Page exception processing method and device, medium and program product

Abstract

The invention discloses a page exception processing method and device, a medium and a program product, and relates to the field of artificial intelligence, and the method comprises the steps: obtaining an exception recognition feature vector; inputting the anomaly identification feature vector into a pre-trained feature encoder to obtain anomaly potential space representation output by the feature encoder; inputting the abnormal potential space representation into a pre-trained page exception analysis model to obtain a page exception detection result and exception category information corresponding to the abnormal potential space representation output by the page exception analysis model; and determining an exception handling mode corresponding to the exception category information according to a preset mapping relationship between the exception category information and the exception handling mode and the exception category information, and handling the page exception according to the exception handling mode. According to the scheme, the time consumption of exception handling is reduced, the exception handling efficiency is improved, the accuracy of the page exception analysis result is improved, and then the use experience of a user on the financial system page is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and in particular to a method, device, medium, and program product for handling page anomalies. Background Technology

[0002] With the acceleration of digital transformation in the banking industry, the application of front-end systems such as online banking and mobile banking is becoming increasingly widespread. As these front-end systems become more complex, the types of front-end anomalies are also increasing, including but not limited to page loading failures, unresponsive buttons, and incorrect data display, which seriously affect the user experience.

[0003] Currently, the analysis and handling of front-end page anomalies are mainly achieved through manual analysis of logs recorded after anomalies occur in the front-end application system or through manual inspection.

[0004] However, manually analyzing the logs recorded after an anomaly occurs in the front-end application system, or conducting manual inspections, to analyze and handle front-end page anomalies, relies on manually retrieving logs and manually analyzing the causes. This is time-consuming and inefficient, affecting the user experience. Furthermore, the accuracy of page anomaly analysis results is low due to the possibility of errors in manual analysis. Summary of the Invention

[0005] This application provides a page anomaly handling method, device, medium, and program product to solve the problems in the prior art where front-end page anomaly analysis and handling relies on manual log retrieval and analysis of causes, which is time-consuming and inefficient, affecting user experience. Furthermore, the accuracy of page anomaly analysis results is low due to the possibility of errors in manual analysis.

[0006] Firstly, this application provides a method for handling page anomalies, the method comprising:

[0007] Obtain the anomaly identification feature vector; wherein, the anomaly identification feature vector is the feature vector of the data that changes when the page is abnormal;

[0008] The anomaly identification feature vector is input into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder.

[0009] The abnormal latent space representation is input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the abnormal latent space representation output by the page anomaly analysis model.

[0010] According to a preset mapping relationship between the abnormal category information and an abnormal processing manner and the abnormal category information, a processing manner corresponding to the abnormal category information is determined, and the page abnormality is processed according to the processing manner.

[0011] In a second aspect, the present application provides a page abnormality processing apparatus, the apparatus comprising:

[0012] An obtaining module is configured to obtain an abnormality recognition feature vector, wherein the abnormality recognition feature vector is a feature vector of data that changes when a page abnormality occurs.

[0013] A first input module is configured to input the abnormality recognition feature vector into a pre-trained feature encoder to obtain an abnormality latent space representation output by the feature encoder.

[0014] A second input module is configured to input the abnormality latent space representation into a pre-trained page abnormality analysis model to obtain a page abnormality detection result corresponding to the abnormality latent space representation and abnormal category information output by the page abnormality analysis model.

[0015] A processing module is configured to determine a processing manner corresponding to the abnormal category information according to a preset mapping relationship between the abnormal category information and the processing manner and the abnormal category information, and process the page abnormality according to the processing manner.

[0016] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the page abnormality processing method according to any of the embodiments of the present application.

[0017] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the program is executable by a processor to implement the page abnormality processing method according to any of the embodiments of the present application.

[0018] In a fifth aspect, the present application provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the page abnormality processing method according to any of the embodiments of the present application.

[0019] The proposed solution involves obtaining an anomaly identification feature vector, where the anomaly identification feature vector is the feature vector of the data that changes when a page anomaly occurs. The anomaly identification feature vector is input into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder. This anomaly latent space representation is then input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model. Based on a pre-defined mapping relationship between anomaly category information and anomaly handling methods, and the anomaly category information itself, the anomaly handling method corresponding to the anomaly category information is determined, and the page anomaly is handled according to the anomaly handling method. In other words, the proposed solution inputs the anomaly identification feature vector into the feature encoder to obtain the anomaly latent space representation, then inputs this representation into the page anomaly analysis model to obtain the page anomaly detection result and anomaly category information, and handles the page anomaly according to the anomaly category information. This avoids reliance on manual log retrieval and manual analysis of causes, reduces the time consumption of anomaly handling, improves the efficiency of anomaly handling, and enhances the accuracy of page anomaly analysis results, thereby improving the user experience of the financial system page. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the page exception handling method provided in this application;

[0022] Figure 2 This is a schematic diagram of the training process of the page anomaly analysis model of the page anomaly handling method provided in this application;

[0023] Figure 3 This is a schematic diagram of the page exception handling device provided in this application;

[0024] Figure 4 This is a schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0025] In the following, the technical solutions in the embodiments of the present application will be described clearly and completely in conjunction with the drawings in the embodiments of the present application, so that those skilled in the art can better understand the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should be within the scope of the present application.

[0026] In the technical solutions of the present application, the collected information is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards of relevant countries and regions, necessary security measures are taken, public order and good customs are not violated, and appropriate operation portals are provided for users to choose authorization or refusal.

[0027] Figure 1 FIG. 1 is a flowchart of a page exception processing method provided by the present application. The method can be executed by a page exception processing device, which can be implemented in software and / or hardware. In a specific embodiment, the device can be applied in an electronic device, which can be a computer. The following embodiments will be described by taking the device applied in an electronic device as an example. Referring to FIG. 1, the method can specifically include the following steps: Figure 1

[0028] Step 101, obtaining an exception recognition feature vector.

[0029] The exception recognition feature vector is a feature vector of data changed when a page exception occurs.

[0030] Specifically, when a page exception occurs, part of the page data is changed. At this time, the changed page data is collected, and the page data is cleaned and processed to obtain a data feature vector, which is the exception recognition feature vector.

[0031] Optionally, step 101 can be implemented through steps 1011 to 1013.

[0032] Step 1011, collecting page access information, page element information and page performance information.

[0033] ​Specifically, page access information refers to data related to user interaction with the page when using the front-end application. This includes user actions on the page and changes to the page based on those actions, reflecting user behavior patterns and page usage. For example, page access information might include the specific element clicked, timestamps, and coordinates. Page element information refers to data related to the page's structure and content, reflecting the page's rendering state and element dependencies. For example, page element information might include all currently included elements and components on the page. Page performance information refers to data related to page performance, reflecting the system's operating status and performance. Page access information, page element information, and page performance information can be collected through information collection interfaces provided by the page.

[0034] Optionally, page access information includes page click count, page dwell time, and page navigation path. Page element information includes page hypertext markup language, page cascading style sheets, page character encoding, and page element dependencies. Page performance information includes network request logs, page memory usage information, and page error logs.

[0035] Specifically, page click count refers to the number of times a user clicks on the page, including the specific element clicked and the timestamp. Page dwell time refers to the time a user enters and exits each page, used to identify issues such as page loading failures or lag. Page navigation path refers to the sequence of pages accessed by the user, used to detect process interruptions or abnormal navigation. Page Hypertext Markup Language (HTML) refers to the page's HTML structure, used to detect rendering anomalies. Page Cascading Style Sheets (CSS) refers to the page's CSS, used to detect style anomalies. Page character encoding refers to the page's character encoding, used to detect encoding problems. Page element dependencies refer to the dependencies between page elements, such as the character encoding file and CSS that a button depends on. Network request logs record the Uniform Resource Locator (URI), status code, latency, and other information for each network request, used to identify network problems. Page memory usage information shows the memory usage during page loading, used to identify issues such as memory leaks. Page error logs record error information during page code execution, used to assist in root cause analysis. Page access information, page element information, and page performance information can comprehensively and accurately display information related to the page, providing a data foundation for subsequent anomaly analysis and processing.

[0036] Step 1012 involves cleaning the page access information, page element information, and page performance information to obtain multiple anomaly identification datasets.

[0037] Specifically, the purpose of data cleaning is to remove noisy data, handle missing values, filter irrelevant information, and extract high-value features related to anomaly detection. Data from different sources is then converted to a unified format and scale. Noisy data removal involves identifying and removing invalid user behavior data, such as erroneous operations when the page is not fully loaded. Handling missing values ​​involves filling in or deleting data records containing missing values ​​to ensure data integrity. Filtering irrelevant information involves removing events from page areas or decorative elements unrelated to page anomalies. After cleaning page access information, page element information, and page performance information, the data is divided into multiple anomaly detection datasets, each containing feature vectors related to page anomalies.

[0038] Step 1013: Perform feature engineering on multiple anomaly detection datasets to obtain anomaly detection feature vectors.

[0039] Specifically, the goal of feature engineering is to extract and transform data to better reflect the characteristics of abnormal behavior and improve the performance of anomaly detection models. This involves extracting anomaly-related features from multiple anomaly detection datasets, reducing data dimensionality through feature selection and dimensionality reduction techniques, and then converting the features to a uniform scale to ensure the stability and accuracy of model training. The extracted and processed features are combined into a feature vector, which comprehensively characterizes the changes when a page exhibits anomalies; this is the anomaly detection feature vector. By cleaning and performing feature engineering on page access information, page element information, and page performance information, anomaly detection feature vectors can be generated, providing data support for subsequent anomaly analysis and processing.

[0040] Step 102: Input the anomaly recognition feature vector into the pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder.

[0041] Specifically, a pre-trained feature encoder can map high-dimensional anomaly recognition feature vectors to a low-dimensional latent space, generating an anomaly latent space representation. The anomaly recognition feature vectors are input into the pre-trained feature encoder, which reduces the feature dimension, thus obtaining the anomaly latent space representation output by the feature encoder.

[0042] Optionally, the feature encoder is trained according to steps 21 to 24.

[0043] Step 21: Input the training feature vector into the encoding module of the encoder to be trained to obtain the training latent space representation output by the encoding module.

[0044] The training feature vector is a feature vector to be trained in a preset training database. The training feature vector in the preset training database has the same data format as the anomaly recognition feature vector. The encoder to be trained includes an encoding module and a decoding module.

[0045] Specifically, the pre-set training database is a dataset containing multiple training feature vectors, which share the same data format as anomaly detection feature vectors. The purpose of the pre-set training database is to provide the feature encoder with sufficient data to learn the inherent structure and features of the data. The encoder to be trained consists of an encoding module and a decoding module. Its purpose is to map high-dimensional feature vectors to a low-dimensional latent space and reconstruct the original feature vectors through the decoding module. The input to the encoding module is the training feature vector, and the output is a low-dimensional representation of the training latent space. The structure of the encoding module can be composed of multiple layers of neural networks, with each layer progressively reducing the feature dimension. Therefore, inputting the training feature vectors into the encoding module of the encoder to be trained yields the training latent space representation output by the encoding module after reducing the feature dimension of the training feature vectors.

[0046] Step 22: Input the training latent space representation into the decoding module to obtain the reconstructed training feature vector output by the decoding module.

[0047] Specifically, the decoding module takes a low-dimensional training latent space representation as input and outputs a reconstructed training feature vector. The decoding module can be structured as a multi-layered neural network, with each layer progressively increasing the feature dimension. Therefore, inputting the training latent space representation into the decoding module yields the reconstructed training feature vector output by the module after increasing the feature dimension of the training feature vector.

[0048] Step 23: Adjust the encoding module parameters and decoding module parameters of the encoder to be trained based on the training feature vector and the reconstructed training feature vector to obtain the encoder to be trained with adjusted parameters.

[0049] Specifically, the goal of training the encoder is to enable the encoding module to learn the latent representation of the data and the decoding module to reconstruct the input data as accurately as possible. Based on the trained feature vectors and the reconstructed trained feature vectors, the parameters of the encoding and decoding modules of the encoder are adjusted. This involves optimizing the encoder and decoder parameters by minimizing the reconstruction error, thereby ensuring the model can effectively learn the intrinsic structure and features of the data. The reconstruction error is a metric that measures the difference between the reconstructed feature vectors and the original training feature vectors. Common methods for calculating the reconstruction error include mean squared error and cross-entropy loss. Based on the calculated reconstruction error, the parameters of the encoding and decoding modules are adjusted using the backpropagation algorithm to minimize the reconstruction error. After adjusting the parameters of the encoding and decoding modules of the encoder, the parameter-adjusted encoder is obtained.

[0050] Step 24: Use the parameter-adjusted encoder to be trained as the new encoder to be trained, and use a feature vector from the preset training database that has not been input into the encoder to be trained as the new training feature vector. Return to step 21 and continue until the first preset iteration end condition is met, and use the parameter-adjusted encoder to be trained as the feature encoder.

[0051] Specifically, the first preset iteration termination condition is the iteration termination condition for determining the end of the feature encoder training. For example, the first preset iteration termination condition could be that the number of iterations reaches a first preset number, or the reconstruction error is less than a preset reconstruction error, etc. The parameter-adjusted encoder to be trained is used as the new encoder to be trained. A feature vector from the preset training database that has not been input into the encoder to be trained is used as the new training feature vector. The step of inputting the training feature vector into the encoder to be trained is returned until the first preset iteration termination condition is reached, and the parameter-adjusted encoder to be trained is used as the feature encoder. The parameters of the encoding and decoding modules are adjusted according to the training feature vector and the reconstructed training feature vector to obtain the parameter-adjusted encoder to be trained. After iterative training, the feature encoder is obtained, providing effective feature representations for subsequent anomaly detection and classification.

[0052] Step 103: Input the anomaly latent space representation into the pre-trained page anomaly analysis model to obtain the page anomaly detection results and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model.

[0053] Specifically, the page anomaly detection result indicates whether an anomaly exists on the page, while the anomaly category information is the category information corresponding to the anomaly when it exists, such as page loading failure or button unresponsiveness. The anomaly latent space representation is input into a pre-trained page anomaly analysis model. The model obtains the page anomaly detection result and anomaly category information based on this representation, thus yielding the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model.

[0054] After performing step 103, steps 31 to 33 can also be performed.

[0055] Step 31: Send the page anomaly detection results and anomaly category information to the user device corresponding to the target staff member, and receive the anomaly handling result evaluation returned by the user device.

[0056] Specifically, the page anomaly detection results and anomaly category information are sent to the user device corresponding to the target staff member. The staff member evaluates the accuracy of the page anomaly detection results and anomaly category information based on the page anomaly detection results and anomaly category information, obtains an anomaly handling result evaluation, and returns the anomaly handling result evaluation to the electronic device executing this embodiment.

[0057] Step 32: If the evaluation of the anomaly handling result does not meet the preset evaluation requirements, then obtain the detection result and category information corresponding to the anomaly potential space representation sent by the user equipment.

[0058] Specifically, the preset evaluation requirements are pre-defined standards that the accuracy of the detected anomaly results and anomaly category information on the determined page should meet. For example, the preset evaluation requirement is that the score of the anomaly handling result evaluation should be greater than a scoring threshold. The anomaly potential space represents the detection results and category information corresponding to them, which are the detection results and category information determined by the staff based on human experience. When the anomaly handling result evaluation does not meet the preset evaluation requirements, the staff sends the detection results and category information corresponding to the anomaly potential space representation to the electronic device executing this embodiment through the user device. The electronic device executing this embodiment receives the detection results and category information corresponding to the anomaly potential space representation.

[0059] Step 33: Adjust the target model parameters of the page anomaly analysis model based on the page anomaly detection results, anomaly category information, detection results, and category information, so that the page anomaly analysis model can obtain the detection results and category information based on the anomaly latent space representation.

[0060] Specifically, the target model parameters of the page anomaly analysis model are adjusted based on the page anomaly detection results, anomaly category information, detection results, and category information. For example, the model's learning rate is adjusted to improve the accuracy of the model's output results, enabling the model to obtain detection results and category information from the anomaly latent space representation. When the anomaly handling result evaluation does not meet the preset evaluation requirements, the model parameters are adjusted based on the detection results and category information sent by the user device. The adjusted model can then obtain detection results and category information from the anomaly latent space representation, further improving the accuracy of the model's output results and thus further enhancing the efficiency of page anomaly analysis and processing based on the model.

[0061] Step 104: Based on the preset mapping relationship between exception category information and exception handling methods, determine the exception handling method corresponding to the exception category information, and handle the page exception according to the exception handling method.

[0062] Specifically, the preset mapping relationship between exception category information and exception handling methods includes each exception category and its corresponding handling method. For example, the preset mapping relationship is as follows: Page loading failure corresponds to checking network requests, server status, and resource paths. Unresponsive buttons correspond to checking button event listeners and page encoding errors. Data display errors correspond to checking data loading and rendering logic. Memory leaks correspond to optimizing memory usage and fixing memory leaks. Network timeouts correspond to retrying network requests or checking network connectivity. After obtaining the exception category information, based on the preset mapping relationship and the exception category information itself, the corresponding exception handling method is determined, and the page exception is handled according to the exception handling method. This means that the system can automatically take appropriate measures based on the detected exception type, thereby improving system stability and user experience.

[0063] The proposed solution involves obtaining an anomaly identification feature vector, where the anomaly identification feature vector is the feature vector of the data that changes when a page anomaly occurs. The anomaly identification feature vector is input into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder. This anomaly latent space representation is then input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model. Based on a pre-defined mapping relationship between anomaly category information and anomaly handling methods, and the anomaly category information itself, the anomaly handling method corresponding to the anomaly category information is determined, and the page anomaly is handled according to the anomaly handling method. In other words, the proposed solution inputs the anomaly identification feature vector into the feature encoder to obtain the anomaly latent space representation, then inputs this representation into the page anomaly analysis model to obtain the page anomaly detection result and anomaly category information, and handles the page anomaly according to the anomaly category information. This avoids reliance on manual log retrieval and manual analysis of causes, reduces the time consumption of anomaly handling, improves the efficiency of anomaly handling, and enhances the accuracy of page anomaly analysis results, thereby improving the user experience of the financial system page.

[0064] Figure 2 This is a schematic diagram of the training process of the page anomaly analysis model of the page anomaly handling method provided in this application. This embodiment... Figure 1 Based on the illustrated embodiments and various optional implementation schemes, the training steps of the page anomaly analysis model are described in detail. For example... Figure 2 As shown, the method may include the following steps:

[0065] Step 201: Input the training recognition feature vector into the feature encoder to obtain the training recognition latent space representation output by the feature encoder.

[0066] The training recognition feature vector is a recognition feature vector to be trained in the preset training recognition database. The data format of the recognition feature vector to be trained in the preset training recognition database is the same as that of the anomaly recognition feature vector.

[0067] Specifically, the preset training recognition database stores a large number of feature vectors to be trained. The feature vectors to be trained in the preset training recognition database have the same data format as the anomaly recognition feature vectors, ensuring that the feature vectors used during model training are consistent with those used in subsequent applications, thereby improving the accuracy of the model's output. The trained recognition feature vectors are input into the feature encoder to obtain the training recognition latent space representation output by the feature encoder. This representation is used in the subsequent training steps of the page anomaly analysis model in this embodiment, thus combining the encoder obtained through unsupervised learning with the model obtained through supervised learning in this embodiment, thereby improving the accuracy of the model's output.

[0068] Step 202: Input the training recognition latent space representation into the model to be trained to obtain the training page detection results and training anomaly category information output by the model to be trained.

[0069] Specifically, the model to be trained can be any deep learning architecture. By inputting the training recognition latent space representation into the model to be trained, the training page detection results and training anomaly category information output by the model can be obtained.

[0070] Step 203: Based on the training page detection results, training anomaly category information, the label detection results corresponding to the training recognition feature vector, and the label anomaly category information, adjust the target model parameters of the model to be trained to obtain the model to be trained after parameter adjustment.

[0071] Specifically, the training identifies the correct label detection results and anomaly category information corresponding to the feature vectors. Therefore, based on the training page detection results, training anomaly category information, and the label detection results and anomaly category information corresponding to the training identify feature vectors, the target model parameters of the model to be trained are adjusted, such as adjusting the model's learning rate, to obtain the parameter-adjusted model to be trained.

[0072] Optionally, the target model parameters include the learning rate and / or the length of the soft cue embedding.

[0073] Specifically, the target model parameters are determined by the learning rate and / or the length of the soft cue embeddings because complex tasks require longer soft cue embeddings and / or smaller learning rates. A smaller learning rate can prevent the model's existing analytical capabilities from being compromised by a larger learning rate. Choosing a smaller learning rate and a longer soft cue embedding can ensure the stability of the model's training process while avoiding overfitting.

[0074] Step 204: Use the parameter-adjusted training model as the new training model, and use a training recognition feature vector from a preset training recognition database that is not input into the training model as the new training recognition feature vector. Return to step 201 until the second preset iteration end condition is reached, and use the parameter-adjusted training model as the page anomaly analysis model.

[0075] Specifically, the parameter-adjusted model to be trained is used as the new model to be trained. A feature vector from a preset training recognition database that is not input into the model to be trained is used as the new training recognition feature vector. The process returns to the step of inputting the training recognition feature vector into the feature encoder to obtain the training recognition latent space representation output by the feature encoder, until the second preset iteration termination condition is reached. For example, the second preset iteration termination condition can be that training stops when a certain cutoff condition is reached, i.e., the return step stops. The cutoff condition can be that the number of training rounds reaches a preset number, or the training loss falls below a certain value, etc. The parameter-adjusted model to be trained at this time is the page anomaly analysis model.

[0076] The proposed solution pre-constructs a training database where the training feature vectors and anomaly detection feature vectors share the same data format. This ensures that the feature vectors used during model training are consistent with those used in subsequent applications, thereby improving the accuracy of the model's output. The training feature vectors are input into a feature encoder to obtain the training latent space representation output by the encoder. This representation is then used in the training steps of the page anomaly analysis model in this embodiment. This combines the encoder obtained through unsupervised learning with the model obtained through supervised learning in this embodiment, further enhancing the accuracy of the model's output. Furthermore, defining the target model parameters as the learning rate and / or the length of the soft cue embedding ensures the stability of the model's training process while preventing overfitting, further improving the accuracy of the model's output.

[0077] Figure 3 This is a schematic diagram of a page exception handling apparatus provided in this application, which is suitable for executing the page exception handling method provided in this application. Figure 3 As shown, the device may specifically include:

[0078] The acquisition module 301 is used to acquire an anomaly identification feature vector; wherein, the anomaly identification feature vector is the feature vector of the data that changes when the page is abnormal.

[0079] The first input module 302 is used to input the anomaly recognition feature vector into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder.

[0080] The second input module 303 is used to input the anomaly latent space representation into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model.

[0081] The processing module 304 is used to determine the exception handling method corresponding to the exception category information based on the preset mapping relationship between exception category information and exception handling method, and to handle the page exception according to the exception handling method.

[0082] In one embodiment, the acquisition module 301 is specifically used to: collect page access information, page element information, and page performance information; perform data cleaning on the page access information, page element information, and page performance information to obtain multiple anomaly identification datasets; and perform feature engineering processing on the multiple anomaly identification datasets to obtain the anomaly identification feature vector.

[0083] In one embodiment, the page access information obtained by module 301 includes the number of page clicks, page dwell time, and page jump path; the page element information includes page hypertext markup language, page cascading style sheets, page character encoding, and page element dependencies; and the page performance information includes network request logs, page memory usage information, and page error logs.

[0084] In one embodiment, the device further includes: an adjustment module, configured to send the page anomaly detection result and the anomaly category information to the user device corresponding to the target staff member, and receive an anomaly handling result evaluation returned by the user device; if the anomaly handling result evaluation does not meet preset evaluation requirements, then obtain the detection result and category information corresponding to the anomaly latent space representation sent by the user device; adjust the target model parameters of the page anomaly analysis model according to the page anomaly detection result, the anomaly category information, the detection result, and the category information, so that the page anomaly analysis model can obtain the detection result and the category information according to the anomaly latent space representation.

[0085] In one embodiment, the apparatus further includes: a feature encoder training module, configured to input a training feature vector into the encoding module of the encoder to be trained, to obtain a training latent space representation output by the encoding module; wherein the training feature vector is a training feature vector in a preset training database, and the training feature vector in the preset training database has the same data format as the anomaly recognition feature vector; the encoder to be trained includes an encoding module and a decoding module; the training latent space representation is input into the decoding module to obtain a reconstructed training feature vector output by the decoding module; the encoding module parameters and decoding module parameters of the encoder to be trained are adjusted according to the training feature vector and the reconstructed training feature vector to obtain a parameter-adjusted encoder to be trained; the parameter-adjusted encoder to be trained is used as a new encoder to be trained, and a training feature vector in the preset training database that is not input into the encoder to be trained is used as a new training feature vector; the process of "inputting the training feature vector into the encoding module of the encoder to be trained to obtain a training latent space representation output by the encoding module" is repeated until a first preset iteration termination condition is met, and the parameter-adjusted encoder to be trained is used as the feature encoder.

[0086] In one embodiment, the apparatus further includes: a page anomaly analysis model training module, configured to input a training recognition feature vector into the feature encoder to obtain a training recognition latent space representation output by the feature encoder; wherein the training recognition feature vector is a training recognition feature vector in a preset training recognition database, and the training recognition feature vector in the preset training recognition database has the same data format as the anomaly recognition feature vector; inputting the training recognition latent space representation into a training model to obtain the training page detection result and training anomaly category information output by the training model; and based on the training page detection result, the training anomaly category information, and the training anomaly category information, the training recognition latent space representation is further configured to be used to obtain a training page detection result and training anomaly category information output by the training model. The training identifies the marker detection results and anomaly category information corresponding to the feature vectors, and adjusts the target model parameters of the model to be trained to obtain the parameter-adjusted model to be trained. The parameter-adjusted model to be trained is used as the new model to be trained, and a feature vector to be trained that is not input into the model to be trained from the preset training identification database is used as the new training identification feature vector. The process returns to the step of "inputting the training identification feature vector into the feature encoder to obtain the training identification latent space representation output by the feature encoder" until the second preset iteration end condition is reached. The parameter-adjusted model to be trained is then used as the page anomaly analysis model.

[0087] In one embodiment, the target model parameters of the page anomaly analysis model training module include the learning rate and / or the length of the soft cue embedding.

[0088] The apparatus of this application acquires anomaly identification feature vectors; wherein, the anomaly identification feature vectors are feature vectors of data that change when a page anomaly occurs; the anomaly identification feature vectors are input into a pre-trained feature encoder to obtain an anomaly latent space representation output by the feature encoder; the anomaly latent space representation is input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model; based on a preset mapping relationship between anomaly category information and anomaly handling methods, and the anomaly category information, the anomaly handling method corresponding to the anomaly category information is determined, and the page anomaly is handled according to the anomaly handling method. In other words, the scheme of this application inputs the anomaly identification feature vector into the feature encoder to obtain the anomaly latent space representation, then inputs the anomaly latent space representation into the page anomaly analysis model to obtain the page anomaly detection result and anomaly category information, and handles the page anomaly according to the anomaly category information. This avoids reliance on manual log retrieval and manual analysis of causes, reduces the time consumption of anomaly handling, improves the efficiency of anomaly handling, and enhances the accuracy of page anomaly analysis results, thereby improving the user experience of the financial system page.

[0089] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the page exception handling method provided in any of the above embodiments.

[0090] This application also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the page exception handling method provided in any of the above embodiments.

[0091] The following is for reference. Figure 4 It shows a schematic diagram of the structure of an electronic device 400 suitable for implementing the present application. Figure 4 The electronic device shown is merely an example and should not impose any limitations on the functionality and scope of this application.

[0092] like Figure 4 As shown, the electronic device 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage section 408 into a random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the electronic device 400. The CPU 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0093] The following components are connected to I / O interface 405: an input section 406 including a keyboard, mouse, etc.; an output section 407 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to I / O interface 405 as needed. A removable medium 411, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 410 as needed so that computer programs read from it can be installed into storage section 408 as needed.

[0094] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 409, and / or installed from removable medium 411. When the computer program is executed by central processing unit (CPU) 401, it performs the functions defined above in the system of this application.

[0095] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0096] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0097] The modules and / or units described in this application can be implemented in software or hardware. The described modules and / or units can also be housed in a processor; for example, a processor can be described as including an acquisition module, a first input module, a second input module, and a processing module. The names of these modules do not necessarily limit the module itself.

[0098] In another aspect, this application also provides a computer-readable medium, which may be included in the device described in the above embodiments; or it may exist independently and not assembled into the device. The computer-readable medium carries one or more programs that, when executed by the device, cause the device to perform the following operations:

[0099] Obtain the anomaly identification feature vector; where the anomaly identification feature vector is the feature vector of the data that changes when the page is abnormal; input the anomaly identification feature vector into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder; input the anomaly latent space representation into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model; determine the anomaly handling method corresponding to the anomaly category information and handle the page anomaly according to the anomaly handling method based on the preset mapping relationship between anomaly category information and anomaly handling method;

[0100] According to the technical solution of this application, an anomaly identification feature vector is obtained; wherein, the anomaly identification feature vector is the feature vector of the data that changes when the page is abnormal; the anomaly identification feature vector is input into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder; the anomaly latent space representation is input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model; based on the preset mapping relationship between anomaly category information and anomaly handling methods, and the anomaly category information, the anomaly handling method corresponding to the anomaly category information is determined, and the page anomaly is handled according to the anomaly handling method. That is, the solution of this application inputs the anomaly identification feature vector into the feature encoder to obtain the anomaly latent space representation, then inputs the anomaly latent space representation into the page anomaly analysis model to obtain the page anomaly detection result and anomaly category information, and handles the page anomaly according to the anomaly category information, thereby avoiding reliance on manual log retrieval and manual analysis of causes, reducing the time consumption of anomaly handling, improving the efficiency of anomaly handling, and improving the accuracy of page anomaly analysis results, thus improving the user experience of the financial system page.

[0101] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the page exception handling method provided in any embodiment of this application.

[0102] In the implementation of the computer program product, computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0103] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

[0104] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for handling page anomalies, characterized in that, The method includes: Obtain the anomaly identification feature vector; wherein, the anomaly identification feature vector is the feature vector of the data that changes when the page is abnormal; The anomaly identification feature vector is input into a pre-trained feature encoder to obtain the anomaly latent space representation output by the feature encoder. The abnormal latent space representation is input into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the abnormal latent space representation output by the page anomaly analysis model. Based on the preset mapping relationship between exception category information and exception handling methods, and the exception category information, the exception handling method corresponding to the exception category information is determined, and the page exception is handled according to the exception handling method.

2. The method according to claim 1, characterized in that, The process of obtaining the anomaly identification feature vector includes: Collect page access information, page element information, and page performance information; Data cleaning is performed on the page access information, the page element information, and the page performance information to obtain multiple anomaly identification datasets; Feature engineering is performed on the multiple anomaly detection datasets to obtain the anomaly detection feature vectors.

3. The method according to claim 2, characterized in that, The page access information includes the number of page clicks, the page dwell time, and the page jump path; The page element information includes the page's Hypertext Markup Language, Cascading Style Sheets, page character encoding, and page element dependencies; The page performance information includes network request logs, page memory usage information, and page error logs.

4. The method according to claim 1, characterized in that, After inputting the anomaly latent space representation into a pre-trained page anomaly analysis model to obtain the page anomaly detection result and anomaly category information corresponding to the anomaly latent space representation output by the page anomaly analysis model, the method further includes: The page anomaly detection results and the anomaly category information are sent to the user device corresponding to the target staff member, and the anomaly handling result evaluation returned by the user device is received. If the evaluation of the anomaly handling result does not meet the preset evaluation requirements, then the detection result and category information corresponding to the anomaly potential space representation sent by the user equipment are obtained; The target model parameters of the page anomaly analysis model are adjusted based on the page anomaly detection results, the anomaly category information, the detection results, and the category information, so that the page anomaly analysis model can obtain the detection results and the category information based on the anomaly latent space representation.

5. The method according to claim 1, characterized in that, The feature encoder is trained according to the following steps: The training feature vector is input into the encoding module of the encoder to be trained to obtain the training latent space representation output by the encoding module; wherein, the training feature vector is a training feature vector in a preset training database, and the training feature vector in the preset training database has the same data format as the anomaly recognition feature vector; the encoder to be trained includes an encoding module and a decoding module. The training latent space representation is input into the decoding module to obtain the reconstructed training feature vector output by the decoding module; The encoding module parameters and decoding module parameters of the encoder to be trained are adjusted based on the training feature vector and the reconstructed training feature vector to obtain the encoder to be trained with adjusted parameters. The parameter-adjusted encoder to be trained is used as the new encoder to be trained. A feature vector from the preset training database that has not been input into the encoder to be trained is used as the new training feature vector. The process of "inputting the training feature vector into the encoding module of the encoder to be trained to obtain the training latent space representation output by the encoding module" is repeated until the first preset iteration end condition is met. The parameter-adjusted encoder to be trained is then used as the feature encoder.

6. The method according to claim 5, characterized in that, The page anomaly analysis model was trained according to the following steps: The training recognition feature vector is input into the feature encoder to obtain the training recognition latent space representation output by the feature encoder; wherein, the training recognition feature vector is a recognition feature vector to be trained in a preset training recognition database, and the data format of the recognition feature vector to be trained in the preset training recognition database is the same as that of the anomaly recognition feature vector; The training identification latent space representation is input into the model to be trained to obtain the training page detection results and training anomaly category information output by the model to be trained. Based on the training page detection results, the training anomaly category information, the label detection results corresponding to the training recognition feature vector, and the label anomaly category information, the target model parameters of the model to be trained are adjusted to obtain the model to be trained with adjusted parameters. The model to be trained after parameter adjustment is used as the new model to be trained. A training recognition feature vector from the preset training recognition database that has not been input into the model to be trained is used as the new training recognition feature vector. The process of "inputting the training recognition feature vector into the feature encoder to obtain the training recognition latent space representation output by the feature encoder" is repeated until the second preset iteration end condition is met. The model to be trained after parameter adjustment is used as the page anomaly analysis model.

7. The method according to claim 6, characterized in that, The target model parameters include the learning rate and / or the length of the soft cue embedding.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the page exception handling method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the page exception handling method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the page exception handling method as described in any one of claims 1 to 7.

Images