Strategy iteration method, device and system based on strategy robot

Abstract

The present invention provides a policy iteration method, device, and system based on a policy robot. The method comprises: performing feature backtracking on the left table of each imported sample based on a feature backtracking automatic processing tool to generate feature wide table data for the policy robot's corresponding service business; requesting processing resources in the policy robot's online operating environment and configuring an algorithm model and rule parameters for policy mining based on the read feature wide table data; executing policy mining for the policy robot's corresponding service business; and iteratively outputting target policy information after performing rule evaluation on the mined policy set. The present invention can automate feature backtracking processing, execute policy mining for the policy robot's corresponding service business, and automate the processing portion of policy mining, thereby shortening processing time and increasing model output speed.

Description

Technical Field

[0001] The present invention relates to the field of computer information processing, and in particular to a policy iteration method, device, system and electronic equipment based on a policy robot. Background Art

[0002] In the existing technology, during the model construction process, the cumulative number of feature quantities or variables in the feature system of different business scenarios is large (for example, the feature quantities of the resource usage period in a certain application scenario have accumulated to more than 6,000 variables, etc.), but only a small part of them are actually applied to the model. In model construction, feature screening, and other data processing, relevant business personnel mainly rely on manual feature screening and model construction, which leads to low efficiency and long time consumption. In addition, for feature-based model construction, there is no tool for automatic strategy mining rules, which cannot realize the visualization and automation of automated mining.

[0003] In order to achieve automation and visualization of model building, maximize the output of feature value in various business scenarios, and generate additional gains on existing models and rule systems, it is necessary to provide an improved policy iteration method. Summary of the Invention

[0004] In order to solve the problems in existing model construction such as the lack of tools for automatic strategy mining rules, the inability to realize visualization and automation of automated mining, low efficiency and long time consumption of the model building process, etc.

[0005] The present invention provides a policy iteration method based on a policy robot, comprising: performing feature backtracking on the left table of each imported sample based on a feature backtracking automatic processing tool to generate feature wide table data of the corresponding service business of the policy robot; requesting processing resources in the online operation environment of the policy robot and configuring an algorithm model and rule parameters for policy mining based on the read feature wide table data; executing policy mining of the corresponding service business of the policy robot; performing rule evaluation on the mined policy set and iteratively outputting target policy information.

[0006] According to an optional implementation, based on the feature backtracking automatic processing tool, the features of the designated sample left table for each import are automatically backtracked to generate feature wide table data of the corresponding service business of the strategy robot, including: receiving the sample left table imported by the user through a pre-created visual interactive data source management page; automatically deploying the workflow according to the feature backtracking automatic processing tool, and performing feature backtracking for strategy mining on the sample left table for each import by triggering the feature backtracking button provided by the data source management page; when the feature backtracking is completed, generating the feature wide table data of the corresponding service business and outputting the running status on the data source management page.

[0007] According to an optional implementation, the sample left table includes at least: customer number, backtracking date and / or application number primary key; outputting the running status includes: displaying the completed status and displaying the name of the generated feature wide table data, wherein displaying the name of the generated feature wide table data includes adding a suffix to the sample left table name imported by the user; when the feature backtracking is completed, it also includes: sending an email and / or message to the user for reminder.

[0008] According to an optional implementation, the data source management page further includes: before the user imports the sample left table, the user searches the sample left table and / or specifies the sample left table based on the interaction of the data source management page and performs a necessity check and specifies the predicted label column for the feature backtracking button to be triggered; when the feature backtracking button is triggered, the submission, running, suspension and output processing of the feature backtracking are executed.

[0009] According to an optional implementation, requesting processing resources in the online running environment of the strategy robot includes: when generating the feature wide table data, entering a pre-created visual interactive data source management page; automatically requesting greater than or equal to a predetermined amount of memory resources in the online running environment of the strategy robot to support automatic reading and cleaning of features and sample data sets corresponding to the sample left table, and automatically requesting greater than or equal to a predetermined amount of multi-threaded computing resources to support the fast computing processing speed of the algorithm model.

[0010] According to an optional implementation, the algorithm model and rule parameters for strategy mining are configured based on the read feature wide table data, including: reading the features in the generated feature wide table data; setting corresponding algorithm models and rule parameters for strategy mining corresponding to the features; executing strategy mining for corresponding service businesses, including: triggering the strategy mining toolkit code to run automatically, and executing strategy mining of the features of the feature wide table data for the corresponding service business of the strategy robot according to the set algorithm model and rule parameters.

[0011] According to an optional implementation method, the algorithm model and rule parameters for strategy mining are configured, including: the user configures the process parameters for strategy mining, the hyperparameters of the algorithm model, the loss function parameters of the algorithm model, and the restriction parameters related to strategy mining and output based on the interaction of the user with the data source management page.

[0012] According to an optional implementation, the data source management page further includes: based on the interaction with the data source management page, the user creates, edits, deletes, runs or pauses a strategy mining project; and / or, based on the interaction with the data source management page, the user specifies, presents or selects feature data based on feature wide table data after feature backtracking.

[0013] According to an optional implementation method, the target policy information is iteratively output after rule evaluation is performed on the mined policy set, including: after rule policy extraction and evaluation are performed on the rule policy mined based on the algorithm model and rule parameters configured based on the read feature wide table data, a chart is generated and displayed on the front-end page; wherein, the chart includes a rule cumulative effect evaluation table, a rule stability evaluation table and a rule logic table through analysis, as preferred rule policy information, i.e., target policy information.

[0014] According to an optional implementation method, outputting the target strategy information includes: according to the target strategy information determined this time, performing iterations on the feature wide table data generated after backtracking the left table features of the next imported sample for configuring the algorithm model and rule parameters for strategy mining and executing the strategy mining of the corresponding service business of the strategy robot; until the user stops the iteration or reaches the preset number of iterations; and outputting the target strategy information after the iteration is completed.

[0015] In addition, the second aspect of the present invention also provides a policy iteration device based on a policy robot, including: a feature backtracking unit, which is used to perform feature backtracking on the left table of each imported sample based on a feature backtracking automatic processing tool to generate feature wide table data of the corresponding service business of the policy robot; a strategy mining unit, which is used to request processing resources in the online running environment of the policy robot and configure the algorithm model and rule parameters for strategy mining based on the read feature wide table data; an execution unit, which is used to execute strategy mining of the corresponding service business of the policy robot; and an iterative output unit, which is used to perform rule evaluation on the mined policy set and iteratively output target policy information.

[0016] In addition, the third aspect of the present invention also provides a strategy robot system, including: pre-creating a visual interactive data source management page and a visual interactive data source management page; the user starts the data source management page and imports the sample left table of this time, executes the feature backtracking method described in the first aspect of the present invention, and generates feature wide table data of the corresponding service business of the strategy robot; and, when the feature wide table data is generated, enters the data source management page and executes the strategy mining method described in the first aspect of the present invention; and, after performing rule evaluation on the mined strategy set, iteratively outputs the target strategy information, executes the output and iteration described in the first aspect of the present invention, and outputs the target strategy information after the iteration is completed.

[0017] In addition, the fourth aspect of the present invention also provides an electronic device, comprising: a processor; and a memory storing computer-executable instructions, wherein the computer-executable instructions, when executed, enable the processor to execute the method according to the present invention.

[0018] In addition, the fifth aspect of the present invention also provides a computer-readable medium, wherein the computer-readable storage medium stores one or more programs, and when the one or more programs are executed by a processor, the method described in the present invention is implemented.

[0019] Beneficial effects

[0020] Compared with the prior art, the present invention performs feature backtracking on the sample left table of each import based on a feature backtracking automatic processing tool to generate feature wide table data of the corresponding service business of the policy robot. The user only needs to input the sample left table each time to complete the feature backtracking, thereby realizing the automation of feature backtracking processing; requesting processing resources in the online operation environment of the policy robot and configuring the algorithm model and rule parameters for policy mining based on the read feature wide table data, after the parameter configuration, executing the policy mining of the corresponding service business of the policy robot, so that the processing part of the policy mining is automated, which can shorten the processing time and improve the output speed of the model. By iteratively outputting the target policy information after the rule evaluation of the mined policy set, the target policy information can be determined, and the processing speed of the entire policy mining can be improved, and the automation of policy mining can also be realized.

[0021] In addition, by deploying feature backtracking processing into a workflow, feature backtracking for strategy mining can be completed with one click, which can realize the automation of feature backtracking; providing a visual interactive data source management page, thereby realizing the visualization and automation of various operations on the visual interactive data source management page.

[0022] In addition, in various data processing processes such as request processing resources, feature wide table data acquisition, feature screening, model output and selection, multi-threaded computing is used to achieve this. While effectively processing data, it can further improve the calculation processing speed and effectively ensure the output speed of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to make the technical problems solved by the present invention, the technical means adopted, and the technical effects achieved more clearly, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, it should be noted that the drawings described below are only drawings of exemplary embodiments of the present invention. Those skilled in the art can derive drawings of other embodiments based on these drawings without inventive effort.

[0024] Figure 1 This is a flowchart of an example of a policy iteration method based on a policy robot according to the first embodiment of the present invention.

[0025] Figure 2 This is a flowchart of another example of the policy iteration method based on the policy robot according to the first embodiment of the present invention.

[0026] Figure 3 This is a flowchart of another example of the policy iteration method based on the policy robot of embodiment 1 of the present invention.

[0027] Figure 4 It is a schematic diagram of an example of a policy iteration device based on a policy robot according to embodiment 2 of the present invention.

[0028] Figure 5 It is a schematic diagram of another example of a policy iteration device based on a policy robot according to embodiment 2 of the present invention.

[0029] Figure 6 is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention.

[0030] Figure 7 is a block diagram of an exemplary embodiment of a computer-readable medium according to the present invention. DETAILED DESCRIPTION

[0031] Exemplary embodiments of the present invention will now be described more fully with reference to the accompanying drawings. However, exemplary embodiments can be implemented in various forms, and it should not be understood that the present invention is limited to the embodiments set forth herein. On the contrary, providing these exemplary embodiments enables the present invention to be more comprehensive and complete, making it easier to fully convey the inventive concept to those skilled in the art. In the figures, the same reference numerals represent the same or similar elements, components or parts, and thus their repeated description will be omitted.

[0032] Under the premise of being consistent with the technical concept of the present invention, the features, structures, characteristics or other details described in a specific embodiment do not exclude that they can be combined in one or more other embodiments in a suitable manner.

[0033] In the description of specific embodiments, the features, structures, characteristics, or other details of the present invention are described to enable those skilled in the art to fully understand the embodiments. However, this does not preclude those skilled in the art from practicing the technical solutions of the present invention without one or more of the specific features, structures, characteristics, or other details.

[0034] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0035] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0036] It should be understood that while the terms "first," "second," and "third" may be used herein to describe various devices, elements, components, or parts, this should not be construed as limiting. These terms are used to distinguish one from another. For example, a first device could also be referred to as a second device without departing from the essential technical solution of the present invention.

[0037] The term "and / or" or "and / or" includes any one and all combinations of one or more of the associated listed items.

[0038] In view of the above problems, the present invention provides a policy iteration method based on a policy robot. This method generates feature wide table data of the corresponding service business of the policy robot by performing feature backtracking. The user only needs to input the sample left table each time to complete the feature backtracking, thereby realizing the automation of feature backtracking processing; by configuring the algorithm model and rule parameters for policy mining, and after the parameter configuration, executing the policy mining of the corresponding service business of the policy robot, the processing part of the policy mining is automated, which can shorten the processing time and increase the output speed of the model. By iteratively outputting the target policy information after performing rule evaluation on the mined policy set, the target policy information can be determined, the processing speed of the entire policy mining can be improved, and the automation of policy mining can be realized.

[0039] The method flow of the present invention will be described in detail below with reference to specific examples.

[0040] Example 1

[0041] Below, we will refer to Figures 1 to 3 An embodiment of the policy iteration method based on a policy robot of the present invention is described.

[0042] Figure 1 FIG. 1 is a flow chart of the policy iteration method based on the policy robot of the present invention. Figure 1 As shown, a policy iteration method based on a policy robot includes the following steps.

[0043] Step S101 : Based on the feature backtracking automatic processing tool, feature backtracking is performed on the left table of each imported sample to generate feature wide table data of the corresponding service business of the strategy robot.

[0044] Step S102, requesting processing resources in the online operating environment of the policy robot and configuring an algorithm model and rule parameters for policy mining based on the read feature wide table data;

[0045] Step S103: Execute strategy mining for corresponding service business.

[0046] Step S104: perform rule evaluation on the mined policy set and iteratively output target policy information.

[0047] First, in step S101, based on the feature backtracking automatic processing tool, feature backtracking is performed on the left table of each imported sample to generate feature wide table data of the corresponding service business of the strategy robot.

[0048] To automate online policy determination, we automate each step of the process. Specifically, we deploy feature backtracking as a workflow. Users only need to enter the sample table to complete feature backtracking. For example, feature backtracking for rule mining can be completed with a single click.

[0049] In one embodiment, a robot system is established, which includes a visual interactive data source management page. The data source management page is pre-created and is a visual interactive data source management page. The data source management page includes user input items, operable trigger buttons, etc., wherein, for example, the user input items include query input items, feature input items, etc. For example, the operable trigger buttons include click buttons, edit buttons, modify buttons, feature backtracking buttons, and delete buttons. Specifically, on the visual interactive data source management page, users input, edit, modify, delete, trigger or perform other operations through user input items and / or operable trigger buttons. Furthermore, based on the interaction with the data source management page, users can create, edit, delete, run or pause policy mining projects. Thus, it is possible to visualize and automate various operations on the visual interactive data source management page.

[0050] Specifically, the system receives the sample left table imported by the user through a pre-created visual and interactive data source management page. Furthermore, the fixed feature backtracking is automatically deployed as a workflow based on the relational database structured query language (SQL). By triggering the feature backtracking button provided on the data source management page, feature backtracking (also known as feature backtracking processing) for strategy mining is performed on each imported sample left table.

[0051] Optionally, the sample left table includes at least: a customer number cust_no, a recall date recall_date and / or an application number appl_no primary key.

[0052] Furthermore, when the feature backtracking is completed (ie, the feature backtracking process is completed), the features in the feature wide table data of the corresponding service business of the strategy robot are generated, and the running status is output on the data source management page.

[0053] It should be noted that in this embodiment, the business data includes business scenario parameters, positive sample data, negative sample data, and feature parameters, etc. However, in other embodiments, the business data may also include evaluation parameters related to model training and optimization, or at least two types of data from the feature wide table data, etc. The above is provided as an optional example only and should not be construed as limiting the present invention.

[0054] Specifically, the output running status includes: displaying the completed status and displaying the name of the generated feature wide table data, wherein displaying the name of the generated feature wide table data includes adding a suffix such as results after the left table name of the sample imported by the user.

[0055] In one embodiment, for example, business person A imports sample left table 1 on the data source management page. The following information from sample left table 1 is displayed on the data source management page: customer number cust_no is CT62***9584, recall date recall_date is 2021-4-11, and application number appl_no_draw is JT2021041100***9503. In addition, the data management page also includes other displays related to queries and operations. After importing sample left table 1, business person A clicks the trigger button, which activates the strategy robot or other related operation modules and devices to automatically perform feature backtracking for strategy mining.

[0056] Optionally, when the feature backtracking is completed, an email or message is sent to the user to remind him / her.

[0057] In another embodiment, before the user imports the sample left table, the user searches for the sample left table and / or specifies the sample left table by inputting on the data source management page, that is, the user searches for the sample left table and / or specifies the sample left table based on the interaction on the data source management page.

[0058] Furthermore, when the feature backtracking button is triggered, the submission, operation, suspension and output processing of the feature backtracking are executed. Thus, the visual search query process and other various operations can be realized by inputting, editing, modifying, deleting, triggering or other operations through user input items and / or operable trigger buttons.

[0059] For example, the user enters business scenario parameters (such as resource usage period, resource return period, etc.) and time parameters (such as start time and end time, or a time period including a specified time period, a time period including a specified user group, etc.) on the data source management page to search for a sample left table or a specified sample left table.

[0060] Preferably, after feature backtracking is complete, the triggering feature backtracking key is checked for necessity and a predicted label column is assigned. For example, based on the determined test results, the original primary key is deleted or a new primary key is added. Another example is assigning labels to sample data and using the assigned labels to define positive and negative samples.

[0061] In addition, when the feature backtracking process is completed, feature wide table data is generated, which includes feature type, feature quantity, sample quantity, missing rate and sparse rate of each feature, deviation, etc.

[0062] Specifically, the feature wide table data is used to determine strategy mining rules or operating conditions.

[0063] It should be noted that in this embodiment, the feature-wide table data may be offline data. The feature-wide table data refers to feature-related data in the feature-wide table, specifically including each feature, feature type, number of samples, missing rate and sparse rate of each feature, etc. The above is for illustration only and is not to be construed as limiting the present invention.

[0064] Furthermore, based on the interaction on the data source management page, the user specifies, presents or selects feature data according to the feature wide table data after feature backtracking.

[0065] It should be noted that the above description is only provided as an example and should not be construed as limiting the present invention.

[0066] Next, in step S102, processing resources are requested in the online operating environment of the policy robot and an algorithm model and rule parameters for policy mining are configured based on the read feature wide table data.

[0067] In this embodiment, the robot system further includes a strategy robot, and the strategy robot is used for strategy mining.

[0068] Specifically, in the online operating environment of the policy robot, the user requests processing resources to obtain the generated feature wide table data, or obtain other resources.

[0069] Preferably, the policy robot automatically requests a predetermined amount of memory resources within its online runtime environment to support the automatic reading and cleaning of the features and sample data sets corresponding to the sample left table. This means sufficient memory allocation. This ensures that, for example, datasets with more than 5,000 features and more than 500,000 sample data points can be automatically read and cleaned.

[0070] It should be noted that, in this embodiment, the cleaning (ie, data cleaning) mainly includes missing value filling, time field processing, category variable encoding, etc. However, it is not limited thereto, and other processing methods are also included in other embodiments.

[0071] Furthermore, an algorithm model (the algorithm model is used for strategy mining) and rule parameters are configured based on the read feature wide table data.

[0072] Specifically, the features in the feature wide table data are read, and the corresponding algorithm model and rule parameters for strategy mining are configured for the features. For example, the feature wide table data during resource usage is read (the feature wide table data is, for example, a wide table including user resource application data, user quality data, transaction data, expenditure data, etc. within a specific time period), and each feature is generated (for example, resource application feature A, user quality feature B, transaction feature C, and expenditure feature D, etc.), and the corresponding rule parameters are configured or set according to each generated feature, and the algorithm model is configured. Wherein, the configured rule parameters include configuring or setting at least one of the following parameters: the number of samples, the number of features, the category / numerical features, the selected features, the missing rate and sparse rate of each feature, the degree of deviation, the dimension threshold and correlation threshold of each feature, setting the maximum allowable value of the difference in the number of positive samples and negative samples, process parameters, hyperparameters and loss function parameters of the algorithm model, and other restriction parameters, etc. In addition, the configuration rule parameters also include configuring or setting multiple rules such as the first rule 1 (rule 1), the second rule (rule 2), and the logical relationship between any two rules such as or, and, etc.

[0073] In one embodiment, the user configures process parameters for strategy mining, hyperparameters of the algorithm model, loss function parameters of the algorithm model, and restriction parameters related to strategy mining and output based on interaction with the data source management page.

[0074] It should be noted that the above is merely provided as an optional example and should not be construed as a limitation to the present invention.

[0075] In order to further improve the computing processing speed while effectively processing data, multi-threaded computing is used to achieve this in various data processing processes, such as requesting processing resources, acquiring feature wide table data, feature screening, model output and selection, for example, supporting more than 100 threads.

[0076] Specifically, the data processing involved in the present invention (e.g., robotic systems, policy robots, etc.) is implemented through multi-threaded computing. Optionally, a predetermined number of multi-threaded computing resources are automatically requested to support the rapid computational processing speed of the algorithm model. Therefore, during the model generation process, multi-threaded computing can effectively ensure the speed of model generation.

[0077] In another embodiment, Figure 2 As shown, the process also includes entering a pre-created visual interactive data source management page to request processing resources when generating the feature wide table data (ie, step S201).

[0078] Specifically, in step S201, when the feature wide table data is generated, a pre-created visual interactive data source management page is entered to request processing resources.

[0079] More specifically, in the online operating environment of the policy robot, the user requests the policy robot to process resources, and the policy robot displays the returned resource data on a visually interactive data source management page for the user to read, further filter or use.

[0080] Next, in step S103, strategy mining of the corresponding service business of the strategy robot is performed.

[0081] Specifically, after configuring the algorithm model and rule parameters for strategy mining, the strategy mining of the corresponding service business of the strategy robot is performed.

[0082] In this embodiment, after completing the configuration of the algorithm model and rule parameters, the strategy mining of the corresponding service business of the strategy robot is executed, that is, the strategy mining toolkit code is triggered to automatically run after the configuration.

[0083] Specifically, the strategy mining based on the features in the feature wide table data of the corresponding service business of the strategy robot is executed according to the configured algorithm model and rule parameters.

[0084] In another embodiment, by clicking a configuration completion confirmation button or an execution button on the data source management page, the policy mining of the corresponding service business of the policy robot is executed.

[0085] In another embodiment, after clicking the configuration completion confirmation button on the data source management page, business scenario parameters are input, that is, strategy mining of the corresponding service business of the strategy robot is executed.

[0086] This automates the process of strategy mining, shortens processing time, and increases the speed of model generation.

[0087] It should be noted that the above is merely provided as an optional example and should not be construed as a limitation to the present invention.

[0088] Next, in step S104, rule evaluation is performed on the mined policy set and target policy information is iteratively output.

[0089] Specifically, after executing the policy mining of the corresponding service business of the policy robot, a policy set is automatically generated, and the policy set includes a visualized process, a rule set, a rule identifier, and the like.

[0090] Furthermore, the mined policy sets (or rule policy sets) are evaluated based on, for example, effectiveness, stability, and logic. This results in a rule cumulative effectiveness evaluation table, a rule stability evaluation table, and a rule logic table. For example, evaluation is performed based on evaluation parameters such as single hit rate, single lift, hit rate, hit failure rate, and number of hit failures.

[0091] Preferably, the rule evaluation process can also be visually displayed on the data source management page.

[0092] Specifically, the mined policy set undergoes rule evaluation and then iteratively outputs target policy information. The rule policies mined based on the algorithm model and rule parameters configured using the read feature wide table data are extracted and evaluated, and a chart is generated and displayed on the front-end page. Furthermore, the iteratively output target policy information can be a single optimal policy or a rule set consisting of multiple preferred policies.

[0093] More specifically, the chart serves as target policy information, and the chart includes a rule accumulation effect evaluation table, a rule stability evaluation table, and a rule logic table.

[0094] Next, perform the following operations to determine the target policy information:

[0095] Step S301: Based on the target strategy information determined this time, the algorithm model and rule parameters for strategy mining are configured for the feature wide table data generated after the feature backtracking of the left table of the next imported sample, and the strategy mining iteration for the corresponding service business of the strategy robot is executed;

[0096] Step S302: until the user stops the iteration or the preset number of iterations is reached;

[0097] Step S303: Output the target strategy information after the iteration is completed.

[0098] Therefore, through the iterative determination process of the above-mentioned policy mining, target policy information can be determined.

[0099] It should be noted that the above description is only provided as an example and should not be construed as limiting the present invention.

[0100] Those skilled in the art will appreciate that all or part of the steps for implementing the above embodiments are implemented as a program (computer program) executed by a computer data processing device. When the computer program is executed, the above method provided by the present invention can be implemented. Moreover, the computer program can be stored in a computer-readable storage medium, which can be a readable storage medium such as a disk, an optical disk, a ROM, a RAM, or a storage array composed of multiple storage media, such as a disk or tape storage array. The storage medium is not limited to centralized storage, and can also be distributed storage, such as cloud storage based on cloud computing.

[0101] Compared with the existing technology, the present invention performs feature backtracking on the sample left table each time it is imported based on a feature backtracking automatic processing tool to generate feature wide table data of the corresponding service business of the strategy robot. The user only needs to input the sample left table each time to complete the feature backtracking, thereby realizing the automation of feature backtracking processing; in the online operation environment of the strategy robot, processing resources are requested and the algorithm model and rule parameters for strategy mining are configured based on the read feature wide table data. After the parameters are configured, the strategy mining of the corresponding service business of the strategy robot is executed, so that the processing part of the strategy mining is automated, which can shorten the processing time and improve the output speed of the model.

[0102] By iteratively outputting target policy information after rule evaluation on the mined policy set, the target policy information can be determined, the processing speed of the entire policy mining can be improved, and the automation of policy mining can be achieved.

[0103] In addition, by deploying feature backtracking processing into a workflow, feature backtracking for mining rules can be completed with one click, which can realize the automation of feature backtracking; providing a visual and interactive data source management page, thereby realizing the visualization and automation of various operations on the visual and interactive data source management page.

[0104] In addition, in various data processing processes such as request processing resources, feature wide table acquisition, feature screening, model output and selection, multi-threaded computing is used to achieve this. While effectively processing data, it can further improve the calculation processing speed and effectively ensure the output speed of the model.

[0105] Example 2

[0106] The following describes an apparatus embodiment of the present invention, which can be used to perform the method embodiment of the present invention. Details described in the apparatus embodiment of the present invention should be considered supplementary to the above-described method embodiment; details not disclosed in the apparatus embodiment of the present invention can be implemented with reference to the above-described method embodiment.

[0107] Reference Figure 4 and Figure 5 , the strategy iteration device 400 based on the strategy robot of the present invention will be described.

[0108] Specifically, if Figure 4 As shown, the strategy iteration device 400 includes a feature backtracking unit 401 , a strategy mining unit 402 , an execution unit 403 and an iteration output unit 404 .

[0109] Specifically, the feature backtracking unit 401 is used to perform feature backtracking on the left table of each imported sample based on the feature backtracking automatic processing tool to generate feature wide table data of the corresponding service business of the policy robot; the strategy mining unit 402 is used to request processing resources in the online running environment of the strategy robot and configure the algorithm model and rule parameters for strategy mining based on the read feature wide table data; the execution unit 403 is used to execute strategy mining of the corresponding service business; the iterative output unit 404 performs rule evaluation on the mined strategy set and iteratively outputs the target strategy information.

[0110] The specific processing functions of feature recall unit 401 correspond to step S101 of the method of Example 1. Specifically, feature recall unit 401 receives a sample left table imported by a user through a pre-created visual and interactive data source management page. The sample left table includes at least the primary keys of customer number cust_no, recall date recall_date, and / or application number appl_no.

[0111] Preferably, fixed feature backtracking is automatically deployed as a workflow based on the relational database structured query language SQL, and feature backtracking for strategy mining is performed on the left table of each imported sample by triggering the feature backtracking button provided on the data source management page.

[0112] When feature backtracking is completed, the features in the feature wide table data of the corresponding service business of the strategy robot are generated and the running status is output on the data source management page. The output running status includes displaying the completed status and displaying the name of the generated feature wide table data; wherein, the name of the generated feature wide table data includes adding the suffix "result" to the left table name of the sample imported by the user.

[0113] Optionally, when the feature backtracking is completed, an email or message is sent to the user to remind him / her.

[0114] In another embodiment, before the user imports the sample left table, the user interactively searches the sample left table and / or specifies the sample left table based on the data source management page, performs necessity check on the feature backtracking button to be triggered, and specifies the predicted label column.

[0115] The specific processing function of the strategy mining unit 402 corresponds to step S102 of the method of Example 1. Specifically, the strategy mining unit 402 is used to automatically request memory resources greater than or equal to a predetermined amount in the online running environment of the strategy robot to support the automatic reading and cleaning of the features and sample data sets corresponding to the sample left table.

[0116] Preferably, a fast operation processing speed of the algorithm model is supported by using multi-threaded computing resources that automatically request more than or equal to a predetermined number.

[0117] In this embodiment, when the feature backtracking button is triggered to execute, the submission, execution, suspension and output processing of the feature backtracking are executed.

[0118] When the feature backtracking is completed, the feature wide table data is generated, and when the feature wide table data is generated, a pre-created visual interactive data source management page is entered.

[0119] Furthermore, on the visual interactive data source management page, the generated feature wide table's feature data is read; corresponding algorithm models and rule parameters for strategy mining are set accordingly to the feature data. Specifically, based on interactions on the data source management page, users configure process parameters for strategy mining, algorithm model hyperparameters, algorithm model loss function parameters, and restriction parameters related to strategy mining and output.

[0120] Next, the execution unit 403 performs policy mining for the policy robot's corresponding service business. After configuration, it triggers the automatic execution of the policy mining toolkit code and executes policy mining for the policy robot's corresponding service business based on the feature data of the feature wide table data according to the configured algorithm model and rule parameters. In other words, the specific processing function of the execution unit 403 corresponds to step S103 of the method of Example 1.

[0121] Furthermore, the policy iteration device 400 further includes an iterative output unit 404, and the iterative output unit 404 is configured to perform rule evaluation on the mined policy set and then iteratively output target policy information.

[0122] The specific processing function of the iterative output unit 404 corresponds to step S104 of the method of embodiment 1.

[0123] After performing rule extraction and evaluation on the policy set mined based on the algorithm model and rule parameters configured based on the read feature wide table data, a chart (as target policy information) is generated and displayed on the front-end page. The chart includes a rule cumulative effect evaluation chart, a rule stability evaluation chart, and a rule logic table analyzed as the optimal rule policy information. In addition, the target policy information output by the iteration can be a single optimal policy or a rule set formed by multiple preferred policies.

[0124] More specifically, the chart serves as target policy information, and the chart includes a rule accumulation effect evaluation table, a rule stability evaluation table, and a rule logic table.

[0125] Then, the following operations are performed: according to the target strategy information determined this time, the feature wide table data generated after the backtracking of the left table features of the next imported sample is configured with the algorithm model and rule parameters for strategy mining and the iteration of strategy mining for the corresponding service business of the strategy robot is executed; until the user stops the iteration or the preset number of iterations is reached; the target strategy information after the iteration is completed is output.

[0126] Therefore, through the iterative determination process of the above-mentioned policy mining, target policy information can be determined.

[0127] It should be noted that the above description is only provided as an example and should not be construed as limiting the present invention.

[0128] In addition, the strategy iteration device 400 also supports data source management, project management, chart and report downloading functions. Accordingly, the strategy iteration device 400 may also include processing units or processing modules corresponding to the data source management, project management, chart and report downloading functions.

[0129] For project management, users can create, edit, delete, run or pause strategy mining projects based on interactions on the data source management page.

[0130] In support of data source management and project management, the user interacts with the data source management page and specifies, presents or selects feature data according to the feature wide table after feature backtracking.

[0131] For the chart and report download function, you can download the target policy information, evaluation charts, evaluation reports, etc. output after rule evaluation.

[0132] In one embodiment, if Figure 5 As shown, the strategy iteration device 400 further includes a downloading unit 501, and the downloading unit 501 is used to download charts and reports.

[0133] Specifically, the downloading unit 501 is used to download target policy information, evaluation charts, evaluation reports, etc. output after rule evaluation, and can also download other information data such as feature wide tables generated by feature backtracking.

[0134] It should be noted that due to Figure 5 The feature backtracking unit 401, strategy mining unit 402, execution unit 403 and iterative output unit 404 in Figure 4 The feature backtracking unit 401, strategy mining unit 402, execution unit 403 and iterative output unit 404 are substantially the same, and therefore, descriptions of the same parts are omitted.

[0135] It should be noted that the above description is only provided as an example and should not be construed as limiting the present invention.

[0136] In addition, the present invention also provides a strategy robot system, which includes a pre-created visual interactive data source management page and a visual interactive data source management page.

[0137] Specifically, the user starts the data source management page and imports the sample left table of this time, performs feature backtracking of the method of embodiment 1 of the present invention, and generates feature wide table data of the corresponding service business of the strategy robot.

[0138] Furthermore, when the feature wide table data is generated, the data source management page is entered to execute the strategy mining of the method of embodiment 1 of the present invention.

[0139] Next, rule evaluation is performed on the mined policy set and target policy information is iteratively output. The output and iteration of the method of embodiment 1 of the present invention are performed, and the target policy information after the iteration is completed is output.

[0140] It should be noted that, in Example 2, descriptions of the same parts as in Example 1 are omitted.

[0141] Those skilled in the art will appreciate that the modules in the above device embodiments may be distributed in the device as described, or may be modified accordingly and distributed in one or more devices different from the above embodiments. The modules in the above embodiments may be combined into one module or further split into multiple submodules.

[0142] Compared with the existing technology, the present invention uses a feature backtracking unit to perform feature backtracking to generate feature wide table data for the corresponding service business of the policy robot. The user only needs to input the sample left table each time to complete the feature backtracking, thereby realizing the automation of feature backtracking processing. The policy mining unit configures the algorithm model and rule parameters for policy mining, and the execution unit executes the policy mining of the corresponding service business of the policy robot, so that the processing part of the policy mining is automated, which can shorten the processing time and increase the output speed of the model. By iteratively outputting the target policy information after performing rule evaluation on the mined policy set, the target policy information can be determined, the processing speed of the entire policy mining can be improved, and the policy mining automation can be realized.

[0143] In addition, by deploying feature backtracking processing into a workflow, feature backtracking for mining rules can be completed with one click, which can realize the automation of feature backtracking; providing a visual interactive data source management page, thereby realizing the visualization and automation of various operations on the visual interactive data source management page.

[0144] In addition, in various data processing processes such as request processing resources, feature wide table acquisition, feature screening, model output and selection, multi-threaded computing is used to achieve this. While effectively processing data, it can further improve the calculation processing speed and effectively ensure the output speed of the model.

[0145] The following describes an electronic device embodiment of the present invention, which can be considered a specific physical implementation of the method and apparatus embodiments of the present invention described above. Details described in the electronic device embodiment of the present invention should be considered supplementary to the above-mentioned method or apparatus embodiments; details not disclosed in the electronic device embodiment of the present invention can be implemented with reference to the above-mentioned method or apparatus embodiments.

[0146] Figure 6 1 is a block diagram of an exemplary embodiment of an electronic device according to the present invention. Figure 6 The electronic device 200 according to this embodiment of the present invention is described.

[0147] like Figure 6 As shown, electronic device 200 is implemented as a general-purpose computing device. Components of electronic device 200 may include, but are not limited to, at least one processing unit 210, at least one storage unit 220, a bus 230 connecting various system components (including storage unit 220 and processing unit 210), a display unit 240, and the like.

[0148] The storage unit stores program codes, which can be executed by the processing unit 210, so that the processing unit 210 performs the steps of various exemplary embodiments of the present invention described in the processing method of the electronic device described above. For example, the processing unit 210 can perform the following steps: Figure 1 Steps shown.

[0149] The storage unit 220 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 2201 and / or a cache memory unit 2202 , and may further include a read-only memory unit (ROM) 2203 .

[0150] The storage unit 220 may also include a program / utility 2204 having a set (at least one) of program modules 2205, such program modules 2205 including but not limited to: an operating system, one or more application programs, other program modules and program data, each of which or some combination may include an implementation of a network environment.

[0151] Bus 230 may represent one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0152] The electronic device 200 can also communicate with one or more external devices 300 (e.g., a keyboard, a pointing device, a Bluetooth device, a router, a modem, etc.). Such communication can be performed via an input / output (I / O) interface 250. Furthermore, the electronic device 200 can also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 260. The network adapter 260 can communicate with other modules of the electronic device 200 via the bus 230. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with the electronic device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0153] Through the description of the above embodiments, when the computer program is executed by a data processing device, the computer-readable medium can implement the above method of the present invention.

[0154] like Figure 7As shown, the computer program can be stored on one or more computer-readable media. The computer-readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. Examples of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0155] The computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein the readable program code is carried. The data signal propagated may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or component. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0156] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the present invention is not inherently related to any specific computer, virtual device, or electronic device, and various general-purpose devices can also implement the present invention. The above description is only a specific embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. A policy iteration method based on a policy robot, characterized in that: include: Based on the feature backtracking automatic processing tool, feature backtracking is performed on the sample left table imported each time to generate feature wide table data of the corresponding service business of the strategy robot, including: receiving the sample left table imported by the user through a pre-created visual interactive data source management page; automatically deploying the workflow according to the feature backtracking automatic processing tool and performing feature backtracking for strategy mining on the sample left table imported each time by triggering the feature backtracking button provided by the data source management page; when the feature backtracking is completed, generating the feature wide table data of the corresponding service business and outputting the running status on the data source management page, and sending an email and / or message to the user for reminder; wherein, the sample left table includes at least the customer number, the backtracking date and / or the application number primary key, outputting the running status includes displaying the completed status and displaying the name of the generated feature wide table data, and displaying the name of the generated feature wide table data includes adding a suffix to the name of the sample left table imported by the user; Requesting processing resources in the online operating environment of the strategy robot and configuring an algorithm model and rule parameters for strategy mining based on the read feature wide table data; Execute strategy mining for corresponding service businesses; After evaluating the rules of the mined policy set, the target policy information is iteratively output.

2. The method according to claim 1, characterized in that The data source management page also includes: Before the user imports the sample left table, the user searches the sample left table and / or specifies the sample left table based on the interaction on the data source management page and performs a necessity check on the feature backtracking button to be triggered and specifies the predicted label column; When the feature backtracking button is triggered, the submission, running, suspension and output processing of the feature backtracking are executed.

3. The method according to claim 1, characterized in that Request processing resources in the online operating environment of the strategy robot, including: When generating the feature wide table data, enter the pre-created visual interactive data source management page; In the online running environment of the strategy robot, it automatically requests memory resources greater than or equal to a predetermined amount to support the automatic reading and cleaning of the features and sample data sets corresponding to the sample left table, and automatically requests multi-threaded computing resources greater than or equal to a predetermined amount to support the fast computing processing speed of the algorithm model.

4. The method according to claim 3, characterized in that Configuring an algorithm model and rule parameters for strategy mining based on the read feature wide table data, including: reading features in the feature wide table data, and setting corresponding algorithm models and rule parameters for strategy mining based on the features; Executing strategy mining for corresponding service businesses includes: triggering the strategy mining toolkit code to automatically run, and executing strategy mining for corresponding service businesses according to the set algorithm model and rule parameters.

5. The method according to claim 4, characterized in that Configure the algorithm model and rule parameters for strategy mining, including: Based on the interaction on the data source management page, the user configures process parameters for strategy mining, hyperparameters of the algorithm model, loss function parameters of the algorithm model, and restriction parameters related to strategy mining and output.

6. The method according to claim 5, characterized in that The data source management page also includes: The user creates, edits, deletes, runs or pauses a strategy mining project based on interaction with the data source management page; and / or, Based on the interaction on the data source management page, the user specifies, presents or selects features according to the feature wide table data after feature backtracking.

7. The method according to claim 1, characterized in that After evaluating the rules of the mined policy set, the target policy information is iteratively output, including: Perform rule strategy extraction and evaluation on the strategy set mined based on the algorithm model and rule parameters, generate charts, and display them on the front-end page; The chart as target strategy information at least includes one or more of a rule accumulation effect evaluation table, a rule stability evaluation table, and a rule logic table.

8. The method according to any one of claims 1 to 7, characterized in that The iterative output target policy information includes: Based on the target strategy information determined this time, the algorithm model and rule parameters for strategy mining are configured for the feature wide table data generated after backtracking the left table features of the next imported sample, and the strategy mining iterations for the corresponding service business are executed; Until the user stops the iteration or the preset number of iterations is reached; Output the target strategy information after the iteration is completed.

9. A strategy iteration device based on a strategy robot, characterized in that: include: A feature backtracking unit is used to perform feature backtracking on each imported sample left table based on a feature backtracking automatic processing tool to generate feature wide table data for the corresponding service business of the strategy robot, including: receiving the sample left table imported by the user through a pre-created visual interactive data source management page; automatically deploying the workflow according to the feature backtracking automatic processing tool and performing feature backtracking for strategy mining on each imported sample left table by triggering the feature backtracking button provided on the data source management page; when the feature backtracking is completed, generating the feature wide table data of the corresponding service business and outputting the running status on the data source management page, and sending an email and / or message to the user for reminder; wherein the sample left table includes at least a customer number, a backtracking date and / or an application number primary key, outputting the running status includes displaying the completed status and displaying the name of the generated feature wide table data, and displaying the name of the generated feature wide table data includes adding a suffix to the name of the sample left table imported by the user; A strategy mining unit, configured to request processing resources in the online operating environment of the strategy robot and configure an algorithm model and rule parameters for strategy mining based on the read feature wide table data; An execution unit, configured to execute strategy mining for the corresponding service business of the strategy robot; The iterative output unit is used to iteratively output the target policy information after evaluating the rules of the mined policy set.

10. A strategic robot system, characterized in that: include: Pre-create a visual interactive data source management page and a visual interactive data source management page; The user launches the data source management page and imports the sample left table, performs feature backtracking according to any one of claims 1 to 2, and generates feature wide table data for the corresponding service business of the strategy robot; as well as, When the feature wide table data is generated, entering the data source management page, and executing the strategy mining method according to any one of claims 1, 3 to 6; and After rule evaluation is performed on the mined policy set, target policy information is iteratively output, the output and iteration method according to claim 1 or 7 is performed, and the target policy information after the iteration is completed is output.

11. A strategic robot system, characterized in that: include: Pre-create a visual interactive data source management page and a visual interactive data source management page; The user launches the data source management page and imports the sample left table, performs feature backtracking according to any one of claims 1 to 2, and generates feature wide table data for the corresponding service business of the strategy robot; as well as, When the feature wide table data is generated, entering the data source management page, and executing the strategy mining method according to any one of claims 1, 3 to 6; and After rule evaluation is performed on the mined policy set, target policy information is iteratively output, and the output and iteration method according to claim 8 is performed to output the target policy information after the iteration is completed.

12. An electronic device comprising: processor; And, a memory storing computer-executable instructions, wherein when the computer-executable instructions are executed, the processor is caused to perform the method according to any one of claims 1-8.

13. A computer-readable medium, wherein: The computer-readable storage medium stores one or more programs, wherein the one or more programs, when executed by a processor, implement the method according to any one of claims 1 to 8.

Images