Data analysis method and device based on artificial intelligence, computer equipment and medium

By perturbating and smoothing the text data of financial exams, the problem of low efficiency in cheating detection in existing technologies is solved, and efficient and accurate cheating analysis results are generated.

CN119047456APending Publication Date: 2024-11-29PING AN TECH (SHENZHEN) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411183211.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Existing technologies for fraud detection in the financial industry require a significant amount of manpower for training data collection, data cleaning, and data labeling, resulting in low efficiency and inaccurate results.

Method used

The first pre-trained model is used to perturb the financial exam text data to generate enhanced text samples. The second pre-trained model is used to calculate the difference in log probability function values ​​between the text data and the enhanced text samples, calculate the smoothness feature, and compare it with a preset threshold to generate cheating analysis results.

Benefits of technology

It achieves fast, intelligent and accurate cheating analysis, saves labor costs, improves detection efficiency and ensures the accuracy of results, and avoids the difficulty of manually finding features.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119047456A_ABST
    Figure CN119047456A_ABST
Patent Text Reader

Abstract

The invention belongs to the field of artificial intelligence and the field of financial science and technology, and relates to an artificial intelligence-based data analysis method, which comprises the following steps of: obtaining financial examination text data; performing disturbance processing on the financial examination text data based on a first pre-training model to obtain a preset number of enhanced text samples; calculating the difference of logarithmic probability function values between the financial examination text data and all the enhanced text samples based on a second pre-training model; calculating a smoothness feature of the financial examination text data based on the difference of the logarithmic probability function values; generating a numerical value comparison result of the smoothness feature and a preset threshold value; and generating a cheating analysis result of the financial examination text data based on the numerical comparison result. The invention further provides a data analysis device based on artificial intelligence, computer equipment and a storage medium. In addition, the cheating analysis result can be stored in the block chain. According to the invention, the processing efficiency of cheating detection is effectively improved, and the data accuracy of the generated cheating analysis result is ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence development technology and financial technology, and in particular to artificial intelligence-based data analysis methods, devices, computer equipment and storage media. Background Art

[0002] The financial industry is a highly sensitive and complex industry. In order to improve their professional quality and competitiveness, practitioners in the financial industry need to obtain financial certificates to prove their professionalism. However, with the continuous development of technology, the results of financial industry examinations are directly related to the operation and security of financial institutions. If there is cheating, it will lead to increased risks in the financial industry and may even cause systemic risks. Anti-cheating can ensure fair competition and promote industry development.

[0003] In the prior art, the anti-fraud technology used in the financial industry is usually based on artificial intelligence detection technology to detect whether there is cheating behavior. The detection technology includes the following steps:

[0004] (1) Data collection and cleaning. First, the test takers’ answer data needs to be collected from the test platform and stored in the database. The collected data is cleaned and preprocessed, such as removing invalid data and standardizing the answers. (2) Feature extraction: Through machine learning technology, feature information in the test takers’ answers is extracted, such as sentence structure, vocabulary usage, grammatical rules, etc. (3) Model building: Based on known cheating samples and normal samples, a machine learning model is built for training. The model judges the answers based on the feature information and identifies possible cheating behaviors. (4) Cheating detection: After the test, all answer sheets are tested, and the model automatically determines whether each answer sheet contains cheating behaviors based on the feature information. However, in the above detection technology steps, a lot of manpower is required for training data collection, data cleaning, and data labeling, and the problem of manually finding features is difficult, which results in the detection of cheating behaviors requiring a lot of cost, low efficiency of cheating detection, and the inability to guarantee the accuracy of the generated cheating detection results. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a data analysis method, device, computer equipment and storage medium based on artificial intelligence to solve the problem that in the step processing of the existing detection technology applied to cheating detection, a large amount of manpower is required for training data collection, data cleaning and data labeling processing, and the processing of manually finding features is difficult, which leads to the technical problems that the detection and processing of cheating behavior requires more costs, the processing efficiency of cheating detection is low, and the accuracy of the generated cheating detection results cannot be guaranteed.

[0006] In order to solve the above technical problems, the present application embodiment provides a data analysis method based on artificial intelligence, which adopts the following technical solutions:

[0007] Get financial exam text data;

[0008] Performing perturbation processing on the financial test text data based on the first pre-trained model to obtain a preset number of enhanced text samples;

[0009] Calculate the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model;

[0010] Calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function values;

[0011] Performing numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result;

[0012] A cheating analysis result corresponding to the financial test text data is generated based on the numerical comparison result.

[0013] Furthermore, the step of performing perturbation processing on the financial test text data based on the first pre-training model to obtain a preset number of enhanced text samples specifically includes:

[0014] Loading the first pre-trained model into memory;

[0015] Get the preset quantity;

[0016] The financial test text data is perturbed based on the first pre-trained model in the memory to obtain enhanced text samples matching the preset quantity.

[0017] Furthermore, the step of calculating the difference between the logarithmic probability function values ​​of the financial test text data and all the enhanced text samples based on the second pre-training model specifically includes:

[0018] Calculating the mean of the logarithmic probabilities of all the enhanced text samples based on the second pre-training model;

[0019] Calculating a specified logarithmic probability of the financial examination text data;

[0020] Calculating the difference between the specified log probability and the mean of the log probabilities of all the enhanced text samples;

[0021] The difference is taken as the difference between the log probability function values.

[0022] Furthermore, the step of calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function values ​​specifically includes:

[0023] Calculating the variance of the log probability of all the enhanced text samples;

[0024] Call the preset feature calculation formula;

[0025] Based on the feature calculation formula, the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value is calculated to obtain a corresponding calculation result;

[0026] The calculation result is used as the smoothness feature of the financial test text data.

[0027] Furthermore, the step of generating a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result specifically includes:

[0028] Determining whether the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold;

[0029] If the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold, a first cheating analysis result is generated indicating that the financial test text data is cheating data;

[0030] If the content of the numerical comparison result is that the smoothness feature is less than the preset threshold, a second cheating analysis result is generated that the financial test text data is normal data.

[0031] Furthermore, after the step of generating a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result, the method further includes:

[0032] If the cheating analysis result indicates that the financial examination text data is cheating data, generating a target warning report corresponding to the financial examination text data based on the cheating analysis result;

[0033] Obtaining communication information of reviewers;

[0034] Based on the communication information, the target early warning report is sent to the reviewer.

[0035] Furthermore, the step of generating a target warning report corresponding to the financial examination text data based on the cheating analysis result specifically includes:

[0036] Acquire target test taker information corresponding to the financial test text data;

[0037] Obtaining data analysis results corresponding to the financial examination text data;

[0038] Call the preset early warning report template;

[0039] Fill the financial test text data, the cheating analysis results, the target test taker information and the data analysis results into corresponding positions in the early warning report template to obtain a filled early warning report template;

[0040] The filled early warning report template is used as the target early warning report.

[0041] In order to solve the above technical problems, the embodiment of the present application further provides a data analysis device based on artificial intelligence, which adopts the following technical solution:

[0042] A first acquisition module is used to acquire financial examination text data;

[0043] A processing module, configured to perform perturbation processing on the financial test text data based on a first pre-training model to obtain a preset number of enhanced text samples;

[0044] A first calculation module, configured to calculate the difference between the logarithmic probability function values ​​of the financial test text data and all the enhanced text samples based on a second pre-training model;

[0045] A second calculation module, used for calculating the smoothness feature of the financial test text data based on the difference of the logarithmic probability function value;

[0046] A comparison module, used to perform numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result;

[0047] The first generating module is used to generate a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result.

[0048] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0049] Get financial exam text data;

[0050] Performing perturbation processing on the financial test text data based on the first pre-trained model to obtain a preset number of enhanced text samples;

[0051] Calculate the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model;

[0052] Calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function values;

[0053] Performing numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result;

[0054] A cheating analysis result corresponding to the financial test text data is generated based on the numerical comparison result.

[0055] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0056] Get financial exam text data;

[0057] Performing perturbation processing on the financial test text data based on the first pre-trained model to obtain a preset number of enhanced text samples;

[0058] Calculate the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model;

[0059] Calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function values;

[0060] Performing numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result;

[0061] A cheating analysis result corresponding to the financial test text data is generated based on the numerical comparison result.

[0062] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0063] The present application first obtains financial test text data; then performs perturbation processing on the financial test text data based on a first pre-trained model to obtain a preset number of enhanced text samples; then calculates the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on a second pre-trained model; subsequently calculates the smoothness feature of the financial test text data based on the difference in logarithmic probability function values; further performs numerical comparison processing on the smoothness feature and a preset threshold to obtain a corresponding numerical comparison result; finally, generates a cheating analysis result corresponding to the financial test text data based on the numerical comparison result. The present application proposes to calculate the smoothness feature of financial examination text data based on the use of a first pre-trained model and a second pre-trained model, and then perform content analysis based on the numerical comparison result obtained by numerically comparing the smoothness feature with a preset threshold, and quickly, intelligently and accurately generate cheating analysis results corresponding to the financial examination text data based on the obtained content analysis results, so that there is no need to collect training data, clean data and label data, thereby saving a lot of manpower costs, effectively improving the processing efficiency of cheating detection, improving the efficiency of generating cheating analysis results, and avoiding the problem of difficulty in manually finding features, thereby ensuring the data accuracy of the generated cheating analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0065] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0066] Figure 2 A flowchart of an embodiment of an artificial intelligence-based data analysis method according to the present application;

[0067] Figure 3 is a schematic structural diagram of an embodiment of an artificial intelligence-based data analysis device according to the present application;

[0068] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0070] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0071] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0072] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0073] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0074] The terminal devices 101, 102, and 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, and the like.

[0075] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0076] It should be noted that the artificial intelligence-based data analysis method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the artificial intelligence-based data analysis device is generally set in the server / terminal device.

[0077] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0078] Continue to refer Figure 2 , showing a flow chart of an embodiment of an artificial intelligence-based data analysis method according to the present application. According to different requirements, the order of the steps in the flow chart can be changed, and some steps can be omitted. The artificial intelligence-based data analysis method provided in the embodiment of the present application can be applied to any scenario that requires data analysis of test data, and the artificial intelligence-based data analysis method can be applied to products in these scenarios, for example, cheating analysis of financial test data in the financial field. The artificial intelligence-based data analysis method comprises the following steps:

[0079] Step S201, obtaining financial examination text data.

[0080] In this embodiment, the data analysis method based on artificial intelligence is executed on the electronic device (e.g. Figure 1The server / terminal device shown in the figure) can obtain the financial examination text data through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection method may include but is not limited to 3G / 4G / 5G connection, Wi-Fi connection, Bluetooth connection, Wi MAX connection, Zigbee connection, UWB (ultra wi deband) connection, and other wireless connection methods currently known or to be developed in the future. The financial industry is a highly sensitive and complex industry, and the test results are directly related to the work ability and professional quality of practitioners in the industry. If there are a large number of cheating behaviors, it will lead to increased industry risks and may even cause systemic risks. Candidates who do not cheat will face the risk of being squeezed out by cheaters, unable to obtain the positions and opportunities they deserve, and fair competition will be undermined. Anti-cheating behavior can ensure fair competition and promote industry development. In the financial industry, financial practitioners need to obtain financial certificates to prove their professionalism. Among them, the above-mentioned financial examination text data may refer to the text form of examination data submitted by financial practitioners when obtaining financial certificates. Financial certificates are a series of qualifications that practitioners in the financial industry obtain in order to improve their professionalism and competitiveness. These certificates cover many aspects of the financial field, including investment, risk management, financial analysis, etc. Financial certificates may include: securities practitioner qualification certificate, fund practitioner qualification certificate, futures practitioner qualification certificate, banking practitioner qualification certificate, certified public accountant (CPA), chartered financial analyst (CFA), financial risk manager (FRM), etc.

[0081] Step S202: performing perturbation processing on the financial test text data based on the first pre-training model to obtain a preset number of enhanced text samples.

[0082] In this embodiment, the above-mentioned disturbance processing refers to text disturbance. Text disturbance generally refers to the introduction of a certain degree of noise or change in the original text to simulate text changes or errors that may occur in the real world, or to test the robustness and generalization ability of the model. In the case of data scarcity, more training samples can be generated by perturbing the original data. In text generation, different text samples can be generated according to different disturbance types and degrees, such as simulating spelling errors or grammatical errors by replacing certain words or sentences in the text. Among them, the above-mentioned specific implementation process of perturbation processing of the financial test text data based on the first pre-training model to obtain a preset number of enhanced text samples will be further described in detail in the subsequent specific embodiments of this application, and will not be elaborated on here.

[0083] Step S203, calculating the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model.

[0084] In this embodiment, the specific implementation process of calculating the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model will be further described in detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0085] Step S204: Calculate the smoothness feature of the financial test text data based on the difference between the logarithmic probability function values.

[0086] In this embodiment, the specific implementation process of calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function value will be further described in detail in subsequent specific embodiments of the present application, and will not be elaborated on here.

[0087] Step S205 , performing numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result.

[0088] In this embodiment, there is no specific limitation on the value of the above-mentioned preset threshold, which can be set according to the actual cheating analysis needs. The way in which this application detects whether there is a cheating problem in the financial examination text data is to detect whether the text is generated by AI, and proposes a detection method based on the "unique hypothesis". The unique hypothesis means that the text samples generated by the source model are usually smoother. This application uses the smoothness of the text to detect whether the text is generated by AI. Therefore, a preset threshold ∈ is set. If the smoothness feature C of the financial examination text data is X If the smoothness feature C of the financial examination text data is greater than this threshold ∈, then the financial examination text data is determined to be generated by AI and is determined to be cheating data. X If it is less than this threshold ∈, it is determined that the financial test text data is not generated by AI but written by humans, and the financial test text data is determined to be normal data.

[0089] Step S206: generating a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result.

[0090] In this embodiment, the specific implementation process of generating the cheating analysis result corresponding to the financial examination text data based on the numerical comparison result will be described in further detail in subsequent specific embodiments of the present application, and will not be elaborated on here.

[0091] The present application first obtains financial test text data; then performs perturbation processing on the financial test text data based on a first pre-trained model to obtain a preset number of enhanced text samples; then calculates the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on a second pre-trained model; subsequently calculates the smoothness feature of the financial test text data based on the difference in logarithmic probability function values; further performs numerical comparison processing on the smoothness feature and a preset threshold to obtain a corresponding numerical comparison result; finally, generates a cheating analysis result corresponding to the financial test text data based on the numerical comparison result. The present application proposes to calculate the smoothness feature of financial examination text data based on the use of a first pre-trained model and a second pre-trained model, and then perform content analysis based on the numerical comparison result obtained by numerically comparing the smoothness feature with a preset threshold, and quickly, intelligently and accurately generate cheating analysis results corresponding to the financial examination text data based on the obtained content analysis results, so that there is no need to collect training data, clean data and label data, thereby saving a lot of manpower costs, effectively improving the processing efficiency of cheating detection, improving the efficiency of generating cheating analysis results, and avoiding the problem of difficulty in manually finding features, thereby ensuring the data accuracy of the generated cheating analysis results.

[0092] In some optional implementations, step S202 includes the following steps:

[0093] Load the first pre-trained model into memory.

[0094] In this embodiment, the first pre-trained model is a pre-trained model. There is no specific limitation on the selection of the first pre-trained model, and it can be set according to the actual business usage requirements. The present application can specifically use the Transformer model as the first pre-trained model. Transformer is a model that uses the attention mechanism to increase the speed of model training. Because of its suitability for parallel computing and the complexity of its own model, it is higher in accuracy and performance than the previously popular recurrent neural network. The standard Transformer structure is an encoder-decoder architecture, in which both the encoder and the decoder have an encoding layer and a plurality of identical Transformer module layers stacked together. Among them, the above-mentioned memory refers to the memory of an electronic device.

[0095] Get the preset quantity.

[0096] In this embodiment, there is no specific limitation on the value of the above preset number, which can be selected according to actual usage requirements, and K is used to represent the preset number.

[0097] The financial test text data is perturbed based on the first pre-trained model in the memory to obtain enhanced text samples matching the preset quantity.

[0098] In this embodiment, the financial test text data is perturbed by using the first pre-trained model in the memory, that is, the financial test text data is perturbed K times, and K enhanced text samples similar to the financial test text data are obtained. In the case of data scarcity, more training samples can be generated by perturbing the original data. In text generation, different text samples can be generated according to different perturbation types and degrees, for example, by replacing certain words or sentences in the text to simulate spelling errors or grammatical errors. The specific perturbation process is shown as follows:

[0099]

[0100] Where F(·) represents the first pre-trained model, N(0,σ 2 ) represents the perturbation function, which is a Gaussian formula, X represents the financial examination text data, In addition, by loading the first pre-trained model into the memory and selecting one or more layers of the first pre-trained model to perturb the financial test text data, the processing efficiency of the perturbation processing of the financial test text data can be effectively improved.

[0101] The present application loads the first pre-trained model into the memory; then obtains a preset number; and subsequently perturbs the financial test text data based on the first pre-trained model in the memory to obtain enhanced text samples that match the preset number. The present application uses the first pre-trained model loaded into the memory to perturb the financial test text data, so as to quickly generate enhanced text samples that match the preset number, thereby improving the generation efficiency and generation intelligence of the enhanced text samples.

[0102] In some optional implementations of this embodiment, step S203 includes the following steps:

[0103] The mean of the logarithmic probabilities of all the enhanced text samples is calculated based on the second pre-training model.

[0104] In this embodiment, the second pre-trained model is a pre-trained model. The selection of the second pre-trained model is not specifically limited and can be set according to the actual business use requirements. The present application can specifically use the ChatGPT model as the second pre-trained model. ChatGPT is based on GPT (Generated Pre-trained Transformer) technology and is a series of generative pre-trained language models launched by OpenAI. The GPT model adopts the Transformer architecture, which is a self-attention mechanism (Self-Attention Mechanism), which is good at processing sequence data, especially in understanding and generating natural language. The working principle of ChatGPT: ChatGPT learns the laws, structures and contextual relationships of language in Internet text data through large-scale unsupervised learning. In the pre-training stage, the model learns the basic patterns of language, such as the co-occurrence probability of vocabulary, syntactic structure and semantic relationship. For specific tasks (such as dialogue generation), the pre-trained model will be fine-tuned to enable it to better understand and generate dialogue content and provide natural, fluent and context-related replies. Specifically, the formula for calculating the mean of the logarithmic probabilities of all the enhanced text samples is expressed as follows:

[0105]

[0106] Wherein, μ represents the mean of the logarithmic probability of all the enhanced text samples, represents an enhanced text sample, and K represents the number of enhanced text samples, that is, the preset number.

[0107] Calculate a specified log probability of the financial examination text data.

[0108] In this embodiment, the formula for calculating the specified logarithmic probability of the financial test text data is expressed as follows: logP θ (X), where X is the financial examination text data.

[0109] The difference between the specified log probability and the mean of the log probabilities of all the enhanced text samples is calculated.

[0110] In this embodiment, the formula for calculating the difference between the specified logarithmic probability and the mean of the logarithmic probabilities of all the enhanced text samples is expressed as follows: X =ogP θ (X)-, where d X represents the difference between the specified logarithmic probability and the mean of the logarithmic probabilities of all the enhanced text samples, logP θ(X) represents the specified logarithmic probability of the financial test text data, and u represents the mean of the logarithmic probabilities of all the enhanced text samples.

[0111] The difference is taken as the difference between the log probability function values.

[0112] The present application calculates the mean of the logarithmic probabilities of all the enhanced text samples based on the second pre-training model; then calculates the specified logarithmic probability of the financial test text data; then calculates the difference between the specified logarithmic probability and the mean of the logarithmic probabilities of all the enhanced text samples; and subsequently uses the difference as the difference in the logarithmic probability function value. The present application calculates the mean of the logarithmic probabilities of all the enhanced text samples by using the second pre-training model, and calculates the specified logarithmic probability of the financial test text data, and then calculates the difference between the specified logarithmic probability and the mean of the logarithmic probability of all the enhanced text samples, so as to achieve rapid and accurate generation of the required difference in the logarithmic probability function value between the financial test text data and all the enhanced text samples, thereby improving the generation efficiency and generation intelligence of the difference in the logarithmic probability function value, and ensuring the data accuracy of the difference in the logarithmic probability function value obtained.

[0113] In some optional implementations, step S204 includes the following steps:

[0114] Calculate the variance of the log probability of all the augmented text samples.

[0115] In this embodiment, the formula for calculating the variance of the logarithmic probability of all the enhanced text samples is expressed as follows:

[0116]

[0117] in, represents the variance of the logarithmic probability of all the enhanced text samples, μ represents the mean of the logarithmic probability of all the enhanced text samples, represents an enhanced text sample, and K represents the number of enhanced text samples, that is, the preset number.

[0118] Call the preset feature calculation formula.

[0119] In this embodiment, the above characteristic calculation formula is expressed as follows:

[0120]

[0121] Among them, C X represents the curvature of the logarithmic probability of the financial test text data, that is, the smoothness feature of the financial test text data, d Xrepresents the difference in logarithmic probability function value between the financial test text data and all the enhanced text samples, represents the variance of the log probability of all the enhanced text samples.

[0122] Based on the feature calculation formula, the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value is calculated to obtain a corresponding calculation result.

[0123] In this embodiment, the feature calculation formula can be calculated based on the variance of the logarithmic probability of all the enhanced text samples. , and then The difference between the value of and the value of the logarithmic probability function, that is, d X Substitute the value of into the characteristic calculation formula for calculation, that is, calculate d X The value of The quotient of the value of is obtained to obtain the corresponding calculation result.

[0124] The calculation result is used as the smoothness feature of the financial test text data.

[0125] The present application calculates the variance of the logarithmic probability of all the enhanced text samples; then calls a preset feature calculation formula; then, based on the feature calculation formula, calculates and processes the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value to obtain a corresponding calculation result; and subsequently uses the calculation result as the smoothness feature of the financial test text data. The present application calculates and processes the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value based on the use of a feature calculation formula, so as to quickly and accurately obtain the smoothness feature of the financial test text data, improve the generation efficiency of the smoothness feature of the financial test text data, and ensure the data accuracy of the smoothness feature of the obtained financial test text data.

[0126] In some optional implementations, step S206 includes the following steps:

[0127] It is determined whether the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold.

[0128] In this embodiment, the content of the above numerical comparison result includes that the smoothness feature is greater than the preset threshold, or the smoothness feature is less than the preset threshold.

[0129] If the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold, a first cheating analysis result is generated that the financial test text data is cheating data.

[0130] In this embodiment, if the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold, it will be determined that the financial examination text data is generated by AI, and a first cheating analysis result will be generated that the financial examination text data is cheating data.

[0131] If the content of the numerical comparison result is that the smoothness feature is less than the preset threshold, a second cheating analysis result is generated that the financial test text data is normal data.

[0132] In this embodiment, if the content of the numerical comparison result is that the smoothness feature is less than the preset threshold, it will be determined that the financial test text data is artificially written, and a second cheating analysis result that the financial test text data is normal data will be generated.

[0133] The present application determines whether the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold; if the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold, a first cheating analysis result is generated that the financial test text data is cheating data; and if the content of the numerical comparison result is that the smoothness feature is less than the preset threshold, a second cheating analysis result is generated that the financial test text data is normal data. The present application performs content analysis on the numerical comparison result obtained by numerically comparing the smoothness feature with the preset threshold, and then can quickly, intelligently and accurately generate cheating analysis results corresponding to the financial test text data based on the obtained content analysis results, effectively improving the efficiency of generating cheating analysis results and ensuring the data accuracy of the generated cheating analysis results.

[0134] In some optional implementations of this embodiment, after step S206, the electronic device may further perform the following steps:

[0135] If the cheating analysis result indicates that the financial examination text data is cheating data, a target warning report corresponding to the financial examination text data is generated based on the cheating analysis result.

[0136] In this embodiment, the specific implementation process of generating the target warning report corresponding to the financial examination text data based on the cheating analysis results will be described in further detail in subsequent specific embodiments of the present application, and will not be elaborated on here.

[0137] Get reviewers' communication information.

[0138] In this embodiment, the reviewer refers to a person who is responsible for reviewing the financial test text data. The communication information of the reviewer may include the reviewer's telephone number, email address, and other information.

[0139] Based on the communication information, the target early warning report is sent to the reviewer.

[0140] In this embodiment, after the communication information of the reviewer is obtained, the target early warning report can be sent to the communication terminal of the reviewer corresponding to the communication information.

[0141] If the present application detects that the content of the cheating analysis result is that the financial test text data is cheating data, then a target warning report corresponding to the financial test text data is generated based on the cheating analysis result; then the communication information of the reviewer is obtained; and subsequently, the target warning report is sent to the reviewer based on the communication information. When the present application detects that the content of the cheating analysis result is that the financial test text data is cheating data, it will also intelligently generate a target warning report corresponding to the financial test text data based on the cheating analysis result, and will send the target warning report to the reviewer according to the communication information of the reviewer, so that the reviewer can timely check the relevant information of the financial test text data with cheating behavior according to the received target warning report, and then can timely take corresponding treatment measures for the cheating behavior corresponding to the financial test text data, which is conducive to improving the work efficiency and work experience of the reviewer. In addition, by automatically and quickly generating a target warning report corresponding to the financial test text data according to the cheating analysis result, there is no need to manually write a report, which improves the generation efficiency of the target warning report.

[0142] In some optional implementations of this embodiment, generating a target warning report corresponding to the financial examination text data based on the cheating analysis result includes the following steps:

[0143] Target test taker information corresponding to the financial test text data is obtained.

[0144] In this embodiment, the target examinee information corresponding to the financial examination text data can be queried from the examination data management system by querying the preset examination data management system, wherein the examination data management system is a pre-constructed database system including the personal information of the examinees and the association relationship between the personal information of the examinees and the financial examination text data.

[0145] A data analysis result corresponding to the financial examination text data is obtained.

[0146] In this embodiment, when a cheating analysis result corresponding to the financial test text data is generated based on the numerical comparison result, a data analysis result matching the cheating analysis result is also generated and recorded. Specifically, if the content of the cheating analysis result is that the financial test text data is cheating data, then the data analysis result corresponding to the cheating analysis result is that the slippage feature of the financial test text data is greater than a preset threshold; and if the content of the cheating analysis result is that the financial test text data is normal data, then the data analysis result corresponding to the cheating analysis result is that the slippage feature of the financial test text data is less than a preset threshold.

[0147] Call the preset early warning report template.

[0148] In this embodiment, the warning report template is a report template file pre-built according to the requirements of exam cheating warning, including areas to be filled that match multiple fields. Specifically, the warning report template includes at least a financial exam text field, a cheating analysis result field, an examinee information field, and a data analysis result field.

[0149] The financial test text data, the cheating analysis results, the target test taker information and the data analysis results are filled into corresponding positions in the early warning report template to obtain a filled early warning report template.

[0150] In this embodiment, the data filling of the warning report template can be completed and the filled warning report template can be obtained by filling the financial test text data into the position in the warning report template that matches the area to be filled in the financial test text field, filling the cheating analysis result into the position in the warning report template that matches the area to be filled in the cheating analysis result field, filling the target test taker information into the position in the warning report template that matches the area to be filled in the test taker information field, and filling the data analysis result into the position in the warning report template that matches the area to be filled in the data analysis result field.

[0151] The filled early warning report template is used as the target early warning report.

[0152] This application obtains the target test taker information corresponding to the financial test text data; then obtains the data analysis results corresponding to the financial test text data; then calls a preset early warning report template; subsequently fills the financial test text data, the cheating analysis results, the target test taker information and the data analysis results into the corresponding positions in the early warning report template to obtain a filled early warning report template, and uses the filled early warning report template as the target early warning report. This application uses the financial test text data, the target test taker information corresponding to the financial test text data, the data analysis results and the cheating analysis results to fill the preset early warning report template, so as to realize the rapid and intelligent generation of a target early warning report that matches the financial test text data, so that there is no need to manually write the report, which improves the generation efficiency and generation intelligence of the target early warning report and ensures the richness of the content of the obtained target early warning report.

[0153] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0154] It should be emphasized that in order to further ensure the privacy and security of the above cheating analysis results, the above cheating analysis results can also be stored in a node of a blockchain.

[0155] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0156] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0157] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0158] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0159] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0160] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a data analysis device based on artificial intelligence. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0161] like Figure 3 As shown, the artificial intelligence-based data analysis device 300 described in this embodiment includes: a first acquisition module 301, a processing module 302, a first calculation module 303, a second calculation module 304, a comparison module 305 and a first generation module 306. Among them:

[0162] The first acquisition module 301 is used to acquire financial examination text data;

[0163] A processing module 302 is used to perform perturbation processing on the financial test text data based on the first pre-training model to obtain a preset number of enhanced text samples;

[0164] A first calculation module 303, configured to calculate the difference between the logarithmic probability function values ​​of the financial test text data and all the enhanced text samples based on a second pre-training model;

[0165] A second calculation module 304 is used to calculate the smoothness feature of the financial test text data based on the difference of the logarithmic probability function value;

[0166] A comparison module 305 is used to perform a numerical comparison process on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result;

[0167] The first generating module 306 is used to generate a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result.

[0168] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0169] In some optional implementations of this embodiment, the processing module 302 includes:

[0170] A loading submodule, used to load the first pre-trained model into the memory;

[0171] A first acquisition submodule is used to acquire a preset quantity;

[0172] The processing submodule is used to perform perturbation processing on the financial test text data based on the first pre-trained model in the memory to obtain enhanced text samples matching the preset number.

[0173] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0174] In some optional implementations of this embodiment, the first calculation module 303 includes:

[0175] A first calculation submodule, used for calculating the mean of the logarithmic probabilities of all the enhanced text samples based on the second pre-training model;

[0176] A second calculation submodule, used to calculate the specified logarithmic probability of the financial examination text data;

[0177] A third calculation submodule, used for calculating the difference between the specified logarithmic probability and the mean of the logarithmic probabilities of all the enhanced text samples;

[0178] The first determination submodule is configured to use the difference as the difference between the logarithmic probability function values.

[0179] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0180] In some optional implementations of this embodiment, the second calculation module 304 includes:

[0181] A fourth calculation submodule, used for calculating the variance of the logarithmic probability of all the enhanced text samples;

[0182] The first calling submodule is used to call a preset feature calculation formula;

[0183] A fifth calculation submodule, configured to calculate the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value based on the feature calculation formula to obtain a corresponding calculation result;

[0184] The second determination submodule is used to use the calculation result as the smoothness feature of the financial examination text data.

[0185] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0186] In some optional implementations of this embodiment, the first generating module 306 includes:

[0187] A judging submodule, used for judging whether the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold;

[0188] A first generating submodule, configured to generate a first cheating analysis result indicating that the financial examination text data is cheating data if the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold value;

[0189] The second generating submodule is used to generate a second cheating analysis result indicating that the financial test text data is normal data if the content of the numerical comparison result is that the smoothness feature is less than the preset threshold.

[0190] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0191] In some optional implementations of this embodiment, the artificial intelligence-based data analysis device further includes:

[0192] A second generating module, configured to generate a target warning report corresponding to the financial examination text data based on the cheating analysis result if the content of the cheating analysis result is that the financial examination text data is cheating data;

[0193] The second acquisition module is used to obtain the communication information of the reviewers;

[0194] A sending module is used to send the target early warning report to the reviewer based on the communication information.

[0195] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0196] In some optional implementations of this embodiment, the second generating module includes:

[0197] A second acquisition submodule is used to acquire target test taker information corresponding to the financial test text data;

[0198] A third acquisition submodule is used to acquire data analysis results corresponding to the financial examination text data;

[0199] The second calling submodule is used to call a preset early warning report template;

[0200] A filling submodule, used to fill the financial test text data, the cheating analysis results, the target test taker information and the data analysis results into corresponding positions in the early warning report template to obtain a filled early warning report template;

[0201] The third determination submodule is used to use the filled warning report template as the target warning report.

[0202] In this embodiment, the operations performed by the above modules or units respectively correspond one-to-one to the steps of the artificial intelligence-based data analysis method in the aforementioned implementation mode, and will not be repeated here.

[0203] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0204] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital processors (DSPs), embedded devices, etc.

[0205] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0206] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Med ia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of data analysis methods based on artificial intelligence, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0207] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions stored in the memory 41 or process data, such as running the computer-readable instructions of the artificial intelligence-based data analysis method.

[0208] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0209] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the artificial intelligence-based data analysis method as described above.

[0210] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0211] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A data analysis method based on artificial intelligence, characterized in that: The steps include: Get financial exam text data; Performing perturbation processing on the financial test text data based on the first pre-trained model to obtain a preset number of enhanced text samples; Calculate the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model; Calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function values; Performing numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result; A cheating analysis result corresponding to the financial test text data is generated based on the numerical comparison result.

2. The artificial intelligence-based data analysis method according to claim 1, characterized in that: The step of performing perturbation processing on the financial test text data based on the first pre-training model to obtain a preset number of enhanced text samples specifically includes: Loading the first pre-trained model into memory; Get the preset quantity; The financial test text data is perturbed based on the first pre-trained model in the memory to obtain enhanced text samples matching the preset quantity.

3. The artificial intelligence-based data analysis method according to claim 1, characterized in that: The step of calculating the difference in logarithmic probability function values ​​between the financial test text data and all the enhanced text samples based on the second pre-training model specifically includes: Calculating the mean of the logarithmic probabilities of all the enhanced text samples based on the second pre-training model; Calculating a specified logarithmic probability of the financial examination text data; Calculating the difference between the specified log probability and the mean of the log probabilities of all the enhanced text samples; The difference is taken as the difference between the log probability function values.

4. The artificial intelligence-based data analysis method according to claim 1, characterized in that: The step of calculating the smoothness feature of the financial test text data based on the difference in the logarithmic probability function value specifically includes: Calculating the variance of the log probability of all the enhanced text samples; Call the preset feature calculation formula; Based on the feature calculation formula, the difference between the variance of the logarithmic probability of all the enhanced text samples and the logarithmic probability function value is calculated to obtain a corresponding calculation result; The calculation result is used as the smoothness feature of the financial test text data.

5. The artificial intelligence-based data analysis method according to claim 1, characterized in that: The step of generating a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result specifically includes: Determining whether the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold; If the content of the numerical comparison result is that the smoothness feature is greater than the preset threshold, a first cheating analysis result is generated indicating that the financial test text data is cheating data; If the content of the numerical comparison result is that the smoothness feature is less than the preset threshold, a second cheating analysis result is generated that the financial test text data is normal data.

6. The artificial intelligence-based data analysis method according to claim 1, characterized in that: After the step of generating a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result, the method further includes: If the cheating analysis result indicates that the financial examination text data is cheating data, generating a target warning report corresponding to the financial examination text data based on the cheating analysis result; Obtaining communication information of reviewers; Based on the communication information, the target early warning report is sent to the reviewer.

7. The artificial intelligence-based data analysis method according to claim 6, characterized in that: The step of generating a target warning report corresponding to the financial examination text data based on the cheating analysis result specifically includes: Acquire target test taker information corresponding to the financial test text data; Obtaining data analysis results corresponding to the financial examination text data; Call the preset early warning report template; Fill the financial test text data, the cheating analysis results, the target test taker information and the data analysis results into corresponding positions in the early warning report template to obtain a filled early warning report template; The filled early warning report template is used as the target early warning report.

8. A data analysis device based on artificial intelligence, characterized in that: include: A first acquisition module is used to acquire financial examination text data; A processing module, configured to perform perturbation processing on the financial test text data based on a first pre-training model to obtain a preset number of enhanced text samples; A first calculation module, configured to calculate the difference between the logarithmic probability function values ​​of the financial test text data and all the enhanced text samples based on a second pre-training model; A second calculation module, used for calculating the smoothness feature of the financial test text data based on the difference of the logarithmic probability function value; A comparison module, used to perform numerical comparison processing on the smoothness feature and a preset threshold value to obtain a corresponding numerical comparison result; The first generating module is used to generate a cheating analysis result corresponding to the financial examination text data based on the numerical comparison result.

9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the artificial intelligence-based data analysis method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the artificial intelligence-based data analysis method as described in any one of claims 1 to 7.