Intelligent analysis processing method and system for enterprise audit data

Abstract

The invention discloses an enterprise audit data intelligent analysis processing method and system, and the method comprises the steps: obtaining a set of a plurality of audit data items of an audited enterprise, and carrying out the time sequence feature analysis of the set of the plurality of audit data items of the audited enterprise through employing a data processing and analysis technology based on a deep learning neural network; and according to the global semantic aggregation features of the set of the plurality of audit data items, an analysis result about whether the audited enterprise is abnormal or not is automatically obtained. In this way, the model can understand the association and context relationship between the data items, which is helpful for discovering potential anomalies or risks and improving the accuracy of anomaly detection.

Description

Technical Field

[0001] The present application relates to the field of intelligent processing, and more specifically, to a method and system for intelligent analysis and processing of enterprise audit data. Background Art

[0002] As the level of informatization of audited units increases, data analysis, as an important audit tool, plays an increasingly important role in the process of audit modernization. Audit data analysis is an indispensable part of audit projects, and it is crucial to ensure the healthy operation of enterprises, enhance transparency, improve efficiency, and comply with regulations.

[0003] However, there is currently a lack of effective data correlation analysis tools and methods for enterprise audit data analysis and processing, which makes it difficult to discover and utilize the correlation between different types of data, limits the potential information of each data item, and leads to low accuracy of analysis results.

[0004] Therefore, an optimized method for intelligent analysis and processing of enterprise audit data is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a method and system for intelligent analysis and processing of enterprise audit data, which obtains a set of multiple audit data items of the audited enterprise, and uses data processing and analysis technology based on deep learning neural networks to perform time series feature analysis on the set of multiple audit data items of the audited enterprise, thereby automatically obtaining the analysis result of whether there are any anomalies in the audited enterprise based on the global semantic aggregation features of the set of multiple audit data items. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection.

[0006] According to one aspect of the present application, a method for intelligent analysis and processing of enterprise audit data is provided, which includes:

[0007] Obtain recruitment requirements posted by enterprises;

[0008] Extracting description information of the first candidate talent from the talent database;

[0009] Semantically encoding the recruitment requirements published by the enterprise to obtain a recruitment requirements semantic encoding vector;

[0010] After performing sentence processing on the description information of the first candidate talent, inter-sentence context semantic coding is performed to obtain a sequence of context first candidate talent description information sentence granularity semantic coding vectors;

[0011] Based on the sequence of the granular semantic coding vectors of the first candidate talent description information sentence in the context, performing one-way interactive matching on the recruitment demand semantic coding vector to obtain a recruitment demand-talent information interactive matching vector as a recruitment demand-talent information interactive matching feature;

[0012] Based on the recruitment demand-talent information interactive matching characteristics, an analysis result is obtained.

[0013] In the above-mentioned enterprise audit data intelligent analysis and processing method, the set of multiple audit data items is data regularized to obtain a time series of multiple identical audit data items, including: based on the data item dimension, the set of multiple audit data items is data regularized to obtain a time series of multiple identical audit data items.

[0014] In the above-mentioned method for intelligent analysis and processing of enterprise audit data, the time series of each of the multiple time series of the same audit data items are respectively subjected to audit item sequence encoding to obtain multiple audit item time series correlation feature vectors, including: passing the time series of each of the multiple time series of the same audit data items through an audit item sequence encoder based on an RNN model to obtain the multiple audit item time series correlation feature vectors.

[0015] In the above-mentioned enterprise audit data intelligent analysis and processing method, the audit information significance fusion is performed on the multiple audit item time-series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature, including: inputting the multiple audit item time-series correlation feature vectors into an audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector.

[0016] In the above-mentioned enterprise audit data intelligent analysis and processing method, the multiple audit item time-series correlation feature vectors are input into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector, including: using a scoring function to measure the significance factors of the first k-1 audit item time-series correlation feature vectors in the multiple audit item time-series correlation feature vectors to obtain k-1 audit item time-series correlation significance factors; using the k-1 audit item time-series correlation significance factors as weights, calculating the position-weighted sum of the first k-1 audit item time-series correlation feature vectors in the multiple audit item time-series correlation feature vectors to aggregate the significance factors of the multiple audit item time-series correlation feature vectors. The first k-1 audit item time series association feature vectors are used to obtain the audit item time series association aggregation representation vector of the rear node set; the kth audit item time series association feature vector among the multiple audit item time series association feature vectors is linearly modulated using preset hyperparameters to obtain the kth audit item time series association feature vector after modulation; the positional sum of the kth audit item time series association feature vector after modulation and the audit item time series association aggregation representation vector of the rear node set is calculated to obtain the audit item time series association full-node aggregation representation vector; and the audit item time series association full-node aggregation representation vector is input into a multi-layer perceptron model to obtain the global audit item semantic aggregation representation vector.

[0017] In the above-mentioned enterprise audit data intelligent analysis and processing method, a scoring function is used to measure the significant factors of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significant factors, including: respectively calculating the matrix product between each weight coefficient matrix and the first k-1 audit item time series correlation feature vectors to obtain the first k-1 audit item time series correlation weighted feature vectors; adding the corresponding bias coefficient matrix to each audit item time series correlation weighted feature vector in the first k-1 audit item time series correlation weighted feature vectors. Set the vector to obtain the first k-1 audit item temporal association bias feature vectors; activate the first k-1 audit item temporal association bias feature vectors to obtain the first k-1 audit item temporal association activation feature vectors; calculate the product of each audit item temporal association activation feature vector in the first k-1 audit item temporal association activation feature vectors and the weight coefficient vector to obtain the first k-1 audit item temporal association degree; and input the first k-1 audit item temporal association degree into the softmax function to obtain the k-1 audit item temporal association significance factor.

[0018] In the above-mentioned enterprise audit data intelligent analysis and processing method, the first k-1 audit item time series association bias feature vectors are activated to obtain the first k-1 audit item time series association activation feature vectors, including: inputting the first k-1 audit item time series association bias feature vectors into the sigmoid function for activation to obtain the first k-1 audit item time series association activation feature vectors.

[0019] In the above-mentioned enterprise audit data intelligent analysis and processing method, an analysis result is obtained based on the global audit item semantic aggregation representation feature, including: passing the global audit item semantic aggregation representation vector through a classifier-based analysis result generator to obtain an analysis result, and the analysis result is used to indicate whether there is any abnormality in the audited enterprise.

[0020] According to another aspect of the present application, there is provided a system for intelligent analysis and processing of enterprise audit data, comprising:

[0021] An audit data item acquisition module, used to acquire a set of multiple audit data items of the audited enterprise;

[0022] An audit data item regularization module, used for performing data regularization on the set of the multiple audit data items to obtain a time series of multiple identical audit data items;

[0023] An audit item sequence encoding module, used for respectively performing audit item sequence encoding on the time series of each of the multiple time series of the same audit data items to obtain multiple audit item time series correlation feature vectors;

[0024] An audit information significant fusion module, used for performing audit information significant fusion on the multiple audit item time series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature;

[0025] The analysis result generation module is used to obtain the analysis result based on the semantic aggregation representation features of the global audit items.

[0026] Compared with the prior art, the present application provides a method and system for intelligent analysis and processing of enterprise audit data, which obtains a set of multiple audit data items of the audited enterprise and uses data processing and analysis technology based on deep learning neural networks to perform time series feature analysis on the set of multiple audit data items of the audited enterprise, thereby automatically obtaining an analysis result of whether the audited enterprise has anomalies based on the global semantic aggregation features of the set of multiple audit data items. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 It is a flowchart of the method for intelligent analysis and processing of enterprise audit data according to an embodiment of the present application.

[0029] Figure 2 Schematic diagram of the architecture of the method for intelligent analysis and processing of enterprise audit data according to an embodiment of the present application.

[0030] Figure 3 A flowchart of inputting the multiple audit item time-series correlation feature vectors into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector in the enterprise audit data intelligent analysis and processing method according to the embodiment of the present application.

[0031] Figure 4 A flowchart of using a scoring function in the enterprise audit data intelligent analysis and processing method according to an embodiment of the present application to measure the significance factors of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significance factors.

[0032] Figure 5 It is a block diagram of an enterprise audit data intelligent analysis and processing system according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0034] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0035] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0036] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0037] As the level of informatization of audited units increases, data analysis, as an important audit tool, plays an increasingly important role in the process of audit modernization. Audit data analysis is an indispensable part of audit projects, and it is crucial to ensure the healthy operation of enterprises, enhance transparency, improve efficiency, and comply with regulations.

[0038] However, there is currently a lack of effective data correlation analysis tools and methods for enterprise audit data analysis and processing, which makes it difficult to discover and utilize the correlation between different types of data, limits the potential information of each data item, and leads to low accuracy of analysis results.

[0039] Therefore, in response to the above technical problems, this application proposes an optimized intelligent analysis and processing method for enterprise audit data, which obtains a set of multiple audit data items of the audited enterprise and uses data processing and analysis technology based on deep learning neural networks to perform time series feature analysis on the set of multiple audit data items of the audited enterprise, so as to automatically obtain the analysis result of whether the audited enterprise has abnormalities based on the global semantic aggregation features of the set of multiple audit data items. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection.

[0040] Figure 1 It is a flowchart of the method for intelligent analysis and processing of enterprise audit data according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the architecture of the enterprise audit data intelligent analysis and processing method according to an embodiment of the present application. Figure 1 and Figure 2As shown, according to the embodiment of the present application, the method for intelligent analysis and processing of enterprise audit data includes: S110, obtaining a set of multiple audit data items of the audited enterprise; S120, performing data regularization on the set of multiple audit data items to obtain multiple time series of the same audit data items; S130, performing audit item sequence encoding on the time series of each of the multiple time series of the same audit data items to obtain multiple audit item time series correlation feature vectors; S140, performing audit information significant fusion on the multiple audit item time series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature; and, S150, obtaining an analysis result based on the global audit item semantic aggregation representation feature.

[0041] In step S110, a set of multiple audit data items of the audited enterprise is obtained. It should be understood that the set of audit data items refers to a set of audit projects obtained from one audit. Specifically, the set of one audit project is a summary of a series of related data collected from different audit projects, and these data items may include financial statements, transaction records, account balances, contract agreements, tax information, internal control reports, etc. Based on this, in the technical solution of the present application, a set of multiple audit data items of the audited enterprise is obtained, that is, the data of different audit projects can provide information from different angles, which is helpful for in-depth understanding of the business processes and potential problems of the enterprise, and by collecting a set of multiple audit data items, a comprehensive analysis of the financial and operating conditions of the enterprise can be conducted, so as to accurately identify possible risks and fraud.

[0042] In step S120, the set of multiple audit data items is data-regulated to obtain a time series of multiple identical audit data items. Specifically, in an embodiment of the present application, the set of multiple audit data items is data-regulated to obtain a time series of multiple identical audit data items, including: based on the data item dimension, the set of multiple audit data items is data-regulated to obtain a time series of multiple identical audit data items. It should be understood that, considering that the set of multiple audit data items contains audit data information of multiple data items, these identical audit data items have characteristic information in the time dimension. Therefore, in order to independently and individually perform feature analysis and comparison based on the time scale on the same audit data, so as to more accurately capture and extract the temporal characteristic information of the audit data, in the technical solution of the present application, based on the data item dimension, the set of multiple audit data items is data-regulated to obtain a time series of multiple identical audit data items. In particular, the data item dimension may include dimensional information such as transaction records, account balances, and contract agreements. That is to say, by regularizing the collection of multiple audit data items, each identical audit data item can be compared and analyzed across time to better identify changing trends, periodicities and abnormal fluctuations in the time dimension, thereby helping to improve the efficiency and quality of the audit work.

[0043] In step S130, audit item sequence encoding is performed on the time series of each of the multiple identical audit data items in the time series to obtain multiple audit item time series correlation feature vectors. Specifically, in an embodiment of the present application, audit item sequence encoding is performed on the time series of each of the multiple identical audit data items in the time series to obtain multiple audit item time series correlation feature vectors, including: passing the time series of each of the multiple identical audit data items in the time series through an audit item sequence encoder based on an RNN model to obtain the multiple audit item time series correlation feature vectors. Accordingly, it is considered that the time series of each of the multiple identical audit data items in the time series have mutual correlation and dependency relationships between the same audit data items in the time dimension. The RNN model is a neural network model suitable for processing time series data, which can capture the time series correlation relationship between data items. Based on this, in order to more clearly understand and analyze the mutual correlation between the same audit data items at different time points, in the technical solution of the present application, the time series of each of the multiple time series of the same audit data items are passed through an audit item sequence encoder based on the RNN model to obtain multiple audit item time series correlation feature vectors. It should be understood that in the audit data, data items at different time points often have certain time series correlations, and the audit item sequence encoder based on the RNN model can effectively model this time series relationship, and capture and mine the mutual influence and correlation between the time series of each same audit data item between different time spans, thereby improving the time series analysis and understanding capabilities of each same audit data.

[0044] In step S140, the multiple audit item temporal correlation feature vectors are subjected to audit information significant fusion to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature. Specifically, in an embodiment of the present application, the multiple audit item temporal correlation feature vectors are subjected to audit information significant fusion to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature, including: inputting the multiple audit item temporal correlation feature vectors into an audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector. It should be understood that, considering that the multiple audit item temporal correlation feature vectors have temporal correlation and influence based on the entire audit data, and the different audit item temporal correlation feature vectors have different temporal importance and influence. Therefore, in order to more effectively capture the significant correlation information between the temporal correlation feature vectors of each audit item in time series, so as to fully understand and analyze the global feature information of the enterprise audit project data, in the technical solution of the present application, the multiple audit item temporal correlation feature vectors are input into an audit information fusion module combined with node significance to obtain a global audit item semantic aggregation representation vector. Specifically, some audit items in the audit data set may contain more significant or more representative information, and have a greater impact on the entire audit result. That is, the audit information fusion module that combines node significance can dynamically adjust the weights of each audit item's time-series correlation feature according to the importance of each audit item's time-series correlation feature vector in the entire audit data set, so that the model can pay more attention to these important audit item semantic information and aggregate and transmit them, thereby improving the accuracy and effectiveness of the audit item semantic features on a global scale. The final global audit item semantic aggregation representation vector can clearly understand the contribution of each audit item to the final result, making audit data processing more efficient and reliable.

[0045] Figure 3 The present invention is a flowchart of inputting the multiple audit item time-series correlation feature vectors into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector in the enterprise audit data intelligent analysis and processing method according to the embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3As shown, the multiple audit item time series correlation feature vectors are input into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector, including: S210, using a scoring function to measure the significance factors of the first k-1 audit item time series correlation feature vectors in the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significance factors; S220, using the k-1 audit item time series correlation significance factors as weights, calculating the position-weighted sum of the first k-1 audit item time series correlation feature vectors in the multiple audit item time series correlation feature vectors to aggregate the first k-1 audit item time series correlation feature vectors in the multiple audit item time series correlation feature vectors. Associating feature vectors to obtain a temporal association aggregation representation vector of the audit item of the rear node set; S230, using preset hyperparameters to linearly modulate the kth audit item temporal association feature vector among the multiple audit item temporal association feature vectors to obtain the kth audit item temporal association feature vector after modulation; S240, calculating the positional sum of the kth audit item temporal association feature vector after modulation and the temporal association aggregation representation vector of the audit item of the rear node set to obtain the audit item temporal association full-node aggregation representation vector; and, S250, inputting the audit item temporal association full-node aggregation representation vector into a multi-layer perceptron model to obtain the global audit item semantic aggregation representation vector.

[0046] Figure 4 The present invention is a flowchart of using a scoring function to measure the significant factors of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significant factors in the enterprise audit data intelligent analysis and processing method according to the embodiment of the present application. More specifically, in the embodiment of the present application, Figure 4As shown, a scoring function is used to measure the significant factors of the first k-1 audit item time series association feature vectors among the multiple audit item time series association feature vectors to obtain k-1 audit item time series association significant factors, including: S310, respectively calculating the matrix product between each weight coefficient matrix and the first k-1 audit item time series association feature vectors to obtain the first k-1 audit item time series association weighted feature vectors; S320, adding the corresponding bias vector to each audit item time series association weighted feature vector in the first k-1 audit item time series association weighted feature vectors to obtain the first k-1 audit item temporal association bias feature vectors; S330, activate the first k-1 audit item temporal association bias feature vectors to obtain the first k-1 audit item temporal association activation feature vectors; S340, calculate the product of each audit item temporal association activation feature vector in the first k-1 audit item temporal association activation feature vectors and the weight coefficient vector to obtain the first k-1 audit item temporal association degrees; and, S350, input the first k-1 audit item temporal association degrees into the softmax function to obtain the k-1 audit item temporal association significance factors.

[0047] More specifically, in an embodiment of the present application, the first k-1 audit item timing association bias feature vectors are activated to obtain the first k-1 audit item timing association activation feature vectors, including: inputting the first k-1 audit item timing association bias feature vectors into a sigmoid function for activation to obtain the first k-1 audit item timing association activation feature vectors.

[0048] In an embodiment of the present application, specifically, the multiple audit item temporal correlation feature vectors are input into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector, including: using the audit information fusion module combined with node significance, the multiple audit item temporal correlation feature vectors are processed with the following aggregation formula to obtain the global audit item semantic aggregation representation vector; wherein the aggregation formula is:

[0049]

[0050] Among them, h i represents the i-th audit item time series correlation feature vector among the multiple audit item time series correlation feature vectors, W i represents the weight coefficient matrix, B i is the bias vector, represents the weight coefficient vector, sigmoid(·) represents the sigmoid function, softmax(·) represents the softmax function, e(h i) represents the audit item time series correlation significant factor at each position in the sequence of audit item time series correlation significant factors, k is the number of feature vectors in the multiple audit item time series correlation feature vectors, h k represents the kth audit item temporal association feature vector among the multiple audit item temporal association feature vectors, ∈ is the preset hyperparameter, MLP(·) represents a multilayer perceptron, h c A semantically aggregated representation vector is generated for the global audit item.

[0051] In step S150, an analysis result is obtained based on the global audit item semantic aggregation representation feature. Specifically, in an embodiment of the present application, an analysis result is obtained based on the global audit item semantic aggregation representation feature, including: passing the global audit item semantic aggregation representation vector through a classifier-based analysis result generator to obtain an analysis result, and the analysis result is used to indicate whether there is an abnormality in the audited enterprise. That is, the global audit item semantic aggregation representation feature obtained by node significance aggregation using the multiple audit item time-series correlation feature vectors is classified and processed, so as to automatically obtain an analysis result of whether there is an abnormality in the audited enterprise. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection.

[0052] It should be understood that in the technical solution of the present application, each audit data item time series correlation feature vector in the multiple audit item time series correlation feature vectors respectively represents the time series correlation feature of the same audit data item. However, considering that the differences between the various audit data items cannot be simply represented by numerical dimensions, this makes the global audit item semantic aggregation representation vector obtained have local feature overflow caused by insufficient node feature representation during the process of inputting the multiple audit item time series correlation feature vectors into the audit information fusion module combined with node significance, thereby affecting the accuracy of the global audit item semantic aggregation representation vector when it is classified by the classifier-based analysis result generator.

[0053] Based on this, preferably, the global audit item semantic aggregation representation vector is passed through a classifier-based analysis result generator to obtain an analysis result, including the following steps: calculating the self-mean matrix and the self-standard deviation matrix of the global audit item semantic aggregation representation vector, wherein the value of the (i, j)th position of the self-mean matrix is ​​the mean of the eigenvalue at the i-th position and the eigenvalue at the j-th position of the global audit item semantic aggregation representation vector, and the value of the (i, j)th position of the self-standard deviation matrix is ​​the standard deviation of the eigenvalue at the i-th position and the eigenvalue at the j-th position of the global audit item semantic aggregation representation vector; performing matrix multiplication on the transposed vector of the global audit item semantic aggregation representation vector as a row feature vector and the self-mean matrix to obtain a first intermediate vector and matrix-multiply the self-standard deviation matrix with the global audit item semantic aggregation representation vector to obtain a second intermediate vector; calculate the dot sum of the transposed vectors of the first intermediate vector and the second intermediate vector to obtain a third intermediate vector; calculate the matrix product of the self-mean matrix and the self-standard deviation matrix, and matrix-multiply the transposed vector of the global audit item semantic aggregation representation vector with the matrix product to obtain a fourth intermediate vector; calculate the dot sum of the transposed vectors of the third intermediate vector and the fourth intermediate vector to obtain an optimized global audit item semantic aggregation representation vector; and, pass the optimized global audit item semantic aggregation representation vector through the classifier-based analysis result generator to obtain the analysis result.

[0054] Accordingly, by using the self-mean matrix and self-standard deviation matrix of the group aggregation statistical evaluation of the local distribution of the feature value granularity of the global audit item semantic aggregation representation vector as the retrieval and response distribution enhancement of the global audit item semantic aggregation representation vector, and constructing a reference-free distribution retrieval response framework under an open domain based on the group aggregation feature distribution of the global audit item semantic aggregation representation vector, the redundancy of the distribution response caused by the local overflow characteristics of the global audit item semantic aggregation representation vector is avoided through response superposition, so as to realize the faithfulness constraint related to the self-aggregation statistics of the response of the optimized global audit item semantic aggregation representation vector to the classification target domain, and improve the accuracy of the analysis result obtained by the classifier-based analysis result generator of the optimized global audit item semantic aggregation representation vector. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection.

[0055] In summary, the enterprise audit data intelligent analysis and processing method based on the embodiment of the present application is explained, which obtains a set of multiple audit data items of the audited enterprise, and uses data processing and analysis technology based on deep learning neural networks to perform time series feature analysis on the set of multiple audit data items of the audited enterprise, so as to automatically obtain the analysis result of whether there are any abnormalities in the audited enterprise based on the global semantic aggregation features of the set of multiple audit data items. In this way, the model can understand the association and contextual relationship between data items, which helps to discover potential anomalies or risks and improve the accuracy of anomaly detection.

[0056] Figure 5 FIG. 1 is a block diagram of an enterprise audit data intelligent analysis and processing system according to an embodiment of the present application. Figure 5 As shown, according to the enterprise audit data intelligent analysis and processing system 100 of the embodiment of the present application, it includes: an audit data item acquisition module 110, which is used to obtain a set of multiple audit data items of the audited enterprise; an audit data item regularization module 120, which is used to perform data regularization on the set of multiple audit data items to obtain multiple time series of the same audit data items; an audit item sequence encoding module 130, which is used to perform audit item sequence encoding on the time series of each of the multiple same audit data items in the time series of the multiple same audit data items respectively to obtain multiple audit item time series correlation feature vectors; an audit information significant fusion module 140, which is used to perform audit information significant fusion on the multiple audit item time series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature; and an analysis result generation module 150, which is used to obtain an analysis result based on the global audit item semantic aggregation representation feature.

[0057] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned enterprise audit data intelligent analysis and processing system have been referred to above. Figures 1 to 4 The method has been introduced in detail in the description of the intelligent analysis and processing method of enterprise audit data, and therefore, its repeated description will be omitted.

[0058] As described above, the enterprise audit data intelligent analysis and processing system 100 according to the embodiment of the present disclosure can be implemented in various wireless terminals, such as a server with an enterprise audit data intelligent analysis and processing algorithm. In one possible implementation, the enterprise audit data intelligent analysis and processing system 100 according to the embodiment of the present disclosure can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the enterprise audit data intelligent analysis and processing system 100 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the enterprise audit data intelligent analysis and processing system 100 can also be one of the many hardware modules of the wireless terminal.

[0059] Alternatively, in another example, the enterprise audit data intelligent analysis and processing system 100 and the wireless terminal may also be separate devices, and the enterprise audit data intelligent analysis and processing system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0060] The foregoing is merely an example of the principles of the present disclosure, and various modifications may be made by those skilled in the art without departing from the scope of the present disclosure. The above embodiments are presented for the purpose of illustration and not limitation. The present disclosure may also take many forms other than those explicitly described herein. Therefore, it is emphasized that the present disclosure is not limited to the methods, systems and devices explicitly disclosed, but is intended to include variations and modifications within the spirit and scope of the appended claims.

Claims

1. A method for intelligent analysis and processing of enterprise audit data, characterized in that: include: Obtain a collection of multiple audit data items of the audited enterprise; Performing data regularization on the set of the multiple audit data items to obtain a time series of multiple identical audit data items; Performing audit item sequence coding on each time series of the same audit data items in the time series of the multiple same audit data items to obtain multiple audit item time series correlation feature vectors; Performing audit information significant fusion on the multiple audit item time-series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature; Based on the semantic aggregation representation features of the global audit items, an analysis result is obtained.

2. The enterprise audit data intelligent analysis and processing method according to claim 1 is characterized in that: Performing data regularization on the set of multiple audit data items to obtain a time series of multiple identical audit data items includes: based on the data item dimension, performing data regularization on the set of multiple audit data items to obtain a time series of multiple identical audit data items.

3. The enterprise audit data intelligent analysis and processing method according to claim 2 is characterized in that: Audit item sequence encoding is performed on the time series of each of the multiple time series of the same audit data items to obtain multiple audit item time series correlation feature vectors, including: passing the time series of each of the multiple time series of the same audit data items through an audit item sequence encoder based on an RNN model to obtain the multiple audit item time series correlation feature vectors.

4. The method for intelligent analysis and processing of enterprise audit data according to claim 3 is characterized in that: Performing audit information significance fusion on the multiple audit item time-series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature, including: inputting the multiple audit item time-series correlation feature vectors into an audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector.

5. The method for intelligent analysis and processing of enterprise audit data according to claim 4 is characterized in that: Inputting the multiple audit item time-series correlation feature vectors into the audit information fusion module combined with node significance to obtain the global audit item semantic aggregation representation vector, including: Using a scoring function to measure the significant factors of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significant factors; Taking the k-1 audit item time series correlation significance factors as weights, calculating the weighted sum of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors by position, so as to aggregate the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain the audit item time series correlation aggregate representation vector of the rear node set; Using a preset hyperparameter, linearly modulate the kth audit item time series association feature vector among the multiple audit item time series association feature vectors to obtain the modulated kth audit item time series association feature vector; Calculate the positional sum of the k-th audit item time series association feature vector after the modulation and the audit item time series association aggregation representation vector of the rear node set to obtain the audit item time series association full-node aggregation representation vector; The audit item temporal association full-node aggregate representation vector is input into a multi-layer perceptron model to obtain the global audit item semantic aggregate representation vector.

6. The method for intelligent analysis and processing of enterprise audit data according to claim 5 is characterized in that: Using a scoring function to measure the significant factors of the first k-1 audit item time series correlation feature vectors among the multiple audit item time series correlation feature vectors to obtain k-1 audit item time series correlation significant factors, including: Calculate the matrix product between each weight coefficient matrix and the first k-1 audit item time series associated feature vectors to obtain the first k-1 audit item time series associated weighted feature vectors; Adding a corresponding bias vector to each audit item time series association weighted feature vector in the first k-1 audit item time series association weighted feature vectors to obtain the first k-1 audit item time series association biased feature vectors; Activating the first k-1 audit item time series associated bias feature vectors to obtain the first k-1 audit item time series associated activation feature vectors; Calculate the product of each audit item time series association activation feature vector in the first k-1 audit item time series association activation feature vectors and the weight coefficient vector to obtain the first k-1 audit item time series association degree; The temporal correlations of the first k-1 audit items are input into the softmax function to obtain the temporal correlation significance factors of the k-1 audit items.

7. The method for intelligent analysis and processing of enterprise audit data according to claim 6 is characterized in that: Activating the first k-1 audit item time series association bias feature vectors to obtain the first k-1 audit item time series association activation feature vectors, including: inputting the first k-1 audit item time series association bias feature vectors into a sigmoid function for activation to obtain the first k-1 audit item time series association activation feature vectors.

8. The method for intelligent analysis and processing of enterprise audit data according to claim 7 is characterized in that: Based on the global audit item semantic aggregation representation feature, an analysis result is obtained, including: passing the global audit item semantic aggregation representation vector through a classifier-based analysis result generator to obtain an analysis result, wherein the analysis result is used to indicate whether there is an abnormality in the audited enterprise.

9. An enterprise audit data intelligent analysis and processing system, characterized in that: include: An audit data item acquisition module, used to acquire a set of multiple audit data items of the audited enterprise; An audit data item regularization module, used for performing data regularization on the set of the multiple audit data items to obtain a time series of multiple identical audit data items; An audit item sequence encoding module, used for respectively performing audit item sequence encoding on the time series of each of the multiple time series of the same audit data items to obtain multiple audit item time series correlation feature vectors; An audit information significant fusion module, used for performing audit information significant fusion on the multiple audit item time series correlation feature vectors to obtain a global audit item semantic aggregation representation vector as a global audit item semantic aggregation representation feature; The analysis result generation module is used to obtain the analysis result based on the semantic aggregation representation features of the global audit items.

10. The enterprise audit data intelligent analysis and processing system according to claim 9 is characterized in that: The inter-sentence context semantic encoding module is used to: pass the time series of each of the multiple time series of the same audit data items through an audit item sequence encoder based on an RNN model to obtain the multiple audit item time series association feature vectors.

Images