Model training method, information generation method, device, equipment, medium and product

Abstract

The embodiment of the invention discloses a model training method and device, an information generation method and device, equipment, a medium and a product. A specific embodiment of the method comprises the steps of obtaining a training data set; for the target training data, executing a training step: performing feature grouping on a feature set corresponding to the target historical customer group feature information to generate at least one feature group; determining at least one feature information group corresponding to the at least one feature group; inputting the at least one feature information group into the initial feature cross combination model to generate feature cross combination information; determining whether the initial customer group recommendation feature association information generation model is trained or not; and determining the initial customer group recommendation feature associated information generation model as a customer group recommendation feature associated information generation model in response to determining that the training is completed. The implementation mode is related to artificial intelligence, the model is generated by utilizing the customer group recommendation feature association information, and the recommendation feature set for the customer group can be accurately screened out for related business scenes.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of computer technology, and particularly, to a model training method, an information generation method, an apparatus, a device, a medium and a product. BACKGROUND

[0002] At present, the value risk detection related to the customer group has become an important development direction in the current financial technology field. The selection of features related to value risk is crucial, which directly determines the accuracy of value risk detection.

[0003] For the selection of features related to value risk, the commonly used way is to select features for the customer group through a filtering method (for example, a variance threshold method, a mutual information method).

[0004] However, the inventors have found that when the above-mentioned way is used to select features, the following technical problems often exist:

[0005] When selecting features, the interaction between feature selection and model training is ignored, resulting in that the selected features for the customer group are not accurate enough.

[0006] The above information disclosed in the background section is only intended to enhance the understanding of the background of the inventive concept, and therefore, it can include information that does not form the prior art known to those of ordinary skill in the art in the country. SUMMARY

[0007] The summary section of the present disclosure is used to introduce the concept in a brief form, which will be described in detail in the specific embodiments section. The summary section of the present disclosure is not intended to identify key or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0008] Some embodiments of the present disclosure propose a model training method, an information generation method, an apparatus, a device, a medium and a product to solve the technical problems mentioned in the background section.

[0009] In a first aspect, some embodiments of the present disclosure provide a model training method, comprising: obtaining a training data set, wherein the training data comprises: historical customer group feature description information, historical customer group feature information, and a historical customer group recommendation feature set; for target training data in the training data set, performing the following training steps: performing feature grouping on a feature set corresponding to the target historical customer group feature information to generate at least one feature group, wherein the target training data comprises: target historical customer group feature information, target historical customer group feature description information, and a target historical customer group recommendation feature set; determining at least one feature information group corresponding to the at least one feature group according to the target historical customer group feature information and the target historical customer group feature description information; inputting the at least one feature information group into an initial feature cross combination model to generate feature cross combination information, wherein the initial customer group recommendation feature association information generation model comprises: the initial feature cross combination model; determining whether the initial customer group recommendation feature association information generation model is trained according to the feature cross combination information and the target historical customer group recommendation feature set; and in response to determining that the training is completed, determining the initial customer group recommendation feature association information generation model as the customer group recommendation feature association information generation model.

[0010] Optionally, the above-mentioned performing feature grouping on a feature set corresponding to the target historical customer group feature information to generate at least one feature group comprises: performing feature grouping of the same category on the feature set corresponding to the target historical customer group feature information to generate at least one feature group.

[0011] Optionally, the above-mentioned initial customer group recommendation feature association information generation model further comprises: an initial feature linear combination model; and the above-mentioned determining whether the initial customer group recommendation feature association information generation model is trained according to the feature cross combination information and the target historical customer group recommendation feature set comprises: inputting the target historical customer group feature information and the target historical customer group feature description information into the initial feature linear combination model to generate feature linear combination information; generating initial customer group recommendation feature association information according to the feature linear combination information and the feature cross combination information; and determining whether the initial customer group recommendation feature generation model is trained according to the initial customer group recommendation feature association information and the target historical customer group recommendation feature set.

[0012] Optionally, the above-mentioned method further comprises: in response to determining that the training is not completed, performing model parameter updating on the initial customer group recommendation feature generation model to obtain an updated customer group recommendation feature generation model; re-extracting training data from the training data set after removing the target training data as candidate training data; taking the updated customer group recommendation feature generation model as the initial customer group recommendation feature generation model and the candidate training data as the target training data, and performing the above-mentioned training steps again.

[0013] Optionally, the initial feature cross combination model is a model for extracting feature correlation information between two features in different feature groups.

[0014] In a second aspect, some embodiments of the present disclosure provide a model training apparatus, comprising: a first obtaining unit configured to obtain a training data set, wherein the training data comprises: historical customer group feature description information, historical customer group feature information, and a historical customer group recommendation feature set; and a training unit configured to perform the following training steps for target training data in the training data set: performing feature grouping on a feature set corresponding to the target historical customer group feature information to generate at least one feature group, wherein the target training data comprises: target historical customer group feature information, target historical customer group feature description information, and a target historical customer group recommendation feature set; determining at least one feature information group corresponding to the at least one feature group according to the target historical customer group feature information and the target historical customer group feature description information; inputting the at least one feature information group into an initial feature cross combination model to generate feature cross combination information, wherein the initial customer group recommendation feature correlation information generation model comprises the initial feature cross combination model; determining whether the initial customer group recommendation feature correlation information generation model is trained according to the feature cross combination information and the target historical customer group recommendation feature set; and in response to determining that the training is completed, determining the initial customer group recommendation feature correlation information generation model as the customer group recommendation feature correlation information generation model.

[0015] Optionally, the training unit can be configured to perform feature grouping of the same category on the feature set corresponding to the target historical customer group feature information to generate at least one feature group.

[0016] Optionally, the initial customer group recommendation feature correlation information generation model further comprises an initial feature linear combination model, and the training unit can be configured to input the target historical customer group feature information and the target historical customer group feature description information into the initial feature linear combination model to generate feature linear combination information, generate initial customer group recommendation feature correlation information according to the feature linear combination information and the feature cross combination information, and determine whether the initial customer group recommendation feature generation model is trained according to the initial customer group recommendation feature correlation information and the target historical customer group recommendation feature set.

[0017] Optionally, the steps further comprise: in response to determining that the training is not completed, performing model parameter updating on the initial customer group recommendation feature generation model to obtain an updated customer group recommendation feature generation model; re-extracting training data from the training data set after removing the target training data as candidate training data; and performing the training steps again by taking the updated customer group recommendation feature generation model as the initial customer group recommendation feature generation model and the candidate training data as the target training data.

[0018] Optionally, the initial feature cross combination model is a model for extracting feature correlation information between two features in different feature groups.

[0019] In a third aspect, some embodiments of the present disclosure provide an information generation method, including: obtaining customer value flow data for a target customer group; extracting customer feature description information and customer feature information for the customer value flow data; inputting the customer feature description information and the customer feature information into a pre-trained customer recommendation feature correlation information generation model to generate customer recommendation feature correlation information, wherein the customer recommendation feature correlation information generation model is generated based on the information generation method in the present disclosure.

[0020] Optionally, the method further includes: filtering a recommended feature set from a feature set corresponding to the customer feature information according to the customer recommendation feature correlation information; determining a recommended feature information set corresponding to the recommended feature set according to the customer value flow data; and performing a value risk prediction information generation operation on the recommended feature information set.

[0021] In a fourth aspect, some embodiments of the present disclosure provide an information generation apparatus, including: a second obtaining unit configured to obtain customer value flow data for a target customer group; an extracting unit configured to extract customer feature description information and customer feature information for the customer value flow data; and a generating unit configured to input the customer feature description information and the customer feature information into a pre-trained customer recommendation feature correlation information generation model to generate customer recommendation feature correlation information, wherein the customer recommendation feature correlation information generation model is generated based on the information generation method in the present disclosure.

[0022] Optionally, the apparatus further includes: filtering a recommended feature set from a feature set corresponding to the customer feature information according to the customer recommendation feature correlation information; determining a recommended feature information set corresponding to the recommended feature set according to the customer value flow data; and performing a value risk prediction information generation operation on the recommended feature information set.

[0023] In a fifth aspect, some embodiments of the present disclosure provide an electronic device, including: one or more processors; and a storage apparatus having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner of the first aspect and the third aspect.

[0024] In a sixth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program is executed by a processor to implement the method described in any implementation manner of the first aspect and the third aspect.

[0025] In a seventh aspect, some embodiments of the present disclosure provide a computer program product, comprising a computer program which, when executed by a processor, implements the method described in any implementation manner of the first aspect and the third aspect.

[0026] The above embodiments of this disclosure have the following beneficial effects: By using the model training methods of some embodiments of this disclosure to generate models using customer group recommendation feature association information, recommendation feature sets for customer groups can be accurately selected for relevant business scenarios. Specifically, the reason why the relevant recommendation feature sets are not accurate enough is that the interaction between feature selection and model training is ignored when selecting features, resulting in the features selected for customer groups being not accurate enough. Based on this, the model training methods of some embodiments of this disclosure first obtain a training dataset, wherein the training data includes: historical customer group feature description information, historical customer group feature information, and historical customer group recommendation feature sets. Here, the obtained training dataset is used for training the subsequent initial customer group recommendation feature association information generation model. Through historical customer group feature description information, the subsequent initial customer group recommendation feature association information generation model can learn more feature information about the specific feature descriptions of relevant features for customer groups, and the subsequent model can learn more about the association between data distribution and customer group recommendation feature sets. Moreover, through historical customer group feature information, the subsequent initial customer group recommendation feature association information generation model can learn more feature information about relevant features for customer groups. Then, for the target training data in the training dataset, the following training steps are performed: First, the feature set corresponding to the target historical customer group feature information is grouped to generate at least one feature group. The target training data includes: target historical customer group feature information, target historical customer group feature description information, and target historical customer group recommendation feature set. Grouping the feature set here facilitates the subsequent model learning more complex cross-relationships between features, thereby improving the accuracy of generating the subsequent customer group recommendation feature set. Second, based on the target historical customer group feature information and target historical customer group feature description information, at least one feature information group corresponding to at least one feature group is determined for use in the subsequent model training. Third, at least one feature information group is input into the initial feature cross-combination model to generate feature cross-combination information, thereby learning more complex cross-relationships between features and increasing the accuracy of generating the subsequent customer group recommendation feature set. The initial customer group recommendation feature association information generation model includes: the initial feature cross-combination model. Fourth, based on the feature cross-combination information and the target historical customer group recommendation feature set, it can be accurately determined whether the initial customer group recommendation feature association information generation model has been successfully trained. The fifth step, in response to confirming training completion, establishes the initial customer group recommendation feature association information generation model as the final customer group recommendation feature association information generation model. In summary, by learning the relevant feature information of the data distribution corresponding to historical customer group feature descriptions through the model, and by performing cross-learning between features through a feature cross-combination model, it is possible to accurately select recommendation feature sets for specific customer groups for relevant business scenarios. Attached Figure Description

[0027] The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by describing in detail some embodiments thereof with reference to the annexed drawings. In the drawings: Like reference numerals refer to like elements throughout the drawings. It should be noted that the elements in the following drawings are illustrated for purposes of example only and represent structures which can be incorporated into a larger system. Further, the drawings are not necessarily drawn to scale, with emphasis instead being placed upon illustrating the principles of the embodiments.

[0028] Figure 1 is a schematic diagram of one application scenario of a model training method according to some embodiments of the present disclosure;

[0029] Figure 2 is a flowchart of some embodiments of a model training method according to the present disclosure;

[0030] Figure 3 is a schematic diagram of feature same-category grouping in some embodiments of a model training method according to the present disclosure;

[0031] Figure 4 is a flowchart of another embodiments of a model training method according to the present disclosure;

[0032] Figure 5 is a flowchart of some embodiments of an information generation method according to the present disclosure;

[0033] Figure 6 is a structural schematic diagram of some embodiments of a model training apparatus according to the present disclosure;

[0034] Figure 7 is a structural schematic diagram of some embodiments of an information generation apparatus according to the present disclosure;

[0035] Figure 8 is a structural schematic diagram of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0036] Embodiments of the present disclosure will be described below in greater detail with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure will be more thoroughly and completely understood. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of the present disclosure.

[0037] It should also be noted that, for the sake of brevity, only some of the pertinent features of the application are shown in the drawings. Embodiments of the present disclosure and features in embodiments can be combined with each other as long as they do not conflict with each other.

[0038] It should be noted that the terms "first", "second", and the like in the present disclosure are merely intended to distinguish different devices, modules or units, and do not imply the sequence of the functions performed by these devices, modules or units or the mutual dependency of these devices, modules or units.

[0039] It should be noted that the terms "one", "multiple" in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that "one" or "multiple" should be understood as "one or more" unless otherwise explicitly indicated in the context.

[0040] The names of the messages or information exchanged between the plurality of devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of the messages or information.

[0041] The collection, storage and use of user personal information (such as historical customer feature information and historical customer feature description information) involved in the present disclosure should be performed by the relevant organization or individual to fulfill the obligations including conducting personal information security impact assessment, fulfilling the notification obligations to the personal information subject, and obtaining the prior authorization consent of the personal information subject.

[0042] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0043] Figure 1 is a schematic diagram of an application scenario of a model training method according to some embodiments of the present disclosure.

[0044] In Figure 1In the application scenario of the electronic device 101, first, the electronic device 101 can obtain a training data set. The training data includes: historical customer group feature description information, historical customer group feature information, and a historical customer group recommendation feature set. Then, for the target training data 102 in the training data set, the electronic device 101 can perform the following training steps: first, the target historical customer group feature information 1022 corresponds to the feature set 103, and at least one feature group 104 is generated. The target training data 102 includes: target historical customer group feature information 1022, target historical customer group feature description information 1021, and target historical customer group recommendation feature set 1023. In this application scenario, the at least one feature group 104 can include: feature group A and feature group B. Second, at least one feature information group 105 corresponding to at least one feature group 103 is determined according to the target historical customer group feature information 1022 and the target historical customer group feature description information 1021. In this application scenario, the at least one feature information group 105 can include: the feature information group A corresponding to the feature group A and the feature information group B corresponding to the feature group B. Third, at least one feature information group 105 is input into an initial feature cross combination model 106 to generate feature cross combination information 108. The initial customer group recommendation feature association information generation model 106 includes: an initial feature cross combination model 107. Fourth, whether the initial customer group recommendation feature association information generation model 106 is trained is determined according to the feature cross combination information 108 and the target historical customer group recommendation feature set 1023. Fifth, in response to determining that the training is complete, the initial customer group recommendation feature association information generation model 106 is determined as the customer group recommendation feature association information generation model.

[0045] It should be noted that the electronic device 101 described above can be hardware or software. When the electronic device is hardware, it can be implemented as a distributed cluster composed of multiple servers or terminal devices, or as a single server or a single terminal device. When the electronic device is software, it can be installed in the hardware devices listed above. It can be implemented as multiple software or software modules for providing distributed services, or as a single software or software module. No specific limitation is made here.

[0046] It should be understood that Figure 1 The number of electronic devices in the above description is only illustrative. According to the needs of implementation, there can be any number of electronic devices.

[0047] Continuing to refer to Figure 2 Flow 200 illustrating some embodiments of a model training method according to the present disclosure is shown. The model training method includes the following steps:

[0048] Step 201, obtaining a training data set.

[0049] In some embodiments, the execution subject of the above model training method (e.g. Figure 1 The electronic device 101 shown in the figure) can obtain the training data set through wired connection or wireless connection. The training data set can be a data set used for subsequent model training of the initial customer group recommendation feature association information generation model. The training data includes historical customer group feature description information, historical customer group feature information, and a historical customer group recommendation feature set. The historical customer group feature description information can be the feature description information of the target customer group in the historical time period. Specifically, the historical time period can be a pre-set time period. For example, the historical time period can be the first half of the current time. The target customer group can be the object used as the training data set in the future. In practice, the target customer group can be a target user group. Each object in the target customer group can be a group of objects with the same feature properties under a certain feature. For example, the objects in the target customer group can be a customer group using the same application software. The feature description information can be information describing the specific details of the features of the customer group. For example, the feature of the customer group is the age of the objects, and the corresponding feature description information can be, but not limited to, at least one of the following: the value range of the age, the variance of the age, whether the age is continuous, the maximum value of the age, and the minimum value of the age. The historical customer group feature information can be the feature information of the target customer group corresponding feature set in the historical time period. Specifically, the target customer group corresponding feature set can include, but not limited to, at least one of the following: the average age of the objects, the number of objects, and the geographical distribution of the objects. In practice, for the credit scenario, the target customer group corresponding feature set can further include, but not limited to, at least one of the following: the interest rate level of the objects, the institution type of the objects, the maximum limit of the objects, and the average limit of the objects. Each training data has corresponding customer group information. That is, each customer group information has corresponding historical customer group feature description information, historical customer group feature information, and historical customer group recommendation feature set. In practice, the customer group information can be the identifier corresponding to the customer group. The historical customer group recommendation feature set can represent the feature set for information recommendation to the customer group corresponding to the customer group information. In practice, the historical customer group recommendation feature set can be a subset of the historical customer group feature information corresponding feature set. Each historical customer group recommendation feature in the historical customer group recommendation feature set is a feature that most interests each object corresponding to the customer group information.

[0050] It should be noted that the above execution subject can encapsulate the customer group information and the historical customer group recommendation feature set in the form of mapping data as the label of the training data. That is, the label of the training data can be {customer group information, historical customer group recommendation feature set}.

[0051] Step 202, for the target training data in the training data set, the following training steps are performed:

[0052] Step 2021, grouping the target historical customer group feature information corresponding feature set by features to generate at least one feature group.

[0053] In some embodiments, the execution subject can perform feature grouping on the feature set corresponding to the target historical customer group feature information to generate at least one feature group. The target training data includes: target historical customer group feature information, target historical customer group feature description information, and target historical customer group recommendation feature set.

[0054] As an example, the execution subject can perform random grouping of different feature categories on each feature in the feature set corresponding to the target historical customer group feature information to generate at least one feature group. That is, the features between each feature group are features of different categories. In practice, the feature categories can include but are not limited to at least one of: feature field description category, feature field effect type, feature field distribution category, and customer group description category.

[0055] In some optional implementations of some embodiments, the execution subject can perform feature grouping of the same category on the feature set corresponding to the target historical customer group feature information to generate at least one feature group. That is, each feature in each feature group belongs to the same feature category.

[0056] As an example, the execution subject can use the feature-feature category association table to perform feature grouping of the same category on the feature set corresponding to the target historical customer group feature information to group features of the same category into a group to generate at least one feature group.

[0057] As an example, as shown in Figure 3 , a schematic diagram of feature grouping of the same category is shown. The feature set of the customer group object includes: feature 1, feature 2, feature 3, feature 4, feature 5, and feature 6. The at least one feature group includes: a feature group corresponding to the feature description category, a feature group corresponding to the feature effect category, a feature group corresponding to the feature distribution category, and a feature group corresponding to the customer group description category. The feature group corresponding to the feature description category can include: feature 1 and feature 2. The feature group corresponding to the feature effect category can include: feature 3 and feature 5. The feature group corresponding to the feature distribution category can include: feature 4. The feature group corresponding to the customer group description category can include: feature 6.

[0058] Step 2022, according to the target historical customer group feature information and the target historical customer group feature description information, determine at least one feature information group corresponding to at least one feature group.

[0059] In some embodiments, the execution subject can determine at least one feature information group corresponding to at least one feature group according to the target historical customer group feature information and the target historical customer group feature description information. The features in the at least one feature group and the feature information in the at least one feature information group have a one-to-one correspondence. The feature information is the specific content of the feature.

[0060] As an example, the execution subject can query the feature information corresponding to each feature in the at least one feature group from the target historical customer group feature information and the target historical customer group feature description information in a manner of information query, and obtain the at least one feature information group.

[0061] In step 2023, the at least one feature information group is input into an initial feature cross combination model to generate feature cross combination information.

[0062] In some embodiments, the execution subject can input the at least one feature information group into the initial feature cross combination model to generate the feature cross combination information. The feature cross combination information can represent the feature cross information between the respective feature information in the at least one feature information group. The initial feature cross combination model can be a feature cross combination model that has not yet completed training. The feature cross combination model can be a neural network model that cross-combines respective feature information. For example, the feature cross combination model can be a multi-layer fully connected layer. The initial customer group recommendation feature association information generation model includes the initial feature cross combination model. The initial customer group recommendation feature association information generation model can be a customer group recommendation feature association information generation model that has not yet completed training. The customer group recommendation feature association information can be information representing the association relationship between respective customer group recommendation features in a customer group recommendation feature set. In practice, the customer group recommendation feature association relationship can be an association parameter value between respective customer group recommendation features. The greater the association parameter value, the closer the feature between the two associated customer group recommendation features. For example, the customer group recommendation feature association information generation model can be a plurality of serially connected fully connected modules. Each fully connected module includes at least one serially connected fully connected layer. In practice, the feature cross combination information can be information in vector form.

[0063] In some optional implementations of some embodiments, the initial feature cross combination model described above is a model that extracts feature association information between two features in different feature groups. Specifically, the initial feature cross combination model includes a feature association parameter between two features in different feature groups. That is, each feature association parameter in the feature association parameter set included in the initial feature cross combination model corresponds to two features in different feature groups.

[0064] In step 2024, it is determined whether the initial customer group recommendation feature association information generation model is trained according to the feature cross combination information and the target historical customer group recommendation feature set.

[0065] In some embodiments, the execution subject can determine whether the initial customer group recommendation feature association information generation model is trained according to the feature cross combination information and the target historical customer group recommendation feature set.

[0066] As an example, first, the execution subject can input the feature cross combination information to the remaining full connection module included in the initial customer group recommendation feature association information generation model to generate initial customer group recommendation feature association information. Then, the initial customer group recommendation feature association information is analyzed to generate a customer group recommendation feature set. Next, the customer group recommendation feature set and the target historical customer group recommendation feature set are used to generate customer group recommendation feature loss information. Finally, in response to determining that the customer group recommendation feature loss information is less than the target loss value, it is determined that the initial customer group recommendation feature association information generation model training is complete. In response to determining that the customer group recommendation feature loss information is greater than or equal to the target loss value, it is determined that the initial customer group recommendation feature association information generation model training is not complete.

[0067] Step 2025, in response to determining that the training is complete, the initial customer group recommendation feature association information generation model is determined as the customer group recommendation feature association information generation model.

[0068] In some embodiments, in response to determining that the training is complete, the execution subject can determine the initial customer group recommendation feature association information generation model as the customer group recommendation feature association information generation model. The customer group recommendation feature association information generation model can be a neural network model after training.

[0069] In some optional implementations of some embodiments, after step 203, the steps further include:

[0070] First, in response to determining that the training is not complete, the initial customer group recommendation feature generation model is updated to obtain an updated customer group recommendation feature generation model.

[0071] As an example, the execution subject can use gradient descent and other algorithms to update the model parameters of the initial customer group recommendation feature generation model to obtain an updated customer group recommendation feature generation model.

[0072] Second, re-extract training data from the training data set after removing the target training data as candidate training data.

[0073] As an example, the execution subject can randomly extract training data from the training data set after removing the target training data as candidate training data

[0074] Third, the updated customer group recommendation feature generation model is used as the initial customer group recommendation feature generation model, and the candidate training data is used as the target training data. The training steps are executed again.

[0075] The above various embodiments of the present disclosure have the following beneficial effects: by the model training method of some embodiments of the present disclosure, the model for generating recommendation feature set for the customer group can be accurately selected for the relevant business scenarios. Specifically, the reason why the relevant recommendation feature set is not accurate is that the interaction between feature selection and model training is ignored during feature selection, resulting in inaccurate features selected for the customer group. Based on this, the model training method of some embodiments of the present disclosure first acquires a training data set, wherein the training data includes: historical customer group feature description information, historical customer group feature information and historical customer group recommendation feature set. Here, the acquired training data set is used for subsequent training of the initial customer group recommendation feature association information generation model. Through the historical customer group feature description information, the subsequent initial customer group recommendation feature association information generation model can learn more specific feature description of the relevant features of the customer group, and the subsequent model can learn more association between the data distribution and the customer group recommendation feature set. And through the historical customer group feature information, the subsequent initial customer group recommendation feature association information generation model can learn more feature information of the relevant features of the customer group. Then, for the target training data in the training data set, the following training steps are performed: first, the target historical customer group feature information corresponding feature set is grouped, and at least one feature group is generated. Wherein, the target training data includes: target historical customer group feature information, target historical customer group feature description information and target historical customer group recommendation feature set. Here, the feature grouping of the feature set can facilitate the subsequent model to learn more complex cross-relationships between features to improve the generation accuracy of the subsequent customer group recommendation feature set. The second step is to determine at least one feature information group corresponding to at least one feature group according to the target historical customer group feature information and the target historical customer group feature description information, for the subsequent model training. The third step is to input at least one feature information group into the initial feature cross-combination model to generate feature cross-combination information to learn more complex cross-relationships between features and increase the generation accuracy of the subsequent customer group recommendation feature set. Wherein, the initial customer group recommendation feature association information generation model includes: the initial feature cross-combination model. The fourth step is to accurately determine whether the initial customer group recommendation feature association information generation model is trained according to the feature cross-combination information and the target historical customer group recommendation feature set. The fifth step is to determine the initial customer group recommendation feature association information generation model as the customer group recommendation feature association information generation model in response to determining that the training is completed. In summary, by learning the relevant feature information of the data distribution corresponding to the historical customer group feature description information, and by cross-learning between features through the feature cross-combination model, the recommendation feature set for the customer group can be accurately selected for the relevant business scenarios.

[0076] Further reference Figure 4, shows the flow 400 of some other embodiments of the model training method according to the present disclosure. The model training method comprises the following steps:

[0077] Step 401, obtaining a training data set.

[0078] Step 402, for target training data in the training data set, performing the following training steps:

[0079] Step 4021, grouping the target historical customer feature information corresponding feature set by feature, to generate at least one feature group.

[0080] Step 4022, determining at least one feature information group corresponding to at least one feature group according to the target historical customer feature information and the target historical customer feature description information.

[0081] Step 4023, inputting the at least one feature information group into the initial feature cross combination model to generate feature cross combination information.

[0082] Step 4024, inputting the target historical customer feature information and the target historical customer feature description information into the initial feature linear combination model to generate feature linear combination information.

[0083] In some embodiments, the execution subject (for example Figure 1 The electronic device 101 shown can input the target historical customer feature information and the target historical customer feature description information into the initial feature linear combination model to generate feature linear combination information. Wherein, the initial feature linear combination model can be a feature linear combination model that has not yet completed training. The feature linear combination model can be a neural network model that linearly combines each feature information. The feature linear combination information can be the linear combination of each feature information between the target historical customer feature information and the target historical customer feature description information. For example, the feature linear combination model can be a fully connected layer. In practice, the feature linear combination information can be information in the form of a vector.

[0084] Step 4025, generating initial customer recommendation feature association information according to the feature linear combination information and the feature cross combination information.

[0085] In some embodiments, the execution subject can generate initial customer recommendation feature association information according to the feature linear combination information and the feature cross combination information.

[0086] As an example, first, the feature linear combination information and the feature cross combination information are fused to generate fusion information. Then, the fusion information is input into a fully connected layer to generate initial customer recommendation feature association information.

[0087] At step 4026, according to the initial customer group recommendation feature association information and the target historical customer group recommendation feature set, it is determined whether the initial customer group recommendation feature generation model is trained.

[0088] In some embodiments, the execution subject of the above can determine whether the initial customer group recommendation feature generation model is trained according to the initial customer group recommendation feature association information and the target historical customer group recommendation feature set.

[0089] As an example, first, the customer group recommendation feature set loss information for the initial customer group recommendation feature association information and the target historical customer group recommendation feature set is determined. Then, in response to determining that the customer group recommendation feature set loss information is less than the target loss value, it is determined that the initial customer group recommendation feature generation model is trained. In response to determining that the customer group recommendation feature set loss information is greater than or equal to the target loss value, it is determined that the initial customer group recommendation feature generation model is not trained.

[0090] At step 4027, in response to determining that the training is completed, the initial customer group recommendation feature association information generation model is determined as the customer group recommendation feature association information generation model.

[0091] In some embodiments, the specific implementation of steps 401, 4021-4023, 4027 and the technical effects brought by them can be referred to Figure 2 The specific implementation of steps 201, 2021-2023, 2025 and the technical effects brought by them in some embodiments corresponding to the above can be referred to

[0092] As can be seen from Figure 4 Compared with the description of some embodiments corresponding to Figure 2 , the model training method in some embodiments corresponding to Figure 4 The flow 300 of the model training method in some embodiments corresponding to the above increases the linear combination feature information for the target historical customer group feature information and the target historical customer group feature description information through the initial feature linear combination model in the initial customer group recommendation feature association information generation model, so that the initial customer group recommendation feature association information generation model learns more linear features, and the subsequently generated customer group recommendation feature association information is more accurate.

[0093] Continuing to refer to Figure 5 , the flow 500 of some embodiments of the model training method according to the present disclosure is shown. The model training method comprises the following steps:

[0094] At step 501, customer group value flow data for a target customer group is obtained.

[0095] In some embodiments, the execution subject of the above model training method (for example Figure 1The electronic device 101 shown can obtain customer value flow data for a target customer group. The target customer group can be a customer group for which a customer recommendation feature set is to be determined. The customer value flow data can be value flow data corresponding to the customer group. In practice, for a credit scenario, the customer value flow data can be consumption data for the customer group.

[0096] At step 502, customer feature description information and customer feature information for the customer value flow data are extracted.

[0097] In some embodiments, the execution subject can extract customer feature description information and customer feature information for the customer value flow data.

[0098] As an example, the execution subject can extract customer feature description information and customer feature information for the customer value flow data using a related feature information extraction method.

[0099] At step 503, the customer feature description information and the customer feature information are input into a pre-trained customer recommendation feature association information generation model to generate customer recommendation feature association information.

[0100] In some embodiments, the execution subject can input the customer feature description information and the customer feature information into a pre-trained customer recommendation feature association information generation model to generate customer recommendation feature association information. The customer recommendation feature association information generation model is generated based on a model training method.

[0101] In some optional implementations of some embodiments, after step 503, the steps further include:

[0102] First, a recommendation feature set is filtered from the feature set corresponding to the customer feature information according to the customer recommendation feature association information. The recommendation features in the recommendation feature set can be features of interest to the target customer group. For example, the recommendation feature set can include: home appliance features, target brand features. The recommendation feature set is a feature set of a recommendation information set for the target customer group. The recommendation information in the recommendation information set can be recommendation content recommended to some of the objects in the customer group. In practice, the recommendation information can be a recommended item, and can also be a recommended advertisement.

[0103] As an example, the execution subject can use a pre-set recommendation feature extraction method to filter a recommendation feature set from the feature set corresponding to the customer feature information according to the customer recommendation feature association information. For example, the recommendation feature extraction method can be to extract features corresponding to a connection value mean greater than a target value as recommendation features.

[0104] In the second step, the recommended feature information set corresponding to the recommended feature set is determined according to the customer value flow data. The recommended features in the recommended feature set and the recommended feature information in the recommended feature information set have a one-to-one correspondence.

[0105] As an example, the execution subject can extract the recommended feature information corresponding to each recommended feature in the recommended feature set from the customer value flow data to obtain the recommended feature information set.

[0106] In the third step, the value risk prediction information generation operation for the recommended feature information set is performed. The value risk prediction information generation operation can be an operation of generating value risk prediction information. In practice, for the credit scenario, the value risk prediction information can be credit risk prediction information.

[0107] The above-mentioned various embodiments of the present disclosure have the following beneficial effects: the information generation method of some embodiments of the present disclosure can utilize the customer recommended feature association information generation model to accurately screen out the recommended feature set for the customer for the relevant business scenario.

[0108] Further reference Figure 5 , as an implementation of the method shown in the above-mentioned various figures, the present disclosure provides some embodiments of a model training device, which device embodiments correspond to those method embodiments shown in Figure 2 , and the model training device can be specifically applied in various electronic devices.

[0109] As Figure 5 shown, a model training device 500 includes a first acquisition unit 501 and a training unit 502. The first acquisition unit 501 is configured to acquire a training data set, wherein the training data includes historical customer feature description information, historical customer feature information, and a historical customer recommended feature set; the training unit 502 is configured to perform feature grouping on a target historical customer feature information corresponding feature set to generate at least one feature group, wherein the target training data includes target historical customer feature information, target historical customer feature description information, and a target historical customer recommended feature set; determine at least one feature information group corresponding to at least one feature group according to the target historical customer feature information and the target historical customer feature description information; input the at least one feature information group into an initial feature cross combination model to generate feature cross combination information, wherein the initial customer recommended feature association information generation model includes the initial feature cross combination model; determine whether the initial customer recommended feature association information generation model is trained according to the feature cross combination information and the target historical customer recommended feature set; in response to determining that the training is completed, the initial customer recommended feature association information generation model is determined as the customer recommended feature association information generation model.

[0110] In some optional implementations of some embodiments, the training unit 502 can be further configured to: group the feature sets corresponding to the target historical customer group feature information into the same category, and generate at least one feature group.

[0111] In some optional implementations of some embodiments, the initial customer group recommendation feature association information generation model further includes an initial feature linear combination model, and the training unit 502 can be further configured to: input the target historical customer group feature information and the target historical customer group feature description information into the initial feature linear combination model to generate feature linear combination information; generate initial customer group recommendation feature association information according to the feature linear combination information and the feature cross combination information; and determine whether the initial customer group recommendation feature generation model is trained according to the initial customer group recommendation feature association information and the target historical customer group recommendation feature set.

[0112] In some optional implementations of some embodiments, the device 500 further includes a parameter updating unit, an extraction unit, and an execution unit (not shown in the figure). The parameter updating unit can be configured to: filter a recommendation feature set from the feature set corresponding to the customer group feature information according to the customer group recommendation feature association information. The extraction unit can be configured to: determine a recommendation feature information set corresponding to the recommendation feature set according to the customer group value flow data. The execution unit can be configured to: perform a value risk prediction information generation operation on the recommendation feature information set.

[0113] It can be understood that the units described in the model training device 500 correspond to the respective steps in the methods described with reference to Figure 2 . Therefore, the operations, features, and beneficial effects described above for the methods also apply to the model training device 500 and the units contained therein, which will not be described here.

[0114] Further reference Figure 6 , as an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of information generation devices, which correspond to the method embodiments shown in Figure 4 , and the information generation devices can be applied in various electronic devices.

[0115] As Figure 6As shown, an information generation apparatus 600 includes a second acquisition unit 601, an extraction unit 602, and a generation unit 603. The second acquisition unit 601 is configured to acquire customer value flow data for a target customer group. The extraction unit 602 is configured to extract customer characteristic description information and customer characteristic information for the customer value flow data. The generation unit 603 is configured to input the customer characteristic description information and the customer characteristic information into a pre-trained customer recommendation characteristic association information generation model to generate customer recommendation characteristic association information, where the customer recommendation characteristic association information generation model is generated based on a model training method.

[0116] In some optional implementations of some embodiments, the apparatus 500 further includes a screening unit, a determination unit, and an operation execution unit (not shown in the figure). The screening unit can be configured to screen a set of recommended characteristics from a set of characteristics corresponding to the customer characteristic information according to the customer recommendation characteristic association information. The determination unit can be configured to determine a set of recommended characteristic information corresponding to the set of recommended characteristics according to the customer value flow data. The operation execution unit can be configured to perform a value risk prediction information generation operation for the set of recommended characteristic information.

[0117] It can be understood that the units described in the information generation apparatus 600 correspond to the respective steps in the method described above. Therefore, the operations, features, and beneficial effects described above for the method also apply to the information generation apparatus 600 and the units included therein, which will not be described here again. Figure 4

[0118] The following refers to Figure 7 , which shows a structural schematic diagram of an electronic device (e.g., the electronic device 101 in Figure 1 ) 700 suitable for implementing some embodiments of the present disclosure. Figure 7 The electronic device shown is merely an example and should not impose any limitation on the functions and use range of the embodiments of the present disclosure.

[0119] As shown in Figure 7 , the electronic device 700 can include a processing apparatus (e.g., a central processor, a graphics processor, etc.) 701, which can perform various appropriate actions and processes according to programs stored in a read-only memory 702 or loaded from a storage apparatus 708 into a random access memory 703. Various programs and data required for the operation of the electronic device 700 are also stored in the random access memory 703. The processing apparatus 701, the read-only memory 702, and the random access memory 703 are connected to each other through a bus 704. An input / output interface 705 is also connected to the bus 704.

[0120] ​Generally, the following devices can be connected to the input / output interface 705: input devices 706, including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, and so on; output devices 707, including, for example, a liquid crystal display (LCD), a speaker, a vibrator, and so on; storage devices 708, including, for example, a magnetic disk, a hard disk, and so on; and communication devices 709. The communication devices 709 can allow the electronic device 700 to exchange data with other devices via a wired or wireless connection. Figure 7 The electronic device 700 is shown with various devices, but it is understood that not all of the shown devices are required to be implemented or present. More or fewer devices can alternatively be implemented or present. Figure 7 Each block shown in the flowcharts can represent a device or multiple devices, as desired.

[0121] In particular, processes described above with reference to the flowcharts can be implemented as a computer software program according to some embodiments of the present disclosure. For example, some embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for executing the methods shown in the flowcharts. In some such embodiments, the computer program can be downloaded and installed from a network via the communication devices 709, or installed from the storage devices 708, or installed from the read-only memory 702. When the computer program is executed by the processing devices 701, the above-described functions defined in the methods of some embodiments of the present disclosure are performed.

[0122] Note that the computer-readable medium or media used to provide the computer program sequence to the computer system can be embedded in a computer program product, which comprises all the respective features, which are provided with the computer program sequence, and which are enumerated above. It is understood that the computer-readable medium or media described herein are included in the computer program product, or are a component of the computer program product. In some embodiments of the disclosure, the computer-readable storage medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a computer-readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In some embodiments of the disclosure, a computer-readable storage medium can be any tangible medium that contains, or stores a program for use by or in connection with an instruction execution system, apparatus, or device. In some embodiments of the disclosure, a computer-readable signal medium can include a computer-readable storage medium in baseband or propagated as a carrier wave in a propagated data signal, which contains a computer-readable program code. Such a propagated signal can take a wide variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium that is not a computer-readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire line, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0123] In some embodiments, the client, server, or both can communicate using any current known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and can be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include local area networks ("LAN"), wide area networks ("WAN"), the Internet, and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any current known or future developed networks.

[0124] The computer readable medium can be included in the electronic device; or can exist separately from the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: obtain a training data set, wherein the training data comprises: historical customer feature description information, historical customer feature information, and a historical customer recommended feature set; for target training data in the training data set, perform the following training steps: perform feature grouping on a feature set corresponding to the target historical customer feature information to generate at least one feature group, wherein the target training data comprises: target historical customer feature information, target historical customer feature description information, and a target historical customer recommended feature set; determine at least one feature information group corresponding to the at least one feature group according to the target historical customer feature information and the target historical customer feature description information; input the at least one feature information group into an initial feature cross combination model to generate feature cross combination information, wherein the initial customer recommended feature association information generation model comprises: the initial feature cross combination model; determine whether the initial customer recommended feature association information generation model is trained according to the feature cross combination information and the target historical customer recommended feature set; in response to determining that the training is completed, determine the initial customer recommended feature association information generation model as a customer recommended feature association information generation model. Obtain customer value flow data for a target customer; extract customer feature description information and customer feature information for the customer value flow data; input the customer feature description information and the customer feature information into the pre-trained customer recommended feature association information generation model to generate customer recommended feature association information, wherein the customer recommended feature association information generation model is generated based on a model training method.

[0125] Computer program code for carrying out operations of some embodiments of the disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0126] The flow and block diagrams in the drawings represent possible architectural, functional, and operational architectures of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block can represent a module, a segment, or a portion of code that comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustrations, and combinations thereof, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or combinations of hardware and software.

[0127] The units described in some embodiments of the present disclosure can be implemented by means of software, or can be implemented by hardware. The described units can also be arranged in a processor, for example, can be described as: a processor comprising a first acquisition unit and a training unit. Among them, the names of these units do not constitute a limitation to the units themselves in some cases, for example, the first acquisition unit can also be described as "acquiring training data set".

[0128] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, non-limiting example types of hardware logic components that can be used include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SOCs), complex programmable logic devices (CPLDs), etc.

[0129] Some embodiments of the present disclosure also provide a computer program product comprising a computer program which, when executed by a processor, implements any of the above-mentioned model training methods or information generation methods.

[0130] The above description is merely some preferred embodiments of the present disclosure and a description of the principles of the technology used. Those skilled in the art should understand that the scope of the application involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or equivalent features without departing from the above inventive concept. For example, the above features can be replaced with technical features disclosed in the embodiments of the present disclosure (but not limited to) having similar functions to form technical solutions.

Claims

1. A model training method, comprising: Obtain the training dataset, which includes: historical customer group feature description information, historical customer group feature information, and historical customer group recommendation feature set; For the target training data in the training dataset, perform the following training steps: The feature set corresponding to the target historical customer group feature information is grouped to generate at least one feature group. The target training data includes: target historical customer group feature information, target historical customer group feature description information, and target historical customer group recommendation feature set. Based on the target historical customer group characteristic information and the target historical customer group characteristic description information, determine at least one characteristic information group corresponding to at least one characteristic group; At least one set of feature information is input into the initial feature cross-combination model to generate feature cross-combination information, wherein the initial customer group recommendation feature association information generation model includes: the initial feature cross-combination model; Based on the feature cross-combination information and the target historical customer group recommendation feature set, determine whether the initial customer group recommendation feature association information generation model has been trained successfully. In response to the confirmation that training is complete, the initial customer group recommendation feature association information generation model is determined as the customer group recommendation feature association information generation model.

2. The method according to claim 1, wherein, The step of grouping the feature set corresponding to the target historical customer group feature information to generate at least one feature group includes: The feature set corresponding to the target historical customer group feature information is grouped into feature groups of the same category to generate at least one feature group.

3. The method according to claim 1, wherein, The initial customer group recommendation feature association information generation model further includes: an initial feature linear combination model; and The step of determining whether the initial customer group recommendation feature association information generation model has been trained successfully based on feature cross-combination information and the target historical customer group recommendation feature set includes: The target historical customer group feature information and target historical customer group feature description information are input into the initial feature linear combination model to generate feature linear combination information; Based on the linear combination information of the features and the cross combination information of the features, initial customer group recommendation feature association information is generated; Based on the initial customer group recommendation feature association information and the target historical customer group recommendation feature set, determine whether the initial customer group recommendation feature generation model has been trained successfully.

4. The method according to claim 1, wherein, The method further includes: In response to the determination that training is not yet complete, the model parameters of the initial customer group recommendation feature generation model are updated to obtain the updated customer group recommendation feature generation model. Re-extract training data from the training dataset after removing the target training data, and use it as candidate training data; The updated customer group recommendation feature generation model is used as the initial customer group recommendation feature generation model, and the candidate training data is used as the target training data. The training steps are then executed again.

5. The method according to claim 1, wherein, The initial feature cross-combination model is a model that extracts feature association information between two features in different feature groups.

6. An information generation method, comprising: Obtain customer value flow data for the target customer group; Extract customer characteristic description information and customer characteristic information from the customer group value flow data; The customer group feature description information and the customer group feature information are input into a pre-trained customer group recommendation feature association information generation model to generate customer group recommendation feature association information, wherein the customer group recommendation feature association information generation model is generated based on the method described in any one of claims 1-5.

7. The method according to claim 6, wherein, The method further includes: Based on the customer group recommendation feature association information, a recommendation feature set is selected from the feature set corresponding to the customer group feature information; Based on the customer value flow data, determine the set of recommendation feature information corresponding to the recommendation feature set; Perform a value risk prediction information generation operation for the recommended feature information set.

8. A model training device, comprising: The first acquisition unit is configured to acquire a training dataset, wherein the training data includes: historical customer group feature description information, historical customer group feature information, and historical customer group recommendation feature set; The training unit is configured to perform the following training steps on the target training data in the training dataset: grouping the feature set corresponding to the target historical customer group feature information to generate at least one feature group, wherein the target training data includes: target historical customer group feature information, target historical customer group feature description information, and target historical customer group recommendation feature set; determining at least one feature information group corresponding to at least one feature group based on the target historical customer group feature information and the target historical customer group feature description information; inputting at least one feature information group into an initial feature cross-combination model to generate feature cross-combination information, wherein the initial customer group recommendation feature association information generation model includes: an initial feature cross-combination model; determining whether the initial customer group recommendation feature association information generation model has been trained successfully based on the feature cross-combination information and the target historical customer group recommendation feature set; and, in response to determining that training is complete, identifying the initial customer group recommendation feature association information generation model as the customer group recommendation feature association information generation model.

9. An information generation device, comprising: The second acquisition unit is configured to acquire customer value flow data for the target customer group; The extraction unit is configured to extract customer characteristic description information and customer characteristic information for the customer group value flow data; The generation unit is configured to input the customer group feature description information and the customer group feature information into a pre-trained customer group recommendation feature association information generation model to generate customer group recommendation feature association information, wherein the customer group recommendation feature association information generation model is generated based on the method described in any one of claims 1-5.

10. An electronic device, comprising: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

11. A computer-readable medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-7.

12. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-7.

Images