AI-based enterprise abnormal information analysis system and method

Abstract

The invention provides an enterprise abnormal information analysis system and method based on AI, relates to the field of intelligent data analysis, and adopts an artificial intelligence analysis technology based on deep learning to obtain abnormal enterprise financial information data from an enterprise financial management system, and analyzing the abnormal enterprise financial information data so as to judge the abnormal type of the abnormal financial information data. In this way, automatic and intelligent enterprise abnormal financial data processing can be realized, and the processing efficiency and accuracy of abnormal financial information can be improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent data analysis, and more specifically, to an AI-based enterprise abnormal information analysis system and method. Background Art

[0002] In business operations, the analysis of abnormal financial information plays a vital role, as it can reveal potential operating risks, management loopholes, etc. By analyzing abnormal financial information of enterprises, enterprises can locate the problem more accurately and take targeted measures to solve it. With the expansion of enterprise scale and the complexity of business, the amount of financial data has exploded, and traditional manual processing methods are no longer able to meet this challenge. When manual processing of these abnormal financial information is carried out, due to the limited ability of manual processing of large amounts of data, it is impossible to fully analyze all relevant information, and manual processing efficiency is low, and it is also easily affected by subjective judgment.

[0003] Therefore, an optimized AI-based enterprise abnormal information analysis solution is needed. Summary of the invention

[0004] In response to the shortcomings of the prior art, the present application provides an AI-based enterprise abnormal information analysis system and method.

[0005] According to one aspect of the present application, there is provided an enterprise abnormal information analysis system based on AI, which includes:

[0006] The enterprise financial abnormal data collection module is used to obtain abnormal enterprise financial data from the enterprise's financial management system;

[0007] An enterprise financial abnormal data feature encoding module is used to perform feature encoding on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector;

[0008] The enterprise financial data classification result generation module is used to determine the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector.

[0009] According to another aspect of the present application, a method for analyzing abnormal enterprise information based on AI is provided, which includes:

[0010] Obtain abnormal enterprise financial data from the enterprise's financial management system;

[0011] Performing feature encoding on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector;

[0012] Based on the multi-scale financial information semantic feature vector, the abnormal type of the abnormal enterprise financial data is determined.

[0013] This application has significant technical effects due to the adoption of the above technical solutions:

[0014] The AI-based enterprise abnormal information analysis system and method provided in this application adopts artificial intelligence analysis technology based on deep learning to obtain abnormal enterprise financial information data from the enterprise financial management system, and analyzes the abnormal enterprise financial information data to determine the abnormal type of the abnormal financial information data. In this way, automated and intelligent enterprise abnormal financial data processing can be achieved, which is conducive to improving the processing efficiency and accuracy of abnormal financial information. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] 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.

[0016] Figure 1 This is a flowchart of an AI-based enterprise abnormal information analysis method according to an embodiment of the present application.

[0017] Figure 2 This is a flowchart of performing feature encoding on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application.

[0018] Figure 3 This is a flowchart of performing context encoding of different distance information on the multiple financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application.

[0019] Figure 4 The present invention is a flowchart of determining the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application.

[0020] Figure 5 This is a flowchart for optimizing the subclass label correlation metric of the multi-scale financial information semantic feature vector based on local distribution differences in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application to obtain an optimized multi-scale financial information semantic feature vector.

[0021] Figure 6 This is a system block diagram of an AI-based enterprise abnormal information analysis system according to an embodiment of the present application.

[0022] Figure 7 is a block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0024] In today's complex and ever-changing business environment, the robustness of corporate operations is inseparable from the health of its financial situation, and the accurate analysis of abnormal financial information has become a key link in corporate risk prevention and control and strategic adjustment. With the continuous expansion of corporate territory and the increasing complexity of business models, the volume and dimensions of financial data have achieved exponential growth, and this trend has posed unprecedented challenges to financial management.

[0025] Faced with massive and complex financial data, traditional manual review and analysis methods have gradually shown their limitations. On the one hand, due to the limited time and energy, it is difficult for humans to comprehensively sort out and deeply analyze the logic and motivations behind each abnormal transaction in a short period of time, which may lead to the omission of key risk points and thus affect the quality of corporate decision-making. On the other hand, the manual processing process is easily affected by subjective judgment, experience differences, and emotional fluctuations, making it difficult to ensure the objectivity and accuracy of the analysis results, increasing the risk of misjudgment and omission. Therefore, an optimized AI-based enterprise abnormal information analysis solution is expected.

[0026] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also demonstrated a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks has provided new solutions and solutions for AI-based enterprise abnormal information analysis.

[0027] Figure 1 Flowchart of the AI-based enterprise abnormal information analysis method according to an embodiment of the present application. Figure 1 As shown, according to the AI-based enterprise abnormal information analysis method of the embodiment of the present application, the method includes: S110, obtaining abnormal enterprise financial data from the enterprise's financial management system; S120, feature encoding the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector; S130, judging the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector.

[0028] Specifically, in the technical solution of the present application, it is first necessary to obtain abnormal corporate financial data from the company's financial management system. It should be well understood that abnormal corporate financial data will contain, for example, financial transaction information, financial statement information, tax-related information, and the like. Different types of anomalies often show different characteristics in the data. For example, account anomalies may be more reflected in account transaction information such as transaction amount, transaction frequency, and counterparty; while financial statement anomalies may be more reflected in financial statement information such as balance sheet, income statement, and cash flow statement. By conducting in-depth analysis and mining of abnormal corporate financial data, the type of financial anomaly can be accurately determined, and strong support can be provided for the company to take corresponding countermeasures.

[0029] Figure 2 The present invention is a flowchart of a method for analyzing abnormal enterprise information based on AI in accordance with an embodiment of the present invention, for encoding the abnormal enterprise financial data to obtain a multi-scale semantic feature vector of financial information. Figure 2 As shown, the feature encoding of the abnormal corporate financial data to obtain a multi-scale financial information semantic feature vector includes: S121, dividing the abnormal corporate financial data to obtain a plurality of financial information keywords; S122, passing the plurality of financial information keywords through an embedding layer to obtain a plurality of financial key information word embedding vectors; S123, context encoding of different distance information on the plurality of financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector.

[0030] Next, the abnormal corporate financial data is divided and processed to obtain multiple financial information keywords. It is not difficult to understand that through division processing, information points that play a key role in abnormal analysis, namely financial information keywords, can be extracted from complex financial data. In this way, a large amount of complex data can be simplified into a series of keywords, which is convenient for subsequent data processing, thereby improving the efficiency of abnormal data classification and analysis. As the key point in financial data, financial information keywords can directly or indirectly reflect the abnormal nature of corporate financial data. Different types of anomalies often show different characteristics on specific financial information keywords. Through in-depth analysis and mining of financial information keywords, a cognitive framework for various types of anomalies can be gradually constructed, and ultimately the abnormal classification of corporate financial data can be achieved.

[0031] Then, the multiple financial information keywords are passed through the embedding layer to obtain multiple financial key information word embedding vectors. The embedding layer is mainly used to process discrete data and convert it into a continuous vector representation, and is widely used in natural language processing, image processing and other fields. The purpose of the embedding layer is to capture the correlation and semantic information between each element (such as vocabulary) in the input data, and map these elements to a low-dimensional, dense vector space. The original financial information keywords are usually high-dimensional and sparse. By inputting multiple financial information keywords into the embedding layer, a vector representation that can reflect the semantic similarity and correlation between the keywords to a certain extent can be obtained.

[0032] In an embodiment of the present application, one possible implementation method of passing the multiple financial information keywords through an embedding layer to obtain multiple financial key information word embedding vectors may be: the embedding layer uses a learnable embedding matrix to embed the multiple financial information keywords to obtain the multiple financial key information word embedding vectors.

[0033] Figure 3 The present invention is a flowchart of performing context encoding of different distance information on the multiple financial key information word embedding vectors in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application to obtain the multi-scale financial information semantic feature vector. Figure 3 As shown, the context encoding of different distance information of the multiple financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector includes: S1231, passing the multiple financial key information word embedding vectors through a medium-short distance financial information context encoder to obtain a first-scale financial information semantic feature vector; S1232, passing the multiple financial key information word embedding vectors through a long-distance financial information context encoder to obtain a second-scale financial information semantic feature vector; S1323, fusing the first-scale financial information semantic feature vector and the second-scale financial information semantic feature vector to obtain the multi-scale financial information semantic feature vector.

[0034] The keywords in financial information do not exist in isolation, and there are often complex contextual relationships between them. In order to capture the contextual information between these keywords and understand their specific meanings and functions in financial data, it is necessary to embed the multiple financial key information words into vectors through a medium- and short-distance financial information context encoder to obtain a first-scale financial information semantic feature vector. It is easy to understand that although the embedding layer can convert financial keywords into vector representations, these vectors themselves do not contain rich semantic information. Through the medium- and short-distance financial information context encoder, these vectors can be further processed and processed to contain more semantic information. The essence of the medium- and short-distance financial information context encoder in this application is a bidirectional long short-term memory neural network model (Bi-LSTM). Bi-LSTM is a special recurrent neural network (RNN) that can consider both forward and backward contextual information when processing sequence data. This structure enables the model to understand the meaning of the input data more comprehensively. LSTM itself can solve the problem of gradient vanishing or gradient explosion that is prone to occur in traditional RNN when processing long sequences, so that long-term dependencies in the sequence can be captured, which is particularly important for analyzing abnormal information in corporate financial data, because abnormal information may be hidden in long sequences of data.

[0035] At the same time, the multiple financial key information word embedding vectors are passed through a long-distance financial information context encoder to obtain a second-scale financial information semantic feature vector. The essence of the long-distance financial information context encoder in this application is a semantic encoder based on a transformer. Bi-LSTM is a variant of a recurrent neural network (RNN), which captures context information in sequence data through two LSTM layers, forward and backward. The LSTM unit contains a forget gate, an input gate, and an output gate, which control the flow of information through a gating mechanism to solve the problem of gradient vanishing or gradient explosion in traditional RNN when processing long sequences. Transformer completely abandons the cyclic structure of RNN and adopts a self-attention mechanism to capture dependencies in the sequence. Transformer consists of an encoder and a decoder, in which the encoder part encodes the input sequence through multiple layers of self-attention layers and feedforward neural network layers to extract rich semantic information. Transformer can simultaneously consider all elements in the input sequence through the self-attention mechanism, thereby capturing more comprehensive context information. This global context perception capability makes Transformer more accurate and in-depth when extracting semantic information. Although Bi-LSTM alleviates the long-distance dependency problem to a certain extent through the gating mechanism, the self-attention mechanism of Transformer is more effective in dealing with long-distance dependencies. It can directly calculate the correlation between any two elements without accumulating information through multiple time steps.

[0036] The first-scale financial information semantic feature vector focuses on capturing the medium- and short-range contextual relationship of financial information. Through the bidirectional processing mechanism of Bi-LSTM, it can better understand the local semantic structure of financial data; the second-scale financial information semantic feature vector focuses on the global financial information context perception. The converter-based semantic encoder can process all elements in the sequence at the same time through the self-attention mechanism to extract global semantic information. In order to integrate medium- and short-range and global semantic information to form a more comprehensive and in-depth understanding of financial information, the technical solution of this application needs to fuse the first-scale financial information semantic feature vector and the second-scale financial information semantic feature vector to obtain a multi-scale financial information semantic feature vector. Fusion can make the model more robust and reduce dependence on specific types of data, thereby showing better generalization capabilities when processing different types of financial data.

[0037] In an embodiment of the present application, a possible implementation method of fusing the first-scale financial information semantic feature vector and the second-scale financial information semantic feature vector to obtain a multi-scale financial information semantic feature vector may be: fusing the first-scale financial information semantic feature vector and the second-scale financial information semantic feature vector using the following fusion formula to obtain the multi-scale financial information semantic feature vector; wherein the fusion formula is:

[0038] V c =Concat[V1, V2]

[0039] Wherein, V1 represents the semantic feature vector of the first-scale financial information, V2 represents the semantic feature vector of the second-scale financial information, Concat[·] represents the cascade function, V c Represents the multi-scale financial information semantic feature vector.

[0040] Figure 4 The present invention is a flowchart of determining the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector in the AI-based enterprise abnormal information analysis method according to an embodiment of the present application. Figure 4 As shown, based on the multi-scale financial information semantic feature vector, determining the abnormal type of the abnormal enterprise financial data includes: S131, optimizing the subclass label related metric based on local distribution difference of the multi-scale financial information semantic feature vector to obtain an optimized multi-scale financial information semantic feature vector; S132, passing the optimized multi-scale financial information semantic feature vector through an abnormal type label classifier to obtain a classification result, and the classification result is used to represent the label of the abnormal type to which the abnormal enterprise financial data belongs.

[0041] In particular, when processing abnormal corporate financial data, different financial information keywords may differ in frequency, position, etc. in the context of the financial text. This will cause the embedding layer and the financial information context encoder to be affected to varying degrees when capturing semantic information, resulting in differences in the local distribution of the obtained financial information semantic feature vectors. Since there are differences in the local distribution of multi-scale financial information semantic feature vectors, when the vector is passed to the abnormal type label classifier, the classifier may be affected by this difference, resulting in a deviation in the classifier's judgment of different types of abnormal data. That is, considering the local distribution differences of the multi-scale financial information semantic feature vectors, when the classification judgment is made by the classifier, there will be a category harmony deviation relative to the predetermined class label of the abnormal type label classifier, affecting the classification iteration effect. Therefore, it is necessary to optimize the sub-class label related metrics of the multi-scale financial information semantic feature vectors based on local distribution differences.

[0042] Specifically, in the embodiments of the present application, Figure 5 The present invention is a flowchart of optimizing the subclass label correlation metric of the multi-scale financial information semantic feature vector based on the local distribution difference in the enterprise abnormal information analysis method based on AI in accordance with the embodiment of the present application to obtain the optimized multi-scale financial information semantic feature vector. Figure 5 As shown, the subclass label related metric optimization based on local distribution difference of the multi-scale financial information semantic feature vector to obtain the optimized multi-scale financial information semantic feature vector includes: S1311, determining the weight learning parameter matrix of the abnormal type label classifier; S1312, extracting each column vector in the weight learning parameter matrix to obtain a set of weight learning parameter column vectors; S1313, respectively calculating the metric correlation prediction factor between each weight learning parameter column vector in the set of weight learning parameter column vectors and the multi-scale financial information semantic feature vector to obtain a metric correlation prediction factor modulation vector; S1314, using the metric correlation prediction factor modulation vector as the weight vector, calculating the position point multiplication between the metric correlation prediction factor modulation vector and the multi-scale financial information semantic feature vector to obtain the optimized multi-scale financial information semantic feature vector.

[0043] More specifically, in an embodiment of the present application, a possible implementation method of respectively calculating the metric correlation prediction factor between each weight learning parameter column vector in the set of the weight learning parameter column vector and the multi-scale financial information semantic feature vector to obtain the metric correlation prediction factor modulation vector is: respectively calculating the metric correlation prediction factor between each weight learning parameter column vector in the set of the weight learning parameter column vector and the multi-scale financial information semantic feature vector using the following formula to obtain the metric correlation prediction factor modulation vector, wherein the formula is:

[0044]

[0045] Among them, V i represents each weight learning parameter column vector in the set of weight learning parameter column vectors, V c represents the multi-scale financial information semantic feature vector, It means subtracting by position. is the square of the bi-norm of the eigenvector, ‖·‖ F is the Frobenius norm of the eigenvector, M p represents the weight learning parameter matrix, exp(·) represents the exponential operation with the natural constant e as the base, V p represents the metric correlation predictor modulation vector.

[0046] In the technical solution of the present application, it is taken into account that the multi-scale financial information semantic feature vector has local distribution differences, which leads to the problem that when it is classified by the abnormal type label classifier, there will be a category harmony offset relative to the predetermined class label of the abnormal type label classifier, thereby affecting the classification iteration effect. Based on this, in the technical solution of the present application, the multi-scale financial information semantic feature vector is optimized for sub-class label related metrics based on local distribution differences, and the superposition paradigm of the two norms between each column vector in the weight learning parameter matrix of the abnormal type label classifier and the multi-scale financial information semantic feature vector is used as the sub-class label related metric integral factor and the matrix norm of the weight learning parameter of the abnormal type label classifier is used as the differential factor, and then the multi-scale financial information semantic feature vector is bidirectionally linearly modulated using the association paradigm between the sub-class label related metric integral factor and the differential factor, thereby improving the local feature distribution consistency of the multi-scale financial information semantic feature vector by adaptively stretching and compressing the multi-scale financial information semantic feature vector, thereby reducing the harmonious deviation of the multi-scale financial information semantic feature vector relative to the predetermined class label of the abnormal type label classifier, thereby improving the accuracy of the classification result.

[0047] Finally, the optimized multi-scale financial information semantic feature vector is passed through an abnormal type label classifier to obtain a classification result, and the classification result is used to represent the label of the abnormal type to which the abnormal corporate financial data belongs. It should be understood that the optimized multi-scale financial information semantic feature vector is a high-dimensional, abstract feature representation extracted after complex processing of the original financial data, and these features can capture the potential abnormal patterns in the original abnormal financial data. In order to further clarify the category of the abnormality to which the financial data belongs, it is necessary to pass the feature vector through an abnormal type label classifier. The abnormal type label classifier is a machine learning model based on supervised learning, and its main task is to predict the corresponding abnormal type label based on the input feature vector. This classifier will learn a mapping relationship from feature space to label space during the training process, that is, it can infer the most likely abnormal type based on the input feature vector. The classification result can clearly point out the type of abnormal financial data, thereby helping enterprises to quickly locate the problem and deal with it in a targeted manner.

[0048] In an embodiment of the present application, the optimized multi-scale financial information semantic feature vector is passed through an anomaly type label classifier to obtain a classification result, and the classification result is used to represent the label of the anomaly type to which the abnormal corporate financial data belongs. One possible implementation method may be: using the fully connected layer of the anomaly type label classifier to fully connect encode the optimized multi-scale financial information semantic feature vector to obtain an encoded multi-scale financial information semantic feature vector; and inputting the encoded multi-scale financial information semantic feature vector into the Softmax classification function of the anomaly type label classifier to obtain the classification result.

[0049] In summary, the AI-based enterprise abnormal information analysis method based on the embodiment of the present application is explained, which adopts artificial intelligence analysis technology based on deep learning to obtain abnormal enterprise financial information data from the enterprise financial management system, and analyzes the abnormal enterprise financial information data to determine the abnormal type of the abnormal financial information data. In this way, automated and intelligent enterprise abnormal financial data processing can be achieved, which is conducive to improving the processing efficiency and accuracy of abnormal financial information.

[0050] Figure 6 : is a system block diagram of an enterprise abnormal information analysis system based on AI according to an embodiment of the present application. Figure 6 As shown, according to the AI-based enterprise abnormal information analysis system 100 of the embodiment of the present application, it includes: an enterprise financial abnormal data collection module 110, which is used to obtain abnormal enterprise financial data from the enterprise's financial management system; an enterprise financial abnormal data feature encoding module 120, which is used to feature encode the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector; an enterprise financial data classification result generation module 130, which is used to determine the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector.

[0051] Here, those skilled in the art can understand that the specific functions and operations of each unit and module in the above-mentioned AI-based enterprise abnormal information analysis system 100 have been referred to above. Figures 1 to 5 The description of the AI-based enterprise abnormal information analysis method has been introduced in detail, and therefore, its repeated description will be omitted.

[0052] In summary, the AI-based enterprise abnormal information analysis system 100 according to the embodiment of the present application is explained, which uses artificial intelligence analysis technology based on deep learning to obtain abnormal enterprise financial information data from the enterprise financial management system, and analyzes the abnormal enterprise financial information data to determine the abnormal type of the abnormal financial information data. In this way, automated and intelligent enterprise abnormal financial data processing can be achieved, which is conducive to improving the processing efficiency and accuracy of abnormal financial information.

[0053] As described above, the AI-based enterprise abnormal information analysis system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for AI-based enterprise abnormal information analysis. In one example, the AI-based enterprise abnormal information analysis system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the AI-based enterprise abnormal information analysis 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 AI-based enterprise abnormal information analysis system 100 can also be one of the many hardware modules of the wireless terminal.

[0054] Alternatively, in another example, the AI-based enterprise abnormal information analysis system 100 and the wireless terminal may also be separate devices, and the AI-based enterprise abnormal information analysis 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.

[0055] Below, reference Figure 7 To describe an electronic device according to an embodiment of the present application. Figure 7 1 is a block diagram of an electronic device 10 according to an embodiment of the present application. For example, the electronic device 10 may be provided as a server or a terminal device. Figure 7 The electronic device 10 includes a processing component 11, which further includes one or more processors, and a memory resource represented by a memory 12 for storing instructions executable by the processing component 11, such as an application. The application stored in the memory 12 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 11 is configured to execute instructions to perform the above method.

[0056] The electronic device 10 may further include a power supply component 13 configured to perform power management of the electronic device 10, a wired or wireless network interface 14 configured to connect the electronic device 10 to a network, and an input / output (I / O) interface 15. The electronic device 10 may operate based on an operating system stored in the memory 12, such as Windows Server™, Mac OSX™, Unix™, Linux™, FreeBSD™ or the like.

[0057] In an embodiment of the present application, a computer-readable storage medium is also provided, such as a memory 12 including computer program instructions. The computer program instructions can be executed by a processing component 11 of an electronic device 10 to complete the above method.

[0058] The present application may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer program instructions for causing a processor to implement various aspects of the present application.

[0059] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0060] The computer program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer program instructions from the network and forwards the computer program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0061] The computer program instructions for performing the operation of the present application can be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The computer program instructions can be executed completely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer, partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet). In some embodiments, by using the state information of the computer program instructions to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA) or a programmable logic array (PLA), the electronic circuit can execute the computer program instructions, thereby realizing various aspects of the present application.

[0062] Various aspects of the present application are described herein with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer program instructions.

[0063] These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0064] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0065] In the several embodiments provided in this application, it should be understood that the disclosed methods, systems, etc. can be implemented in other ways. For example, the system embodiments described above are only illustrative, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0066] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0067] In addition, each functional module in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0068] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application.

[0069] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present application is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present application. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0070] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical solution of the present application may be modified or replaced by equivalents without departing from the spirit of the technical solution of the present application.

Claims

1. An AI-based enterprise abnormal information analysis method, It is characterized in that include: Obtain abnormal enterprise financial data from the enterprise's financial management system; Performing feature encoding on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector; Based on the multi-scale financial information semantic feature vector, the abnormal type of the abnormal enterprise financial data is determined.

2. According to the AI-based enterprise abnormal information analysis method of claim 1, It is characterized in that Feature encoding is performed on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector, including: The abnormal enterprise financial data is divided and processed to obtain a plurality of financial information keywords; Passing the plurality of financial information keywords through an embedding layer to obtain a plurality of financial key information word embedding vectors; Context encoding of different distance information is performed on the multiple financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector.

3. According to claim 2, the AI-based enterprise abnormal information analysis method, It is characterized in that The multiple financial information keywords are passed through an embedding layer to obtain multiple financial key information word embedding vectors, including: the embedding layer uses a learnable embedding matrix to embed the multiple financial information keywords to obtain the multiple financial key information word embedding vectors.

4. According to the AI-based enterprise abnormal information analysis method of claim 3, It is characterized in that Performing context encoding of different distance information on the multiple financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector includes: Passing the multiple financial key information word embedding vectors through a medium- and short-distance financial information context encoder to obtain a first-scale financial information semantic feature vector; Passing the multiple financial key information word embedding vectors through a long-distance financial information context encoder to obtain a second-scale financial information semantic feature vector; The first-scale financial information semantic feature vector and the second-scale financial information semantic feature vector are fused to obtain the multi-scale financial information semantic feature vector.

5. According to the AI-based enterprise abnormal information analysis method of claim 4, It is characterized in that The medium- and short-range financial information context encoder is a bidirectional long short-term memory neural network model.

6. The AI-based enterprise abnormal information analysis method according to claim 5, It is characterized in that The long-distance financial information context encoder is a transformer-based semantic encoder.

7. The AI-based enterprise abnormal information analysis method according to claim 6, It is characterized in that Based on the multi-scale financial information semantic feature vector, determining the abnormal type of the abnormal enterprise financial data includes: Performing subclass label related metric optimization based on local distribution difference on the multi-scale financial information semantic feature vector to obtain an optimized multi-scale financial information semantic feature vector; The optimized multi-scale financial information semantic feature vector is passed through an abnormal type label classifier to obtain a classification result, and the classification result is used to represent a label of the abnormal type to which the abnormal corporate financial data belongs.

8. The AI-based enterprise abnormal information analysis method according to claim 7, It is characterized in that The multi-scale financial information semantic feature vector is optimized based on the subclass label correlation metric of the local distribution difference to obtain an optimized multi-scale financial information semantic feature vector, including: Determine a weight learning parameter matrix of the abnormal type label classifier; Extracting each column vector in the weight learning parameter matrix to obtain a set of weight learning parameter column vectors; Respectively calculating the metric correlation prediction factor between each weight learning parameter column vector in the set of weight learning parameter column vectors and the multi-scale financial information semantic feature vector to obtain a metric correlation prediction factor modulation vector; The metric relevance prediction factor modulation vector is used as a weight vector, and the position point multiplication between the metric relevance prediction factor modulation vector and the multi-scale financial information semantic feature vector is calculated to obtain the optimized multi-scale financial information semantic feature vector.

9. An AI-based enterprise abnormal information analysis system, It is characterized in that include: The enterprise financial abnormal data collection module is used to obtain abnormal enterprise financial data from the enterprise's financial management system; An enterprise financial abnormal data feature encoding module is used to perform feature encoding on the abnormal enterprise financial data to obtain a multi-scale financial information semantic feature vector; The enterprise financial data classification result generation module is used to determine the abnormal type of the abnormal enterprise financial data based on the multi-scale financial information semantic feature vector.

10. The AI-based enterprise abnormal information analysis system according to claim 9, It is characterized in that The enterprise financial abnormal data feature coding module includes: An abnormal enterprise financial data preprocessing unit, used for dividing and processing the abnormal enterprise financial data to obtain a plurality of financial information keywords; A financial keyword embedding unit, used to pass the plurality of financial information keywords through an embedding layer to obtain a plurality of financial key information word embedding vectors; Financial information keyword context semantic coding is used to perform context coding of different distance information on the multiple financial key information word embedding vectors to obtain the multi-scale financial information semantic feature vector.

Images