Transaction risk intelligent prevention and control method and device based on emotion recognition

Abstract

The application discloses a transaction risk intelligent prevention and control method and device based on emotion recognition, equipment and medium, the method comprises the following steps: judging whether the transaction request of a user meets the identification condition, if yes, obtaining video information of the user answering a preset question, identifying the user image in the video information to obtain an expression recognition result; identifying the user voice in the video information to obtain a voice recognition result, comprehensively analyzing the expression recognition result and the voice recognition result according to an emotion comprehensive analysis model to obtain an identification result of whether there is a transaction risk, and if there is a transaction risk, intercepting the transaction request. The transaction risk intelligent prevention and control method based on emotion recognition can identify the emotion of a user by comprehensively analyzing the user image and the user voice, analyze the expression and language of the user to determine whether the emotion of the user is stable, and obtain the identification result of whether there is a transaction risk; and the transaction request with a risk is intercepted, so that the security of the user transfer transaction is improved.

Description

Technical Field

[0001] This invention relates to the field of smart financial technology, and in particular to a method and device for intelligent prevention and control of transaction risks based on emotion recognition. Background Technology

[0002] Financial institutions often face requests for large-sum transfers from users. However, some of these transactions carry inherent risks, such as users being asked to transfer money after receiving calls from unknown numbers. While financial institution staff typically question and verify the account details before the transfer, they often fail to effectively identify the risks associated with such transactions, leading to financial losses for the user after the transfer is completed. Therefore, existing technological methods suffer from an inability to accurately identify the risks associated with user transfer transactions. Summary of the Invention

[0003] This invention provides a method and apparatus for intelligent risk prevention and control of transactions based on emotion recognition, aiming to solve the problem that existing methods cannot accurately identify the risks of user transfer transactions.

[0004] In a first aspect, embodiments of the present invention provide a method for intelligent prevention and control of transaction risks based on emotion recognition, wherein the method includes:

[0005] If a transaction request is received from the user, it is determined whether the transaction request meets the preset identification conditions;

[0006] If the transaction request meets the identification conditions, obtain video information of the user answering preset questions;

[0007] The user image in the video information is identified according to the preset face recognition strategy to obtain the corresponding expression recognition result;

[0008] The user's voice in the video information is identified according to the preset voice risk recognition strategy to obtain the corresponding voice recognition result;

[0009] The facial expression recognition results and the speech recognition results are comprehensively analyzed based on a pre-set emotion comprehensive analysis model to obtain an analysis result on whether there is a transaction risk.

[0010] If the analysis results indicate that there is a risk, the transaction request will be blocked.

[0011] Secondly, embodiments of the present invention also provide a transaction risk intelligent prevention and control device based on emotion recognition, wherein the device is used to execute the transaction risk intelligent prevention and control method based on emotion recognition as described in the first aspect above, and the device includes:

[0012] The judgment unit is used to determine whether a transaction request input by a user meets preset recognition conditions if a transaction request is received.

[0013] The video information acquisition unit is used to acquire video information of the user answering preset questions if the transaction request meets the identification conditions.

[0014] The facial expression recognition result acquisition unit is used to recognize the user image in the video information according to a preset face recognition strategy, and obtain the corresponding facial expression recognition result;

[0015] The speech recognition result acquisition unit is used to recognize the user's speech in the video information according to a preset speech risk recognition strategy, and obtain the corresponding speech recognition result;

[0016] The comprehensive analysis unit is used to perform a comprehensive analysis of the facial expression recognition results and the speech recognition results based on a preset emotion comprehensive analysis model, so as to obtain an analysis result on whether there is a transaction risk.

[0017] The request interception unit is used to intercept the transaction request if the analysis result indicates that there is a risk.

[0018] Thirdly, embodiments of the present invention also provide a computer device, wherein the device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0019] Memory, used to store computer programs;

[0020] When the processor executes the program stored in the memory, it implements the steps of the intelligent transaction risk prevention and control method based on emotion recognition described in the first aspect above.

[0021] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the intelligent transaction risk prevention and control method based on emotion recognition as described in the first aspect above.

[0022] This invention provides a method, apparatus, device, and medium for intelligent transaction risk prevention and control based on emotion recognition. The method includes: determining whether a user's transaction request meets recognition conditions; if so, acquiring video information of the user answering preset questions; recognizing the user's image in the video information to obtain an expression recognition result; recognizing the user's voice in the video information to obtain a voice recognition result; and comprehensively analyzing the expression recognition result and voice recognition result according to an emotion comprehensive analysis model to obtain an identification result indicating whether a transaction risk exists. If a transaction risk exists, the transaction request is intercepted. This method for intelligent transaction risk prevention and control based on emotion recognition, by comprehensively recognizing the user's emotions through user images and voice, analyzing the user's expressions and language to determine whether the user's emotions are stable to obtain an identification result indicating whether a transaction risk exists, and intercepting risky transaction requests, improves the security of user transfer transactions. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 A flowchart illustrating the intelligent risk prevention and control method for transactions based on emotion recognition provided in this embodiment of the invention;

[0025] Figure 2 A schematic block diagram of a transaction risk intelligent prevention and control device based on emotion recognition provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0029] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0030] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0031] This invention provides an embodiment of an intelligent transaction risk control method based on emotion recognition. This method is applied to a management server, which executes stored software programs to implement the aforementioned intelligent transaction risk control method based on emotion recognition. The management server is the server-side component configured within a financial institution for intelligent transaction risk control; it can be a cluster server. The management server can process user transaction requests. Users or financial institution staff can submit corresponding transaction requests to the management server through transaction terminals, such as desktop computers, laptops, or automated teller machines (ATMs).

[0032] like Figure 1 As shown, the method includes steps S110 to S160.

[0033] S110. If a transaction request is received from the user, determine whether the transaction request meets the preset identification conditions.

[0034] The management server can receive transaction requests from users, whether submitted by the user or by staff. The transaction information (such as transaction amount and password) is entered by the user, thus the transaction request corresponds to a user. Upon receiving a transaction request, the management server determines whether it meets pre-set identification conditions. Specifically, these conditions can include setting a supervising bank and a transaction amount threshold. The server then determines whether the receiving account in the request belongs to a supervising bank (such as an overseas bank or a non-state-owned bank) or whether the transaction amount exceeds the threshold. If the receiving account belongs to a supervising bank or the transaction amount exceeds the threshold, the transaction request is deemed to meet the identification conditions; otherwise, it is deemed not to meet the conditions, and the transaction request is submitted directly.

[0035] S120. If the transaction request meets the identification conditions, obtain video information of the user answering preset questions.

[0036] If the transaction request meets the recognition criteria, a preset question can be played, and the user answers it. The image and audio of the user's answers to the preset question are captured to obtain corresponding video information, which includes the user's image and audio. The preset question could be something like "Describe the purpose of your transfer" or "Describe the recipient information of your transfer."

[0037] S130. The user image in the video information is identified according to the preset face recognition strategy to obtain the corresponding expression recognition result.

[0038] The facial recognition strategy is used to identify user images contained in video information, and the corresponding facial expression recognition results are obtained by recognizing user facial expression information.

[0039] In a specific embodiment, step S130 includes the following sub-steps: extracting the corresponding user facial features from the user image according to the facial feature extraction rules in the facial recognition strategy; and recognizing the user facial features according to the feature analysis model in the facial recognition strategy to obtain the corresponding expression recognition result.

[0040] Specifically, user facial features can be extracted from user images according to the facial feature extraction rules in the facial recognition strategy. User images can be extracted from continuous video frames according to the extraction interval set in the facial feature extraction rules. For example, if the extraction interval is set to 0.2 seconds, one image is extracted from the continuous video frames every 0.2 seconds as the user image. Based on the portrait template in the portrait feature extraction rules, a local portrait corresponding to the portrait template is extracted from the user image. The background information of the user image is removed from the local portrait, resulting in a local portrait containing only facial information. Therefore, one local portrait can be extracted from each user image. Further, each local portrait is convolved using a convolutional neural network according to the facial feature extraction specifications, thereby converting the information contained in the local portrait into a one-dimensional feature vector. For example, a 600×800 resolution local portrait is converted into a 1×512 one-dimensional feature vector using a convolutional neural network. Each local portrait can be converted into a set of one-dimensional feature vectors, and the resulting multiple one-dimensional feature vectors are combined to form the user's facial features. Furthermore, the feature analysis model in the face recognition strategy can be used to comprehensively identify the user's facial features, thereby obtaining the expression recognition result.

[0041] In a specific embodiment, the step of identifying whether the user's facial features have risks based on the feature analysis model in the facial recognition strategy to obtain the corresponding expression recognition result includes: inputting the user's facial features into the long short-term memory network in the feature analysis model to obtain the expression feature vector corresponding to the user's facial features; calculating the similarity value between the expression feature vector and the preset standard feature vector to obtain the corresponding expression recognition result.

[0042] The obtained user facial features can be input into the Long Short-Term Memory Network (LSTM) configured in the feature analysis model. LSTM is a type of Recurrent Neural Network (RNN). Based on LSTM, user facial features containing any number of one-dimensional feature vectors can be converted into fixed-length vectors. The fixed-length vector output by the LSTM is then the expression feature vector. The process of the LSTM processing the input one-dimensional feature vector includes: ① Calculating the forget gate output information: f (t) =σ(W f ×h (t_1) +U f ×X (t)+ b f ), where f (t) Forget gate parameter values, 0 ≤ f (t) ≤1; σ is the activation function calculation symbol, and σ can be specifically represented as Then W f ×h (t_1) +U f ×X (t)+ b f The calculation result can be used as the input x to activate the function σ to obtain f. (t) W f U f and b f All of these are parameter values ​​for the formula in this cell; h (t_1) This is the output gate information of the previous cell; X (t) Given the one-dimensional feature vector (1×M-dimensional vector) of the current cell as input, if the current cell is the first cell in the Long Short-Term Memory network, then h (t_1) ① Set to "0". ② Calculate the input gate information: i (t) =σ(W i ×h (t_1) +U i ×X (t) +b i );a (t) =tanh(W a ×h (t-1) +U a ×X (t) +ba ), where i (t) The input gate parameter values ​​are 0 ≤ i (t) ≤1; W i U i b i W a U a and b a All of these are parameter values ​​for the formula in this cell, a (t) Let a be the calculated input gate vector value. (t) It is a 1×M dimensional vector. ③ Update cellular memory information: C (t) =C (t_1) ⊙f (t) +i (t) ⊙a (t) C represents the cellular memory information accumulated in each calculation process. (t) C represents the cellular memory information currently output by the cell. (t_1) The cell memory information output by the previous cell, ⊙ is the vector operator, C (t_1) ⊙f (t) The calculation process is to convert vector C (t_1) Each dimension value is related to f. (t) Multiplying, the calculated vector dimension is the same as that of vector C. (t_1) The dimensions are the same. ④ Calculate the output gate information: o (t) =σ(W o ×h (t_1) +U o ×X (t) +b o );h (t) =o (t) ⊙tanh(C (t) ), o (t) The output gate parameter value, 0 ≤ o (t) ≤1; W o U o and b o All of these are parameter values ​​for the formula in this cell, h (t) For the output gate information of this cell, h (t) Let h be a 1×M dimensional vector. (t) As the current output, it is also passed to the next time step. ⑤ Input each one-dimensional feature vector from the user's facial features sequentially into the Long Short-Term Memory (LSTM) network. Each input of a one-dimensional feature vector updates the cell memory information in the LTM network. After all one-dimensional feature vectors have been input, calculate the output information of each cell sequentially: y (t) =σ(V×h) (t)+c), where V and c are parameter values ​​of the formula in this cell. Each cell calculates an output value, and the combined output information of N cells yields the memory network output information of a feature information. The memory network output information of a feature information is a 1×N dimensional vector, which corresponds to the expression feature vector. Therefore, the dimension of the obtained expression feature vector is not related to the number of one-dimensional feature vectors in the user's facial features, but only to the number of cells in the Long Short-Term Memory network.

[0043] The similarity value between the facial expression feature vector and the standard feature vector is calculated, where the standard feature vector is the feature vector corresponding to a local portrait of the user in a calm state. If the number of dimensions of the facial expression feature vector is equal to the number of dimensions of the standard feature vector, then the cosine similarity between the two vectors can be calculated and normalized to obtain the corresponding similarity value. For example, the numerical range of the cosine similarity is [-1, 1], and the numerical range of the normalized similarity value is [0, 1]. The Long Short-Term Memory (LSTM) network can be pre-trained before use. After pre-training, local portraits of multiple users in a calm state are collected and input into the LSM network to obtain the output vector for each user. The average value of the output vectors is calculated as the standard feature vector. A larger similarity value between the facial expression feature vector and the standard feature vector indicates a calmer and more stable facial expression; a smaller similarity value indicates a less calm facial expression, suggesting the user may be panicked or agitated.

[0044] S140. The user's voice in the video information is identified according to the preset voice risk recognition strategy to obtain the corresponding voice recognition result.

[0045] To further improve the accuracy of user emotion recognition, user voice in video information can be recognized based on voice analysis and recognition strategies. By recognizing user voice, corresponding voice recognition results can be obtained. By combining voice recognition results with facial expression recognition results, comprehensive recognition of user emotions can be achieved.

[0046] In a specific embodiment, step S140 includes the following sub-steps: extracting corresponding basic feature parameters from the user's speech according to the basic speech feature extraction rules of the speech risk recognition strategy; performing text recognition on the user's speech according to the text recognition model in the speech risk recognition strategy to obtain the corresponding text recognition result; and performing combined recognition on the text recognition result and the basic feature parameters according to the combined recognition model in the speech risk recognition strategy to obtain the corresponding speech recognition result.

[0047] Specifically, the speech analysis and recognition strategy is configured with basic speech feature extraction rules. Based on these rules, corresponding basic feature parameters can be extracted from the user's speech. According to the target frequency range set in the basic speech feature extraction rules, effective audio corresponding to the target frequency range is extracted from the user's speech. The number of peaks in the effective audio is extracted as the number of words, and the duration of the effective audio is extracted as the speaking duration. The ratio of the number of words to the speaking duration is calculated as the speech rate parameter. The average loudness of the effective audio is extracted as the loudness parameter. Then, the speech rate parameter and loudness parameter are obtained as the corresponding basic feature parameters.

[0048] Furthermore, based on the text recognition model, the valid audio corresponding to the user's speech is subjected to text recognition to convert the speech information into corresponding text information as the corresponding text recognition result. The text recognition model is also a recognition model that converts audio information into text information. The combined recognition model in the speech risk recognition strategy combines the text recognition result and basic feature parameters for recognition, thereby obtaining the speech recognition result.

[0049] In a specific embodiment, the step of combining the text recognition result and the basic feature parameters according to the combined recognition model in the speech risk recognition strategy to obtain the corresponding speech recognition result includes: performing disorder recognition on the basic feature parameters and the text recognition result according to the disorder recognition model in the combined recognition model to obtain the corresponding disorder; calculating the speech emotion coefficient between the basic feature parameters and the standard pronunciation feature parameters configured in the combined recognition model; and combining the disorder and the speech emotion coefficient to obtain the speech recognition result.

[0050] Obtaining speech recognition results specifically involves identifying the disorder level of basic feature parameters and text recognition results using a disorder level recognition model. Specifically, the speech rate parameter from the basic feature parameters can be combined with the text recognition result and input into the disorder level recognition model. The disorder level recognition model includes multiple input nodes, intermediate layers, and one output node. Each intermediate layer contains one or more intermediate nodes, and each layer consists of multiple linearly arranged intermediate nodes. The first input node inputs the speech rate parameter, and subsequent input nodes correspond to a single text character from the text recognition result. The output node outputs the corresponding disorder level. Disorder level can be used to represent the degree of disorder in a user's speech; a higher disorder level indicates a more disordered speech expression, while a lower disorder level indicates a clearer speech expression. For example, fast speech, repeated words (e.g., "I," "I think," etc.), and contradictory expressions are all considered to indicate speech disorder. Therefore, during training, multiple text segments with fast speech, repeated words, or contradictory expressions can be used as training data to train the disorder level recognition model.

[0051] If a user experiences psychological fluctuations during the inquiry process, their speech rate and volume will change. To quantify these changes, a corresponding voice emotion coefficient can be obtained. Specifically, the voice emotion coefficient between the basic feature parameters and the standard pronunciation feature parameters is calculated. The voice emotion coefficient reflects the degree of emotionality in the user's voice. A larger voice emotion coefficient indicates a higher-pitched voice, while a smaller voice emotion coefficient indicates a lower-pitched voice. For example, the formula for calculating the voice emotion coefficient can be expressed as formula (1):

[0052]

[0053] Q is the calculated voice emotion coefficient, Y s The speech rate parameter, Y, is one of the basic feature parameters. h The loudness parameter, B, is one of the basic characteristic parameters. s B is the standard speech rate parameter in the standard pronunciation feature parameters. h is the standard loudness parameter in the standard pronunciation characteristic parameters, and e is the base of the natural logarithm.

[0054] By combining the obtained disorder level with the speech emotion coefficient, the corresponding speech recognition result can be obtained.

[0055] S150. The facial expression recognition result and the speech recognition result are comprehensively analyzed according to the preset emotion comprehensive analysis model to obtain the analysis result of whether there is a transaction risk.

[0056] By comprehensively analyzing the results of facial expression recognition and speech recognition using an emotion analysis model, the analysis can integrate user facial expressions and speech to determine whether there is any risk in a transaction by identifying whether the user's emotions are abnormal. If the comprehensive analysis identifies abnormal user emotions, the result indicates that there is a transaction risk; if the comprehensive analysis shows that the user's emotions are normal, the result indicates that there is no transaction risk.

[0057] In a specific embodiment, step S150 includes the following sub-steps: performing a comprehensive calculation on the facial expression recognition result and the speech recognition result according to the comprehensive calculation rules in the emotion comprehensive analysis model to obtain the corresponding comprehensive emotion coefficient; determining whether the comprehensive emotion coefficient is within the coefficient range set by the emotion comprehensive analysis model to obtain the analysis result of whether there is a transaction risk.

[0058] The emotion comprehensive analysis model is configured with comprehensive calculation rules. These rules allow for the comprehensive calculation of facial expression recognition and speech recognition results to obtain the corresponding comprehensive emotion coefficient. The comprehensive emotion coefficient is a parameter value that characterizes whether a user's overall emotion is abnormal during the transfer process. A larger comprehensive emotion coefficient indicates a more agitated user, while a smaller coefficient indicates a depressed user. The comprehensive calculation rules can be expressed using formula (2):

[0059]

[0060] Where Z is the comprehensive emotion coefficient, Q is the speech emotion coefficient in the speech recognition result, the value of which is not less than zero, R is the disorder degree in the speech recognition result, the value range is [0,1], and X is the similarity value in the expression recognition result, the value range is [0,1].

[0061] Further determine whether the comprehensive sentiment coefficient is within the coefficient range set by the sentiment comprehensive analysis model. For example, the coefficient range can be set to [0.6, 1.4]. If it is within the coefficient range, it is determined that the user's sentiment is normal and the analysis result is that there is no trading risk. If it is not within the coefficient range, it is determined that the user's sentiment is abnormal and the analysis result is that there is trading risk.

[0062] S160. If the analysis result indicates that there is a risk, the transaction request shall be intercepted.

[0063] If the analysis indicates a risk, the management server will intercept the transaction request and not process it, causing the user's current transfer transaction to fail. By intercepting high-risk transaction requests, financial losses can be avoided due to emotional trading (such as trading under abnormal emotional states caused by fraud or coercion).

[0064] In a specific embodiment, after step S150, the method further includes the step of: if the analysis result indicates that there is no risk, submitting the transaction request.

[0065] If the analysis results indicate that there is no risk, the transaction request can be submitted directly for processing. Once the settlement center receives the transaction request, it will complete the corresponding transfer deduction.

[0066] The intelligent transaction risk prevention and control method based on emotion recognition disclosed in the above embodiments includes: determining whether a user's transaction request meets the recognition conditions; if so, acquiring video information of the user answering preset questions; recognizing the user image in the video information to obtain an expression recognition result; recognizing the user's voice in the video information to obtain a voice recognition result; and comprehensively analyzing the expression recognition result and voice recognition result according to an emotion comprehensive analysis model to obtain an identification result indicating whether a transaction risk exists; if a transaction risk exists, the transaction request is intercepted. This intelligent transaction risk prevention and control method based on emotion recognition identifies user emotions by comprehensively analyzing user images and user voice, analyzes user expressions and language to determine whether the user's emotions are stable to obtain an identification result indicating whether a transaction risk exists, and intercepts risky transaction requests, thereby improving the security of user transfer transactions.

[0067] This invention also provides an intelligent trading risk control device based on emotion recognition. This device can be configured in a management server and is used to execute any embodiment of the aforementioned intelligent trading risk control method based on emotion recognition. Specifically, please refer to... Figure 2 , Figure 2 This is a schematic block diagram of a transaction risk intelligent prevention and control device based on emotion recognition provided in an embodiment of the present invention.

[0068] like Figure 2 As shown, the intelligent transaction risk prevention and control device 100 based on emotion recognition includes a judgment unit 110, a video information acquisition unit 120, an expression recognition result acquisition unit 130, a voice recognition result acquisition unit 140, a comprehensive analysis unit 150, and a request interception unit 160.

[0069] The judgment unit 110 is used to judge whether the transaction request received from the user meets the preset identification conditions.

[0070] The video information acquisition unit 120 is used to acquire video information of the user answering preset questions if the transaction request meets the identification conditions.

[0071] The expression recognition result acquisition unit 130 is used to recognize the user image in the video information according to the preset face recognition strategy and obtain the corresponding expression recognition result.

[0072] The speech recognition result acquisition unit 140 is used to recognize the user's speech in the video information according to a preset speech risk recognition strategy, and obtain the corresponding speech recognition result.

[0073] The comprehensive analysis unit 150 is used to perform comprehensive analysis on the facial expression recognition results and the speech recognition results according to a preset emotion comprehensive analysis model, so as to obtain the analysis results of whether there is a transaction risk.

[0074] The request interception unit 160 is used to intercept the transaction request if the analysis result indicates that there is a risk.

[0075] The intelligent transaction risk prevention and control device based on emotion recognition provided in this embodiment of the invention applies the aforementioned intelligent transaction risk prevention and control method based on emotion recognition. It determines whether a user's transaction request meets the recognition conditions. If so, it acquires video information of the user answering preset questions, identifies the user's image in the video information to obtain an expression recognition result, and identifies the user's voice in the video information to obtain a voice recognition result. Based on an emotion comprehensive analysis model, it comprehensively analyzes the expression recognition result and the voice recognition result to obtain an identification result indicating whether a transaction risk exists. If a transaction risk exists, the transaction request is intercepted. The aforementioned intelligent transaction risk prevention and control method based on emotion recognition identifies user emotions by comprehensively analyzing user images and user voice, analyzes user expressions and language to determine whether the user's emotions are stable to obtain an identification result indicating whether a transaction risk exists, and intercepts risky transaction requests, thereby improving the security of user transfer transactions.

[0076] The aforementioned intelligent risk control device for transactions based on emotion recognition can be implemented as a computer program, which can, for example... Figure 3 It runs on the computer device shown.

[0077] Please see Figure 3 , Figure 3 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. The computer device may be a management server for executing an emotion-based intelligent risk control method for transaction risk.

[0078] See Figure 3 The computer device 500 includes a processor 502, a memory, and a communication interface 505 connected via a communication bus 501. The memory may include a storage medium 503 and internal memory 504.

[0079] The storage medium 503 may store the operating system 5031 and the computer program 5032. When the computer program 5032 is executed, it enables the processor 502 to execute a transaction risk intelligent control method based on emotion recognition. The storage medium 503 may be a volatile storage medium or a non-volatile storage medium.

[0080] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.

[0081] The internal memory 504 provides an environment for the operation of the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a transaction risk intelligent prevention and control method based on emotion recognition.

[0082] This communication interface 505 is used for network communication, such as providing data transmission. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device 500 to which the present invention is applied. The specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0083] The processor 502 is used to run the computer program 5032 stored in the memory to implement the corresponding functions in the above-mentioned intelligent prevention and control method for transaction risks based on emotion recognition.

[0084] Those skilled in the art will understand that Figure 3 The embodiments of the computer device shown do not constitute a limitation on the specific configuration of the computer device. In other embodiments, the computer device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. For example, in some embodiments, the computer device may include only memory and a processor. In such embodiments, the structure and function of the memory and processor are different from those shown. Figure 3 The embodiments shown are consistent and will not be repeated here.

[0085] It should be understood that, in this embodiment of the invention, the processor 502 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0086] In another embodiment of the invention, a computer-readable storage medium is provided. This computer-readable storage medium may be volatile or non-volatile. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps included in the above-described intelligent transaction risk control method based on emotion recognition.

[0087] Those skilled in the art will readily understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0088] In the embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Units with the same function may be grouped into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, or it may be an electrical, mechanical, or other form of connection.

[0089] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.

[0090] Furthermore, the functional units in the various embodiments of the present invention 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 integrated unit can be implemented in hardware or as a software functional unit.

[0091] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks.

[0092] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for intelligent prevention and control of transaction risks based on emotion recognition, characterized in that, The method includes: If a transaction request is received from the user, it is determined whether the transaction request meets the preset identification conditions; If the transaction request meets the identification conditions, obtain video information of the user answering preset questions; The user image in the video information is identified according to the preset face recognition strategy to obtain the corresponding expression recognition result; The user's voice in the video information is identified according to the preset voice risk recognition strategy to obtain the corresponding voice recognition result; The facial expression recognition results and the speech recognition results are comprehensively analyzed based on a pre-set emotion comprehensive analysis model to obtain an analysis result on whether there is a transaction risk. If the analysis results indicate that there is a risk, the transaction request will be blocked. The step of recognizing user speech in the video information according to a preset speech risk recognition strategy to obtain the corresponding speech recognition result includes: Based on the basic speech feature extraction rules of the speech risk recognition strategy, corresponding basic feature parameters are extracted from the user's speech, including: extracting effective audio corresponding to the target frequency range set in the basic speech feature extraction rules from the user's speech; extracting the number of peaks in the effective audio as the number of words spoken; extracting the duration of the effective audio as the speaking duration; calculating the ratio of the number of words spoken to the speaking duration to obtain the speech rate parameter; and extracting the mean loudness of the effective audio as the loudness parameter; and combining the speech rate parameter and the loudness parameter to form the basic feature parameters. The user's speech is text-recognized according to the text recognition model in the speech risk recognition strategy to obtain the corresponding text recognition result; The text recognition result and the basic feature parameters are combined and recognized according to the combined recognition model in the speech risk recognition strategy to obtain the corresponding speech recognition result; The step of combining the text recognition result and the basic feature parameters according to the combined recognition model in the speech risk recognition strategy to obtain the corresponding speech recognition result includes: Based on the disorder degree recognition model in the combined recognition model, the disorder degree of the basic feature parameters and the text recognition result is identified to obtain the corresponding disorder degree. Calculate the speech emotion coefficient between the basic feature parameters and the standard pronunciation feature parameters configured in the combined recognition model; the speech emotion coefficient is used to reflect the emotionality of the user's speech, and the speech emotion coefficient is... ; Q To calculate the voice emotion coefficient, Y s The speech rate parameter is one of the basic feature parameters. Y h The loudness parameter is one of the basic characteristic parameters. B s The standard speech rate parameter is one of the standard pronunciation feature parameters. B h Here, e represents the standard loudness parameter among the standard pronunciation feature parameters, where e is the base of the natural logarithm. The disorder level is combined with the voice emotion coefficient to form the voice recognition result.

2. The transaction risk intelligent prevention and control method based on emotion recognition according to claim 1, characterized in that, The step of recognizing the user image in the video information according to a preset face recognition strategy to obtain the corresponding expression recognition result includes: According to the facial feature extraction rules in the facial recognition strategy, the corresponding user facial features are extracted from the user image; The user's facial features are identified using the feature analysis model in the facial recognition strategy to obtain the corresponding expression recognition result.

3. The intelligent transaction risk prevention and control method based on emotion recognition according to claim 2, characterized in that, The step of identifying whether the user's facial features pose a risk based on the feature analysis model in the facial recognition strategy, and obtaining the corresponding expression recognition result, includes: The user's facial features are input into the long short-term memory network in the feature analysis model to obtain the expression feature vector corresponding to the user's facial features; The similarity value between the facial expression feature vector and the preset standard feature vector is calculated to obtain the corresponding facial expression recognition result. 4.The emotion recognition based transaction risk intelligent prevention and control method according to any one of claims 1-3, characterized in that, The step of comprehensively analyzing the facial expression recognition results and the speech recognition results based on a pre-set emotion comprehensive analysis model to obtain an analysis result on whether there is transaction risk includes: The facial expression recognition results and the speech recognition results are comprehensively calculated according to the comprehensive calculation rules in the comprehensive emotion analysis model to obtain the corresponding comprehensive emotion coefficient. Determine whether the comprehensive sentiment coefficient falls within the coefficient range set by the comprehensive sentiment analysis model to obtain the analysis result of whether there is a trading risk.

5. The emotion recognition-based transaction risk intelligent prevention and control method according to claim 4, characterized in that, The step of comprehensively analyzing the facial expression recognition results and the speech recognition results based on a pre-set emotion comprehensive analysis model to obtain an analysis result on whether there is transaction risk includes: If the analysis results indicate that there is no risk, the transaction request will be submitted.

6. An emotion recognition-based transaction risk intelligent prevention and control device, characterized in that, The device is used to execute the intelligent transaction risk prevention and control method based on emotion recognition as described in any one of claims 1-5, the device comprising: The judgment unit is used to determine whether a transaction request input by a user meets preset recognition conditions if a transaction request is received. The video information acquisition unit is used to acquire video information of the user answering preset questions if the transaction request meets the identification conditions. The facial expression recognition result acquisition unit is used to recognize the user image in the video information according to a preset face recognition strategy, and obtain the corresponding facial expression recognition result; The speech recognition result acquisition unit is used to recognize the user's speech in the video information according to a preset speech risk recognition strategy, and obtain the corresponding speech recognition result; The comprehensive analysis unit is used to perform a comprehensive analysis of the facial expression recognition results and the speech recognition results based on a preset emotion comprehensive analysis model, so as to obtain an analysis result on whether there is a transaction risk. The request interception unit is used to intercept the transaction request if the analysis result indicates that there is a risk.

7. A computer device, comprising: The device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements the steps of the intelligent transaction risk prevention and control method based on emotion recognition as described in any one of claims 1-5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent transaction risk prevention and control method based on emotion recognition as described in any one of claims 1-5.

Images