Frontiers in Business, Economics and Management 
ISSN: 2766-824X | Vol. 11, No. 1, 2023 

 

239 

Audit Data and Management Platform Based on 
Intelligent Algorithm 

Yiqi Yang 

Hunan Agricultural University, Changsha, Hunan, China 

Hauyangyiqi@163.com 

 

Abstract: Due to the uncertainty of the external economic environment and the changes caused by new technologies in the 
new era, there are still many deficiencies in the current financial statement audit. The use of new technology can optimize t he 

audit process, reduce audit practice risks and improve audit efficiency. Therefore, it is necessary to study the intelligent audit of 

financial statements. The purpose of this paper is to study the audit data analysis and management platform based on intelligent 

algorithms. This paper focuses on the construction purpose and content of the audit data analysis and management platform, and 

analyzes the specific requirements of some functional modules in the platform according to the audit workflow and business 

requirements. At the same time, this paper also describes the overall design architecture of the platform, which mainly uses the 

isolated point mining method for enterprise intelligent audit. This paper takes it as an effective method to analyze the problems 

existing in the internal audit of enterprises, and constructs the audit model of benefit-cost matching, and verifies the effectiveness 

of the algorithm through an example. There are four isolated points in the gross profit rate of catering of an enterprise.  

Keywords: Intelligent Algorithm; Audit Data; Analysis Management; Management Platform. 

 

1. Introduction 

In the audit of financial statements, there are a lot of simple, 

trivial and repetitive work, which are all done manually [1]. 

With the development of modern technology, the scale of the 

company is continuously expanding, and the business scope 

of the company is also constantly expanding, which has 

virtually increased the workload and audit time of financial 

statement auditors, leading to the decline of audit efficiency. 

In this simple, trivial and repetitive work, proper use of 

intelligent technology to replace manual work can effectively 

improve the quality of audit work; Secondly, when the audit 

is conducted at the end of the year, the company does not need 

to hire many interns to do some simple and repetitive work, 

which can reduce the company's operating costs; Using 

intelligent technology can enable auditors to get rid of tedious 

work and engage in more high-tech auditor work. Intelligent 

audit is not only a combination of technology and methods, 

but also an innovation of audit ideas [2-3]. This not only 

provides new development opportunities for auditors, but also 

further promotes the theory and practice of auditors. In the 

audit, business intelligence is applied to the audit, using the 

extraction of business data, real-time response to data changes 

in the data warehouse, and various data analysis methods to 

analyze the business data of enterprises, so as to achieve the 

purpose of monitoring economic activities [4]. In this way, the 

post-audit can be transformed into prior and in-process 

supervision, the centralized audit can be transformed into 

real-time audit, and the risk warning can be implemented in 

the audit process, so as to improve the efficiency of audit [5]. 

With the continuous growth of the content and business 

data of the communication enterprise audit, the traditional 

audit method is mainly offline audit, the audit work is heavy 

and temporary, the number of auditors is scarce, and the 

support of data market and analysis model is lacking. Andre é 

M. Carrington put forward an efficient management method 

based on data domain of multiple systems in view of various 

problems faced by traditional audit methods. On this basis, he 

defined the data of multiple systems uniformly and defined it 

as a unified database. At the same time, the platform also has 

certain scalability, which can gradually achieve the coverage 

of data, and can provide support for more management areas 

and business systems. On this basis, he uses the K-means 

method to find the center point of a group of continuous 

attribute clusters, and fuzzies it with triangular fuzzy numbers. 

On this basis, combined with the minimum fuzzy degree 

theory, a visual fuzzy decision system is established [6]. The 

existing audit big data analysis methods mainly include the 

lack of mining the audit information hidden in the big data, 

the detection of more fraud technologies in the context of 

informatization and networking, and the quality defects of the 

audit big data. The association analysis algorithm in data 

mining technology is an effective method to extract 

knowledge from a large amount of data. It can complete, hide, 

clean up and reduce defect data, and analyze these data with 

the help of experts or analysts, so as to obtain the 

corresponding audit results. Based on this, Fouad Ben 

Abdelaziz first reviewed the world's relevant literature, and 

then reviewed the world's relevant literature; This paper 

introduces the background, current situation and challenges 

of relevant research in the world; It mainly expounds some 

basic principles and methods of data mode, mode 

transformation, mode establishment, etc; Be responsible for 

the collection and arrangement of audit data; Pre-process the 

audit data and explain the algorithm; A new correlation 

analysis method is proposed. The activation value of hidden 

nodes and the extraction of audit rules are analyzed; On this 

basis, an audit data processing method based on association 

analysis is presented; Finally, the quality of audit data and 

audit risk are discussed. This paper summarizes the current 

research situation in the field of audit data processing in the 

world. The results show that, on the basis of the analysis of 

association rules, he proposed a correlation analysis method 

based on association rules, which can identify a large amount 

of audit data, and can use probabilistic methods to describe 

the occurrence of association events at the same time or 



 

240 

successively [7]. 

This paper discusses the current situation of the audit data 

analysis and management platform and the main problems 

faced by auditors, and introduces the business intelligence 

technology and data mining programming technology, aiming 

at clarifying the actual needs of building the audit data 

analysis and management platform and the feasibility of 

analyzing and developing software systems. Then, this paper 

proposes a financial audit mining model based on outlier 

mining algorithm, and gives the concrete implementation 

method of the model. By analyzing the theory of business 

intelligence and studying the application of relevant 

technologies, applying business intelligence technology to 

complex financial data audit and establishing an audit data 

analysis and management platform can make auditors more 

convenient to audit and reduce the workload. 

2. Audit Data and Management 
Platform Based on Intelligent 
Algorithm 

2.1. Online Analytical Processing Technology 

On-line Analytical Processing (OLAP) is a very critical 

analysis method in data warehouse. Online analysis and 

processing is a complex analysis method based on large-scale 

data. It can analyze and process data in data warehouse in 

multi-dimensional and efficient ways from multiple 

perspectives [8-9]. 

The main feature of OLAP is that it can respond quickly to 

various complex queries, and has the ability of 

multidimensional modeling. Its main technologies include: 

multidimensional data model, multidimensional analysis, 

multidimensional query and display, data cube, etc. [10-11]. 

The main characteristics of online analytical processing are: 

fast speed, and the ability to respond to the analysis requests 

of most users in very little time; Users do not need a program 

to determine specific new operations in the way that users 

expect, and perform different logic and statistical analysis 

related to the application in the way that users expect; Multi-

dimensional, providing users with multi-dimensional 

observation and analysis; Timeliness, no matter how large the 

number of data is or where it is stored, you can get timely 

information and manage these large amounts of information 

[12-13]. The use of OLAP technology has greatly expanded 

the ability of data warehouse to process data, but in this era of 

information explosion, the limitations of OLAP technology 

have also become increasingly prominent. In order to better 

adapt to data processing and analysis, people have begun to 

conduct in-depth research on data mining technology [14-15]. 

2.2. Outlier Algorithm 

(1) Outlier definition 

Definition: In the S set, only when the distance between the 

object with p in S is larger than d, will an isolated point appear 

[16]. 

(2) Measurement of distance 

In distance-based isolation point recognition methods, 

absolute value, Euclidean value and Chebyshev value are 

usually used. The so-called Manhattan distance is a geometric 

academic term used to express the sum of absolute lengths 

between two coordinates [17]. Its definition is: 


=

−=
m

k

xjkxikdij
1

             (1) 

In the formula, m is the number of attributes of the data 

object, and xij is the value of the jth attribute of the i-th object 

[18]. 

Euclidean distance, also known as Euclidean distance, 

refers to the real distance between two points on two 

dimensions. The formula is: 


=

−=
m

k

xjkxikdij
1

2)(             (2) 

(3) Distance-based outlier detection algorithm 

The distance-based outlier detection algorithm is mainly 

used to calculate the distance between each object and other 

objects based on the number of outliers that users want to find 

before this, and then calculate the distance between each 

object and all other objects, so that the M objects with the 

largest distance sum are outliers [19]. 

Definition: xi and xj represent the objects in a data set X, n 

represents the number of objects, dij represents the spacing of 

xi and xj, then the spacing matrix R of X is as follows: 



















=

dnndndn

nddd

nddd

R

...21

...

2...2221

1...1211

          (3) 

Definition: xi is the ith data object in the dataset, and the 

deviation degree Devi of xi is defined as: 


=

=
n

j

dijDevi
1

                 (4) 

That is, Devi is the sum of row i in matrix R. 

From the definition, we can see that the larger Devi is, the 

more likely object i is to become an isolated point. If M is the 

number of outliers the user wants to obtain, then the largest 

outlier is outliers [20]. 

2.3. Overview of Platform Requirements 

The functional requirements of the audit data analysis and 

management platform are as follows: 

(1) The platform uses the data collection server to collect 

data related to the budget implementation of the financial 

department on a regular basis. It can also collect data in other 

electronic formats, and automatically schedule data 

processing templates according to the audit data plan. The 

collected financial data is processed and converted into data 

in standard format; 

(2) The model can be used to monitor all businesses and 

financial data in the financial information system in real time, 

and also to set up auditors appropriately, so as to achieve the 

early warning function of financial data; 

(3) Based on the project implementation plan of the 

financial budget implementation audit, the auditors select a 

corresponding audit mode for the corresponding audit items. 

This platform will automatically extract the basic data of the 

audit background, so as to realize multidimensional and 

multi-level audit query analysis. 

(4) This platform will feed back the audit questions 

analyzed by each module to the auditors, who will further 

implement the audit questions after the analysis. 



 

241 

3. Investigation and Audit Data and 
Management Platform Based on 
Intelligent Algorithms 

3.1. Development Language and Tools 

(1) Development language 

The languages needed for the development of the system 

include C #, Microsoft, SQL, MDX, DMX, and ADOMD 

NET API, etc. In DMX, you can complete the structure of data 

mining, establish data mining models, test models and 

forecast data, and combine DMX language with practical 

applications. NET API interface is used to connect with 

analysis services server, and MDXprediction statement is 

used to predict data based on data mining. MDX provides the 

ability to process multi-dimensional data. It can embed MDX 

declarations into applications, enabling them to interact with 

multi-dimensional databases, and users can perform 

corresponding conditional retrieval according to their own 

needs and needs. The results will be displayed in a chart or 

table. 

(2) Development tools 

SQL SERVER 2008 is not only a relational database and 

data warehouse server, but also an Analysis Services server. 

Through the connection with SQLSERVER2008, users can 

use the software to mine, predict and analyze data in multi-

dimension. 

3.2. Audit Data and Management Platform 
Architecture 

The audit data analysis and management platform include 

data warehouse, multidimensional analysis (OLAP) and data 

mining technologies. With these technologies, data can be 

collected and integrated, and data can be analyzed, so as to 

make a reasonable assessment and prediction, and provide a 

basis for auditors to make decisions. OLAP is a kind of 

confirmatory analysis. It can analyze multidimensional data 

sets based on data warehouse at different levels and 

dimensions, so that auditors can have a comprehensive 

understanding of the overall situation, observe the 

development trend, and select the key points. You can use 

drilling to find some clues before you can conduct an audit. 

Data mining is a kind of mining analysis. It uses various 

mining algorithms to let auditors extract the information and 

knowledge hidden in the massive financial data. Data mining 

improves the application of data from low-level and simple 

queries to mining more knowledge from data to help people 

make decisions. The audit data analysis platform in this paper 

mainly includes four modules: audit data collection, audit 

analysis, suspicious problem handling, and audit 

collaboration. The overall functional framework is shown in 

Figure 1: 

 

Figure 1. Platform functional architecture 

3.3. Financial Audit Mining Model Based on 
Outlier Mining Algorithm 

During the audit process, outlier mining technology can be 

used to obtain abnormal data, which is very important for the 

audit work and may be an important breakthrough. During the 

audit process, the specific model of outlier mining technology 

for anomaly detection is shown in Figure 2. 

(1) Determine audit requirements: analyze the 

characteristics of the audited unit according to the assigned 

audit tasks and requirements, clarify the audit direction and 

audit focus, understand the content to be audited, and 

transform it into an expression that can use outliers for data 

mining. 

(2) Data understanding and preparation: understand and 

analyze the data structure of the audited entity, select the 

appropriate data source according to the needs of the audit 

business, identify the data that can be analyzed using outliers, 

transform them, and load them into the data warehouse for 

mining. 

(3) Modeling and outlier mining: on this basis, the 

proposed outlier mining method is modeled. 

(4) Evaluation: After building the model, we should 

analyze and evaluate the operation results of the model, and 

test and evaluate the knowledge and rationality of the 

operation results. If the results cannot accurately reflect the 

matters to be audited, then we need to revise the model several 

times until the model can meet the needs of the auditors. 

(5) Release: The results of mining should be presented to 

Audit data analysis platform 

Data acquisition module 

Audit analysis module 

Doubt management module 

Business collaboration module 



 

242 

auditors in an intuitive, concise and easy-to-understand way 

through visual technology, so that auditors can quickly 

determine audit clues and make audit judgments. 

 

Figure 2. Financial audit mining model based on outlier mining algorithm 

4. Audit Data and Management 
Platform Based on Intelligent 
Algorithm 

4.1. Implementation of Audit Function 

(1) OLAP-based audit process 

Create a database for analysis: create a database in the 

Analysis Server for analysis, and connect it with the source 

database by ODBC (Open Database Connectivity) or OLEDB 

(Object Linking and Embedding, Database), and use it as the 

data source of the Analysis database. 

Create Dimension: in Shared Dimension, create a 

dimension for analyzing data through the Dimension Wizard 

or Dimension Editor. 

Create a cube: Create a cube through the Cube Wizard or 

Cube Editor. After building a cube, you can use the analysis 

manager to browse the data directly, use Excel or Web as the 

front-end analysis tool to analyze the data, and develop client 

applications to analyze and process the data. 

(2) OLAP analysis 

OLAP has effectively processed complex analysis 

operations, and its focus is on assisting decision-making, 

while providing users with an intuitive and easy-to-

understand retrieval result. In programming, its 

implementation method is as follows: 

using Microsoft.AnalysisServices.AdomdClient;  

using System.Data.SqlClient;  

connection=new AdomdConnection();  

connection.ConnectionString=”Data Source=localhost; 

Initial  

Catalog=Data Warehouse "; 

connection.  Open ();  

AdomdCommand cmd=connection.CreateCommand();  

Cmd. CommandText=MDX query statement; 

//execute the command and display the prediction result  

AdomdDataReader reader=cmd.ExecuteReader();  

if (reader. Read ())  

string predictedGeneration=reader.  GetValue 

(0).ToString(); 

(3) Report management 

The report is a kind of XML (Extensible Markup Language) 

format data. It needs to go through the steps of variable setting, 

cell editing, SQL statement setting, file parsing and so on to 

realize the display of the query results and tables. 

4.2. Data Mining Model and Data  

In this paper, the revenue and cost of a company's 

restaurant are extracted from a database. After conversion, 

cleaning, verification and loading into the data warehouse, the 

profit of the restaurant can be obtained. In order to facilitate 

calculation, we use a positive exponential linear 

normalization method to normalize these data. The formula is: 

Audit needs 

Isolated point analysis problem 

Outlier mining model 

Data warehouse 

Evaluate mining results 

Publish results 

Whether satisfied 
N 

Y 

Data source 



 

243 

)min()max(

)min(
,

,

jj

jji

ji
xx

xx
y

−

−
=             (5) 

The two-dimensional scatter diagram of the obtained data 

is shown in Figure 3: 

 

Figure 3. Scatter Chart of Standardized Catering Revenue and Catering Gross Profit Rate 

The x-axis is the food and beverage income after 

standardization, and the y-axis is the food and beverage gross 

profit rate after standardization. Using the above distance-

based outlier algorithm, a total of four outliers have been 

mined, as shown in Table 1 and Figure 4. 

 

 

 

Table 1. Results of outlier mining 

Date Income after 

standardization 

Catering gross 

profit rate after 

standardization 

November 2022 0.9 0.81 

December 2022 0.05 0.21 

January 2023 0.2 0.2 

February 2023 0.33 0.25 

 

Figure 4. Mining results of distance-based outlier algorithm 

Through the analysis of these three isolated points, it can be seen that three of these three isolated points are because 

0 0.2 0.4 0.6 0.8 1

Nov-22

Dec-22

Jan-23

Feb-23

value

d
a
te

Income after standardization Catering gross profit rate after standardization

Catering income after standardization 

Catering gross 

profit rate after 

standardization 

0 0.2 

0.2 

0.4 

0.4 

0.6 

0.8 

1 

0.6 0.8 1 



 

244 

the operation is in the off-season and the marketing strategy 

of reducing prices to attract customers has been adopted, 

resulting in a relatively low gross profit rate of catering. 

Another isolated point is that part of the revenue of the 

previous month was deferred to November 2022, resulting in 

a relatively high income in November 2022 and a relatively 

high gross profit rate. To this end, we submitted an audit 

report to the company. 

5. Conclusions 

Nowadays, AI, big data and other new technologies have 

been widely used, and the application of new technologies in 

audit analysis is also imminent. The audit data analysis and 

management platform studied in this paper has promoted the 

informatization of the audit work to a certain extent, thus 

further improving the quality and efficiency of the audit work, 

and the audit method has also become more convenient and 

flexible. However, we also found that there are still some 

defects in the platform itself and its use. Therefore, we have 

some prospects for the future development and use of the 

platform: on this basis, we will further strengthen the security 

of the platform's network and data. In order to enhance the 

prevention of network and data security risks, the data of the 

platform can be synchronized and hot standby by two 

computers, and the audit data analysis and management 

platform can be run using the government intranet or private 

network, which has stronger confidentiality and higher 

security performance. At the same time, a complete 

emergency plan should be prepared to respond quickly and 

efficiently in case of safety problems. If possible, you can also 

find a professional data confidentiality engineer to achieve 

professional protection. Therefore, in the follow-up work, we 

should do a good job in the network and data security related 

to the platform. 

References 

[1] Joseph Johnston, Joseph H. Zhang: Auditor Style and Financial 
Reporting Similarity. J. Inf. Syst. 35(1): 79-99 (2021) 

[2] Diane J. Janvrin, Maureen Francis Mascha, Melvin A. 
Lamboy-Ruiz: SOX 404(b) Audits: Evidence from Auditing 
the Financial Close Process of the Accounting System. J. Inf. 
Syst. 34(3): 77-103 (2020) 

[3] Jakob Mökander, Maria Axente: Ethics-based auditing of 
automated decision-making systems: intervention points and 
policy implications. AI Soc. 38(1): 153-171 (2023) 

[4] Inioluwa Deborah Raji, Joy Buolamwini: Actionable Auditing 
Revisited: Investigating the Impact of Publicly Naming Biased 
Performance Results of Commercial AI Products. Commun. 
ACM 66(1): 101-108 (2023) 

[5] R. Hariharan, G. Komarasamy, S. Daniel Madan Raja: 
Implementation of ID-based Audit Protocols to Enhance 
Security and Productivity. Comput. Syst. Sci. Eng. 46(1): 873-
882 (2023) 

[6] Andre M. Carrington, Douglas G. Manuel, Paul W. Fieguth, 
Tim Ramsay, Venet Osmani, Bernhard Wernly, Carol Bennett, 
Steven Hawken, Olivia Magwood, Yusuf Sheikh, Matthew 
McInnes, Andreas Holzinger: Deep ROC Analysis and AUC as 

Balanced Average Accuracy, for Improved Classifier Selection, 
Audit and Explanation. IEEE Trans. Pattern Anal. Mach. Intell. 
45(1): 329-341 (2023) 

[7] Fouad Ben Abdelaziz, Souhir Neifar, Khamoussi Halioui: 
Multilevel optimal managerial incentives and audit fees to limit 
earnings management practices. Ann. Oper. Res. 311(2): 587-
610 (2022) 

[8] Oscar Danilo Gavilánez Alvarez, Glen Dario Rodriguez Rafael: 
VoIP security auditing model based on COBIT 4.1. Int. J. Secur. 
Networks 17(2): 63-76 (2022) 

[9] Sergio Davalos, Ehsan H. Feroz: A textual analysis of the US 
Securities and Exchange Commission's accounting and 
auditing enforcement releases relating to the Sarbanes-Oxley 
Act. Intell. Syst. Account. Finance Manag. 29(1): 19-40 (2022) 

[10] Dimitrios Kydros, Michail Pazarskis, Athanasia Karakitsiou: A 
framework for identifying the falsified financial statements 
using network textual analysis: a general model and the Greek 
example. Ann. Oper. Res. 316(1): 513-527 (2022) 

[11] Hyun Woong (Daniel) Chang, Steven Kaszak, Peter C. Kipp, 
Jesse C. Robertson: The Effect of iXBRL Formatted Financial 
Statements on the Effectiveness of Managers' Decisions When 
Making Inter-Firm Comparisons. J. Inf. Syst. 35(2): 149-177 
(2021) 

[12] David Broneske, Anna Drewes, Bala Gurumurthy, Imad Hajjar, 
Thilo Pionteck, Gunter Saake: In-Depth Analysis of OLAP 
Query Performance on Heterogeneous Hardware. Datenbank-
Spektrum 21(2): 133-143 (2021) 

[13] Bhagyaraj Thangavel, Sabarathinam Srinivasan, Kathamuthu 
Thamilmaran, Guisheng Zhai, Nallappan Gunasekaran: 
Dynamical Analysis of T-S Fuzzy Financial Systems: A 
Sampled-Data Control Approach. Int. J. Fuzzy Syst. 24(4): 
1944-1957 (2022) 

[14] Margareta Gardijan, Zrinka Lukac: The financial efficiency of 
small food and drink producers across selected European 
Union countries using data envelopment analysis. Eur. J. Oper. 
Res. 291(2): 586-600 (2021) 

[15] Emna Mnif, Anis Jarboui, M. Kabir Hassan, Khaireddine 
Mouakhar: Big data tools for Islamic financial analysis. Intell. 
Syst. Account. Finance Manag. 27(1): 10-21 (2020) 

[16] Amrita Chatterjee: Financial inclusion, information and 
communication technology diffusion, and economic growth: a 
panel data analysis. Inf. Technol. Dev. 26(3): 607-635 (2020) 

[17] Manas Tripathi, Arunabha Mukhopadhyay: Financial Loss due 
to a Data Privacy Breach: An Empirical Analysis. J. Organ. 
Comput. Electron. Commer. 30(4): 381-400 (2020) 

[18] Dat Thanh Tran, Alexandros Iosifidis, Juho Kanniainen, 
Moncef Gabbouj: Temporal Attention-Augmented Bilinear 
Network for Financial Time-Series Data Analysis. IEEE Trans. 
Neural Networks Learn. Syst. 30(5): 1407-1418 (2019) 

[19] Rui Pedro Brito, Pedro Júdice: Asset classification under the 
IFRS 9 framework for the construction of a banking investment 
portfolio. Int. Trans. Oper. Res. 29(4): 2613-2648 (2022) 

[20] Ahmed Jasim Hameed, Anfal S. Shareef, Sameer Imad Shaban: 
Effect of Fair Value Based on IFRS 13 on the Qualitative 
Characteristics of Accounting Information: An Exploratory 
Study in the Iraqi Environment. J. Cases Inf. Technol. 24(2): 1-
12 (2021).