Distributed service flow cutting recommendation method and device

Abstract

The application discloses a distributed service flow cutting recommendation method and device, and relates to the field of artificial intelligence, wherein the method comprises the following steps: obtaining transaction log data; integrating the transaction log data to obtain integrated data; adding a service effective coverage field in the integrated data; extracting integrated data features to obtain an input data matrix; importing the input data matrix into a business scenario classification model to predict the channel transaction code and the business scenario description of the input data matrix, and obtaining a prediction result matrix; filling the prediction result matrix with a flow cutting suggestion field to obtain a supplementary record matrix; and generating a business scenario list according to the supplementary record matrix, wherein the business scenario list comprises the channel transaction code, the business scenario description, the service effective coverage field and the flow cutting suggestion field. The application can predict the business scenario, monitor the service effective coverage on the transaction link, generate the flow cutting suggestion, and guarantee the effectiveness and reliability of the distributed service flow cutting under the architecture transformation.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a distributed service flow recommendation method and apparatus. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] Since the architectural transformation, the process of shifting from a centralized host system to a distributed platform system has been a gradual switchover, with each stage of the transaction chain achieving effective platform service coverage and service flow switching enabled. Currently, the test environment arranges service flow switching tests according to specific business scenarios. With the service flow switching switch enabled, business testers initiate the corresponding business scenario. If the transaction is successfully closed, the platform service is considered ready for production deployment, and service flow switching is scheduled. However, this distributed platform service flow switching method has certain inefficiencies and risks. Specifically, on the one hand, in a distributed platform architecture, the processing chain of a transaction becomes longer, flowing between multiple distributed platform applications. Each distributed platform application may have an architectural transformation service flow switching switch, resulting in numerous flow switching switches across the entire chain. Therefore, manually confirming the switch status on the chain is costly, inefficient, and unreliable. On the other hand, relying solely on business testers to initiate a specific transaction only confirms the final transaction result. It's difficult to perceive the actual data flow behind each transaction, or to determine whether the transaction effectively covers the platform service. This approach demands a high level of testing experience from business testers, who must be familiar with the full flow switching status in the test scenario. If any flow switching switch is closed, the transaction actually proceeds to the host branch. Even if the transaction is eventually closed successfully, the functionality of the distributed platform service remains unverified. Therefore, this approach carries the risk of ineffective service coverage due to the difficulty in effectively tracing the process.

[0004] In summary, predicting business scenarios, monitoring the effective service coverage on the transaction chain, and then generating recommendations on whether to switch traffic is a major technical challenge in effectively ensuring the quality of traffic switching during architecture transformation. Summary of the Invention

[0005] This invention provides a distributed service traffic switching recommendation method to predict business scenarios, monitor the effective service coverage on the transaction link, generate traffic switching suggestions, and ensure the effectiveness and reliability of distributed service traffic switching under architecture transformation. The method includes:

[0006] Obtain transaction log data;

[0007] Based on the transaction time, within a preset transaction time, the transaction log data is integrated and processed to obtain integrated data;

[0008] Based on the status of the flow switching switch and its record status in the integrated data, add a field indicating whether the service is effectively covered.

[0009] A predetermined number of message fields are extracted from the integrated data as feature value fields, and an input data matrix is ​​obtained based on the feature value fields.

[0010] The input data matrix is ​​imported into the business scenario classification model to predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and these are used as labels to fill in the input data matrix to obtain the prediction result matrix.

[0011] Fill each row of the prediction result matrix with a "whether to switch flow" field to obtain the supplementary recording matrix;

[0012] Based on preset fields, a business scenario list is generated according to the supplementary entry matrix. The business scenario list includes: channel transaction code, business scenario description, field indicating whether the service is effectively covered, and field indicating whether to switch traffic.

[0013] This invention also provides a distributed service flow-switching intelligent recommendation device, wherein the device includes:

[0014] The data acquisition module is used to obtain transaction log data;

[0015] The data processing module is used to integrate and process transaction log data within a preset transaction time, based on the transaction time, to obtain integrated data; and to add a field indicating whether the service is effectively covered based on the status of the flow switching switch and the record status of the flow switching switch in the integrated data.

[0016] The feature extraction module is used to extract a preset number of message fields from the integrated data as feature value fields, and obtain the input data matrix based on the feature value fields;

[0017] The business scenario identification module is used to import the input data matrix into the business scenario classification model, predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fill them into the input data matrix as labels to obtain the prediction result matrix.

[0018] The flow-cutting suggestion supplementation module is used to fill each row of the prediction result matrix with a flow-cutting suggestion field to obtain the supplementation matrix.

[0019] The scenario list generation module is used to generate a business scenario list based on a supplementary matrix according to preset fields. The business scenario list includes: channel transaction code, business scenario description, service coverage field, and traffic switching suggestion field.

[0020] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described distributed service flow switching recommendation method.

[0021] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described distributed service flow switching recommendation method.

[0022] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described distributed service flow switching recommendation method.

[0023] In this embodiment of the invention, the distributed service flow switching recommendation method, compared with the existing technology that relies on manual confirmation of the on / off status of the link, resulting in high cost, low efficiency, poor reliability, and difficulty in effectively tracking the specific data flow behind each transaction, leading to ineffective service coverage, obtains transaction log data; integrates the transaction log data within a preset transaction time based on the transaction time to obtain integrated data; adds a service coverage field based on the flow switching switch on / off status and recording status in the integrated data; extracts a preset number of message fields from the integrated data as feature value fields; obtains an input data matrix based on the feature value fields; imports the input data matrix into a business scenario classification model to predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fills them into the input data matrix as labels to obtain a prediction result matrix; fills each row of the prediction result matrix with a flow switching suggestion field to obtain a supplementary entry matrix; and generates a business scenario list based on the supplementary entry matrix according to preset fields, the business scenario list including: channel transaction code, business scenario description, service coverage field, and flow switching suggestion field. It can predict business scenarios, monitor the effective service coverage on the transaction chain, generate suggestions on whether to switch traffic, and ensure the effectiveness and reliability of distributed service traffic switching under architecture transformation. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0025] Figure 1 This is a flowchart illustrating the distributed service flow switching recommendation method in an embodiment of the present invention.

[0026] Figure 2 This is a flowchart illustrating the feature extraction steps in an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram of the business scenario classification model in an embodiment of the present invention;

[0028] Figure 4 This is a schematic diagram of the distributed service flow switching and recommendation device in an embodiment of the present invention;

[0029] Figure 5 This is a schematic diagram of the data acquisition module of the distributed service flow switching recommendation device in an embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0031] In this document, the term "and / or" merely describes a relationship, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0032] In the description of this specification, the terms "comprising," "including," "having," and "containing" are open-ended terms, meaning that they include but are not limited to. The terms "an embodiment," "a specific embodiment," "some embodiments," and "for example," etc., refer to specific features, structures, or characteristics described in connection with that embodiment or example that are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. The order of steps involved in the various embodiments is used to illustrate the implementation of this application, and the order of steps is not limited and can be adjusted appropriately as needed.

[0033] As mentioned earlier, existing distributed platform service flow switching methods suffer from inefficiencies and risks. Firstly, in a distributed platform architecture, the processing chain for a single transaction becomes longer, flowing between multiple distributed platform applications. Each application may have a flow switching switch for architecture transformation services, resulting in numerous flow switching switches across the entire chain. Therefore, manually verifying the status of these switches is costly, inefficient, and unreliable. Secondly, relying solely on business testers to initiate a specific transaction only confirms the final transaction result, making it difficult to perceive the specific data flow behind each transaction and determine whether the transaction effectively covers the platform service. This requires a high level of testing experience from business testers, who need to be familiar with all flow switching switches in the test scenario. If any flow switching switch is closed, the transaction actually proceeds to the host branch, and even if the transaction is eventually closed normally, the functionality of the distributed platform service is not verified. Therefore, this approach carries the risk of ineffective service coverage due to difficulty in effective process tracking.

[0034] To address the aforementioned issues, this invention provides a distributed service switching recommendation method. This method can predict business scenarios, monitor the effective service coverage on the transaction link, and generate a suggestion on whether to switch services.

[0035] Figure 1 This is a flowchart illustrating the distributed service flow switching recommendation method in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes:

[0036] Step 101: Obtain transaction log data;

[0037] Step 102: Based on the transaction time, within the preset transaction time, integrate the transaction log data to obtain integrated data;

[0038] Step 103: Based on the status of the flow switching switch and the record status of the flow switching switch in the integrated data, add a field indicating whether the service is effectively covered;

[0039] Step 104: Extract a preset number of message fields from the integrated data as feature value fields, and obtain the input data matrix based on the feature value fields;

[0040] Step 105: Import the input data matrix into the business scenario classification model, predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fill them into the input data matrix as labels to obtain the prediction result matrix;

[0041] Step 106: Fill each row of the prediction result matrix with the "whether to switch flow" field to obtain the supplementary recording matrix;

[0042] Step 107: Generate a business scenario list based on the supplementary entry matrix according to the preset fields. The business scenario list includes: channel transaction code, business scenario description, service coverage field and traffic switching suggestion field.

[0043] Depend on Figure 1 As can be seen from the process shown, the embodiments of the present invention can predict business scenarios, monitor the effective service coverage on the transaction link, generate suggestions on whether to switch traffic, and ensure the effectiveness and reliability of distributed service traffic switching under architecture transformation.

[0044] In one embodiment, obtaining transaction log data in step 101 above may include: collecting data routing information, flow switching status, and full transaction log information of multiple transactions at each stage of the processing link, and combining them into transaction log source data; processing the transaction log source data to obtain transaction log data.

[0045] In practice, data routing information, flow switching status, and full transaction log information for multiple transactions at each stage of the processing chain are collected, which may include:

[0046] The data routing information of multiple transactions across various platform applications in the processing chain can be stored locally, and the data routing information can be stored according to the dimensions of "application name, database IP, database instance, database user, database password, and transaction log register name".

[0047] The flow switching switches of the distributed platform services are stored as dynamic variables in the configuration center system. By calling the application programming interface provided by the configuration center system, the flow switching switch status of each distributed platform service in the processing link of multiple transactions can be obtained in real time. The flow switching switch status can be stored in the dimension of "application name, service name, method name, switch name, switch value, and setting time".

[0048] Based on the data routing information stored locally, and according to the data routing strategy, a JDBC (JavaDataBase Connectivity) connection can be established to obtain the full transaction log information registered in the transaction logbook of various platform applications within a certain time range in near real-time. The full transaction log information can be stored according to the dimension of "application name + database IP + full transaction logbook information".

[0049] Processing the source data of transaction logs can include: performing data dimensionality reduction and data cleaning on multiple full transaction log information in the source data of transaction logs;

[0050] The data dimensionality reduction includes removing irrelevant fields such as "authorized region, authorized teller, channel IP, counterparty code, and customer information," and / or retaining key fields such as "event number, associated event number, service name, method name, region code, teller number, branch, currency, transaction date, transaction time, transaction code, uploading application name, account, and record status," thereby reducing data dimensionality. Data cleaning may include removing irrelevant information such as spaces and punctuation marks from key fields, and / or standardizing and simplifying field formats.

[0051] In one embodiment, step 102 above, which involves integrating transaction log data within a preset transaction time based on transaction time to obtain integrated data, may include:

[0052] Transaction log data is divided into product data and backend processing data;

[0053] Based on the transaction time, within a preset transaction time, all product data are concatenated with all corresponding backend processed data to obtain multiple concatenated data sets.

[0054] All concatenated data is deduplicated to obtain integrated data.

[0055] Product data includes, but is not limited to: corporate loan applications, corporate deposit applications, personal deposit applications, and personal loan applications; back-end processing data includes, but is not limited to: settlement applications, accounting applications, and clearing applications; product data corresponds to one or more back-end processing data.

[0056] In one embodiment, a product application is a corporate loan application. Based on the transaction time, within a preset transaction time, the product application is spliced ​​together with all the corresponding backend processing data to obtain multiple spliced ​​data, which can be "corporate loan application - settlement application - accounting application - clearing application".

[0057] In one embodiment, step 103 above, adding a field indicating whether the service is effectively covered based on the flow switching switch status and the switch's recorded state in the integrated data, may include:

[0058] Check one by one whether the flow switching switch of all the integrated data is turned on, and record whether the status is successful;

[0059] When the data integration flow switching switch is on and the recording status is successful, add a service effective overwrite field after the original data field of the integrated data, and set the service effective overwrite field to yes.

[0060] When the flow switching switch for integrated data is off, or the recording status is failed, add a "Whether the service is effectively covered" field after the original data field of the integrated data, and set the "Whether the service is effectively covered" field to "No".

[0061] In one embodiment, step 104 above, extracting a preset number of message fields from the integrated data as feature value fields, and obtaining the input data matrix based on the feature value fields, may include:

[0062] A preset number of message fields are extracted from the integrated data as feature value fields to obtain a feature value field matrix; labels are added to the feature value field matrix, format conversion is performed, and normalization is carried out to obtain the input data matrix.

[0063] In specific implementation, adding labels includes: adding a label field to the end of each row of the feature value field matrix; format conversion includes converting Boolean-type feature value fields to binary value format, and / or converting text-type feature value fields using lexical assumptions; normalization processing includes:

[0064] The feature value fields after each format conversion are normalized according to the following formula:

[0065]

[0066] Where x represents the eigenvalue after format conversion, μ represents the expected value of the eigenvalue after format conversion, and δ represents the standard deviation of the eigenvalue after format conversion. This represents the feature value field after standard normalization.

[0067] Figure 2 This is a flowchart illustrating the feature extraction steps in an embodiment of the present invention. In one embodiment, such as... Figure 2 As shown, the feature extraction steps include:

[0068] First, extract g message fields from each of the h integrated data as feature value fields to obtain a feature value field matrix. Feature value fields can be divided into text fields and Boolean fields. Table 1 below shows a partial display of the extracted feature value fields.

[0069] Table 1

[0070]

[0071]

[0072] like Figure 2As shown, Boolean type feature value fields are converted to binary value format and used as input data. Text type feature value fields can be formatted using the Bag of Words (BoW) method. A label field 'y' is added to the end of each row of the feature value field matrix. The label field 'y' can record tags such as channel transaction code and business scenario description. For example, the channel transaction code is 2093, and the business scenario description is corporate loan settlement with retention and entrusted payment.

[0073] The feature value fields after each format conversion are normalized according to the following formula:

[0074]

[0075] Where x represents the eigenvalue after format conversion, μ represents the expected value of the eigenvalue after format conversion, and δ represents the standard deviation of the eigenvalue after format conversion. This represents the feature value field after standard normalization.

[0076] Obtain the input data matrix X. When training the business scenario classification model, the label field y in the input data matrix X already records labels such as channel transaction codes and business scenario descriptions. Using this historical data to train the model, when using the model to predict business scenarios, the label field y in the input data matrix X is empty. The trained business scenario classification model predicts the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fills the label field y of the input data matrix to obtain the prediction result matrix.

[0077] In one embodiment, the business scenario classification model in step 105 above includes:

[0078] First convolutional layer, second convolutional layer, first pooling layer, second pooling layer, fully connected output layer.

[0079] Figure 3 This is a schematic diagram of the structure of a business scenario classification model in an embodiment of the present invention. In one embodiment, such as... Figure 3 As shown, the model includes a preprocessing layer, convolutional layers C1 and C2, pooling layers P1 and P2, and a fully connected output layer. The training process of the business scenario classification model includes:

[0080] After feature extraction, the feature value field matrix yields the input data matrix. The convolutional layer C1 learns local features from the sequence of word vectors in the input data matrix. During operation, the convolutional layer uses a convolutional window to traverse the sequence with a fixed stride. Elements within the window are multiplied by multiple convolutional kernels. After each kernel completes its operation with the entire sequence, a new sequence is generated, whose values ​​reflect the features exhibited by the input data under that kernel. The mathematical model can be described as follows:

[0081] s = f(W c ·d+b)

[0082] Among them: W c It is the c-th convolutional kernel among the n convolutional kernels of length p in convolutional layer C1; d represents the submatrix composed of each row of the input data matrix; b∈R represents the bias value; f is the activation function; s is the convolutional kernel W. c The result of the dot product with the i-th submatrix.

[0083] Convolutional layer C2 repeats the operation of convolutional layer C1, with the aim of learning the features of the sequence from a higher span.

[0084] Pooling layers are used to sample data. They scan the input data matrix using a window of length q, replacing the entire window's data with the maximum or average value of the elements within the window, thus shortening the data length. This allows lower convolutional layers to observe data features over a wider span. Pooling layer P1 uses max pooling, where the max function processes the submatrix formed by each row of data, calculating the maximum value in the submatrix to obtain a vector composed of these maximum values. Pooling layer P2 operates similarly to P1, using global max pooling, where the maximum value in each column of the matrix represents the entire column. Essentially, it builds upon P1's pooling algorithm by setting the window length q to equal the number of rows in the matrix. After global max pooling, the matrix generates a one-dimensional vector z that can be processed by fully connected layers, containing high-order features of the original input data.

[0085] The fully connected output layer is used to learn the mapping relationship between the vector z and k known business scenario categories. The mathematical model is:

[0086] y = f(vz + B)

[0087] Where v is the weight matrix; B is the bias vector; and f is the activation function. The operation will output y, representing the probability that the original input sample is classified into a certain data stream category.

[0088] In one embodiment, the business scenario classification model uses cross-entropy loss as the loss function and the root mean square propagation (RMSprop) algorithm as the optimizer. Before training begins, the weights of each layer in the model are randomized. During training, backpropagation and mini-batch gradient descent algorithms are used to progressively update the weights. After multiple rounds of training and testing, the optimal weights are selected as the model parameters.

[0089] The trained business scenario classification model supports importing input data matrices with empty label fields. The model determines which business scenario the input data belongs to, matches the corresponding channel transaction code and business scenario description, and obtains a prediction result matrix.

[0090] In one embodiment, step 106 above, filling each row of the prediction result matrix with a flow-cutting suggestion to obtain a supplementary recording matrix, may include:

[0091] In each row of the prediction result matrix, check whether the channel transaction code field and the business scenario description field are not empty, and whether the service effectively covers the field.

[0092] If the channel transaction code field and business scenario description field in a row are not empty, and the service effective coverage field is yes, then fill in the whether to recommend switching traffic field and set the whether to recommend switching traffic field to yes;

[0093] If the channel transaction code field and business scenario description field in a row are not empty, and the service effective coverage field is no, then fill in the whether to recommend switching traffic field and set the whether to recommend switching traffic field to no.

[0094] In one embodiment, step 107 above, generating a business scenario list based on a supplementary entry matrix according to preset fields, may include: generating a business scenario list based on a supplementary entry matrix according to the value of a "whether to suggest switching traffic" field, and transmitting the business scenario list to the front end for display.

[0095] In summary, the distributed service switching recommendation method provided in this embodiment of the invention can predict business scenarios, monitor the effective service coverage on the transaction link, generate suggestions on whether to switch traffic, and ensure the effectiveness and reliability of distributed service switching under architecture transformation.

[0096] This invention also provides a distributed service flow switching recommendation device, as described in the following embodiments. Since the principle by which this device solves the problem is similar to that of the distributed service flow switching recommendation method, the implementation of this device can refer to the implementation of the distributed service flow switching recommendation method; repeated details will not be elaborated further.

[0097] Figure 4 This is a schematic diagram of the distributed service flow switching and recommendation device in an embodiment of the present invention, as shown below. Figure 4 As shown, the device includes:

[0098] Data acquisition module 01 is used to obtain transaction log data;

[0099] Data processing module 02 is used to integrate transaction log data within a preset transaction time, based on the transaction time, to obtain integrated data; and to add a field indicating whether the service is effectively covered based on the status of the flow switching switch and the record status of the flow switching switch in the integrated data.

[0100] The feature extraction module 03 is used to extract a preset number of message fields from the integrated data as feature value fields, and obtain the input data matrix based on the feature value fields;

[0101] The business scenario identification module 04 is used to import the input data matrix into the business scenario classification model, predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fill them into the input data matrix as labels to obtain the prediction result matrix.

[0102] The flow cut suggestion supplementary module 05 is used to fill each row of the prediction result matrix with a flow cut suggestion field to obtain the supplementary matrix;

[0103] The scenario list generation module 06 is used to generate a business scenario list based on a supplementary matrix according to preset fields. The business scenario list includes: channel transaction code, business scenario description, service coverage field, and traffic switching suggestion field.

[0104] In one embodiment, the data acquisition module 01 is specifically used for:

[0105] Collect data routing information, flow switching status, and full transaction log information for multiple transactions at each stage of the processing chain, and combine them into transaction log source data;

[0106] The source data of the transaction logs is processed to obtain the transaction log data.

[0107] Figure 5 This is a schematic diagram of the structure of the data acquisition module 01 of the distributed service flow switching recommendation device in one embodiment of the present invention. Figure 5 As shown, the data acquisition module 01 includes:

[0108] The data routing acquisition submodule 011 is used to locally store the data routing information of multiple transactions across various platform applications in the processing chain.

[0109] The flow switching acquisition submodule 012 is used to obtain the flow switching status of various distributed platform services on the processing link of multiple transactions in real time by calling the application programming interface provided by the configuration center system.

[0110] The transaction log collection submodule 013 is used to establish a connection and obtain the full transaction log information registered in the transaction logbook of each platform application within a certain time range in near real-time, based on the data routing information stored locally and according to the data routing strategy.

[0111] In one embodiment, the data processing module 02 is specifically used for:

[0112] Perform data dimensionality reduction and data cleaning on multiple full transaction log information in the transaction log source data;

[0113] The data dimensionality reduction includes: removing irrelevant fields and / or retaining key fields;

[0114] The data cleaning includes: removing irrelevant information from key fields and / or unifying and simplifying field formats.

[0115] In one embodiment, the data processing module 02 is specifically used for:

[0116] Transaction log data is divided into product data and backend processing data;

[0117] Based on the transaction time, within a preset transaction time, all product data are concatenated with all corresponding backend processed data to obtain multiple concatenated data sets.

[0118] The concatenated data is then deduplicated to obtain multiple integrated datasets.

[0119] In one embodiment, the data processing module 02 is specifically used for:

[0120] Check one by one whether the flow switching switch of all the integrated data is turned on, and record the status as successful;

[0121] When the data integration flow switch is on and the recording status is successful, add a "whether to effectively cover" field after the original data field of the integrated data, and set the "whether to effectively cover" field to "yes".

[0122] When the stream switching switch for integrated data is off, or the recording status is failed, add a "whether it is effectively overwritten" field after the original data field of the integrated data, and set the "whether it is effectively overwritten" field to "no".

[0123] In one embodiment, the feature extraction module 03 is specifically used for:

[0124] Extract a predetermined number of message fields from the integrated data as feature value fields to obtain a feature value field matrix;

[0125] Labels are added to the feature value field matrix, the format is converted, and the normalization process is performed to obtain the input data matrix.

[0126] In one embodiment, the feature extraction module 03 is specifically used for:

[0127] Add a label field to the end of each row of the eigenvalue field matrix;

[0128] The conversion formats include converting Boolean feature value fields to binary values, and / or converting text-type feature value fields using a lexical hypothesis approach.

[0129] In one embodiment, the feature extraction module 03 is specifically used for:

[0130] The feature value fields after each conversion format are normalized according to the following formula:

[0131]

[0132] Where x represents the eigenvalue after format conversion, μ represents the expected value of the eigenvalue after format conversion, and δ represents the standard deviation of the eigenvalue after format conversion. This represents the feature value field after standard normalization.

[0133] In one embodiment, the flow switching suggestion and supplementation module 05 is specifically used for:

[0134] Check each field in the matching result matrix to see if the channel transaction code field and the business scenario description field are not empty, and whether the effective coverage field is yes.

[0135] If the channel transaction code field and business scenario description field in a row are not empty, and the whether to effectively cover the field is yes, then fill in the whether to recommend switching traffic field and set the whether to recommend switching traffic field to yes;

[0136] If the channel transaction code field and business scenario description field in a row are not empty, and the whether to effectively cover field is no, then fill in the whether to recommend switching traffic field and set the whether to recommend switching traffic field to no.

[0137] In one embodiment, the scenario list generation module 06 is specifically used for:

[0138] Based on the value of the "Whether to recommend switching traffic" field, a list of business scenarios is generated according to the supplementary entry matrix, and the list of business scenarios is sent to the front end for display.

[0139] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described distributed service flow switching recommendation method.

[0140] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described distributed service flow switching recommendation method.

[0141] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described distributed service flow switching recommendation method.

[0142] Compared with existing technologies that rely on manual verification of link switching status, leading to high costs, low efficiency, poor reliability, and difficulty in effectively tracking the specific data flow behind each transaction, resulting in ineffective service coverage, this invention addresses these issues by: obtaining transaction log data; integrating the transaction log data within a preset transaction timeframe to obtain integrated data; adding a "service coverage" field based on the switching switch status and recording state in the integrated data; extracting a preset number of message fields from the integrated data as feature value fields; obtaining an input data matrix based on these feature value fields; importing the input data matrix into a business scenario classification model to predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and filling these as labels into the input data matrix to obtain a prediction result matrix; filling each row of the prediction result matrix with a "whether to switch traffic" suggestion field to obtain a supplementary entry matrix; and generating a business scenario list based on the supplementary entry matrix according to preset fields. The business scenario list includes: channel transaction code, business scenario description, "service coverage" field, and "whether to switch traffic" suggestion field. It can predict business scenarios, monitor the effective service coverage on the transaction chain, generate suggestions on whether to switch traffic, and ensure the effectiveness and reliability of distributed service traffic switching under architecture transformation.

[0143] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0145] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0146] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0147] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A distributed service flow switching recommendation method, characterized in that, include: Obtain transaction log data; Based on the transaction time, within a preset transaction time, the transaction log data is integrated and processed to obtain integrated data; Based on the status of the flow switching switch and its record status in the integrated data, add a field indicating whether the service is effectively covered. Specifically, based on the status of the flow switching switches in the integrated data and the record status of the flow switching switches, a field for whether the service is effectively covered is added, including: Check one by one whether the flow switching switch of all the integrated data is turned on, and record whether the status is successful; When the data integration flow switching switch is on and the recording status is successful, add a service effective overwrite field after the original data field of the integrated data, and set the service effective overwrite field to yes. When the flow switching switch for integrated data is off, or the recording status is failed, add a "Whether the service is effectively covered" field after the original data field of the integrated data, and set the "Whether the service is effectively covered" field to "No". A predetermined number of message fields are extracted from the integrated data as feature value fields, and an input data matrix is ​​obtained based on the feature value fields. The input data matrix is ​​imported into the business scenario classification model to predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and these are used as labels to fill in the input data matrix to obtain the prediction result matrix. Fill each row of the prediction result matrix with a "whether to switch flow" field to obtain the supplementary recording matrix; The process of filling each row of the prediction result matrix with a flow cut suggestion yields a supplementary recording matrix, including: In each row of the prediction result matrix, check whether the channel transaction code field and the business scenario description field are not empty, and whether the service effectively covers the field. If the channel transaction code field and the business scenario description field in a row are not empty, and the service effective coverage field is yes, then fill in the whether to switch traffic suggestion field and set the whether to switch traffic suggestion field to yes; If the channel transaction code field and the business scenario description field in a row are not empty, and the service effective coverage field is no, then fill in the whether to switch traffic suggestion field and set the whether to switch traffic suggestion field to no. Based on preset fields, a business scenario list is generated according to the supplementary entry matrix. The business scenario list includes: channel transaction code, business scenario description, field indicating whether the service is effectively covered, and field indicating whether to switch traffic.

2. The method as described in claim 1, characterized in that, Obtain transaction log data, including: Collect data routing information, flow switching status, and full transaction log information for multiple transactions at each stage of the processing chain, and combine them into transaction log source data; The source data of the transaction logs is processed to obtain the transaction log data.

3. The method as described in claim 2, characterized in that, Collect data routing information, flow switching status, and full transaction log information for multiple transactions at each stage of the processing chain, including: Localize and store the data routing information of multiple transactions across various platform applications in the processing chain; By calling the application programming interface provided by the configuration center system, the flow switching status of multiple transactions on various distributed platform services in the processing link can be obtained in real time. Based on the data routing information stored locally, and according to the data routing strategy, a connection is established to obtain, in near real-time, the full transaction log information registered in the transaction logbook of each platform application within a certain time range.

4. The method as described in claim 2, characterized in that, Processing the source data of the transaction logs includes: Perform data dimensionality reduction and data cleaning on multiple full transaction log information in the transaction log source data; The data dimensionality reduction includes: removing irrelevant fields and / or retaining key fields; The data cleaning includes: removing irrelevant information from key fields and / or unifying and simplifying field formats.

5. The method as described in claim 1, characterized in that, Based on the transaction time, within a preset transaction period, transaction log data is integrated and processed to obtain integrated data, including: Transaction log data is divided into product data and backend processing data; Based on the transaction time, within a preset transaction time, all product data are concatenated with all corresponding backend processed data to obtain multiple concatenated data sets. All concatenated data is deduplicated to obtain integrated data.

6. The method as described in claim 5, characterized in that, The product data includes, but is not limited to: corporate loan applications, corporate deposit applications, personal deposit applications, and personal loan applications; The back-end processed data includes, but is not limited to: settlement applications, accounting applications, and clearing applications; The product data corresponds to one or more backend processing data.

7. The method as described in claim 1, characterized in that, Extract a predetermined number of message fields from the integrated data as feature value fields. Based on the feature value fields, obtain the input data matrix, including: Extract a predetermined number of message fields from the integrated data as feature value fields to obtain a feature value field matrix; Labels are added to the feature value field matrix, the format is converted, and the normalization process is performed to obtain the input data matrix.

8. The method as described in claim 7, characterized in that, The addition of tags includes: Add a label field to the end of each row of the eigenvalue field matrix; The format conversion includes: converting Boolean type feature value fields into binary value formats, and / or converting text type feature value fields using a lexical hypothesis method.

9. The method as described in claim 8, characterized in that, The normalization process includes: The feature value fields after each format conversion are normalized according to the following formula: in, This represents the feature value after format conversion. This represents the expected value of the eigenvalues ​​after format conversion. This represents the standard deviation of the eigenvalues ​​after format conversion. This represents the feature value field after standard normalization.

10. The method as described in claim 1, characterized in that, The business scenario classification model includes: First convolutional layer, second convolutional layer, first pooling layer, second pooling layer, fully connected output layer.

11. The method as described in claim 1, characterized in that, Based on preset fields, a list of business scenarios is generated from the supplementary data matrix, including: Based on the value of the "Whether to switch traffic" suggestion field, a list of business scenarios is generated according to the supplementary entry matrix, and the list of business scenarios is sent to the front end for display.

12. A distributed service flow-switching intelligent recommendation device, characterized in that, include: The data acquisition module is used to obtain transaction log data; The data processing module is used to integrate and process transaction log data within a preset transaction time, based on the transaction time, to obtain integrated data; and to add a field indicating whether the service is effectively covered based on the status of the flow switching switch and the record status of the flow switching switch in the integrated data. The data processing module is specifically used for: Check one by one whether the flow switching switch of all the integrated data is turned on, and record the status as successful; When the data integration flow switch is on and the recording status is successful, add a "whether to effectively cover" field after the original data field of the integrated data, and set the "whether to effectively cover" field to "yes". When the stream switching switch for integrated data is off, or the recording status is failed, add a "whether it is effectively overwritten" field after the original data field of the integrated data, and set the "whether it is effectively overwritten" field to "no". The feature extraction module is used to extract a preset number of message fields from the integrated data as feature value fields, and obtain the input data matrix based on the feature value fields; The business scenario identification module is used to import the input data matrix into the business scenario classification model, predict the channel transaction code and business scenario description corresponding to each row of the input data matrix, and fill them into the input data matrix as labels to obtain the prediction result matrix. The flow-cutting suggestion supplementation module is used to fill each row of the prediction result matrix with a flow-cutting suggestion field to obtain the supplementation matrix. The stream switching suggestion supplement module is specifically used for: Check each field in the matching result matrix to see if the channel transaction code field and the business scenario description field are not empty, and whether the effective coverage field is yes. If the channel transaction code field and business scenario description field in a row are not empty, and the whether to effectively cover field is yes, then fill in the whether to switch traffic suggestion field and set the whether to switch traffic suggestion field to yes; If the channel transaction code field and business scenario description field in a row are not empty, and the whether to effectively cover field is no, then fill in the whether to switch traffic suggestion field and set the whether to switch traffic suggestion field to no. The scenario list generation module is used to generate a business scenario list based on a supplementary matrix according to preset fields. The business scenario list includes: channel transaction code, business scenario description, service coverage field, and traffic switching suggestion field.

13. A computer 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 computer program, it implements the method of any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 11.

15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 11.

Images