and 77%, respectively.
Sadineni (2020) also conducted a similar study by utilizing the Kaggle data
repository, which has 150,000 stored transactions. The research resulted in machine
learning accuracy for detecting fraud, which was above 95%. While Alghofaili et al,
(2020) developed methods for increasing computational speed, handling big data, and
identifying unknown attack patterns. The deep learning technique used in this research
is Long Short-Term Memory (LSTM) which can achieve 99.95% accuracy in less than
one minute. At the same time, Thennakoon et al. (2019) evaluated a series of machine
learning models to choose the optimal algorithm for the type of fraud. The data used in
this study are files with 200 fraud case records, and transaction log files have 917,781
records with four different fraudulent transaction patterns. The results suggest a system
for detecting fraud in real-time consisting of three central units: API module, fraud
detection models, and data warehouse.
Exploring different variables, P. Singh & Singh (2015) built a fraud detection
model by monitoring behavior and geographic location used the Bayesian Learning
Approach. This model can assess the possibility of fraudulent activity by providing a
risk score where a score greater than 0.5 will be considered fraud. Not only detection,
but this model also applies advanced mechanisms to prevent fraud. When a possible
fraud is detected, the system will authenticate the user by asking the user to input the
correct verification code.
Vynokurova et al. (2020) examined Hybrid Machine Learning systems to
overcome anomaly detection in user geolocation (GPS spoofing, Wi-Fi spoofing,
location jumping). This hybrid system consists of two subsystems: an anomaly
detection subsystem and an anomaly type interpretation subsystem, which uses a
decision tree technique with supervised learning and unsupervised learning. This hybrid
system produces a fraud detection accuracy rate of above 90%. The advantage of this
hybrid system is higher data processing speed when given real-time data.
In the insurance sector, Dhieb et al. (2020) developed a Smart Insurance System
based on Blockchain and AI (SISBAR), a fraud detection system used extreme gradient
boosting (XGBoost), and Very Fast Decision Tree (VFDT) as dynamic fraud detection
and classification model for insurance companies. Based on the test, it was concluded
that the XGBoost method has the highest accuracy and prediction. Roy & George
(2017) used more than 500 insurance claim data to detect fraud in insurance claims
using machine learning. Machine learning techniques are used in the form of the
decision tree, random forest, and Naive Bayes. Decision tree and random forest
algorithm classification techniques perform better than naive Bayes.
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
112
Although all systems mentioned above are not used directly by the auditors, the
information obtained from these systems can be used to evaluate performance and
internal control carried out by management, for example, in detecting insurance claim
fraud. If the system finds a high level of fraud, the auditor needs to audit the cause
further, whether it comes from inadequate control, poor team member performance, or
pure customer error. If the cause is found in the internal organization, the auditor must
provide recommendations on the problem. In addition, these models can be used as a
blueprint for developing new fraud detection models that can be used directly in audits.
This development mainly explores new variables that have not been used in audit
automation, such as behavior patterns, location, and geolocation.
Audit activities certainly do not stop when fraud is discovered. A series of audit
work programs still need to be carried out to produce an audit report containing the
auditor's opinion. In this case, Soeprajitno (2019) analyzed the potential of Artificial
Intelligence in issuing auditor opinions. Machine learning Artificial Neural Network
(ANN) tools were used in this study. Combining capabilities on big data, data mining,
artificial neural networks, and final analysis through calculating total irregularities
obtained using Fuzzy will help AI have the ability to detect fraud and issue audit
opinions. Professional judgment, usually done manually, can be obtained from the
ability of AI as the basis for issuing explanatory paragraphs.
Analysis of the Application of Data Analytics and Artificial Intelligence for Fraud
Detection
The results of testing the fraud detection model with the highest accuracy in each
article can be seen in the following table:
Table 1. The Summary Of Fraud Detection Model Test Results
Testing Accuracy Rate
Number of Models
Algorithm Used
90-99%
10
ANN[38], XGBoost+VFDT[15],
ANN[32], ANN[35], SVM[45], LSTM[6],
HMLS[47], RF[30], DDMP[2],
ANN+CART[14]
80-89%
3
NLP+SVM[20], NLP+SVM[21],
Ontology+C4.5[10]
70-79%
4
BNN+ANN[31], RF[5]
SVM[18], Beneish M-Score+LR[27]
60-69%
2
BLR[26], BSN-NN[22]
Source: Data processed
Based on the accuracy table of the fraud detection model above, ten research
results (52.63%) show that the accuracy of tools is above 90%, three research results
(15.79%) show tool accuracy is in the 80%-89% range, four research results (21.05%) )
shows tools are in the range of 70%-79%. Two research results (10.53%) show tools are
in the range of 60%-69%. Data analytics and artificial intelligence accuracy in detecting
fraud are very high. More than half of the samples used showed an accuracy rate above
90%, and the rest were within an average accuracy of 85.55%. It shows that data
analytics and artificial intelligence can be very accurate in detecting fraud and
successful in their application.
In addition, several exciting studies review the success of data analytics and
artificial intelligence methods in identifying fraud. Syahputra & Afnan (2020) found
that big data is positively associated with detecting fraud and conducting forensic
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
113
audits. Roy & George (2017) found that the decision tree and random forest algorithms
have the best accuracy, precision, and recall performance. Furthermore, P. Singh &
Singh (2015) used the Bayesian Learning Approach to detect fraud and perform
authentication for security if anomalies are found in transactions. The fraud risk
management model created by the Alibaba company uses big data, which was later
examined by Chen et al. (2015) and demonstrated its ability to detect fraud in real-time.
In contrast, Bao et al. (2020) used raw data processed with the NDCG@k algorithm
considering flexibility.
All these studies offer an overview of how fraud detection can use various
algorithms with various types of variables and input data. This condition shows
flexibility in the design of fraud detection systems. Auditors can adjust the types of
variables and input data according to organizational conditions. The data with adequate
availability can be selected for analysis from the various types of data that can be used.
In addition, the algorithm used can adjust the ability of the system builder because there
is no rigid limit to only using specific algorithms.
The variable selection also plays an essential role in obtaining the best level of
accuracy. Auditors can test different variable combination types to find the most ideal
for the organization. In addition, since fraud techniques continue to develop in line with
technology, auditors must be keen to see gaps that can become indicators of new fraud.
Therefore, the variables used in the system also need to be updated regularly. It is where
the auditor's professional judgment is needed to assess which areas need more attention
as potential new variables.
The success and usefulness of implementing data analytics and artificial
intelligence are beneficial as a reference and guide for internal auditors to be able to
carry out supervisory duties. The rapid development of information has broadened the
scope of organizational audits and is more diverse. In this case, the auditor cannot
continue to rely on manual control techniques. Along with the development of
information technology, fraud patterns continue to develop, so more advanced,
automated, fast, accurate monitoring techniques are needed and can monitor an
enormous scope in identifying fraud.
Based on the studies that have been carried out, it is known that the application of
data analytics and artificial intelligence will benefit the world of internal audits.
Implementing big data analytics and artificial intelligence can reduce the potential for
audit delays due to its ability to detect fraud in real-time. Audit delay is the period for
completing the audit from the date of the company's reporting until the audit report is
issued (Widati & Septy, 2008).
Data analytics and artificial intelligence can detect fraud in real-time and are
effectively used by organizations. Using these two tools can also reduce: manual fraud
detection activities, increase the scope of detection, speed up the process, and have a
better future prediction system. In addition, artificial intelligence can also assist auditors
in reducing physical work, where previously auditors had to examine various
documents with many pages. Artificial intelligence plays a role in automatically
recognizing and processing all documents so that manual processes can be reduced and
auditors have more time to analyze. This intelligent audit automation can also help
improve audit quality because it has a high level of accuracy for detecting fraud and
reduces costs for conducting audits.
Without reducing the benefits of artificial intelligence, it should still be noted that
based on the analysis, not all artificial intelligence methods can detect fraud in real time
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
114
or are fully automated. Artificial intelligence has been able to conclude/information like
humans, but in most models, the process is still semi-automated. The type of input data
used can not all be captured directly by the system, or it still needs to be inputted first
by the officer. Integrating the fraud detection system with the operating system that
produces data as material for fraud analysis is necessary to automate the process. This
integration allows the fraud detection system to analyze each new data entry during an
activity/transaction. The development of this real-time system needs to be carried out
continuously because the faster the early warning system works, the intervention can be
carried out immediately so that losses can be minimized.
Previously, it was also mentioned how initial investment was an obstacle to
adopting data analytics and artificial intelligence. Although it is very clear how many
benefits will be obtained, there are not many studies that directly mention the number of
costs required. This information is helpful for organizations to conduct a cost-benefit
analysis. The existence of more explicit cost information can further convince
organizations to facilitate the use of data analytics and artificial intelligence.
The many benefits derived from data analytics and artificial intelligence do not
mean that these two tools are easy to implement. In addition to costs, auditors certainly
need to consider the risks that must be mitigated so as not to harm the achievement of
organizational goals. In addition, auditors face challenges in implementing data
analytics and artificial intelligence to detect fraud optimally.
Kovanen (2020) identified several risks that accompany audit automation. First,
there are technological risks in data errors, algorithm and system malfunctions such as
producing incorrect output or system crashes. Second, the regulation on the use of data
needs to be a concern, especially in the ethical use of data and the demand to maintain
data security that concerns many parties. Another risk that needs to be considered is
cyber attacks. Much data continuously processed by the system is sensitive data, such as
financial and employee information. This condition further increases the vulnerability to
cyber-attacks. This attack can be in the form of unauthorized access, extraction,
deletion, and destruction of data, as well as other activities that result in disruption,
operations, and losses, both financial and non-financial. In addition, detection accuracy
also depends on the adequacy and authenticity of the processed data. Thus, internal
auditors must ensure that the organization has adequate internal controls to produce
quality data.
Apart from risks, there are challenges faced in using intelligent automation. The
biggest challenge is the demand for auditor competence to keep pace with technological
developments. The most significant demands are technical capability, regulatory
understanding, and accountability for audit confidence and accuracy (Kovanen, 2020).
The transparency of this intelligent system is also a concern of its own. Audit
automation makes it easier for auditors to audit activities that utilize artificial
intelligence and data analytics and the organization's operational activities. The
algorithms used may also be different from those used for audit automation. Auditors
must also understand how this system works, requiring management's openness in
conveying it.
High expectations from regulators and organizations as auditees are also
challenging (Soeprajitno, 2019). Stakeholders have always expected the assurance level
to reach 100%, although it is difficult to achieve. Previously, conducting conventional
audits by sampling could cause missing critical areas to be examined. Nevertheless,
now with the help of intelligent audit automation with high accuracy, the expected
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
115
standard for audit confidence levels is getting higher. There is no longer any excuse for
auditors not to meet stakeholder expectations, so auditors must rush to master these new
techniques for better audit quality.
CONCLUSION
This study concludes that various fraud detection models based on data analytics
and artificial intelligence have a high accuracy value in improving audit quality. These
models with various algorithms, variables, and input data can be a blueprint for
developing a fraud detection system following the organization's characteristics.
This research is limited to the scoping review method on the literature that has
reviewed and tested fraud detection based on data analytics and artificial intelligence. In
this regard, there are still many things that need to be explored regarding techniques in
data analytics and artificial intelligence, which also continue to develop in line with the
development of information technology. Not all data analytics and artificial intelligence
techniques are suitable for every characteristic of fraud. Further research is expected to
be able to examine more deeply related to the techniques of applying data analytics and
artificial intelligence to detect each type and characteristic of fraud.
This research indicates that the Indonesia Financial and Development Supervisory
Agency, as the supervisory agency for the Government's Internal Supervisory
Apparatus, is expected to encourage the urgency of using technology, including data
analytics and artificial intelligence, to detect fraud optimally. In addition, the Indonesia
Agency for the Assessment and Application of Technology needs to improve the
competence of internal auditors and technology experts to cooperate in developing
artificial intelligence for fraud detection.
REFERENCES
ACFE. (2020). Report to The Nations on Occupational Fraud and Abuse: 2020 Global
Fraud Study. https://acfepublic.s3-us-west-2.amazonaws.com/2020-Report-to-
the-Nations.pdf
ACFE. (2022). Occupational Fraud 2022: A Report to The Nations.
https://acfepublic.s3.us-west-
2.amazonaws.com/2022+Report+to+the+Nations.pdf
ACFE Indonesia. (2020). Indonesian Fraud Survey 2019. https://acfe-
indonesia.or.id/wp-content/uploads/2021/02/SURVEI-FRAUD-INDONESIA-
2019.pdf
Albizri, A., Appelbaum, D., & Rizzotto, N. (2019). Evaluation of Financial Statements
Fraud Detection Research: A Multi-Disciplinary Analysis. International Journal
of Disclosure and Governance, 16(4), 206241. https://doi.org/10.1057/s41310-
019-00067-9
Alghofaili, Y., Albattah, A., & Rassam, M. A. (2020). A Financial Fraud Detection
Model Based on LSTM Deep Learning Technique. Journal of Applied Security
Research, 15(4), 498516. https://doi.org/10.1080/19361610.2020.1815491
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
116
Arksey, H., & Malley, L. O. (2005). Scoping Studies: Towards A Methodological
Framework. International Journal of Social Research Methodology, 8(1), 1932.
https://doi.org/10.1080/1364557032000119616
Bao, Y., Ke, Bi., Li, B., Yu, Y. J., & Zhang, J. (2020). Detecting Accounting Fraud in
Publicly Traded U.S. Firms Using a Machine Learning Approach. Journal of
Accounting Research, 58(1), 199235. https://doi.org/10.1111/1475-679X.12292
Button, M., Lewis, C., & Tapley, J. (2009). Fraud typologies and victims of fraud.
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attac
hment_data/file/118469/fraud-typologies.pdf
Chen, J., Tao, Y., Wang, H., & Chen, T. (2015). Big Data Based Fraud Risk
Management at Alibaba. Journal of Finance and Data Science, 1(1), 110.
https://doi.org/10.1016/j.jfds.2015.03.001
Craja, P., Kim, A., & Lessmann, S. (2020). Deep Learning For Detecting Financial
Statement Fraud. Decision Support Systems, 139, 113421.
https://doi.org/10.1016/j.dss.2020.113421
Dhieb, N., Ghazzai, H., Besbes, H., & Massoud, Y. (2020). A Secure AI-Driven
Architecture for Automated Insurance Systems: Fraud Detection and Risk
Measurement. IEEE Access, 8, 5854658558.
https://doi.org/10.1109/ACCESS.2020.2983300
Dong, W., Liao, S., & Zhang, Z. (2018). Leveraging Financial Social Media Data for
Corporate Fraud Detection. Journal of Management Information Systems, 35(2),
461487. https://doi.org/10.1080/07421222.2018.1451954
Fonna, N. (2019). Development of the Industrial Revolution 4.0 in Various Fields.
Bogor: Guepedia.
Fortress, H. (2020). The Effect of Using Big Data Analytics on The Work of Auditors
[Universitas Katolik Parahyangan].
https://repository.unpar.ac.id/handle/123456789/11669
Goel, S., Gangolly, J., Faerman, S. R., & Uzuner, O. (2010). Can Linguistic Predictors
Detect Fraudulent Financial Filings? Journal of Emerging Technologies in
Accounting, 7(1), 2546. https://doi.org/10.2308/jeta.2010.7.1.25
Goel, S., & Uzuner, O. (2016). Do Sentiments Matter in Fraud Detection? Estimating
Semantic Orientation of Annual Reports. Intelligent Systems in Accounting,
Finance and Management, 23(3), 215239. https://doi.org/10.1002/isaf.1392
Hakami, T. A., Rahmat, M. M., Yaacob, M. H., & Saleh, N. M. (2020). Fraud Detection
Gap between Auditor and Fraud Detection Models: Evidence from Gulf
Cooperation Council. Asian Journal of Accounting and Governance, 13, 113.
https://doi.org/10.17576/ajag-2020-13-01
Handoko, B. L., Mulyawan, A. N., Tanuwijaya, J., & Tanciady, F. (2020). Big Data in
Auditing for Tthe Future of Data Driven Fraud Detection. International Journal
of Innovative Technology and Exploring Engineering, 9(3), 29022907.
https://doi.org/10.35940/ijitee.b7568.019320
Jan, C. (2018). An Effective Financial Statements Fraud Detection Model for the
Sustainable Development of Financial Markets: Evidence from Taiwan.
Sustainability, 10(2), 513. https://doi.org/10.3390/su10020513
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
117
Kovanen, A. (2020). Risks of Intelligent Automation and Their Impact on Internal Audit
[Tampere University]. https://trepo.tuni.fi/handle/10024/121440
Kurniawati, G. N. (2020). Big Data Analytics and Its Uses for Your Business
Development. DO Lab. https://www.dqlab.id/big-data-analytics-dan-
kegunaannya-untuk-perkembangan-bisnis
McCarthy, J., Minsky, M. L., Rochester, N., & Shannon, C. E. (1955). A Proposal for
The Dartmouth Summer Research Project on Artificial Intelligence. 113.
http://jmc.stanford.edu/articles/dartmouth/dartmouth.pdf
Muawanah, A., Adawiyah, D., Maisarah, I., Ali, M. R. A., & Widiastuti, N. P. E.
(2022). Auditor Behavior Responding to the Emergence of Artificial Intelligence
in the Audit Process. Jurnal Publikasi Ekonomi Dan Akuntansi , 2(1), 5260.
http://ejurnal.stie-trianandra.ac.id/index.php/jupea/article/view/152
Mui, G. Y. (2018). Defining Auditor Expertise in Fraud Detection. Journal of Forensic
and Investigative Accounting, 10(2), 168186.
https://s3.amazonaws.com/web.nacva.com/JFIA/Issues/JFIA-2018-No2-2.pdf
Mustika, N. I., Nenda, B., & Ramadhan, D. (2021). Machine Learning Algorithms in
Fraud Detection: Case Study on Retail Consumer Financing Company. Asia
Pacific Fraud Journal, 6(2), 213221.
https://doi.org/10.21532/apfjournal.v6i2.216
Naqvi, A. (2020). Artificial Intelligence for Audit, Forensic Accounting, and Valuation.
New Jersey: John Wiley & Sons, Inc. https://doi.org/10.1002/9781119601906
Oracle.com. (2022). What is Big Data? Oracle Cloud Infrastructure (OCI).
https://www.oracle.com/big-data/what-is-big-data/
Pamungkas, U. D., & Firmansyah, A. (2021). How is The Regulation of Cryptocurrency
Ownership by Companies Based on Financial Accounting Standards? Jurnal
Ilmiah Akuntansi Kesatuan, 9(3), 489510.
https://doi.org/10.37641/jiakes.v9i3.895
Patil, S., Nemade, V., & Soni, P. K. (2018). Predictive Modelling for Credit Card Fraud
Detection Using Data Analytics. Procedia Computer Science, 132, 385395.
https://doi.org/10.1016/j.procs.2018.05.199
Raj, S. B. E., & Portia, A. A. (2011). Analysis on Credit Card Fraud Detection Methods.
2011 International Conference on Computer, Communication and Electrical
Technology (ICCCET), 152156. https://doi.org/10.1109/ICCCET.2011.5762457
Richardson, V., Terrell, K., & Teeter, R. (2020). Data Analytics for Accounting (2nd
ed.). New York: McGraw-Hill. https://www.amazon.com/Data-Analytics-
Accounting-Vernon-Richardson/dp/1260837831
Rizki, A. A., Surjandari, I., & Wayasti, R. A. (2017). Data Mining Application to Detect
Financial Fraud in Indonesia’s Public Companies. 2017 3rd International
Conference on Science in Information Technology (ICSITech), 206211.
https://doi.org/10.1109/ICSITech.2017.8257111
Roy, R., & George, K. T. (2017). Detecting Insurance Claims Fraud using Machine
Learning Techniques. 2017 International Conference on Circuit, Power and
Computing Technologies (ICCPCT), 16.
https://doi.org/10.1109/ICCPCT.2017.8074258
Wishmy Meinawa Ikhsan 1), Elzami Haqie Ednoer 2), Winanda Setyaning Kridantika 3), Amrie
Firmansyah 4), Fraud Detection Automation Through Data Analytics and Artificial Intelligence
118
Sadineni, P. K. (2020). Detection of Fraudulent Transactions in Credit Card Using
Machine Learning Algorithms. 2020 Fourth International Conference on I-
SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), 659660.
https://doi.org/10.1109/I-SMAC49090.2020.9243545
Satyawan, M. D., Triani, N. N. A., Yanthi, M. D., Siregar, C. S., & Kusumaningsih, A.
(2021). Accelerating the Role of Technology in Audit During Covid-19
Pandemic. Jurnal Akuntansi Multiparadigma, 12(1), 186206.
https://doi.org/10.21776/ub.jamal.2021.12.1.11
Singh, N., Lai, K., Vejvar, M., & Cheng, T. C. E. (2019). Data‐Driven Auditing: A
Predictive Modeling Approach to Fraud Detection and Classification. Journal of
Corporate Accounting & Finance, 30(3), 6482.
https://doi.org/10.1002/jcaf.22389
Singh, P., & Singh, M. (2015). Fraud Detection by Monitoring Customer Behavior and
Activities. International Journal of Computer Applications, 111(11), 2332.
https://doi.org/10.5120/19584-1340
Soeprajitno, R. R. W. N. (2019). Artificial Intelligence (AI) Potencial Issue an
Auditor’s Opinion? Jurnal Riset Akuntansi Dan Bisnis Airlangga, 4(1).
https://doi.org/10.31093/jraba.v4i1.142
Syahputra, B. E., & Afnan, A. (2020). Fraud Detection: The Role of Big Data and
Forensic Audit. Jurnal ASET (Akuntansi Riset), 12(2), 301316.
https://doi.org/doi.org/10.17509/jaset.v12i2.28939
Tang, X.-B., Liu, G.-C., Yang, J., & Wei, W. (2018). Knowledge-based Financial
Statement Fraud Detection System: Based on an Ontology and a Decision Tree.
Knowledge Organization, 45(3), 205219. https://doi.org/10.5771/0943-7444-
2018-3-205
Tarjo, & Herawati, N. (2015). Application of Beneish M-Score Models and Data
Mining to Detect Financial Fraud. Procedia - Social and Behavioral Sciences,
211, 924930. https://doi.org/10.1016/j.sbspro.2015.11.122
Temponeras, G. S., Alexandropoulos, S.-A. N., Kotsiantis, S. B., & Vrahatis, M. N.
(2019). Financial Fraudulent Statements Detection through a Deep Dense
Artificial Neural Network. 2019 10th International Conference on Information,
Intelligence, Systems and Applications (IISA), 15.
https://doi.org/10.1109/IISA.2019.8900741
The Institute of Internal Auditors. (2022). About Internal Audit. The Institute of Internal
Auditors. https://www.theiia.org/en/about-us/about-internal-audit/
Thennakoon, A., Bhagyani, C., Premadasa, S., Mihiranga, S., & Kuruwitaarachchi, N.
(2019). Real-Time Credit Card Fraud Detection Using Machine Learning.
Proceedings of the 9th International Conference On Cloud Computing, Data
Science and Engineering, Confluence 2019, 488493.
https://doi.org/10.1109/CONFLUENCE.2019.8776942
Triepels, R., Daniels, H., & Feelders, A. (2018). Data-Driven Fraud Detection in
International Shipping. Expert Systems with Applications, 99, 193202.
https://doi.org/10.1016/j.eswa.2018.01.007
Vynokurova, O., Peleshko, D., Bondarenko, O., Ilyasov, V., Serzhantov, V., &
Peleshko, M. (2020). Hybrid Machine Learning System for Solving Fraud
Riset : Jurnal Aplikasi Ekonomi, Akuntansi dan Bisnis Vol. 4 No. 2, September 2022, Hal 103 119
119
Detection Tasks. 2020 IEEE Third International Conference on Data Stream
Mining & Processing (DSMP), 15.
https://doi.org/10.1109/DSMP47368.2020.9204244
West, J., & Bhattacharya, M. (2016). Intelligent Financial Fraud Detection: A
Comprehensive Review. Computers & Security, 57, 4766.
https://doi.org/10.1016/j.cose.2015.09.005
Widati, L. W., & Septy, F. (2008). Factors Affecting the Time Span of Financial
Statements to the Public (Empirical Study on LQ 45 Companies Listed on the
Indonesia Stock Exchange). Fokus Ekonomi (FE), 7(3), 173187.
Zamzami, F., Faiz, I. A., & Mukhlis. (2013). Internal Audit: Concepts and Practices.
Gadjah Mada University Press.
https://ugmpress.ugm.ac.id/userfiles/product/daftar_isi/Audit_Internal.pdf
... Several studies have shown that data analytics and machine learning can improve fraud prediction. Perols, et al. [4] and Ikhsan, et al. [5] concluded that these technologies could dramatically improve fraud detection. Ikhsan, et al. [5] highlighted these technologies' role in assisting with internal audits. ...
... Perols, et al. [4] and Ikhsan, et al. [5] concluded that these technologies could dramatically improve fraud detection. Ikhsan, et al. [5] highlighted these technologies' role in assisting with internal audits. Singla and Jangir [6] emphasizes the significance of these tools in real-time financial data while Patil [7] offers a complete review of machine-learning approaches for fraud prediction in the insurance industry. ...
Article
Full-text available
This research aims to reduce fraud risks in Indonesian banks and non-bank financial institutions providing microloan services. The study employs data analytics and machine learning techniques using employee data from Bank X spanning 2017-2019 with samples of 28,004 workers (2017-2018) and 27,274 employees (2019). Confirmatory factor analysis and XGBoost predictive modelling are applied within the fraud triangle framework to identify critical fraud risk indicators related to employee pressure. An algorithmic approach categorizes personnel based on fraud risk ratings enabling the detection of potentially suspicious activities for proactive intervention. The analysis reveals that incorporating data analytics into governance, risk management and compliance (GRC) systems can accurately forecast fraud probability by focusing on factors associated with employee pressure and opportunities. This facilitates targeted fraud prevention solutions by integrating control mechanisms, risk processes and auditing standards. The predictive model provides valuable insights for policymakers to combat fraud by enhancing governance and risk management practices specific to microloans. This research concludes that the predictive model is a pragmatic decision-making tool for banks offering micro-loans. It mitigates dangers by detecting high-risk personnel and transactions. Integrating data analytics with robust GRC frameworks enables financial institutions to uphold integrity through proactive fraud monitoring and targeted preventive interventions tailored to identify risk profiles. The study offers an integrated technological organizational approach to protect microlending activities.
... This directly contributes to increased transparency of financial statements, as auditors can present their findings more clearly and accurately, reducing the likelihood of errors or omissions of important information. This increased transparency is critical in building trust with stakeholders, including investors and regulators, who increasingly expect accuracy and thoroughness in financial statements, which is in line with previous research (Al Lawati & Sanad, 2023;Appelbaum et al., 2017;Ikhsan et al., 2022;Mohammed Shani & Al-Tameemi, 2021;Tawfeeq Yousif Alabdullah, 2023). ...
... Furthermore, the results obtained also indicate that the application of AI can facilitate auditors in performing more accurate and predictive analyses of financial data. This supports the findings of (Al-Sayyed et al., 2021;Appelbaum et al., 2017;Ikhsan et al., 2022;Meitasari & Audrey, 2023) who found that AI technologies impact financial report evidence. Although their study indicated that neural network technology did not have a significant effect on financial report evidence, this finding suggests that variability in the effectiveness of certain AI tools may depend on the application context and user expertise. ...
Article
In today's digital age, artificial intelligence (AI) has revolutionized many sectors, including auditing and finance, with the potential to improve the efficiency and effectiveness of the audit process. However, there is limited understanding of the impact of AI implementation on financial statement transparency and external auditor reputation. This study aims to fill that knowledge gap by exploring how AI can strengthen financial statement transparency and enhance auditor reputation. The methodology used is descriptive-analytical with an empirical normative approach, collecting data through literature review and documentation. The results finds that the integration of AI in auditing significantly improves financial statement transparency, facilitates auditors in conducting more in-depth and accurate analyses, and potentially enhances the reputation of external auditors in the eyes of stakeholders. The findings confirm that AI not only improves the efficiency of the audit process but also plays a strategic role in building trust and integrity in financial reporting. The implications of this research are significant in demonstrating the importance of adapting the latest technologies to meet and exceed evolving financial and reputational expectations in the digital age.
... The use of artificial intelligence (AI) in forensic accounting and auditing marks a significant milestone in the field's recent developments. AI has become central to transforming how we understand and manage audit and fraud detection processes (Ikhsan et al., 2022). Its ability to analyze large datasets quickly and accurately has opened the door to enhanced efficiency in handling the vast volumes of data that auditors and accounting professionals encounter (Collins Kindzeka, 2023). ...
Article
Full-text available
This study provides a bibliometric analysis of ethical considerations in AI applications for fraud detection based on data from ScienceDirect spanning 2020 to 2024. The analysis identifies “artificial intelligence” as a core focus in the literature, alongside a marked increase in attention to ethical concerns, including data privacy, transparency, and accountability. Additionally, the study reveals progress in applying advanced technologies like blockchain, ChatGPT, and fintech within fraud detection frameworks, which increasingly demand ethical scrutiny. Key findings emphasize the necessity for comprehensive ethical frameworks to ensure transparency, accountability, and public trust in AI-driven fraud detection systems. Practical implications suggest that organizations should prioritize ethical dimensions within AI strategies, enhancing both trust and the effectiveness of detection mechanisms. By using bibliometric analysis, this study finds new trends and gaps in the ethical aspects of using AI to find fraud, which adds new information that hasn’t been fully explored in other studies.
... This automation frees up auditors to focus on more strategic, judgment-based aspects of the audit, such as interpreting findings and advising clients on risk management and internal controls. According to the findings of Xing et al. (2020), Ikhsan et al. (2022), Munoko et al. (2020), AI-driven tools enhance the objectivity of audits by reducing human biases and ensuring consistent application of audit procedures across different clients and industries. The integration of these technologies into auditing practices not only improves the efficiency and reliability of audits but also allows auditors to provide more timely and insightful recommendations to their clients. ...
Article
Full-text available
This study explores the impact of technological innovations on audit transparency, objectivity, and assurance. The study employs a systematic literature review methodology, analyzing a wide range of scholarly articles, research papers, and reports to synthesize the findings. The methodology involved identifying keywords, conducting comprehensive searches in academic databases, and evaluating the selected literature. The study identifies key themes on how technological innovations impact audit practices through analysis of the literature. The impacts of technology include enhanced audit transparency through improved documentation capabilities, real-time reporting, and increased stakeholder engagement. Technological advancements bolster audit objectivity by automating repetitive tasks, facilitating advanced data analysis, and promoting standardized audit procedures. However, the analysis highlighted challenges associated with the use of technology in audits including complex technology implementation and the potential for biases. This research study contributes to the existing body of knowledge by consolidating relevant research and insights on the subject matter.
... In the dynamic scene of life that continues to grow, the development of technology and knowledge is a matter of course (Ikhsan et al., 2022). Almost every realm of human activity is now inseparable from the role of technology (Issa et al., 2016). ...
Article
Full-text available
The development of information technology has significantly impacted various fields, including audit practice. A notable advancement is the ATLAS application, an electronic working paper designed to assist auditors in filling out work papers. This research aimed to examine the application of ATLAS and assess its effectiveness in the financial statement audit process at KAP Hari Purnomo & Jaswadi. Using a descriptive qualitative approach, data were collected through interviews with auditors. The findings revealed that KAP Hari Purnomo & Jaswadi had comprehensively implemented the ATLAS application, covering all stages from pre-engagement, risk assessment, and risk response to reporting. The ATLAS application proved effective, as it streamlined and facilitated auditors' work in inputting audit data. The integration of all audit stages within the ATLAS application demonstrated its efficiency and effectiveness in the financial statement audit process, making it a valuable tool for auditors at KAP Hari Purnomo & Jaswadi.
... The satisfaction of farmers is strongly linked to perceived security, which in turn is influenced by various factors such as the confidentiality of personal information, payment methods, and privacy protection. These factors are crucial for maintaining the integrity of agricultural data and ensuring the privacy of farmers' personal information [43]. The recent studies of RWELA show a connection between service quality and customer satisfaction in agricultural marketing cooperative societies (AMCOS) [44]. ...
Article
Full-text available
In the Philippines, where agriculture plays a vital role in the economy, nearly half of the country’s 30 million hectares of land is used for agricultural purposes. Despite government efforts to support rice farmers, challenges, such as production failures and profit losses, remain prevalent. This study evaluates the quality of government services provided to rice farmers through a service quality-based framework (SERVQUAL), assessing five key dimensions: assurance (AS), empathy (EM), reliability (RL), responsiveness (RS), and tangibility (TA). Structural equation modeling (SEM) is employed to examine the relationships among these service quality dimensions and other latent variables, including service quality (SQ), farmers’ satisfaction (FS), farmers’ trust (FT), perceived security (PS), and perceived effectiveness (PE). Data were gathered from 400 respondents using a 55-item survey. The findings indicate that perceived security (PS) is the most critical factor in enhancing the overall quality of government services. The study emphasizes the importance of adopting a SERVQUAL-based approach, with a focus on reliability and responsiveness to improve farmers’ satisfaction and trust. Additionally, fostering trust between farmers and the government is crucial to increasing the effectiveness of government programs and strengthening farmer–government collaboration. The study further advocates for prioritizing sustainable agricultural practices to ensure long-term productivity and environmental resilience. By improving service quality with a focus on sustainability, the Philippines can work towards achieving broader sustainable development goals.
Chapter
The role of artificial intelligence (AI) in ad fraud detection within the blockchain and programmatic advertising (PA) ecosystem is becoming increasingly crucial. As digital advertising continues to evolve, the complexities and scale of real-time transactions pose significant challenges for advertisers and publishers. Ad fraud, including click fraud, results in substantial financial losses, projected to reach $172 billion by 2028. AI technologies, such as machine learning and deep learning, have proven effective in identifying fraudulent activities and enhancing the accuracy and reliability of ad campaigns. Additionally, blockchain technology offers transparency and security by recording and verifying each ad interaction, ensuring data integrity and trust. This book chapter explores the significance of AI and blockchain in addressing ad fraud, highlighting their potential to transform the digital advertising landscape and improve the efficiency and effectiveness of advertising investments.
Article
Full-text available
Perkembangan Kecerdasan Buatan (AI) telah menawarkan peluang transformatif dalam berbagai sektor, termasuk sistem peradilan. Dalam konteks Indonesia, implementasi AI sebagai hakim menjanjikan potensi untuk mengatasi tantangan integritas dan efisiensi yang dihadapi oleh sistem peradilan, yang telah tercoreng oleh kasus korupsi dan kerentanan terhadap bias subjektif. Dengan memanfaatkan kemampuan AI dalam mengolah data besar secara cepat dan objektif, diharapkan dapat menciptakan proses pengambilan keputusan yang lebih transparan, mengurangi bias, dan meminimalkan potensi korupsi. Penelitian ini membahas implementasi AI dalam peran sebagai hakim dalam sistem peradilan pidana Indonesia, mengkaji potensi efisiensi, objektivitas, dan transparansi yang ditawarkan oleh AI, serta mengeksplorasi sinergi antara AI dan hakim manusia dalam meningkatkan kualitas layanan hukum. Melalui analisis teoriti, penelitian ini menggarisbawahi pentingnya pengembangan standar dan protokol, transparansi, pelatihan dan pendidikan, serta evaluasi berkala dalam integrasi AI. Kerjasama antara AI dan hakim manusia tidak hanya memperkaya proses pengambilan keputusan dalam peradilan tetapi juga mempertahankan inti humanistik hukum. Penelitian ini menunjukkan bahwa dengan pendekatan yang hati-hati dan etis, integrasi AI dalam sistem peradilan dapat memperkuat keadilan, meningkatkan efisiensi, dan memastikan bahwa teknologi mendukung, bukan menggantikan, kebijaksanaan hakim manusia, membuka era baru dalam peradilan yang lebih adil, efisien, dan bermartabat.
Article
During audit planning, auditors examine the business of different firms. Still, the target is to minimize the discrepancy in the real planned financial statement of inspection and summary reports of internal audits. On the other hand, expenditures on artificial intelligence have been increasing in Turkish firms; according to the National Artificial Strategy document, AI will be part of every organizational process, including internal audits. Moreover, the literature supports a positive relationship between internal audits and firms’ decreasing capital loss. So, this research aims to analyze the relationship between AI expenditures, internal audit reports, and the firms’ historical loss. To reach this aim, suitable data was analyzed from 732 incorporated companies that are members of the Chamber of Trade and Industry/Tekirdag/Turkey. Structural equation modeling results show that AI investments decrease the discrepancy between financial statements and internal audit reports (β=-0.045). On the other hand, discrepancies found in the internal audit reports compared to real financial statements are increasing firms’ financial losses by almost 10% (β=.118). In other words, investing in AI contributes to more realistic financial reports, resulting in fewer financial losses. From this perspective, this study is one of the leading studies that connects AI investment to internal audits and the financial performance of Turkish firms.
Article
Full-text available
Prior research in the fields of accounting and information systems has shed some light on the significant effects of financial reporting fraud on multiple levels of the economy. In this paper, we compile prior multi-disciplinary literature on financial statement fraud detection. Financial reporting fraud detection efforts and research may be more impactful when the findings of these different domains are combined. We anticipate that this research will be valuable for academics, analysts, regulators, practitioners, and investors.
Article
Full-text available
As the use of the internet is growing exponentially, more and more businesses such as the financial sector are operationaliz-ing their services online. Consequently, financial frauds are increasing in number and forms around the world, which results in tremendous financial losses which make financial fraud a major problem. Unauthorized access and irregular attacks are examples of threats that should be detected by means of financial fraud detection systems. Machine learning and data mining techniques have been widely used to tackle this issue over the past few years. However, these methods still need to be improved in terms of speed computation, dealing with big data, and identify the unknown attack patterns. Therefore, in this paper, a deep learning-based method is proposed for the detection of financial fraud based on the Long Short-Term Memory (LSTM) technique. This model is aimed at enhancing the current detection techniques as well as enhancing the detection accuracy in the light of big data. To evaluate the proposed model, a real dataset of credit card frauds is utilized and the results are compared with an existing deep learning model named Auto-encoder model and some other machine learning techniques. The experimental results illustrated a prefect performance of LSTM where it achieved 99.95% of accuracy in less than a minute.
Article
Full-text available
: Industry revolution 4.0 makes people face change, the auditor profession is no exception. Auditors no longer conduct audits using the manual method but use computerized systems such as big data analytics. Our research aims to find out how auditors must change, when facing new technology approach audit. Very complicated and various data kinds that was too large to be audited manually. The research method in this research is descriptive qualitative, the method of data collection uses interviews with informants. The informant is the auditor partner of the public accounting firm. Big data has advantage and disadvantage to audit and fraud detection profession. The results of this study state that audit firm must be aware of the obstacles that come from internal and external for the implementation of big data analytics audit. Obstacles will hinder the adoption of technology for the auditor, while the auditor must change rapidly following the demands of technological development
Article
Full-text available
The private insurance sector is recognized as one of the fastest-growing industries. This rapid growth has fueled incredible transformations over the past decade. Nowadays, there exist insurance products for most high-value assets such as vehicles, jewellery, health/life, and homes. Insurance companies are at the forefront in adopting cutting-edge operations, processes, and mathematical models to maximize profit whilst servicing their customers claims. Traditional methods that are exclusively based on human-in-theloop models are very time-consuming and inaccurate. In this paper, we develop a secure and automated insurance system framework that reduces human interaction, secures the insurance activities, alerts and informs about risky customers, detects fraudulent claims, and reduces monetary loss for the insurance sector. After presenting the blockchain-based framework to enable secure transactions and data sharing among different interacting agents within the insurance network, we propose to employ the extreme gradient boosting (XGBoost) machine learning algorithm for the aforementioned insurance services and compare its performances with those of other state-of-the-art algorithms. The obtained results reveal that, when applied to an auto insurance dataset, the XGboost achieves high performance gains compared to other existing learning algorithms. For instance, it reaches 7% higher accuracy compared to decision tree models when detecting fraudulent claims. The obtained results reveal that, when applied to an auto insurance dataset, the XGboost achieves high performance gains compared to other existing learning algorithms. For instance, it reaches 7% higher accuracy compared to decision tree models when detecting fraudulent claims. Furthermore, we propose an online learning solution to automatically deal with real-time updates of the insurance network and we show that it outperforms another online state-of-the-art algorithm. Finally, we combine the developed machine learning modules with the hyperledger fabric composer to implement and emulate the artificial intelligence and blockchain-based framework.
Article
Full-text available
This study investigates an auditor’s fraud detection gap (FDG) in Gulf Cooperation Council (GCC) companies by comparing the result of the fraud detection models (namely the Beneish M-score, Dechow F-score, and Altman Z-score) with an actual of audit opinion given by the auditors. Prior scholars documented that, financial models are accurate and important measurements in fraud detection. However, the majority of accidentally fraud cases in the region are revealed which mean the unclear role of the internal and external auditor. The data consists of 365 companies operated in the GCC for the period from 2015 to 2017 with a total of 1,095 observations. The study found that the success rate of detecting financial statement frauds for Dechow F model is much higher than Beneish M or Altman Z models. The result also indicated that the highest FDG-accuracy results were obtained by the Dechow F model. However, the Beneish M model can detect financial statements’ fraud better for companies associated to local audit firms as compared to international audit firms. Additionally, Big 4 audit firms are associated with a lower FDG in Beneish M model but increase FDG in Altman Z model. Hence, the study supported the inclusion of statistical models, to a certain extent, as an alternative or supplementary method that assisted in making better decision-making for companies within the Gulf States. The regulators, policy maker, and practitioners, mainly the audit firms must concern that the ability to detect financial statement’s fraud can be enhanced by utilizing the appropriate fraud detection model.
Article
This study aims to implement a machine learning algorithm in detecting fraud based on historical data set in a retail consumer financing company. The outcome of machine learning is used as samples for the fraud detection team. Data analysis is performed through data processing, feature selection, hold-on methods, and accuracy testing. There are five machine learning methods applied in this study: Logistic Regression, K-Nearest Neighbor (KNN), Decision Tree, Random Forest, and Support Vector Machine (SVM). Historical data are divided into two groups: training data and test data. The results show that the Random Forest algorithm has the highest accuracy with a training score of 0.994999 and a test score of 0.745437. This means that the Random Forest algorithm is the most accurate method for detecting fraud. Further research is suggested to add more predictor variables to increase the accuracy value and apply this method to different financial institutions and different industries.
Article
Financial statement fraud is an area of significant consternation for potential investors, auditing companies, and state regulators. The paper proposes an approach for detecting statement fraud through the combination of information from financial ratios and managerial comments within corporate annual reports. We employ a hierarchical attention network (HAN) to extract text features from the Management Discussion and Analysis (MD &A) section of annual reports. The model is designed to offer two distinct features. First, it reflects the structured hierarchy of documents, which previous approaches were unable to capture. Second, the model embodies two different attention mechanisms at the word and sentence level, which allows content to be differentiated in terms of its importance in the process of constructing the document representation. As a result of its architecture, the model captures both content and context of managerial comments, which serve as supplementary predictors to financial ratios in the detection of fraudulent reporting. Additionally, the model provides interpretable indicators denoted as "red-flag" sentences, which assist stakeholders in their process of determining whether further investigation of a specific annual report is required. Empirical results demonstrate that textual features of MD&A sections extracted by HAN yield promising classification results and substantially reinforce financial ratios.