Method, device and system for creating machine learning solution template

Abstract

A method and apparatus for creating and using a machine learning scheme template are disclosed. A template scheme is obtained for describing at least part of a machine learning process for at least one input source marker. Input source configuration definition information about the template scheme is obtained, wherein the input source configuration definition information is used to generate an input source configuration interface, so that at least one configuration input source configured via the input source configuration interface replaces at least one input source marker in the template scheme. A template file of the machine learning scheme template is generated based on the obtained template scheme and input source configuration definition information. Thus, the modeling threshold and modeling time can be reduced by reusing the created machine learning scheme template, and the data matching problem between actual business data and inherent data in the template scheme can be solved, so that the machine learning scheme template can obtain better modeling effect when applied to business data with different data structures in the same business direction.

Description

Technical Field

[0001] This invention generally relates to the field of artificial intelligence, and more specifically, to a method, method and apparatus for creating a machine learning scheme template. Background Technology

[0002] Machine learning is an inevitable product of the development of artificial intelligence. It aims to improve system performance by leveraging experience through computational means. In computer systems, "experience" usually exists in the form of "data." Through machine learning algorithms, "models" can be generated from this data. In other words, by providing experiential data to machine learning algorithms, models can be generated based on this data. When faced with new situations, the model will provide corresponding judgments, i.e., predict the outcome. It can be seen that how to generate models based on experiential data (i.e., the machine learning modeling process) is the key to machine learning technology.

[0003] Currently, when applying machine learning technology to specific business scenarios, modelers typically need to conduct modeling research from scratch based on the characteristics of the business scenario. The process from data preparation to model debugging is time-consuming and requires a high level of skill from the modelers. For businesses, this is not only time-consuming but also has a high barrier to entry. Summary of the Invention

[0004] The exemplary embodiments of the present invention aim to overcome the shortcomings of the prior art in that the machine learning modeling process is time-consuming and has high threshold requirements.

[0005] According to a first aspect of the present invention, a method for creating a machine learning scheme template is provided, comprising: obtaining a template scheme for describing at least a portion of a machine learning process for at least one input source label, wherein the machine learning process involves model training and / or model application; obtaining input source configuration constraint information regarding the template scheme, wherein the input source configuration constraint information is used to generate an input source configuration interface such that at least one configured input source configured via the input source configuration interface replaces at least one input source label in the template scheme; and generating a template file of the machine learning scheme template based on the obtained template scheme and the input source configuration constraint information.

[0006] Optionally, the step of obtaining input source configuration limitation information about the template scheme includes: generating a control for setting input source configuration limitation information based on the obtained template scheme; displaying the generated control to the first user; and receiving the input source configuration limitation information set by the first user through the control.

[0007] Optionally, the input source marker is used to identify the input tables and / or fields that can be replaced in the template scheme, and the input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured displayed on the input source configuration interface, the processing node corresponding to each input table, the name of each field that needs to be configured under each input table, and indication information on whether each field is displayed as an optional field on the input source configuration interface.

[0008] Optionally, the input source configuration limitation information also includes the field format corresponding to each field, wherein the field format corresponding to at least one field is set to allow configuration of one or more fields in the actual business data for a single field, so that the configured one or more fields are processed in the same way as the single field in the template scheme.

[0009] Optionally, the input source configuration limitation information also includes a processing item for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme.

[0010] Optionally, the processing item includes validation items for each field, wherein the validation items include the allowed format and / or allowed value range for each field.

[0011] Optionally, the verification item may also include indication information on whether verification should be performed.

[0012] Optionally, the template scheme includes at least one parameter placeholder, and the method further includes: obtaining parameter configuration constraint information about the template scheme, wherein the parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme, and the step of generating a template file of a machine learning scheme template based on the obtained template scheme and input source configuration constraint information includes: generating a template file of a machine learning scheme template based on the obtained template scheme, parameter configuration constraint information and input source configuration constraint information.

[0013] Optionally, the parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder.

[0014] Optionally, the type information indicates script parameters and / or runtime parameters.

[0015] Optionally, the parameter configuration limitation information also includes a processing item for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme.

[0016] Optionally, the processing item includes a verification item for verifying the at least one configuration parameter.

[0017] Optionally, the step of obtaining parameter configuration limitation information about the template scheme includes: generating a control for setting parameter configuration limitation information based on the obtained template scheme; displaying the generated control to the first user; and receiving the parameter configuration limitation information set by the first user through the control.

[0018] Optionally, the parameter configuration limitation information further includes: classification information for limiting the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the step of obtaining parameter configuration limitation information about the template scheme further includes: displaying a control for classifying parameter placeholders to the first user; and obtaining classification information based on the classification of parameter placeholders by the first user through the control.

[0019] Optionally, the method further includes: displaying a control for uploading instructional documents to a first user; receiving the instructional documents uploaded by the first user through the control; and merging the instructional documents into a template file.

[0020] Optionally, the method further includes: displaying a control for setting resource configuration information to a first user; receiving resource configuration information set by the first user through the control, the resource configuration information being used to characterize the resource configuration for performing at least part of the machine learning process; and merging the resource configuration into a template file.

[0021] Optionally, the method further includes: displaying an input source configuration interface generated based on input source configuration constraint information to a third user; obtaining at least one configuration input source configured by the third user through the input source configuration interface based on a test data table corresponding to the test scenario; replacing at least one input source marker in the template scheme with the at least one configuration input source to obtain a modified machine learning scheme template; executing at least a portion of the machine learning process based on the modified machine learning scheme template to obtain the execution result of the at least a portion of the machine learning process; evaluating the execution result to obtain a test result; and determining whether to release the machine learning scheme template or debug the machine learning scheme template based on the test result.

[0022] According to a second aspect of the present invention, a method for performing a machine learning process based on a machine learning scheme template is proposed, comprising: obtaining a template file of a machine learning scheme template, wherein the template file includes a template scheme and input source configuration constraint information, the template scheme being used to describe at least a portion of a machine learning process for at least one input source label, the machine learning process involving model training and / or model application, the input source configuration constraint information being used to generate an input source configuration interface; displaying the input source configuration interface generated based on the input source configuration constraint information to a second user; obtaining at least one configured input source configured by the second user via the input source configuration interface; replacing at least one input source label in the template scheme with the obtained at least one configured input source to obtain a modified template scheme; and performing the machine learning process based on the modified template scheme.

[0023] Optionally, the input source configuration interface includes a control for setting a configuration input source, and the step of obtaining at least one configuration input source configured by the second user through the input source configuration interface includes: receiving the configuration input source set by the second user through the control.

[0024] Optionally, the input source marker is used to identify the input table and / or field that can be replaced in the template scheme. The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured displayed on the input source configuration interface, the processing node corresponding to each input table, the name of each first field that needs to be configured under each input table, and indication information on whether each first field is displayed as an optional field on the input source configuration interface. The input source configuration interface also includes at least one of the following: the name of at least one input table that needs to be configured, the name of each first field that needs to be configured under each input table, and indication information on whether each first field is an optional field. The configured input source includes at least one of the following: the business data table configured for the input table, and the second field under the business data table configured for the first field.

[0025] Optionally, the input source configuration limitation information also includes the field format corresponding to each first field, wherein the field format corresponding to at least one first field is set to allow one or more second fields in the actual business data to be configured for a single first field, so that the configured one or more second fields are processed in the same way as the single first field in the template scheme. The control in the input source configuration interface for setting the second field for the first field is generated based on the field format corresponding to the first field, so that the second user can configure the second field for the first field according to the field format corresponding to the first field through the control.

[0026] Optionally, the input source configuration limitation information further includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme. The method further includes: processing the configuration input source according to the processing items; and displaying the processing results in the input source configuration interface.

[0027] Optionally, the processing item includes a validation item for each first field, wherein the validation item includes the allowed format and / or allowed value range of each first field, and the step of processing the configuration input source according to the processing item includes: validating the format and / or value of the second field configured for the first field according to the validation item of the first field, wherein the processing result is used to indicate whether the format and / or value of the second field configured for the first field conforms to the validation item.

[0028] Optionally, the verification item may also include indication information on whether verification should be performed.

[0029] Optionally, the template scheme includes at least one parameter placeholder, and the template file further includes parameter configuration limitation information, which is used to generate a parameter configuration interface. The method further includes: displaying a parameter configuration interface generated based on the parameter configuration limitation information to a second user; obtaining at least one configuration parameter configured by the second user through the parameter configuration interface; and replacing at least one parameter placeholder in the template scheme with the obtained at least one configuration parameter to obtain a modified template scheme.

[0030] Optionally, the parameter configuration interface includes controls for setting configuration parameters, and the step of obtaining at least one configuration parameter configured by the second user through the parameter configuration interface includes: receiving the configuration parameters set by the second user through the controls.

[0031] Optionally, the parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value of the parameter placeholder that needs to be configured, as displayed on the parameter configuration interface. The parameter configuration interface also includes at least one of the following: type information, input method information, display information, and default value of the parameter placeholder that needs to be configured.

[0032] Optionally, the parameter configuration limitation information further includes a processing item for limiting the processing of at least one configuration parameter configured via the parameter configuration interface before replacing at least one parameter placeholder in the template scheme. The method further includes: processing the configuration parameter according to the processing item; and displaying the processing result in the input source configuration interface.

[0033] Optionally, the processing item includes a verification item for verifying the at least one configuration parameter.

[0034] Optionally, the type information indicates script parameters and / or runtime parameters.

[0035] Optionally, the parameter configuration limitation information further includes: classification information for limiting the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the parameter configuration interface displays the parameter placeholders to be configured according to the classification information, and parameter placeholders of different categories are displayed in different classification areas.

[0036] Optionally, the template file may further include a description document to assist the second user in understanding and / or configuring the template scheme, and the method may further include providing the description document to the second user.

[0037] Optionally, the template file further includes resource configuration information, which is used to characterize the resource configuration for performing at least part of the machine learning process. The step of performing the machine learning process based on the modified template scheme includes: performing the machine learning process using the resource configuration characterized by the resource configuration information based on the modified template scheme, or the step of performing the machine learning process based on the modified template scheme includes: predicting the resource configuration required by the modified template scheme in the machine learning process, and performing the machine learning process using the predicted resource configuration.

[0038] According to a third aspect of the present invention, an apparatus for creating a machine learning scheme template is provided, comprising: a first acquisition module for acquiring a template scheme describing at least a portion of a machine learning process for at least one input source label, wherein the machine learning process involves model training and / or model application; a second acquisition module for acquiring input source configuration constraint information regarding the template scheme, wherein the input source configuration constraint information is used to generate an input source configuration interface such that at least one configured input source configured via the input source configuration interface replaces at least one input source label in the template scheme; and a generation module for generating a template file of the machine learning scheme template based on the acquired template scheme and the input source configuration constraint information.

[0039] Optionally, the second acquisition module: generates a control for setting input source configuration limitation information based on the acquired template scheme; displays the generated control to the first user; and receives the input source configuration limitation information set by the first user through the control.

[0040] Optionally, the input source marker is used to identify the input tables and / or fields that can be replaced in the template scheme, and the input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured displayed on the input source configuration interface, the processing node corresponding to each input table, the name of each field that needs to be configured under each input table, and indication information on whether each field is displayed as an optional field on the input source configuration interface.

[0041] Optionally, the input source configuration limitation information also includes the field format corresponding to each field, wherein the field format corresponding to at least one field is set to allow configuration of one or more fields in the actual business data for a single field, so that the configured one or more fields are processed in the same way as the single field in the template scheme.

[0042] Optionally, the input source configuration limitation information also includes a processing item for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme.

[0043] Optionally, the processing item includes validation items for each field, wherein the validation items include the allowed format and / or allowed value range for each field.

[0044] Optionally, the verification item may also include indication information on whether verification should be performed.

[0045] Optionally, the template scheme includes at least one parameter placeholder, and the device further includes: a third acquisition module, configured to acquire parameter configuration constraint information about the template scheme, wherein the parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme, and the generation module generates a template file of a machine learning scheme template based on the acquired template scheme, parameter configuration constraint information, and input source configuration constraint information.

[0046] Optionally, the parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder.

[0047] Optionally, the type information indicates script parameters and / or runtime parameters.

[0048] Optionally, the parameter configuration limitation information also includes a processing item for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme.

[0049] Optionally, the processing item includes a verification item for verifying the at least one configuration parameter.

[0050] Optionally, the third acquisition module: generates a control for setting parameter configuration limitation information based on the acquired template scheme; displays the generated control to the first user; and receives the parameter configuration limitation information set by the first user through the control.

[0051] Optionally, the parameter configuration limitation information further includes: classification information for limiting the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the third acquisition module further includes: displaying a control for classifying parameter placeholders to the first user; and acquiring classification information based on the classification of parameter placeholders by the first user through the control.

[0052] Optionally, the device further includes: a first display module for displaying a control for uploading instructional documents to a first user; a first receiving module for receiving instructional documents uploaded by the first user through the control; and a first merging module for merging the instructional documents into a template file.

[0053] Optionally, the device further includes: a second display module for displaying a control for setting resource configuration information to a first user; a second receiving module for receiving the resource configuration information set by the first user through the control, the resource configuration information being used to characterize the resource configuration for executing at least part of the machine learning process; and a second merging module for merging the resource configuration into a template file.

[0054] Optionally, the device further includes: a third display module for displaying an input source configuration interface generated based on input source configuration constraint information to a third user; a fourth acquisition module for acquiring at least one configuration input source configured by the third user through the input source configuration interface based on a test data table corresponding to the test scenario; a replacement module for replacing at least one input source marker in the template scheme with the at least one configuration input source to obtain a modified machine learning scheme template; an execution module for executing at least a portion of the machine learning process based on the modified machine learning scheme template to obtain the execution result of the at least a portion of the machine learning process; an evaluation module for evaluating the execution result to obtain a test result; and a release or debugging module for determining whether to release the machine learning scheme template based on the test result, or to debug the machine learning scheme template based on the test result.

[0055] According to a fourth aspect of the present invention, an apparatus for executing a machine learning process based on a machine learning scheme template is provided, comprising: a first acquisition module for acquiring a template file of a machine learning scheme template, wherein the template file includes a template scheme and input source configuration constraint information, the template scheme describing at least a portion of a machine learning process for at least one input source label, the machine learning process involving model training and / or model application, and the input source configuration constraint information for generating an input source configuration interface; a first display module for displaying the input source configuration interface generated based on the input source configuration constraint information to a second user; a second acquisition module for acquiring at least one configured input source configured by the second user via the input source configuration interface; a first replacement module for replacing at least one input source label in the template scheme with the acquired at least one configured input source to obtain a modified template scheme; and an execution module for executing the machine learning process based on the modified template scheme.

[0056] Optionally, the input source configuration interface includes a control for setting the configuration input source, and the second acquisition module receives the configuration input source set by the second user through the control.

[0057] Optionally, the input source marker is used to identify the input table and / or field that can be replaced in the template scheme. The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured displayed on the input source configuration interface, the processing node corresponding to each input table, the name of each first field that needs to be configured under each input table, and indication information on whether each first field is displayed as an optional field on the input source configuration interface. The input source configuration interface also includes at least one of the following: the name of at least one input table that needs to be configured, the name of each first field that needs to be configured under each input table, and indication information on whether each first field is an optional field. The configured input source includes at least one of the following: the business data table configured for the input table, and the second field under the business data table configured for the first field.

[0058] Optionally, the input source configuration limitation information also includes the field format corresponding to each first field, wherein the field format corresponding to at least one first field is set to allow one or more second fields in the actual business data to be configured for a single first field, so that the configured one or more second fields are processed in the same way as the single first field in the template scheme. The control in the input source configuration interface for setting the second field for the first field is generated based on the field format corresponding to the first field, so that the second user can configure the second field for the first field according to the field format corresponding to the first field through the control.

[0059] Optionally, the input source configuration limitation information further includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme. The device further includes: a first processing module for processing the configuration input source according to the processing items; and a second display module for displaying the processing results in the input source configuration interface.

[0060] Optionally, the processing item includes a validation item for each first field, wherein the validation item includes the allowed format and / or allowed value range of each first field, and the step of processing the configuration input source according to the processing item includes: validating the format and / or value of the second field configured for the first field according to the validation item of the first field, wherein the processing result is used to indicate whether the format and / or value of the second field configured for the first field conforms to the validation item.

[0061] Optionally, the verification item may also include indication information on whether verification should be performed.

[0062] Optionally, the template scheme includes at least one parameter placeholder, and the template file further includes parameter configuration limitation information, which is used to generate a parameter configuration interface. The device further includes: a third display module for displaying the parameter configuration interface generated based on the parameter configuration limitation information to a second user; a third acquisition module for acquiring at least one configuration parameter configured by the second user through the parameter configuration interface; and a second replacement module for replacing at least one parameter placeholder in the template scheme with the acquired at least one configuration parameter to obtain a modified template scheme.

[0063] Optionally, the parameter configuration interface includes controls for setting configuration parameters, and the third acquisition module receives the configuration parameters set by the second user through the controls.

[0064] Optionally, the parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value of the parameter placeholder that needs to be configured, as displayed on the parameter configuration interface. The parameter configuration interface also includes at least one of the following: type information, input method information, display information, and default value of the parameter placeholder that needs to be configured.

[0065] Optionally, the parameter configuration limitation information further includes a processing item for limiting the processing of at least one configuration parameter configured via the parameter configuration interface before replacing at least one parameter placeholder in the template scheme. The device further includes: a second processing module for processing the configuration parameter according to the processing item; and a fourth display module for displaying the processing result in the input source configuration interface.

[0066] Optionally, the processing item includes a verification item for verifying the at least one configuration parameter.

[0067] Optionally, the type information indicates script parameters and / or runtime parameters.

[0068] Optionally, the parameter configuration limitation information further includes: classification information for limiting the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the parameter configuration interface displays the parameter placeholders to be configured according to the classification information, and parameter placeholders of different categories are displayed in different classification areas.

[0069] Optionally, the template file further includes a description document to assist the second user in understanding and / or configuring the template scheme, and the device further includes a providing module for providing the description document to the second user.

[0070] Optionally, the template file further includes resource configuration information, which is used to characterize the resource configuration for performing at least part of the machine learning process. The execution module performs the machine learning process based on the modified template scheme and uses the resource configuration characterized by the resource configuration information. Alternatively, the execution module predicts the resource configuration required by the modified template scheme in the machine learning process and uses the predicted resource configuration to perform the machine learning process.

[0071] According to a fifth aspect of the invention, a system is provided that includes at least one computing device and at least one storage device for storing instructions, wherein the instructions, when executed by at least one computing device, cause at least one computing device to perform the method as described in the first or second aspect of the invention.

[0072] According to a sixth aspect of the invention, a computer-readable storage medium for storing instructions is provided, wherein when the instructions are executed by at least one computing device, they cause at least one computing device to perform the method as described in the first or second aspect of the invention.

[0073] In the method, method and apparatus for creating, using and implementing machine learning solution templates according to exemplary embodiments of the present invention, the modeling threshold and modeling time can be reduced by reusing machine learning solution templates. The input source configuration limitation information in the template file of the machine learning solution template can be used to solve the data matching problem between actual business data and inherent data in the template solution, so that the machine learning solution template can achieve better modeling results when applied to business data with different data structures under the same business direction. Attached Figure Description

[0074] These and / or other aspects and advantages of the present invention will become clearer and more readily understood from the following detailed description of embodiments of the invention taken in conjunction with the accompanying drawings, wherein:

[0075] Figure 1 A flowchart is shown for a method of creating a machine learning scheme template according to an exemplary embodiment of the present invention;

[0076] Figure 2 A schematic diagram of the interface for creating a machine learning solution template is shown;

[0077] Figure 3 A flowchart is shown illustrating a method for performing a machine learning process based on a machine learning scheme template according to an exemplary embodiment of the present invention;

[0078] Figure 4 A schematic diagram of the configuration interface for a machine learning solution template is shown;

[0079] Figure 5 A structural block diagram of an apparatus for creating a machine learning scheme template according to an exemplary embodiment of the present invention is shown;

[0080] Figure 6 A structural block diagram of an apparatus for creating a machine learning scheme template according to an exemplary embodiment of the present invention is shown. Detailed Implementation

[0081] To enable those skilled in the art to better understand the present invention, exemplary embodiments of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0082] Figure 1 A flowchart of a method for creating a machine learning scheme template according to an exemplary embodiment of the present invention is shown. Figure 1 The method shown can be implemented entirely in software via computer programs, or it can be executed via a specially configured computing device. Figure 1 The method shown.

[0083] See Figure 1 In step S110, a template scheme is obtained to describe at least a portion of the machine learning process for at least one input source label.

[0084] Machine learning modeling solutions often differ significantly across different business scenarios. For example, machine learning modeling solutions in time series scenarios focus on building time series windows and are more sensitive to time series data, while machine learning modeling solutions in marketing scenarios pay more attention to data such as product and user tags.

[0085] In this invention, the template scheme can be regarded as a set of general machine learning modeling schemes for a business scenario, summarized and defined by a user (for ease of distinction, it can be referred to as the first user) based on the business scenario. Here, the first user can refer to a scientist with rich experience in machine learning modeling.

[0086] A template scheme may include configurations for at least some steps of a machine learning process. For example, a template scheme may describe one or more configuration information such as the processing objects, processing methods, and processing results involved in each step of at least some machine learning processes.

[0087] The machine learning process involves model training and / or model application. Model training refers to the process of training a machine learning model, which may include, but is not limited to, at least one of the following steps: data import, data splitting, feature extraction, model training, model testing, and model evaluation. Detailed descriptions of each step can be found in existing machine learning knowledge, and will not be repeated here. Model application refers to the process of applying the machine learning model. For example, it can refer to the process of using a trained machine learning model to predict data to obtain prediction results. Examples include packaging the application, deploying it online, and providing services.

[0088] A template scheme can be a file written in a specific language to describe at least part of a machine learning process. For example, a template scheme can be a DAG (Directed Acyclic Graph) file, which describes the configuration information of the machine learning steps represented by each node (i.e., the processing node mentioned below) in the directed acyclic graph.

[0089] A template scheme includes one or more input sources, which refer to the input sources used in the machine learning process described by the template scheme. These input sources may include, but are not limited to, input tables and / or fields. An input table is the data table used in the machine learning process described by the template scheme, and a field is the field within the input table.

[0090] Template-based solutions can be viewed as general machine learning modeling schemes applicable to specific business scenarios. However, even within the same business scenario, different data structures may be used for the business data. This means that the input sources in the template solution and the actual business data may differ in data structure. To ensure that the input sources in the template solution can adapt to business data with different data structures, the input sources in the template solution can refer to input sources that can be replaced by the actual business data. Specific implementation details regarding this replacement can be found in the relevant descriptions below.

[0091] To facilitate differentiation, this invention uses input source markers to characterize replaceable input sources in the template scheme. The input source markers are only used to refer to replaceable input sources in the template scheme; this invention does not limit the specific form of the input source in the template scheme. That is, replaceable input sources in the template scheme may or may not be identified by special markers.

[0092] In step S120, input source configuration constraint information for the template scheme is obtained.

[0093] The input source configuration constraint information is used to generate the input source configuration interface, so that at least one configured input source configured through the input source configuration interface replaces at least one input source marker in the template scheme. The input source configuration interface refers to the interface displayed to the user using the machine learning scheme template (for ease of distinction, it can be referred to as the second user), used to assist the second user in mapping actual business data to the input source markers in the template scheme.

[0094] The input source configuration limitation information can be set by a first user, who can set this information for one or more input sources in the template scheme. This invention can obtain the input source configuration limitation information set by the first user in various ways. For example, the first user can generate a file including the input source configuration limitation information by editing a document, but not limited to this method. This invention can provide a file upload interface to the first user and obtain the input source configuration limitation information from the file uploaded by the first user through this interface. As another example, this invention can also provide a visual operation interface to the first user, and obtain the input source configuration limitation information set by the first user based on the operations performed by the first user on the visual operation interface.

[0095] As an example, controls for setting input source configuration limitations can be generated based on the obtained template scheme. These controls are then displayed to the first user, and the input source configuration limitations set by the first user through these controls are received. Taking an input source indicator input table as an example, the obtained template scheme can be parsed to determine the input tables involved. Controls for adding input tables are then provided to the first user, who can use these controls to add input tables and set relevant input source configuration limitations for those tables based on other controls.

[0096] The input source configuration constraint information may include any information that can help a second user match the actual business data with the input sources in the template scheme. Taking the input source tag used to identify the input tables and / or fields that can be replaced in the template scheme as an example, the input source configuration constraint information may include, but is not limited to, at least one of the following: the name of at least one input table that needs to be configured as displayed on the input source configuration interface, the processing node corresponding to each input table, the name of each field that needs to be configured under each input table, and indication information on whether each field is displayed as an optional field on the input source configuration interface.

[0097] The input table name refers to the table name displayed to the second user. The first user can use the name of the input table in the template scheme as the table name displayed to the second user, or they can name the input table according to its business meaning, so that the second user can configure business data tables with the same or similar business meanings for the input table.

[0098] The processing nodes corresponding to the input table are used to represent the nodes that process the input table in the machine learning process described by the template scheme, that is, in which machine learning step the input table is processed. As an example, the template scheme can be the DAG file mentioned above. Each processing node in the DAG file can have a corresponding node ID, which can be used to represent the processing node corresponding to the input table.

[0099] The field name refers to the field name displayed to the second user. The first user can use the original names of the fields in the input table as the field names displayed to the second user, or they can rename the field names according to the business meaning of the fields, so that the second user can match the fields in the business data table with the fields in the input table according to the business meaning. It should be noted that the fields that need to be configured under each input table can include not only the original fields that exist in the input table, but also fields that do not exist in the input table, i.e., extended fields. For example, assuming that input table A includes fields a, b, and c, the fields that need to be configured under input table A can include not only fields a, b, and c, but also field d that does not exist in input table A. Field d is the extended field. Extended fields can be set by the first user. For example, the first user can start from the business scenario and add one or more extended fields that do not exist in the input table to the input table according to the business data structure that may exist in the business scenario, and set the field name of the extended fields. When adding extended fields, the processing method of the extended fields can also be set. As an example, each extended field added to the input table can also be regarded as a field category, and different extended fields correspond to different field categories. By setting extended fields, support can be provided for matching additional fields in actual business data that exceed the fields involved in the template solution. In other words, additional fields can be mapped to extended fields, and the processing method of extended fields is pre-defined, so that additional fields can also participate in machine learning and realize their data value, thereby enhancing the data adaptability of the template solution.

[0100] The indication information is used to indicate whether a field is optional. Optional fields are those for which the second user can decide whether to configure a field from the actual business data, based on the actual situation. Non-optional fields, also known as mandatory fields, are those for which the second user needs to configure at least one field from the actual business data. This invention can set fields as optional or mandatory based on the characteristics of the fields in the business scenario (such as universality and importance). For example, the first user can set a few commonly used fields in the industry as mandatory fields and set other fields as optional fields. Optional fields include not only fields present in the template scheme but also extended fields not present in the template scheme. For details on extended fields, please refer to the relevant description above; it will not be repeated here. Therefore, by classifying commonly used or highly important fields as mandatory fields, the modeling effect can be guaranteed to be acceptable. By classifying optional fields, the adaptability of the template scheme under different data structures can be optimized. Specifically, the field format corresponding to at least one field is set to allow configuration of one or more fields from the actual business data for a single field, ensuring that the configured one or more fields are processed in the same way as a single field in the template scheme. Therefore, through this "one-to-many" field correspondence method, even if there are a large number of fields in the actual business data, these fields can be matched with the fields in the input table. This allows all fields in the actual business data to participate in the machine learning process (such as feature construction), realize their data value, and influence the final result.

[0101] In addition, the input source configuration constraint information may also include the field format corresponding to each field.

[0102] Additionally, the input source configuration constraint information may also include processing items for defining the processing that at least one configured input source, configured via the input source configuration interface, undergoes before at least one input source tag in the replacement template scheme. These processing items may include, but are not limited to, validation items for each field. Validation items may include allowed formats and / or allowed value ranges for each field, and may also include indications of whether validation should be performed.

[0103] In step S130, a template file for a machine learning scheme template is generated based on the obtained template scheme and input source configuration constraint information.

[0104] The template file for the machine learning solution template includes the template solution and input source configuration constraints. The template solution can be viewed as a set of general machine learning modeling solutions defined by experienced scientists based on business scenarios. It includes machine learning modeling know-how accumulated by these scientists, which second users can reuse to lower the modeling threshold and reduce modeling time. The input source configuration constraints help second users map actual business data to the input source markers in the template solution, solving the data matching problem between actual business data and the inherent data in the template solution. This ensures that the template solution achieves good modeling results when applied to business data with different data structures within the same business context.

[0105] After generating the template file, you can test it to see if it meets your expectations. If it does, you can publish the machine learning solution template; if it does not, you can debug the machine learning solution template.

[0106] As an example, an input source configuration interface generated based on input source configuration constraints can be shown to a third user, where the third user can refer to a tester. At least one configured input source is obtained from the third user's test data table corresponding to the test scenario, configured via the input source configuration interface. Here, configuring an input source refers to the input source configured by the third user based on the input source markers in the template scheme according to the test data. Configuring an input source can include a test data table configured for an input table and / or fields in the test data table configured for fields in the input table. For a detailed configuration process, please refer to the following text. Figure 3 The relevant descriptions are not repeated here; at least one input source marker in the template scheme is replaced with at least one configuration input source to obtain a modified machine learning scheme template; at least part of the machine learning process is executed based on the modified machine learning scheme template to obtain the execution results of at least part of the machine learning process; the execution results are evaluated to obtain test results, and based on the test results, it is determined whether to release the machine learning scheme template, or to debug the machine learning scheme template based on the test results. For example, if the test results meet expectations, the machine learning scheme template can be released; if the test results do not meet expectations, the machine learning scheme template can be debugged, such as by modifying the input source configuration constraint information in the template file.

[0107] The template scheme may also include at least one parameter placeholder. The parameter placeholder can be any pre-defined language not commonly used for writing code, such as "{$placeholder$}". The parameter represented by the parameter placeholder refers to a parameter that can be determined by the second user, and may include, but is not limited to, script parameters and / or runtime parameters. Therefore, when setting up the template scheme, the first user can also modify the template scheme based on experience, replacing variable parameters with placeholders. For example, variable parameters such as feature combination methods, hyperparameters, and runtime resources can be set as parameter placeholders.

[0108] This invention can also obtain parameter configuration constraint information regarding the template scheme. The parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme. The parameter configuration interface refers to an interface displayed to a second user, used to assist the second user in configuring the parameters represented by the parameter placeholders.

[0109] The parameter configuration constraints can be set by the first user, who can set these constraints for each parameter placeholder in the template scheme. This invention can obtain the parameter configuration constraints in various ways. For example, the first user can generate a file containing the parameter configuration constraints by editing a document, but not limited to this method. This invention can provide a file upload interface to the first user and obtain the parameter configuration constraints from the file uploaded by the first user through this interface. Alternatively, this invention can also provide a visual operation interface to the first user, and obtain the parameter configuration constraints set by the first user based on the operations performed on the visual operation interface.

[0110] As an example, a control for setting parameter configuration limitations can be generated based on the obtained template scheme. This generated control is then displayed to the first user, and the parameter configuration limitations set by the first user through the control are received. For instance, after obtaining the template scheme, it can be parsed to identify parameter placeholders within the scheme, and a control for setting parameter configuration limitations related to these placeholders can be displayed to the first user. This allows the first user to obtain the parameter configuration limitations set by the first user through this control.

[0111] The parameter configuration constraint information may include any information that can assist the second user in configuring the parameters represented by the parameter placeholders. For example, the parameter configuration constraint information may include, but is not limited to, at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that requires configuration for the parameter placeholders. Specifically, the type information may indicate script parameters and / or runtime parameters; the input method information may include fill-in method and / or selection method, where fill-in method refers to inputting by filling in, and selection method refers to inputting by selecting from multiple provided options; the display information may include prompts to help the second user understand the parameters that need to be input, such as the name of the parameter represented by the parameter placeholder; the default value may refer to the displayed value of the selected option when the input method is selection; the actual value may refer to the actual value of the provided option, and the displayed value may differ from the actual value. The displayed value may refer to translated content that is easy for the user to understand, thus allowing the actual value of the selected option to be hidden while exposing the translated display value to the user when providing options. This approach makes it easy for users to understand, while also protecting the actual values ​​set by the first user as know-how.

[0112] The parameter configuration constraint information may also include processing items used to limit the processing of at least one configuration parameter configured via the parameter configuration interface before replacing at least one parameter placeholder in the template scheme. These processing items may include, but are not limited to, validation items for validating at least one configuration parameter. Validation items may include allowed formats and / or allowed value ranges for the configuration parameter, and may also include indication information regarding whether validation should be performed.

[0113] When the template scheme includes parameter configuration constraint information, this invention can generate a template file for a machine learning scheme template based on the acquired template scheme, parameter configuration constraint information, and input source configuration constraint information. That is, the template file can include not only the template scheme and input source configuration constraint information, but also parameter configuration constraint information. The parameter configuration constraint information can assist a second user in configuring the parameters represented by the parameter placeholders in the template scheme, ensuring that the overall modeling effect is adapted to the actual business scenario and improving the application effect of the template scheme in the actual business scenario.

[0114] The parameter configuration constraint information may also include classification information for limiting the configuration of parameter placeholders according to category areas on the parameter configuration interface. During the process of obtaining parameter configuration constraint information about the template scheme, the present invention can also display a control for classifying parameter placeholders to the first user, and obtain classification information based on the classification of parameter placeholders by the first user through the control. Therefore, the generated parameter configuration interface can display the parameter placeholders to be configured according to the classification information, with parameter placeholders of different categories displayed in different category areas, making the parameter placeholder configuration more logical for the second user.

[0115] This invention can also display a control for uploading instructional documents to a first user, receive the instructional documents uploaded by the first user through the control, and merge the instructional documents into the template file. The instructional documents can be documents instructing users on how to configure the template scheme. These documents can be described in business language that is easy for a second user to understand, allowing the second user to use the machine learning scheme template in most cases without needing to understand machine learning concepts. For example, the instructional documents may include instructions on setting sample labels, so that users do not need to understand the concepts of positive and negative samples in machine learning. They only need to reflect the situation truthfully according to business needs. In anti-fraud business, users only need to identify which transactions are problematic and which are normal, and can operate the entire modeling template through normal business understanding, thus lowering the barrier to template use.

[0116] The present invention can also display a control for setting resource configuration information to a first user, receive the resource configuration information set by the first user through the control, the resource configuration information being used to characterize the resource configuration for executing at least part of the machine learning process, and merge the resource configuration into a template file. The resource configuration information mentioned herein can be regarded as runtime resource settings provided by the first user.

[0117] Figure 1 The method shown can be executed by a machine learning platform used to implement machine learning-related business. Figure 2 The diagram illustrates the creation interface for a machine learning solution template presented to a user by a machine learning platform. The user referred to here is the one who creates the machine learning solution template, i.e., the first user mentioned above, such as a scientist with extensive experience in machine learning modeling.

[0118] like Figure 2 As shown, the creation of a machine learning solution template can be divided into four parts: basic information configuration, input configuration, placeholder configuration, and classification configuration.

[0119] 1. Basic Information Configuration

[0120] Basic information may include, but is not limited to, modeling template name, modeling template visibility status, modeling template DAG, and modeling template configuration documentation.

[0121] The modeling template name refers to the name of the machine learning solution template displayed to the user (the template user, also known as the second user mentioned above). Scientists can name the machine learning solution template according to the applicable business scenario and the function of the machine learning solution template. For example, scientists can name a machine learning solution template created for a marketing scenario a general marketing modeling template.

[0122] The visibility status of the modeling template refers to whether the modeling template is visible to other users. If it is selected as invisible, users cannot view it on the front end; otherwise, users can view it through the front end entry point.

[0123] A DAG modeling template refers to a DAG file containing a machine learning solution template. Scientists can upload DAG files by clicking an upload control. After the upload is complete, the backend can automatically scan the placeholder information in the DAG file and display controls on the interface based on the scan results, allowing scientists to configure the placeholders.

[0124] The modeling template configuration documentation is a document that informs users how to configure the modeling template. It may include detailed explanations of each configuration field to help users understand it.

[0125] 2. Input configuration

[0126] Input configuration refers to the information on input source configuration limits set by scientists.

[0127] Initially, the interface may only display one "Add Input Table" control. After the first user clicks the "Add Input Table" control, the interface will display input table module information.

[0128] The input table module includes controls for entering the input table name, controls for entering the node ID (i.e., the node ID shown in the diagram), controls for adding fields, and a field display table. The field display table includes controls for entering field names, controls for setting whether field validation is required, and controls for selecting field types.

[0129] Scientists can set the input table name according to its business meaning, so that users can select the appropriate data table from the actual business data for matching. Correspondingly, scientists can also rename field names according to their meaning to ensure that the field names seen by users are easy to understand. Scientists can also set whether field validation is required; if field validation is selected, the field type also needs to be set to verify whether the field types provided by the user are consistent with the scientist's requirements.

[0130] 3. Placeholder Configuration

[0131] It can read placeholders from DAGs uploaded by scientists and display the corresponding placeholder modules. Each placeholder module contains an option box, a placeholder name, and the corresponding option content. Placeholder variables include both script variables and runtime parameters.

[0132] When selecting a script variable as a placeholder, the option box offers two choices: selectable items and input fields. Selectable items present a choice-based input requirement; selectable items only support single selection and have corresponding options. A minimum of two options are allowed, and users can click "Add Options" to add more. Input fields present a text box with the input field name, where users can enter their information.

[0133] Option names help users understand the information required for each option. Option names should be consistent with the modeling template configuration documentation as much as possible. If a field type validation is selected, the content must be validated based on the field type set by the scientist. If no field type validation is selected, no validation is required. Field type refers to the field type set by the scientist, facilitating verification of whether the user-provided field type matches the scientist's requirements. This can include, but is not limited to, enums and String. If a field threshold validation is selected, validation must be performed based on the field threshold. If not selected, the user-entered data will not be validated. Field thresholds are optional; for example, regular expressions or closed intervals (representing intervals greater than or equal to, less than or equal to) can be used to represent field thresholds. After filling in the threshold, the data under the corresponding form field column will be validated to ensure it meets the interval requirements. If a regular expression is used, regular expression hints will be provided, explaining what field information needs to be filled in.

[0134] When selecting run parameters using placeholders, only users are allowed to fill in the information. Scientists can specify operators and explicitly name parameters (English names within the code) through node aliases (node ​​names, such as node IDs). They can also set the parameter names displayed to the user, default parameter items, parameter field types, corresponding threshold regular expressions, and prompts for the regular expressions. Run parameter field types can include, but are not limited to, String, Int, Double, Boolean, and Long.

[0135] 4. Categorized Configuration

[0136] Under each placeholder variable, there is an "Add Category Line" button. The category line above the previous one is displayed as a module when shown to the user. If there are no other category line modules set above the current category line, then the category line up to the first placeholder variable forms a module.

[0137] Scientists need to input category names for the category display. These category names represent the category information of the placeholder variables displayed as a module. For example, if the placeholder variables between two category lines are the marketing time-series window configuration and the financial purchase time-series window configuration, scientists can set the category names as time-series parameter settings.

[0138] After completing the settings, scientists click the "Save Plan" control, and the platform will save the current modeling template. Scientists can also view the created modeling templates on the modeling template management page. The creation time is the time when the "Save Plan" button is clicked. If a scientist clicks "Cancel," a warning message will be displayed: "Cancellation will not retain all current settings; please cancel with caution."

[0139] Thus concluded Figure 1 , Figure 2 The present invention provides a detailed description of the method for creating machine learning scheme templates. The present invention also provides a flowchart of a method for executing a machine learning process based on a machine learning scheme template. The machine learning scheme template can be generated based on the method for creating machine learning scheme templates of the present invention. Therefore, the method for executing a machine learning process based on a machine learning scheme template of the present invention may further include... Figure 1 The steps shown.

[0140] Figure 3 A flowchart illustrating a method for performing a machine learning process based on a machine learning scheme template according to an exemplary embodiment of the present invention is shown. Figure 3 The method shown can be implemented entirely in software via computer programs, or it can be executed via a specially configured computing device. Figure 3 The method shown.

[0141] See Figure 3 In step S310, obtain the template file of the machine learning solution template.

[0142] The template file includes a template scheme and input source configuration constraints. The template scheme describes at least part of the machine learning process for at least one input source label, which involves model training and / or model application. The input source configuration constraints are used to generate the input source configuration interface. For details on the template scheme, input source labels, and input source configuration constraints, please refer to the relevant descriptions above; they will not be repeated here.

[0143] In step S320, an input source configuration interface, generated based on the input source configuration constraint information, is displayed to the second user. This interface assists the second user in matching actual business data with the input source markers in the template solution. The second user refers to the user using the machine learning solution template. The second user can be a scientist with extensive machine learning modeling experience or a business professional lacking such experience.

[0144] In step S330, at least one configuration input source configured by the second user via the input source configuration interface is obtained. The input source configuration interface may include controls for setting configuration input sources and can receive configuration input sources set by the second user through the controls.

[0145] The configuration input source originates from actual business data. A configuration input source refers to an input source configured by a second user based on actual business data and the input source markers in the template scheme. Taking the input source markers used to identify replaceable input tables and / or fields in the template scheme as an example, a field under an input table in the template scheme can be called a first field, and a field under a business data table can be called a second field. The configuration input source can include a business data table configured for the input table and / or a second field under a business data table configured for the first field. Here, the business data table refers to a data table generated in the actual business scenario, and the business data table represents the actual business data.

[0146] Taking the input source tag used to identify replaceable input tables and / or fields in a template scheme as an example, the input source configuration constraint information may include at least one of the following: the name of at least one input table that needs to be configured, the processing node corresponding to each input table, the name of each first field that needs to be configured under each input table, and indication information indicating whether each first field is displayed as an optional field on the input source configuration interface. The input source configuration interface may also include at least one of the following: the name of at least one input table that needs to be configured, the name of each first field that needs to be configured under each input table, and indication information indicating whether each first field is an optional field.

[0147] As an example, the second user can first configure the corresponding business data table according to the input table name displayed on the input source configuration interface, and then configure the corresponding second field for the first field according to the name of each first field that needs to be configured under each input table and the indication information of whether it is an optional field, so as to match the second field in the business data table with the first field in the input table.

[0148] For a first field that is optional, the second user can determine whether a second field exists in the business data table that matches the first field. If a matching second field exists, the second user sets the corresponding second field for the first field; otherwise, the second field does not need to be set for the first field. For a first field that is not optional (i.e., a required field), the second user needs to configure at least one second field in the business data table for that field.

[0149] As mentioned above, whether the first field is optional can be determined based on the generality or importance of the field in the business scenario. For example, fields that are general or of high importance can be set as required fields, while fields that are not general or of low importance can be set as optional fields. Therefore, when the second user matches the second field in the business data table with the first field in the input table, at least one corresponding second field exists for the required first field. This ensures that the modeling effect is not too bad when the template solution is applied to the actual business scenario. Whether a corresponding second field exists for an optional first field is set by the second user according to the actual situation. That is, the second user does not need to configure a corresponding second field for every first field in the input table, thus optimizing the adaptability of the data table while ensuring the modeling effect.

[0150] As mentioned above, the first field to be configured under each input table can include not only the original fields existing in the input table, but also fields that do not exist in the input table, i.e., extended fields. Extended fields can be added to the input table based on the data structure that may exist in the business scenario, and these extended fields can also have corresponding processing methods. Therefore, when the second user matches the second field in the business data table with the first field in the input table, they can also map the second field to the field that does not exist in the input table (i.e., the extended field). Thus, even if the number of second fields in the business data table is large, exceeding the fields under the input table in the template solution, the second user can still map the excess fields to the extended fields, allowing these excess fields to participate in machine learning, realize data value, and influence the final result. For example, as mentioned above, each extended field added to the input table can also be regarded as a field category, with different extended fields corresponding to different field categories. For additional fields in the business data table that exceed the fields involved in the input table, the second user can classify the additional fields under the corresponding category of extended fields, allowing the additional fields to also participate in machine learning and realize their data value.

[0151] For example, the field format corresponding to at least one first field is set to allow one or more second fields in the actual business data to be configured for a single first field, so that the configured one or more second fields are processed in the same way as the single first field in the template scheme. The control in the input source configuration interface used to set the second field for the first field is generated based on the field format corresponding to the first field, so that the second user can configure the second field for the first field according to the field format corresponding to the first field through the control.

[0152] In other words, for the first field that needs to be configured under the input table in the input source configuration interface, the field format of this first field can be either allowing only one second field to be configured for this field, or allowing multiple second fields to be configured for this field. That is, the first field that needs to be configured under the input table in the input source configuration interface can include either a first field that supports "one-to-one" configuration or a first field that supports "one-to-many" configuration. For a first field that supports "one-to-one" configuration, a second user is allowed to configure at most one second field for this field; for a first field that supports "one-to-many" configuration, a second user is allowed to configure multiple second fields for this field.

[0153] Therefore, even if there are a large number of fields in the business data table, this "one-to-many" field correspondence method can be used to match these fields with the first field in the input table, so that all fields in the business data table can participate in the modeling process, realize their data value, and affect the final modeling result.

[0154] As an example of the present invention, the input table displayed in the input source configuration interface may include at least one optional form, which may include multiple column names. The column names in the optional form may be basic fields already present in the template scheme (e.g., card number, account opening time, etc.), and these basic fields already have corresponding processing methods in the template scheme. For additional fields in the actual business data that exceed the fields covered by the template scheme, the second user can fill in such additional fields under the corresponding column names in the optional form. For example, if the additional field is activation time, then the field name of activation time can be filled in under "account opening time," so that the activation time can be processed according to the processing method for account opening time.

[0155] The column names in the optional form can also be field categories (such as the extended fields mentioned above), with each field category having a corresponding processing method. For additional fields not covered in the template solution, the second user can fill in such additional fields under the corresponding column names in the optional form. For example, if the additional field is activation time, then the field name of activation time can be filled in under the "Time" category, so that the activation time can be processed according to the processing method of the "Time" category.

[0156] In addition, the input source configuration constraint information may also include the field format corresponding to each first field.

[0157] Additionally, the input source configuration limitation information may also include processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before replacing at least one input source tag in the template scheme. The present invention can also process the configuration input source according to the processing items and display the processing results in the input source configuration interface.

[0158] As an example, the processing items may include validation items for each first field, which may include the allowed formats and / or allowed value ranges for each first field. Then, the format and / or values ​​of the second field configured for the first field can be validated according to the validation items for the first field. The processing result is used to indicate whether the format and / or values ​​of the second field configured for the first field conform to the validation items. Optionally, the validation items may also include indication information on whether validation should be performed.

[0159] In step S340, at least one input source marker in the template scheme is replaced with at least one obtained configuration input source to obtain the modified template scheme.

[0160] Taking the configuration input source, which includes a business data table configured for the input table and a second field under the business data table configured for the first field, as an example, the input table in the template scheme can be replaced with the configured business data table. The processing methods for the input table and its first field in the template scheme are known. When replacing the input source marker in the template scheme with the configuration input source, the processing method of the replaced business data table can be set according to the processing method of the input table, and the processing method of the configured second field can be set according to the processing method of the first field. Therefore, the modified template scheme describes the machine learning process for the configuration input source.

[0161] Therefore, when using the machine learning solution template, the second user does not need to generate a corresponding table according to the field requirements of the template solution. They only need to match the fields with the same business meaning to complete the import of actual business data, thus avoiding the user's repeated modification of the data structure.

[0162] In step S350, the machine learning process is performed based on the modified template scheme.

[0163] The modified template describes the machine learning process for a configured input source, which represents actual business data. Therefore, executing the machine learning process based on the modified template can yield machine learning results that conform to the actual business scenario.

[0164] In summary, a template solution can be viewed as a set of general machine learning modeling solutions for a business scenario, summarized and defined by experienced scientists based on that scenario. The template solution includes machine learning modeling know-how accumulated by scientists. Secondary users can reuse this know-how by using the machine learning modeling solution to lower the modeling threshold and reduce modeling time. Furthermore, when using the template, secondary users do not need to understand the modeling principles of the machine learning solution template, nor do they need to understand the modeling process. They can obtain machine learning results that match the business scenario simply by visually matching the actual business data with the data in the template solution.

[0165] As mentioned above, the template scheme may also include at least one parameter placeholder, and the template file may also include parameter configuration constraint information for generating the parameter configuration interface. For details regarding parameter placeholders, parameter configuration interfaces, and parameter configuration constraint information, please refer to the relevant descriptions above; they will not be repeated here.

[0166] This invention can also display a parameter configuration interface generated based on parameter configuration constraints to a second user, obtain at least one configuration parameter configured by the second user through the parameter configuration interface, and replace at least one parameter placeholder in the template scheme with the obtained at least one configuration parameter to obtain a modified template scheme. Thus, some configurable parameters can be made available to the second user. These configurable parameters can be parts that need to be adjusted according to business requirements. In this way, the execution effect of the machine learning process can be ensured to adapt to the actual business scenario, improving the usability of the machine learning scheme template.

[0167] The parameter configuration interface can include controls for setting configuration parameters and can receive configuration parameters set by a second user through the controls.

[0168] The parameter configuration constraints may also include, but are not limited to, at least one of the following: the type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder. The parameter configuration interface may also include, but is not limited to, at least one of the following: the type information, input method information, display information, and default value of the parameter placeholder that needs to be configured. For details regarding type information, input method information, display information, default value, and actual value, please refer to the relevant descriptions above; they will not be repeated here.

[0169] The parameter configuration constraint information may further include processing items for defining the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme. The present invention can also process the configuration parameters according to the processing items and display the processing results in the input source configuration interface. The processing items may include validation items for validating at least one configuration parameter. Validation items may include, but are not limited to, the allowed formats and / or allowed value ranges of the configuration parameter, and may also include indication information regarding whether validation is performed.

[0170] The parameter configuration constraints can also include category information to restrict the configuration of parameter placeholders according to category areas on the parameter configuration interface. The parameter configuration interface can display the parameter placeholders to be configured according to the category information, with parameter placeholders of different categories displayed in different category areas. Categorized display means that parameters belonging to the same category are displayed as a group, making the user's input more logical.

[0171] The template file may also include explanatory documentation to assist a second user in understanding and / or configuring the template scheme. This invention can also provide explanatory documentation to the second user. The explanatory documentation may be described in business language that is easy for the second user to understand, so that in most cases the second user does not need to understand machine learning concepts to use the machine learning scheme template.

[0172] As an example of the present invention, the template file may further include resource configuration information, which characterizes the resource configuration for performing at least part of the machine learning process, wherein the resource configuration information may be set by a first user. When performing step S350, the machine learning process can be performed based on the modified template scheme, using the resource configuration characterized by the resource configuration information.

[0173] As another example of the present invention, during step S350, the resource configuration required for the modified template scheme during the machine learning process can also be predicted, and the predicted resource configuration can be used to execute the machine learning process. The required resource configuration can be predicted using, but is not limited to, rule formulas or prediction models. For example, the required resource configuration can be calculated based on rule formulas, combined with the amount of sample data, feature extraction methods, and model training algorithms. Alternatively, various data volumes for the modeling task can be provided under different resource configurations (e.g., resource configurations increasing from small to large). The data volumes and resource configurations at the time of successful and failed trials can be used as samples to pre-train a machine learning model for predicting the required resource configuration, and the machine learning model can be used to predict the required resources.

[0174] Figure 3 The method shown can be executed by a machine learning platform used to implement machine learning-related business. Figure 4 The diagram illustrates the configuration interface of a machine learning solution template displayed to a user by the machine learning platform. The user mentioned here refers to the user of the machine learning solution template, i.e., the second user mentioned above.

[0175] like Figure 4 As shown, the top left corner displays the name of the modeling template, and the modeling template configuration instructions are the documentation uploaded by the first user when creating this modeling template. The modeling template configuration instructions explain which data tables and fields need to be configured for the current modeling template, and their corresponding business meanings.

[0176] The user table, product table, and behavior table are the names of the input tables that need to be configured in the template solution. Users need to fill in the corresponding business data tables. Users can directly enter the table name of the business data table. During the user's filling process, the system can search for business data tables that match the user's input, and a drop-down list will pop up the matching items. The matching items can be sorted in reverse chronological order, with 5 business data tables displayed per page. Users can view them by scrolling down.

[0177] Taking the user table as an example, user_id, age, and sex are fields that need to be configured in the user table. After selecting the corresponding business data table for the user table, users can select the corresponding fields in the business data table for the fields that need to be configured in the user table. After the fields are filled in, the user can verify whether the entered fields meet the requirements. The user can verify the field type and field threshold. If the field type does not meet the requirements, the user can be prompted with "#field name# requires a string field. The current configuration field does not meet the requirements. Please reconfigure." The corresponding error field is displayed with a red star. If the field threshold does not meet the requirements, the user can be prompted with "#field name# only accepts data from #lower threshold# to #upper threshold#. Please check the corresponding table data."

[0178] Timing parameter settings and sample parameter settings refer to placeholders for the parameters that need to be configured. The corresponding names and input content can be displayed based on the parameter configuration constraints. If the parameter configuration constraints include one or more options, these options will be displayed for the user to choose from. As an example, default values ​​that can be adjusted by the user can be displayed.

[0179] After the user completes the form, the replaceable parts of the template (such as input source tags and parameter placeholders) can be replaced based on the user's input, and the machine learning process can be executed using the replaced template.

[0180] The method for creating machine learning scheme templates of the present invention can also be implemented as an apparatus for creating machine learning scheme templates. Figure 5A structural block diagram of an apparatus for creating machine learning scheme templates according to an exemplary embodiment of the present invention is shown. The functional units of the apparatus for creating machine learning scheme templates can be implemented by hardware, software, or a combination of hardware and software that implement the principles of the present invention. Those skilled in the art will understand that... Figure 5 The described functional units can be combined or divided into sub-units to achieve the principles of the invention described above. Therefore, the description herein can support any possible combination, division, or further limitation of the functional units described herein.

[0181] The following is a brief description of the functional units that a device for creating machine learning scheme templates may have, and the operations that each functional unit can perform. For details, please refer to the relevant descriptions above, which will not be repeated here.

[0182] See Figure 5 The apparatus 500 for creating machine learning scheme templates includes a first acquisition module 510, a second acquisition module 520, and a generation module 530.

[0183] The first acquisition module 510 is used to acquire a template scheme for describing at least a portion of a machine learning process for at least one input source label, wherein the machine learning process involves model training and / or model application.

[0184] The second acquisition module 520 is used to acquire input source configuration constraint information about the template scheme, wherein the input source configuration constraint information is used to generate an input source configuration interface, such that at least one configuration input source configured through the input source configuration interface replaces at least one input source mark in the template scheme.

[0185] As an example, the second acquisition module 520 can generate a control for setting input source configuration limitation information based on the acquired template scheme, display the generated control to the first user, and receive the input source configuration limitation information set by the first user through the control.

[0186] For information on template schemes, input source tags, input source configuration limitations, and input source configuration interfaces, please refer to the relevant descriptions above; they will not be repeated here.

[0187] The generation module 530 is used to generate a template file for a machine learning scheme template based on the acquired template scheme and input source configuration constraint information.

[0188] The template scheme may include at least one parameter placeholder. The apparatus 500 for creating the machine learning scheme template may also include a third acquisition module. The third acquisition module is used to acquire parameter configuration constraint information about the template scheme, wherein the parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme. The generation module 530 can generate a template file for the machine learning scheme template based on the acquired template scheme, parameter configuration constraint information, and input source configuration constraint information. For details regarding the parameter configuration constraint information and parameter configuration interface, please refer to the relevant descriptions above; they will not be repeated here.

[0189] As an example, the third acquisition module generates a control for setting parameter configuration limitation information based on the acquired template scheme, displays the generated control to the first user, and receives the parameter configuration limitation information set by the first user through the control. The parameter configuration limitation information may also include category information for limiting the configuration of parameter placeholders according to category areas on the parameter configuration interface. The third acquisition module also displays a control for classifying parameter placeholders to the first user, and obtains classification information based on the classification of parameter placeholders by the first user through the control.

[0190] The apparatus 500 for creating machine learning solution templates may further include a first display module, a first receiving module, and a first merging module. The first display module displays a control for uploading documentation to a first user; the first receiving module receives the documentation uploaded by the first user via the control; and the first merging module merges the documentation into the template file.

[0191] The apparatus 500 for creating machine learning scheme templates may further include a second display module, a second receiving module, and a second merging module. The second display module displays controls for setting resource configuration information to a first user; the second receiving module receives the resource configuration information set by the first user through the controls, the resource configuration information representing the resource configuration for performing at least part of the machine learning process; and the second merging module merges the resource configuration into the template file.

[0192] The apparatus 500 for creating machine learning solution templates may further include a third display module, a fourth acquisition module, a replacement module, an execution module, an evaluation module, and a publishing or debugging module. The third display module displays an input source configuration interface generated based on input source configuration constraint information to a third user; the fourth acquisition module acquires at least one configuration input source configured by the third user via the input source configuration interface based on a test data table corresponding to the test scenario; the replacement module replaces at least one input source marker in the template solution with the at least one configuration input source to obtain a modified machine learning solution template; the execution module executes at least a portion of the machine learning process based on the modified machine learning solution template to obtain the execution result of the at least a portion of the machine learning process; the evaluation module evaluates the execution result to obtain a test result; and the publishing or debugging module determines whether to publish the machine learning solution template based on the test result, or to debug the machine learning solution template based on the test result.

[0193] It should be understood that specific implementations of the apparatus 500 for creating machine learning scheme templates according to exemplary embodiments of the present invention can be referred to in conjunction with... Figure 1 , Figure 2 The description of the methods used to create machine learning scheme templates is provided below and will not be repeated here.

[0194] The method for executing a machine learning process based on a machine learning scheme template of the present invention can also be implemented as an apparatus for executing a machine learning process based on a machine learning scheme template. Figure 6 A structural block diagram of an apparatus for creating a machine learning scheme template according to an exemplary embodiment of the present invention is shown. The functional units of the apparatus for performing a machine learning process based on the machine learning scheme template can be implemented by hardware, software, or a combination of hardware and software that implement the principles of the present invention. Those skilled in the art will understand that... Figure 6 The described functional units can be combined or divided into sub-units to achieve the principles of the invention described above. Therefore, the description herein can support any possible combination, division, or further limitation of the functional units described herein.

[0195] The following is a brief description of the functional units that a device for executing a machine learning process based on a machine learning scheme template can have, and the operations that each functional unit can perform. For details involved, please refer to the relevant descriptions above, which will not be repeated here.

[0196] See Figure 6 The apparatus 600 for executing a machine learning process based on a machine learning scheme template includes a first acquisition module 610, a first display module 620, a second acquisition module 630, a first replacement module 640, and an execution module 650.

[0197] The first acquisition module 610 is used to acquire a template file of a machine learning scheme template. The template file includes a template scheme and input source configuration constraint information. The template scheme describes at least a portion of the machine learning process for at least one input source label. The machine learning process involves model training and / or model application. The input source configuration constraint information is used to generate an input source configuration interface. For details regarding the template scheme, input source label, and input source configuration constraint information, please refer to the relevant descriptions above; they will not be repeated here.

[0198] The first display module 620 is used to display the input source configuration interface generated based on the input source configuration constraint information to the second user. For details regarding the input source configuration interface, please refer to the relevant description above; it will not be repeated here.

[0199] The second acquisition module 630 is used to acquire at least one configuration input source configured by the second user through the input source configuration interface. The input source configuration interface may include controls for setting configuration input sources, and the second acquisition module 630 can receive the configuration input sources set by the second user through the controls. For details on configuration input sources, please refer to the relevant description above; further details will not be repeated here.

[0200] The first replacement module 640 is used to replace at least one input source marker in the template scheme with at least one acquired configuration input source to obtain a modified template scheme.

[0201] Execution module 650 is used to execute a machine learning process based on the modified template scheme. The template file may also include resource configuration information, which characterizes the resource configuration required to execute at least part of the machine learning process. Execution module 650 can execute the machine learning process based on the modified template scheme and using the resource configuration characterized by the resource configuration information. Alternatively, execution module 650 can predict the resource configuration required by the modified template scheme during the machine learning process and use the predicted resource configuration to execute the machine learning process. Specific implementation details of execution module 650 can be found in the relevant description above.

[0202] The input source configuration constraint information also includes processing items for defining the processing that at least one configuration input source configured via the input source configuration interface undergoes before replacing at least one input source tag in the template scheme. The apparatus 600 for performing a machine learning process based on the machine learning scheme template may further include a first processing module and a second display module. The first processing module is used to process the configuration input source according to the processing items; the second display module is used to display the processing results in the input source configuration interface. For details on the processing items, please refer to the relevant description above; they will not be repeated here.

[0203] The template scheme may also include at least one parameter placeholder, and the template file may also include parameter configuration constraint information. This parameter configuration constraint information is used to generate a parameter configuration interface. The apparatus 600 for executing a machine learning process based on the machine learning scheme template may also include a third display module, a third acquisition module, and a second replacement module. The third display module is used to display the parameter configuration interface generated based on the parameter configuration constraint information to a second user; the third acquisition module is used to acquire at least one configuration parameter configured by the second user through the parameter configuration interface; the second replacement module is used to replace at least one parameter placeholder in the template scheme with the acquired at least one configuration parameter to obtain a modified template scheme. The parameter configuration interface may include controls for setting configuration parameters, and the third acquisition module can receive the configuration parameters set by the second user through the controls. For details regarding parameter configuration constraint information and the parameter configuration interface, please refer to the relevant descriptions above; they will not be repeated here.

[0204] The parameter configuration constraint information may further include processing items used to define the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme. The apparatus 600 for executing a machine learning process based on the machine learning scheme template may further include a second processing module and a fourth display module. The second processing module is used to process the configuration parameters according to the processing items; the fourth display module is used to display the processing results in the input source configuration interface. For details on the processing items, please refer to the relevant description above; they will not be repeated here.

[0205] The template file also includes instructional documentation to assist a second user in understanding and / or configuring the template scheme. The apparatus 600 for performing a machine learning process based on the machine learning scheme template may also include a providing module for providing instructional documentation to the second user.

[0206] It should be understood that the specific implementation of the apparatus 600 for performing a machine learning process based on a machine learning scheme template according to an exemplary embodiment of the present invention can be referred to in conjunction with the above description. Figure 3 , Figure 4 The description of the method for executing the machine learning process based on the machine learning scheme template will not be repeated here.

[0207] Reference above Figures 1 to 6 A method, method, and apparatus for creating a machine learning scheme template according to exemplary embodiments of the present invention are described. It should be understood that the above method can be implemented by a program recorded on a computer-readable medium. For example, according to exemplary embodiments of the present invention, a computer-readable storage medium storing instructions can be provided, wherein the computer-readable medium records a method for executing the present invention for creating a machine learning scheme template (e.g., ...). Figure 1(as shown) or methods that perform machine learning processes based on machine learning scheme templates (e.g.) Figure 3 The computer program shown.

[0208] The computer program in the aforementioned computer-readable medium can run in an environment deployed in computer devices such as clients, hosts, agent devices, and servers. It should be noted that the computer program can be used to perform actions other than... Figure 1 or Figure 3 In addition to the steps shown, additional steps may be performed besides those described above, or more specific processing may be performed while executing the steps described above. The details of these additional steps and further processing have been referenced. Figure 1 , Figure 3 The above has already been described, and will not be repeated here to avoid repetition.

[0209] It should be noted that the apparatus for creating a machine learning scheme template and the apparatus for performing a machine learning process based on the machine learning scheme template according to the exemplary embodiments of the present invention can rely entirely on the operation of a computer program to realize the corresponding functions. That is, each apparatus corresponds to each step in the functional architecture of the computer program, so that the entire apparatus is called through a special software package (e.g., a lib library) to realize the corresponding functions.

[0210] on the other hand, Figure 5 , Figure 6 The various devices shown can also be implemented using hardware, software, firmware, middleware, microcode, or any combination thereof. When implemented in software, firmware, middleware, or microcode, the program code or code segment used to perform the corresponding operation can be stored in a computer-readable medium such as a storage medium, so that a processor can perform the corresponding operation by reading and running the corresponding program code or code segment.

[0211] For example, an exemplary embodiment of the present invention can also be implemented as a computing device, which includes a storage component and a processor. The storage component stores a set of computer-executable instructions. When the set of computer-executable instructions is executed by the processor, a method for creating a machine learning scheme template or a method for performing a machine learning process based on the machine learning scheme template is executed.

[0212] Specifically, the computing device can be deployed on a server or client, or on a node device in a distributed network environment. Furthermore, the computing device can be a PC, tablet, personal digital assistant, smartphone, web application, or other device capable of executing the aforementioned set of instructions.

[0213] Here, the computing device is not necessarily a single computing device, but can be any collection of devices or circuits capable of executing the above instructions (or instruction sets) individually or in combination. The computing device can also be part of an integrated control system or system manager, or can be configured to interconnect with a portable electronic device locally or remotely (e.g., via wireless transmission) through an interface.

[0214] In the computing device, the processor may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, the processor may also include analog processors, digital processors, microprocessors, multi-core processors, processor arrays, network processors, etc.

[0215] Some operations described in the method for creating a machine learning scheme template or the method for performing a machine learning process based on a machine learning scheme template according to exemplary embodiments of the present invention can be implemented in software, some operations can be implemented in hardware, and these operations can also be implemented in a combination of software and hardware.

[0216] The processor can execute instructions or code stored in one of its storage components, which can also store data. The instructions and data can also be sent and received over a network via a network interface device, which can employ any known transport protocol.

[0217] Storage components can be integrated with the processor, for example, by placing RAM or flash memory within an integrated circuit microprocessor. Alternatively, storage components can include separate devices, such as external disk drives, storage arrays, or other storage devices that can be used by any database system. Storage components and the processor can be operatively coupled, or can communicate with each other, for example, via I / O ports, network connections, etc., enabling the processor to read files stored in the storage component.

[0218] In addition, the computing device may also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, mouse, touch input device, etc.). All components of the computing device may be interconnected via a bus and / or network.

[0219] The operations involved in the method for creating a machine learning scheme template or the method for performing a machine learning process based on a machine learning scheme template according to exemplary embodiments of the present invention can be described as various interconnected or coupled functional blocks or functional diagrams. However, these functional blocks or functional diagrams can be equally integrated into a single logic device or operate according to non-precise boundaries.

[0220] For example, as described above, an apparatus for creating a machine learning scheme template or an apparatus for performing a machine learning process based on a machine learning scheme template according to an exemplary embodiment of the present invention may include a storage component and a processor, wherein the storage component stores a set of computer-executable instructions, and when the set of computer-executable instructions is executed by the processor, the method for creating a machine learning scheme template or the method for performing a machine learning process based on a machine learning scheme template described above is executed.

[0221] The foregoing has described various exemplary embodiments of the present invention. It should be understood that the above description is merely exemplary and not exhaustive, and the present invention is not limited to the disclosed exemplary embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for creating machine learning scheme templates, comprising: Obtain a template scheme for describing at least a portion of a machine learning process, wherein the template scheme includes at least one input source with input source tags, the machine learning process involves model training and / or model application, and the input source tags characterize input tables and / or fields in the template scheme that can be replaced by a second user; Obtain input source configuration constraint information for the template scheme, wherein the input source configuration constraint information is set by a first user, and includes information that assists a second user in mapping actual business data to input sources in the template scheme. The input source configuration constraint information includes the name of the input table and the names of each field that needs to be configured under the input table. The fields that need to be configured under the input table include original fields existing in the input table and extended fields that do not exist in the input table. The input source configuration constraint information is used to generate an input source configuration interface, which is used by the second user to configure at least one input source to replace at least one input source marker in the template scheme; and A template file for generating a machine learning scheme template is generated based on the obtained template scheme and input source configuration constraint information.

2. The method according to claim 1, wherein, The steps for obtaining input source configuration constraint information for the template scheme include: The obtained template scheme is used to generate controls for setting input source configuration constraints. The generated controls are displayed to the first user; and Receive the input source configuration limitation information set by the first user through the control.

3. The method according to claim 2, wherein, The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured, the processing node corresponding to each input table, and the indication information of whether each field is displayed as an optional field on the input source configuration interface.

4. The method according to claim 3, wherein, The input source configuration limitation information also includes the field format corresponding to each field. At least one field is set to allow one or more fields in the actual business data to be configured for a single field, so that the configured one or more fields are processed in the same way as the single field in the template scheme.

5. The method according to claim 3, wherein, The input source configuration constraint information also includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme.

6. The method according to claim 5, wherein, The processing items include validation items for each field, wherein the validation items include the allowed formats and / or allowed value ranges for each field.

7. The method according to claim 6, wherein, The verification item also includes indications of whether or not to perform a verification.

8. The method according to claim 1, wherein, The template scheme includes at least one parameter placeholder, and the method further includes: Obtain parameter configuration constraint information regarding the template scheme, wherein the parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme. The step of generating a template file for a machine learning solution template based on the acquired template scheme and input source configuration constraint information includes: generating a template file for a machine learning solution template based on the acquired template scheme, parameter configuration constraint information, and input source configuration constraint information.

9. The method according to claim 8, wherein, The parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder.

10. The method according to claim 9, wherein, The type information indicates script parameters and / or runtime parameters.

11. The method according to claim 10, wherein, The parameter configuration constraint information also includes a processing item for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme.

12. The method according to claim 11, wherein, The processing item includes a verification item for verifying the at least one configuration parameter.

13. The method according to claim 8, wherein, The steps for obtaining parameter configuration limitation information regarding the template scheme include: The obtained template scheme is used to generate controls for setting parameter configuration constraints. The generated controls are displayed to the first user; and Receive parameter configuration limitation information set by the first user through the control.

14. The method according to claim 13, wherein, The parameter configuration limitation information further includes: classification information used to limit the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the step of obtaining parameter configuration limitation information about the template scheme further includes: Show the first user a control for categorizing parameter placeholders; and The classification information is obtained based on the classification of the parameter placeholders by the first user through the control.

15. The method according to claim 1, further comprising: Show the first user the control for uploading the documentation; Receive the instruction document uploaded by the first user through the control; as well as Merge the documentation into the template file.

16. The method according to claim 1, further comprising: Display controls for setting resource configuration information to the first user; Receive resource configuration information set by a first user through the control, the resource configuration information being used to characterize the resource configuration for executing at least part of the machine learning process; as well as Merge resource configurations into the template file.

17. The method according to any one of claims 1 to 16, further comprising: Display the input source configuration interface, generated based on input source configuration constraint information, to third-party users; Obtain at least one configuration input source configured by a third user through the input source configuration interface based on the test data table corresponding to the test scenario; Replace at least one input source marker in the template scheme with the at least one configuration input source to obtain the modified machine learning scheme template; The machine learning process is executed based on the modified machine learning scheme template to obtain the execution result of the machine learning process. The execution results are evaluated to obtain test results; Based on the test results, determine whether to release the machine learning solution template, or debug the machine learning solution template based on the test results.

18. A method for executing a machine learning process based on a machine learning scheme template, comprising: A template file for obtaining a machine learning solution template is provided. The template file includes a template solution and input source configuration constraint information. The template solution describes at least part of the machine learning process. The template solution includes at least one input source with an input source tag. The input source tag represents an input table and / or field that can be replaced in the template solution. The machine learning process involves model training and / or model application. The input source configuration constraint information is used to generate an input source configuration interface. The input source configuration constraint information is set by a first user. The input source configuration constraint information includes information that can be used to assist a second user in mapping actual business data to the input sources in the template solution. The input source configuration constraint information includes the name of the input table and the names of each first field that needs to be configured under the input table. The first fields that need to be configured under the input table include original fields that exist in the input table and extended fields that do not exist in the input table. Display the input source configuration interface, generated based on the input source configuration constraint information, to the second user; Obtain at least one configuration input source configured by the second user via the input source configuration interface; Replace at least one input source tag in the template scheme with at least one obtained configuration input source to obtain the modified template scheme; The machine learning process is performed based on the modified template scheme.

19. The method according to claim 18, wherein, The input source configuration interface includes controls for setting configuration input sources. The step of obtaining at least one configuration input source configured by the second user through the input source configuration interface includes: Receive the configuration input source set by the second user through the control.

20. The method according to claim 19, wherein, The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured, the processing node corresponding to each input table, and indication information on whether each first field is displayed as an optional field on the input source configuration interface. The input source configuration interface also includes at least one of the following items: the name of at least one input table to be configured, the name of each first field to be configured under each input table, and indication information on whether each first field is an optional field. The configuration input source includes at least one of the following: a business data table configured for the input table, and a second field under the business data table configured for the first field.

21. The method according to claim 20, wherein, The input source configuration limitation information also includes the field format corresponding to each first field. At least one first field's field format is set to allow configuration of one or more second fields from actual business data for a single first field, ensuring that the configured one or more second fields are processed in the same way as the single first field in the template scheme. The control in the input source configuration interface used to set the second field for the first field is generated based on the field format corresponding to the first field, so that the second user can use the control to... Configure the second field according to the field format corresponding to the first field.

22. The method according to claim 20, wherein, The input source configuration constraint information also includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme. The method further includes: processing the configuration input source according to the processing item; and displaying the processing result in the input source configuration interface.

23. The method according to claim 22, wherein, The processing items include validation items for each first field, wherein the validation items include the allowed formats and / or allowed value ranges for each first field. The steps of processing the configuration input source according to the processing items include: validating the format and / or value of the second field configured in the first field according to the validation items of the first field. The processing result is used to indicate whether the format and / or value of the second field configured for the first field conforms to the validation item.

24. The method according to claim 23, wherein, The verification item also includes indications of whether or not to perform a verification.

25. The method according to claim 18, wherein, The template scheme includes at least one parameter placeholder, and the template file also includes parameter configuration constraint information, which is used to generate a parameter configuration interface. The method further includes: Display the parameter configuration interface generated based on the parameter configuration constraint information to the second user; Obtain at least one configuration parameter configured by the second user through the parameter configuration interface; Replace at least one parameter placeholder in the template scheme with at least one obtained configuration parameter to obtain the modified template scheme.

26. The method of claim 25, wherein, The parameter configuration interface includes controls for setting configuration parameters. The step of obtaining at least one configuration parameter configured by the second user through the parameter configuration interface includes: Receive configuration parameters set by the second user through the control.

27. The method according to claim 26, wherein, The parameter configuration constraint information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder. The parameter configuration interface also includes at least one of the following items: type information of the parameter placeholder to be configured, input method information, display information, and default value.

28. The method according to claim 27, wherein, The parameter configuration constraint information also includes processing items for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme. The method further includes: processing the configuration parameters according to the processing item; and displaying the processing result in the input source configuration interface.

29. The method according to claim 28, wherein, The processing item includes a verification item for verifying the at least one configuration parameter.

30. The method according to claim 27, wherein, The type information indicates script parameters and / or runtime parameters.

31. The method according to claim 25, wherein, The parameter configuration limitation information also includes: classification information used to limit the configuration of parameter placeholders according to classification areas on the parameter configuration interface. The parameter configuration interface displays the parameter placeholders that need to be configured according to the classification information, and the parameter placeholders of different categories are displayed in different category areas.

32. The method according to claim 18, wherein, The template file also includes a description document to assist the second user in understanding and / or configuring the template scheme, and the method further includes providing the description document to the second user.

33. The method according to claim 18, wherein, The template file also includes resource configuration information, which characterizes the resource configuration for performing at least part of the machine learning process. The step of performing the machine learning process based on the modified template scheme includes: performing the machine learning process using the resource configuration characterized by the resource configuration information, based on the modified template scheme, or... The steps for performing a machine learning process based on the modified template scheme include: predicting the resource configuration required by the modified template scheme in the machine learning process, and using the predicted resource configuration to perform the machine learning process.

34. An apparatus for creating machine learning scheme templates, comprising: The first acquisition module is used to acquire a template scheme for describing at least part of a machine learning process, wherein the template scheme includes at least one input source with input source tags, the machine learning process involves model training and / or model application, and the input source tags characterize input tables and / or fields in the template scheme that can be replaced by a second user; The second acquisition module is used to acquire input source configuration limitation information regarding the template scheme. This input source configuration limitation information is set by the first user and includes information that assists the second user in mapping actual business data to input sources in the template scheme. The input source configuration limitation information includes the name of the input table and the names of each field that needs to be configured under the input table. The fields that need to be configured under the input table include original fields existing in the input table and extended fields not existing in the input table. The input source configuration limitation information is used to generate an input source configuration interface, which allows the second user to configure at least one input source to replace at least one input source marker in the template scheme. The generation module is used to generate template files for machine learning scheme templates based on the acquired template scheme and input source configuration constraint information.

35. The apparatus according to claim 34, wherein, Second acquisition module: The obtained template scheme is used to generate controls for setting input source configuration constraints. The generated controls are displayed to the first user; and Receive the input source configuration limitation information set by the first user through the control.

36. The apparatus according to claim 35, wherein, The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured, the processing node corresponding to each input table, and the indication information of whether each field is displayed as an optional field on the input source configuration interface.

37. The apparatus according to claim 36, wherein, The input source configuration limitation information also includes the field format corresponding to each field. At least one field is set to allow one or more fields in the actual business data to be configured for a single field, so that the configured one or more fields are processed in the same way as the single field in the template scheme.

38. The apparatus according to claim 36, wherein, The input source configuration constraint information also includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme.

39. The apparatus according to claim 38, wherein, The processing items include validation items for each field, wherein the validation items include the allowed formats and / or allowed value ranges for each field.

40. The apparatus according to claim 39, wherein, The verification item also includes indications of whether or not to perform a verification.

41. The apparatus according to claim 34, wherein, The template scheme includes at least one parameter placeholder, and the device further includes: The third acquisition module is used to acquire parameter configuration constraint information about the template scheme. This parameter configuration constraint information is used to generate a parameter configuration interface, such that at least one configuration parameter configured via the parameter configuration interface replaces at least one parameter placeholder in the template scheme. The generation module generates a template file for a machine learning scheme template based on the acquired template scheme, parameter configuration limitation information, and input source configuration limitation information.

42. The apparatus according to claim 41, wherein, The parameter configuration limitation information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder.

43. The apparatus according to claim 42, wherein, The type information indicates script parameters and / or runtime parameters.

44. The apparatus according to claim 43, wherein, The parameter configuration constraint information also includes a processing item for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme.

45. The apparatus according to claim 44, wherein, The processing item includes a verification item for verifying the at least one configuration parameter.

46. ​​The apparatus according to claim 41, wherein, The third acquisition module: The obtained template scheme is used to generate controls for setting parameter configuration constraints. The generated controls are displayed to the first user; and Receive parameter configuration limitation information set by the first user through the control.

47. The apparatus according to claim 46, wherein, The parameter configuration limitation information also includes: classification information used to limit the configuration of parameter placeholders according to classification areas on the parameter configuration interface, wherein the third acquisition module further includes: Show the first user a control for categorizing parameter placeholders; and The classification information is obtained based on the classification of the parameter placeholders by the first user through the control.

48. The apparatus of claim 34, further comprising: The first display module is used to show the first user the controls for uploading instruction documents; The first receiving module is used to receive the instruction document uploaded by the first user through the control; as well as The first merging module is used to merge documentation into the template file.

49. The apparatus of claim 34, further comprising: The second display module is used to show the first user controls for setting resource configuration information; The second receiving module is used to receive resource configuration information set by the first user through the control, wherein the resource configuration information is used to characterize the resource configuration for executing at least part of the machine learning process; as well as The second merging module is used to merge resource configurations into the template file.

50. The apparatus according to any one of claims 34 to 49, further comprising: The third display module is used to display the input source configuration interface generated based on the input source configuration limitation information to a third user. The fourth acquisition module is used to acquire at least one configuration input source configured by the third user through the input source configuration interface based on the test data table corresponding to the test scenario. A replacement module is used to replace at least one input source tag in the template scheme with the at least one configuration input source to obtain a modified machine learning scheme template; The execution module is used to execute at least a portion of the machine learning scheme template. The learning process is used to obtain the execution results of at least a portion of the machine learning process; An evaluation module is used to evaluate the execution results to obtain test results; The release or debug module is used to determine whether to release the machine learning solution template based on the test results, or to debug the machine learning solution template based on the test results.

51. An apparatus for executing a machine learning process based on a machine learning scheme template, comprising: The first acquisition module is used to acquire a template file of a machine learning solution template. The template file includes a template solution and input source configuration constraint information. The template solution describes at least part of the machine learning process. The template solution includes at least one input source with an input source tag. The input source tag represents an input table and / or field in the template solution that can be replaced. The machine learning process involves model training and / or model application. The input source configuration constraint information is used to generate an input source configuration interface. The input source configuration constraint information is set by a first user. The input source configuration constraint information includes information that can be used to assist a second user in matching actual business data with the input sources in the template solution. The input source configuration constraint information includes the name of the input table and the names of each first field that needs to be configured under the input table. The first fields that need to be configured under the input table include original fields that exist in the input table and extended fields that do not exist in the input table. The first display module is used to display the input source configuration interface generated based on the input source configuration constraint information to the second user; The second acquisition module is used to acquire at least one configuration input source configured by the second user through the input source configuration interface; The first replacement module is used to replace at least one input source tag in the template scheme with at least one obtained configuration input source to obtain the modified template scheme; The execution module is used to perform the machine learning process based on the modified template scheme.

52. The apparatus according to claim 51, wherein, The input source configuration interface includes controls for setting and configuring the input source. The second acquisition module receives input from the second user through the controls. The configuration input source is set.

53. The apparatus according to claim 52, wherein, The input source configuration limitation information includes at least one of the following: the name of at least one input table that needs to be configured, the processing node corresponding to each input table, and indication information on whether each first field is displayed as an optional field on the input source configuration interface. The input source configuration interface also includes at least one of the following items: the name of at least one input table to be configured, the name of each first field to be configured under each input table, and the first field is... No, this indicates that the field is optional. The configuration input source includes at least one of the following: a business data table configured for the input table, and a second field under the business data table configured for the first field.

54. The apparatus according to claim 53, wherein, The input source configuration limitation information also includes the field format corresponding to each first field. At least one first field's field format is set to allow configuration of one or more second fields from actual business data for a single first field, ensuring that the configured one or more second fields are processed in the same way as the single first field in the template scheme. The control in the input source configuration interface used to set the second field for the first field is generated based on the field format corresponding to the first field, so that the second user can configure the second field for the first field according to the field format corresponding to the first field through the control.

55. The apparatus according to claim 53, wherein, The input source configuration constraint information also includes processing items for limiting the processing that at least one configuration input source configured via the input source configuration interface undergoes before at least one input source tag in the replacement template scheme. The device further includes: a first processing module for processing the configuration input source according to the processing item; and a second display module for displaying the processing result in the input source configuration interface.

56. The apparatus according to claim 55, wherein, The processing items include validation items for each first field, wherein the validation items include the allowed formats and / or allowed value ranges for each first field. The steps of processing the configuration input source according to the processing items include: validating the format and / or value of the second field configured in the first field according to the validation items of the first field. The processing result is used to indicate whether the format and / or value of the second field configured for the first field conforms to the validation item.

57. The apparatus according to claim 56, wherein, The verification item also includes indications of whether or not to perform a verification.

58. The apparatus according to claim 51, wherein, The template scheme includes at least one parameter placeholder, and the template file further includes parameter configuration constraint information, which is used to generate a parameter configuration interface. The device also includes: The third display module is used to show the second user a parameter configuration interface generated based on parameter configuration limitation information; The third acquisition module is used to acquire at least one configuration parameter configured by the second user through the parameter configuration interface; The second replacement module is used to replace at least one parameter placeholder in the template scheme with at least one obtained configuration parameter to obtain the modified template scheme.

59. The apparatus according to claim 58, wherein, The parameter configuration interface includes controls for setting configuration parameters, and the third acquisition module receives the configuration parameters set by the second user through the controls.

60. The apparatus according to claim 59, wherein, The parameter configuration constraint information includes at least one of the following: type information, input method information, display information, default value, and actual value displayed on the parameter configuration interface that require configuration for the parameter placeholder. The parameter configuration interface also includes at least one of the following items: type information of the parameter placeholder to be configured, input method information, display information, and default value.

61. The apparatus according to claim 60, wherein, The parameter configuration constraint information also includes processing items for limiting the processing that at least one configuration parameter configured via the parameter configuration interface undergoes before replacing at least one parameter placeholder in the template scheme. The device further includes: a second processing module for processing the configuration parameters according to the processing items; and a fourth display module for displaying the processing results in the input source configuration interface.

62. The apparatus according to claim 61, wherein, The processing item includes a verification item for verifying the at least one configuration parameter.

63. The apparatus according to claim 60, wherein, The type information indicates script parameters and / or runtime parameters.

64. The apparatus according to claim 58, wherein, The parameter configuration limitation information also includes: classification information used to limit the configuration of parameter placeholders according to classification areas on the parameter configuration interface. The parameter configuration interface displays the parameter placeholders that need to be configured according to the classification information, and the parameter placeholders of different categories are displayed in different category areas.

65. The apparatus according to claim 51, wherein, The template file also includes a description document to assist the second user in understanding and / or configuring the template scheme. The device also includes a providing module for providing the description document to the second user.

66. The apparatus according to claim 51, wherein, The template file also includes resource configuration information, which characterizes the resource configuration for executing at least part of the machine learning process. The execution module, based on the modified template scheme, executes the machine learning process using the resource configuration characterized by the resource configuration information, or... The execution module predicts the resource configuration required by the modified template scheme during the machine learning process, and uses the predicted resource configuration to execute the machine learning process.

67. A system comprising at least one computing device and at least one storage device for storing instructions, wherein, When the instruction is executed by the at least one computing device, it causes the at least one computing device to... The apparatus performs the method as described in any one of claims 1 to 33.

68. A computer-readable storage medium for storing instructions, wherein, When the instruction is executed by at least one computing device, it causes the at least one computing device to perform the method as described in any one of claims 1 to 33.

Images