Target object determination method and device, equipment, medium and program product

Abstract

The invention provides a target object determination method which can be applied to the fields of big data, artificial intelligence and financial science and technology. The method comprises the following steps: respectively inputting to-be-mined data of each object in an initial object group into a prediction model, and outputting a predicted value of each object; determining a mining object group from the initial object groups based on the predicted value of each object and the importance degree of the to-be-mined data in the mining task; performing grade division on the mining object groups based on a preset grade division index to obtain mining object groups of different grades; pushing contents to the mining object groups of different grades based on pushing strategies matched with the mining object groups of different grades so as to obtain feedback conditions of the mining object groups of different grades on the contents; and determining a target object from the mining object group based on the feedback condition. The invention further provides a target object determination device and equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present disclosure relates to the fields of big data, artificial intelligence and financial technology, and specifically to a target object determination method, device, equipment, medium and program product. Background Art

[0002] With the rapid development of Internet technology, application promotion through content is becoming more and more important. However, how to determine the target audience that needs the content during content promotion has become a difficult problem.

[0003] In the process of implementing this embodiment, it was found that as the content becomes increasingly rich, using a single rule to determine the target object, or using a single target model to determine the target object, this object mining method cannot meet the promotion needs, resulting in the mined target object not matching the promoted content and low accuracy of target object mining. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a target object determination method, apparatus, device, medium and program product.

[0005] According to a first aspect of the present disclosure, a method for determining a target object is provided, comprising: inputting the data to be mined of each object in an initial object group into a prediction model respectively, and outputting a prediction value of each object, wherein the prediction model is determined based on the data type to which the data to be mined belongs; determining a mining object group from the initial object group based on the prediction value of each object and the importance of the data to be mined in the mining task; grading the mining object group based on a preset grading index to obtain mining object groups of different levels; pushing content to mining object groups of different levels based on a push strategy matching mining object groups of different levels, so as to obtain feedback from mining object groups of different levels on the content; and determining a target object from the mining object group based on the feedback.

[0006] According to an embodiment of the present disclosure, the data to be mined includes at least basic attribute data, behavioral data and social media data, and the data types include at least discrete type, nonlinear type and sequence type with time dependence; the target object determination method also includes: based on the discrete type, nonlinear type and sequence type, the basic attribute data, behavioral data and social media data are classified to obtain classification results; the classification results are matched with multiple pre-trained models to obtain matching results, wherein the functions of the multiple pre-trained models are different and the multiple pre-trained models are obtained by joint training; and the prediction model is determined according to the matching results.

[0007] According to an embodiment of the present disclosure, multiple pre-trained models include at least a first model, a second model, and a third model; wherein the joint training method includes: classifying the sample mining data of each sample object in the sample object group to obtain discrete sample data, nonlinear sample data, and sequence sample data corresponding to each sample object; for each sample object: inputting the discrete sample data into the first prediction model to output the sample first prediction value; inputting the nonlinear sample data into the second prediction model to output the sample second prediction value; inputting the sequence sample data into the third prediction model to output the sample third prediction value; weightedly summing the sample first prediction value, the sample second prediction value, and the sample third prediction value with the importance of the discrete sample data, the nonlinear sample data, and the sequence sample data in the mining task to obtain a sample evaluation value; determining an evaluation error based on the sample evaluation value corresponding to each sample object and the label matching each sample object; and adjusting the model parameters of the first prediction model, the second prediction model, and the third prediction model based on the evaluation error to obtain multiple pre-trained models.

[0008] According to an embodiment of the present disclosure, a target object is determined from a mining object group based on feedback, including: determining an object behavior indicator based on the feedback, wherein the object behavior indicator is used to indicate the acceptance of the mining object group to the content; adjusting the importance based on the object behavior indicator to obtain an adjusted importance; and determining the target object from the mining object group based on the adjusted importance and a predicted value.

[0009] According to an embodiment of the present disclosure, the importance is adjusted based on the object behavior indicator to obtain the adjusted importance, including: inputting the object behavior indicator into a reinforcement learning model, and outputting an adjustment action for the importance, wherein the reinforcement learning model is pre-trained with sample object behavior indicators as input and sample adjustment actions for sample importance as output; and adjusting the importance based on the adjustment action to obtain the adjusted importance.

[0010] According to an embodiment of the present disclosure, data types, prediction values, and importance levels have a mapping relationship; based on the prediction value of each object and the importance of the data to be mined in the mining task, a mining object group is determined from an initial object group, including: for each object: respectively multiplying the prediction value and importance level corresponding to the discrete type, nonlinear type, and sequential type to obtain a sub-evaluation value corresponding to the discrete type, nonlinear type, and sequential type; summing the sub-evaluation values ​​corresponding to the discrete type, nonlinear type, and sequential type to obtain an evaluation value; and based on the evaluation value corresponding to each object, determining a mining object group from the initial object group.

[0011] According to an embodiment of the present disclosure, the preset grade division index includes an evaluation value; based on the preset grade division index, the mining object group is graded to obtain mining object groups of different levels, including: dividing the evaluation values ​​corresponding to each object in the mining object group into different levels; and determining mining object groups of different levels based on the evaluation values ​​of different levels.

[0012] The second aspect of the present disclosure provides a target object determination device, characterized in that the device includes: an extraction module, which is used to input the to-be-mined data of each object in the initial object group into a prediction model respectively, and output a prediction value of each object, wherein the prediction model is determined based on the data type to which the to-be-mined data belongs; a first determination module, which is used to determine the mining object group from the initial object group based on the prediction value of each object and the importance of the to-be-mined data in the mining task; a classification module, which is used to classify the mining object group based on a preset classification index to obtain mining object groups of different levels; a push module, which is used to push content to mining object groups of different levels based on a push strategy matching the mining object groups of different levels, so as to obtain feedback from the mining object groups of different levels on the content; and a second determination module, which is used to determine the target object from the mining object group based on the feedback.

[0013] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0014] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0015] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor.

[0016] According to an embodiment of the present disclosure, a prediction model is determined according to the data type of the data to be mined of each object in the initial object group, and then the initial mining group is determined based on the prediction value and the importance of the data to be mined in the mining task, and the initial mining group is graded, and then the push strategy is matched to obtain the feedback of these groups, and the target object is further determined from the mining object group based on the feedback. Since the prediction model is determined according to the data type, the characteristics of the model can be used to improve the accuracy of the prediction, thereby improving the accuracy of the initial mining group determined for the first time. In addition, since the push strategy is matched for mining object groups of different levels, useful feedback information can be obtained, and then the target object is determined again from the initial mining group based on the feedback information, further improving the accuracy of target object mining. The present disclosure uses a layered progressive approach to jointly solve the problem of low accuracy in target object mining through two minings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0018] Figure 1 The application scenario diagram of the target object determination method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown;

[0019] Figure 2 A flowchart of a target object determination method according to an embodiment of the present disclosure is schematically shown;

[0020] Figure 3 A schematic diagram schematically shows a method of determining a mining object group from an initial object group according to an embodiment of the present disclosure;

[0021] Figure 4 A schematic diagram of determining a target object from a mining object group according to an embodiment of the present disclosure is schematically shown;

[0022] Figure 5 A schematic diagram shows a structural block diagram of a target object determination device according to an embodiment of the present disclosure; and

[0023] Figure 6 A block diagram of an electronic device suitable for implementing a target object determination method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0024] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0025] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0026] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0027] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0028] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0029] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0030] In the process of implementing this embodiment, it was found that as the content becomes increasingly rich, using a single rule to determine the target object, or using a single target model to determine the target object, this object mining method cannot meet the promotion needs, resulting in the mined target object not matching the promoted content and low accuracy of target object mining.

[0031] An embodiment of the present disclosure provides a method for determining a target object, comprising: inputting the data to be mined of each object in an initial object group into a prediction model respectively, and outputting a prediction value of each object, wherein the prediction model is determined based on the data type to which the data to be mined belongs; determining a mining object group from the initial object group based on the prediction value of each object and the importance of the data to be mined in the mining task; classifying the mining object group based on a preset grade classification index to obtain mining object groups of different grades; pushing content to mining object groups of different grades based on a push strategy matching mining object groups of different grades, so as to obtain feedback from mining object groups of different grades on the content; and determining a target object from the mining object group based on the feedback.

[0032] Figure 1 The application scenario diagram of the target object determination method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0033] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0034] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0035] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0036] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0037] It should be noted that the target object determination method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the target object determination device provided in the embodiment of the present disclosure can generally be set in the server 105. The target object determination method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the target object determination device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0038] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0039] The following will be based on Figure 1 The scene described by Figure 2~Figure 4 The target object determination method of the disclosed embodiment is described in detail.

[0040] Figure 2 The flowchart of the target object determination method according to the embodiment of the present disclosure is schematically shown.

[0041] like Figure 2 As shown, the target object determination method of this embodiment includes operations S210 to S250.

[0042] In operation S210, the data to be mined of each object in the initial object group is respectively input into the prediction model, and the prediction value of each object is output.

[0043] In operation S220 , a mining object group is determined from the initial object group based on the predicted value of each object and the importance of the to-be-mined data in the mining task.

[0044] In operation S230, the mining object groups are graded based on a preset grade division index to obtain mining object groups of different grades.

[0045] In operation S240, content is pushed to mining object groups of different levels based on push strategies matching mining object groups of different levels, so as to obtain feedback from mining object groups of different levels on the content.

[0046] In operation S250 , a target object is determined from the mining object group based on the feedback situation.

[0047] According to an embodiment of the present disclosure, the initial object group may include multiple objects. The initial object group may be screened from a large number of objects according to the expert's specific requirements for the specific content to be pushed. The data to be mined may include data associated with the object in multiple databases with access rights.

[0048] According to an embodiment of the present disclosure, the prediction model is determined based on the data type to which the data to be mined belongs. For example, each type of data may correspond to a prediction model, and different prediction models may be used for different data types.

[0049] According to an embodiment of the present disclosure, the prediction value may include prediction values ​​corresponding to multiple data types. The importance of the to-be-mined data of different data types in the mining task is different. The importance of the to-be-mined data in the mining task can be determined according to the influence of the type of the to-be-mined data on the content to be pushed.

[0050] According to an embodiment of the present disclosure, the mining object group may include multiple mining objects. Objects corresponding to predicted values ​​whose importances meet a threshold may be determined as mining objects.

[0051] According to an embodiment of the present disclosure, the preset level division index can be determined according to the influence of the to-be-mined data of each mining object in the mining object group on the content to be pushed. The push strategy can be predetermined according to the level of the mining object. Since the push strategy of each level is different, the form of pushing content to mining object groups of different levels is different.

[0052] According to the embodiments of the present disclosure, the content may include but is not limited to text, pictures, videos, audio, etc. The feedback may include but is not limited to the clicks, views, and evaluations of the mining object on the content. The click rate and the viewing rate may be counted based on the clicks and views, and the inclination of the object to accept the content may be determined based on the evaluation. Based on the click rate, the viewing rate, the inclination, etc., the target object is determined from the mining object group. The target object may include multiple objects.

[0053] According to an embodiment of the present disclosure, a prediction model is determined according to the data type of the data to be mined of each object in the initial object group, and then the initial mining group is determined based on the prediction value and the importance of the data to be mined in the mining task, and the initial mining group is graded, and then the push strategy is matched to obtain the feedback of these groups, and the target object is further determined from the mining object group based on the feedback. Since the prediction model is determined according to the data type, the characteristics of the model can be used to improve the accuracy of the prediction, thereby improving the accuracy of the initial mining group determined for the first time. In addition, since the push strategy is matched for mining object groups of different levels, useful feedback information can be obtained, and then the target object is determined again from the initial mining group based on the feedback information, further improving the accuracy of target object mining. The present disclosure uses a layered progressive approach to jointly solve the problem of low accuracy in target object mining through two minings.

[0054] According to an embodiment of the present disclosure, the target object determination method may include the following steps: Figure 2 In addition to the operations S210 to S250 shown, the operation may also include: obtaining the to-be-mined data of each object in the initial object group from the target database set. The target database set may include multiple databases.

[0055] In an embodiment of the present disclosure, before obtaining the data to be mined, the consent or authorization of the initial subject group may be obtained. For example, before obtaining the data to be mined, a request to obtain the data to be mined may be issued to the initial subject group. When the initial subject group agrees or authorizes the data to be mined, the data to be mined is obtained.

[0056] For example, with respect to the content of online courses, by obtaining the permission of each object in the initial object group and the target database access and acquisition rights, crawlers can be used to obtain basic attribute data such as the occupations of these objects, data related to their educational background, and comment data on online courses from these databases.

[0057] According to an embodiment of the present disclosure, the data to be mined may include at least basic attribute data, behavior data, and social media data, and the data types may include at least discrete, nonlinear, and sequence types with time dependency.

[0058] In this embodiment, the target object determination method may include the following steps: Figure 2 In addition to the operations S210 to S250 shown, the following operations may also be included: classifying basic attribute data, behavior data, and social media data based on discrete, nonlinear, and sequential types to obtain classification results; matching the classification results with multiple pre-trained models to obtain matching results; and determining a prediction model based on the matching results.

[0059] According to the embodiments of the present disclosure, basic attribute data may include, but are not limited to, the occupation, asset size, income, etc. of the object. Behavioral data may include, but are not limited to browsing behavior, financial-related behavior, transaction behavior, etc. Social media data may include, but are not limited to, the interactive data of the object in social media, etc. Discrete data may include, but are not limited to, occupation, etc. Non-linear data may include, but are not limited to, income, asset size, etc. Sequential data may include, but are not limited to, browsing behavior, financial-related behavior, transaction behavior, interactive data, etc. It should be noted that all of the above data are obtained with the permission of the user, comply with relevant laws, regulations and standards, and take necessary confidentiality measures.

[0060] According to the embodiments of the present disclosure, the functions of the multiple pre-trained models are different, and the multiple pre-trained models are obtained by joint training. The sample features can be divided into different subsets according to the characteristics of each model, and input into the corresponding model, and the model performance can be evaluated using methods such as cross-validation, and the joint training can obtain multiple pre-trained models with different functions.

[0061] For example, decision trees are characterized by being able to process data with clear classification boundaries or threshold features, being able to intuitively display the decision-making process, and being easy to understand and explain. Therefore, they can be used to process discrete data. Random forests are characterized by being able to process high-dimensional data, being able to capture complex nonlinear relationships, and being robust to noise and overfitting. Therefore, they can be used to process nonlinear data. Recurrent neural networks are good at processing sequence data and can capture long-term dependencies in time series. Therefore, they can be used to process sequence data with time dependencies.

[0062] According to the embodiments of the present disclosure, by pre-training prediction models of different functional types, the ability of the model to process different types of data can be utilized to improve the accuracy of prediction, thereby improving the accuracy of the initial determination of the mining object.

[0063] According to an embodiment of the present disclosure, the multiple pre-trained models include at least a first prediction model, a second prediction model, and a third prediction model.

[0064] The joint training method may include the following operations: classifying the sample mining data of each sample object in the sample object group to obtain discrete sample data, nonlinear sample data and sequence sample data corresponding to each sample object. For each sample object: inputting the discrete sample data into the first prediction model to output the first sample prediction value; inputting the nonlinear sample data into the second prediction model to output the second sample prediction value; inputting the sequence sample data into the third prediction model to output the third sample prediction value; weighted summing the feature values ​​corresponding to the first sample prediction value, the second sample prediction value and the third sample prediction value with the weight values ​​corresponding to the first prediction model, the second prediction model and the third prediction model to obtain the sample evaluation value. Based on the sample evaluation values ​​corresponding to each sample object and the labels matching each sample object, determining the evaluation error; and adjusting the model parameters of the first prediction model, the second prediction model and the third prediction model based on the evaluation error to obtain multiple pre-trained models.

[0065] According to an embodiment of the present disclosure, the first prediction model, the second prediction model, and the third prediction model each have different capabilities for processing different types of data. The sample object group may be a fixed object group for content.

[0066] For example, for the content of an online course, the sample object group may include all objects that use the online course. The sample mining data may be the registration data of the sample objects that register when using the online course, the behavior record data of the sample objects using the online course, and the comment data of the sample objects on the online course. The registration data may be input into the first prediction model to obtain a first prediction value. The behavior record data may be input into the second prediction model to obtain a second prediction value. The comment data may be input into the third prediction model to obtain a third prediction value. The mining task may be the task of mining objects that have the potential to use the online course. The importance of the sample mining data in the mining task may be obtained based on expert experience.

[0067] According to the embodiments of the present disclosure, the sample evaluation values ​​corresponding to the sample objects can be compared with the evaluation threshold, and the sample objects greater than the evaluation threshold can be determined as prediction mining objects; the evaluation error can be determined according to the ratio of the prediction mining objects to the sample objects. The model parameters of the first prediction model, the second prediction model, and the third prediction model can be adjusted with the goal of minimizing the evaluation error to obtain multiple pre-trained models.

[0068] According to the embodiments of the present disclosure, by jointly training multiple pre-trained models, it is possible to consider the mutual influence between the features of different data types while reasonably utilizing the inherent characteristics of each model, thereby preventing overfitting of the model.

[0069] According to an embodiment of the present disclosure, the data type, the predicted value, and the importance have a mapping relationship. Figure 2 The operation S220 shown, based on the predicted value of each object and the importance of the data to be mined in the mining task, determines the mining object group from the initial object group, and may include the following operations: for each object: respectively multiplying the predicted value and the importance corresponding to the discrete type, nonlinear type and sequential type to obtain the sub-evaluation value corresponding to the discrete type, nonlinear type and sequential type; summing the sub-evaluation values ​​corresponding to the discrete type, nonlinear type and sequential type to obtain the evaluation value; and determining the mining object group from the initial object group based on the evaluation value corresponding to each object.

[0070] Figure 3 The diagram schematically shows a schematic diagram of determining a mining object group from an initial object group according to an embodiment of the present disclosure.

[0071] like Figure 3 As shown, the initial object group may include N objects, where N is an integer greater than 2. The data to be mined may include data 1 to be mined of object 1, and so on, up to data N to be mined of object N. The data to be mined of each object may be classified into discrete data, nonlinear data, and sequence data, such as data 1 to be mined is divided into discrete data 1, nonlinear data 1, and sequence data 1, and data N to be mined is divided into discrete data N, nonlinear data N, and sequence data N. Each type of data of each object may be input into the first prediction model M1, the second prediction model M2, and the third prediction model M3, respectively, and the corresponding prediction values ​​are output respectively, such as discrete data 1 corresponds to the first prediction value 1, nonlinear data 1 corresponds to the second prediction value 1, and sequence data 1 corresponds to the third prediction value 1. Discrete data N corresponds to the first prediction value N, nonlinear data N corresponds to the second prediction value N, and sequence data N corresponds to the third prediction value N. Each object can be evaluated by weighted summing the importance w corresponding to the data type (e.g., the importance w1 for discrete data, w2 for nonlinear data, and w3 for sequence data) and the corresponding predicted value to obtain an evaluation value, i.e., evaluation value 1, evaluation value 2, ... evaluation value N. The evaluation value can be compared with the evaluation threshold to select a target evaluation value greater than the evaluation threshold. The object corresponding to the target evaluation value is determined as the mining object, and the mining objects form a mining object group.

[0072] According to the embodiments of the present disclosure, the predicted values ​​of different types of data and the importance of different types of data are combined to determine the mining object. Based on the importance of the data, the mining object can be accurately determined to complete the initial mining task.

[0073] In the process of implementing this embodiment, it is found that when a single rule is used to determine the target object, or a single target model is used to determine the target object, the mined target object does not match the promoted content, resulting in a poor target object experience and difficulty in obtaining real feedback information.

[0074] According to an embodiment of the present disclosure, the preset grading index may include an evaluation value.

[0075] For the above Figure 2 The operation S230 shown, based on a preset grade classification index, grades the mining object group to obtain mining object groups of different grades, which may include the operations of: dividing the evaluation values ​​corresponding to each object in the mining object group into different grades; and determining mining object groups of different grades based on the evaluation values ​​of different grades.

[0076] According to the embodiments of the present disclosure, the evaluation values ​​corresponding to the objects in the mining object group can be divided according to the threshold of the division level to obtain evaluation values ​​of different levels. Since the evaluation values ​​have a mapping relationship with the objects in the mining object group, the mining object group can be divided into different levels according to the evaluation values ​​of different levels.

[0077] According to the embodiments of the present disclosure, objects are graded based on evaluation values, which is conducive to matching different strategies for different grades, enhancing the target object's experience, and thus improving the reference value of feedback.

[0078] According to an embodiment of the present disclosure, a preset level division index may also be determined according to a mining task. The mining task may be to mine push objects for push content. Based on the preset level division index, level division is performed through a level prediction model.

[0079] For example, for the mining task of potential objects of private banks, the preset level classification indicators can be determined according to asset scale, asset growth rate, transaction frequency and transaction assets, risk tolerance, transaction target and demand matching, etc. Asset scale, such as the total assets or current asset scale of the object, can be used to indicate the asset level of the object. Asset growth rate, such as the growth rate of the object's assets in the past period of time, can be used to indicate the asset growth trend of the object. Transaction frequency and transaction assets can be used to indicate the transaction activities of the object, transaction activity and asset liquidity demand, etc. Risk tolerance can be used to indicate the risk preference and acceptance of financial products of the object, and determine the transaction products and strategies suitable for the object. The transaction target and demand matching can be used to indicate the matching degree between the transaction target of the object and the services provided by the private bank. Different levels from high to low can include the first level, the second level and the third level. The first level can be, for example, an object with large asset scale, high asset growth rate, frequent transactions and large transaction assets, moderate risk tolerance, and a high matching degree between the transaction target and the private banking service. The second level can be, for example, an object with medium asset scale, stable asset growth rate, moderate transaction activity, acceptable risk tolerance, and a certain matching degree between the transaction target and the private banking service. The third level may include, for example, those with smaller assets, lower asset growth rates, infrequent trading activities, lower or unclear risk tolerance, and those whose trading goals have a low match with private banking services.

[0080] Differentiated push strategies can be formulated according to different levels of mining objects. For example, for the first-level objects, push strategies such as one-to-one exclusive services can be customized. For the second-level objects, push strategies such as regular discounts and discount reminders can be customized. For the third-level objects, push strategies such as transaction consulting services and asset allocation services can be customized.

[0081] According to the embodiments of the present disclosure, customer experience can be improved by dividing corresponding object levels.

[0082] According to the embodiments of the present disclosure, for the above Figure 2 The operation S250 shown, based on the feedback, determines the target object from the mining object group, and may include the operations of: determining the object behavior indicator according to the feedback; adjusting the importance based on the object behavior indicator to obtain the adjusted importance; and determining the target object from the mining object group based on the adjusted importance and the predicted value.

[0083] According to an embodiment of the present disclosure, the object behavior index can be used to indicate the acceptance of the mining object group to the content. The acceptance of the mining object group to the content can be determined based on the click rate, viewing rate, and tendency, etc. If it is determined that the acceptance does not meet the preset degree, the importance can be adjusted.

[0084] According to the embodiments of the present disclosure, in order to better improve the accuracy of mining, after pushing content to different levels of mining object groups, it is necessary to obtain feedback in real time. It is possible to pre-bury points in the channels of each pushed content in order to obtain active or passive feedback information from the mining object. Feedback information may include, but is not limited to, the degree of acceptance of the form of pushed content, the time of pushed content, the degree of acceptance of the channel, etc. Feedback information may also include, but is not limited to, the evaluation, suggestions, and requirements of the pushed content by the mining object after receiving the pushed content. The obtained feedback can be integrated into the data warehouse for cleaning, sorting and normalization. Use association analysis, cluster analysis and other technologies to mine the correlation between object behavior and content. Use statistical methods such as regression models and classification models to determine object behavior indicators. Possible errors can be identified and processed through data verification and outlier detection. According to business logic and data characteristics, errors are corrected to ensure the accuracy of calculated object behavior indicators.

[0085] According to an embodiment of the present disclosure, an adjusted evaluation value can be obtained based on the weighted sum of the predicted value corresponding to each type of data of each object in the mining object group and the adjusted importance corresponding to the type of data, and the target object can be determined from the mining object group based on the adjusted evaluation value.

[0086] According to the embodiments of the present disclosure, the importance is adjusted based on the feedback, thereby optimizing the initial mining results, which can optimize the determination of the degree of influence on the mining task, further optimize the initial mining object, and improve the accuracy of the mining task.

[0087] According to an embodiment of the present disclosure, adjusting the importance based on the object behavior indicator to obtain the adjusted importance may include the following operations: inputting the object behavior indicator into a reinforcement learning model, and outputting an adjustment action for the importance, wherein the reinforcement learning model is pre-trained with sample object behavior indicators as input and sample adjustment actions for sample importance as output; and adjusting the importance based on the adjustment action to obtain the adjusted importance.

[0088] Figure 4 The schematic diagram of determining a target object from a mining object group according to an embodiment of the present disclosure is schematically shown.

[0089] like Figure 4As shown, for example, the mining object group may include P mining objects. According to the level division of the P mining objects, content can be pushed to mining objects of different levels, and then feedback can be obtained, such as feedback situation 1, feedback situation 2...feedback situation P, etc. The object behavior index can be determined based on these feedback situations, and then the object behavior index value can be calculated, and the object behavior index shown is input into the reinforcement learning model M4 to output the adjustment action. The degree of influence can be adjusted according to the adjustment action. Then, the adjusted evaluation value is obtained according to the weighted sum of the predicted value corresponding to each type of data of each mining object in the mining object group and the adjusted importance corresponding to the type of data. According to the evaluation threshold, the adjusted target evaluation value is screened from the adjusted evaluation value, and then the mining object corresponding to the adjusted target evaluation value is determined as the target object.

[0090] According to an embodiment of the present disclosure, the reinforcement learning model is pre-trained with sample object behavior indicators as input and sample adjustment actions for sample importance as output. For example, historical feedback data of the content can be collected. The historical feedback data is cleaned, integrated and preprocessed to obtain historical object behavior indicators. The historical object behavior indicators are represented by vectors and defined as the state representation of the reinforcement learning model. The adjustment of the importance of historical mining data in the mining task is defined as an action. Each action should have a clear adjustment method and expected effect so that it can be selected and evaluated in the reinforcement learning process. The reward function can be defined according to the goal of the mining object, such as maximizing the click-through rate, view rate, etc. The reward value should be dynamically adjusted according to the gap between the actual execution and the expected goal to encourage the model to explore a better strategy. The reinforcement learning algorithm can be selected according to the complexity of the problem and the characteristics of the data. Factors such as the training speed, generalization ability and decision accuracy of the model can be considered.

[0091] Historical feedback data can be used to pre-train the reinforcement learning model to obtain preliminary policy knowledge. In the actual operation process, new feedback data is continuously collected to perform online learning and optimization of the model. Regularization, early stopping and other technologies are introduced to prevent model overfitting and improve the generalization ability of the model.

[0092] According to the embodiments of the present disclosure, the importance is adjusted based on reinforcement learning, and reinforcement learning can be used to achieve dynamic adjustment, real-time feedback and automatic adjustment, and optimize the mining object.

[0093] Based on the above target object determination method, the present disclosure also provides a target object determination device. Figure 5 The device is described in detail.

[0094] Figure 5 The structural block diagram of the target object determination device according to an embodiment of the present disclosure is schematically shown.

[0095] like Figure 5 As shown, the target object determination device 500 of this embodiment includes an extraction module 510 , a first determination module 520 , a division module 530 , a push module 540 and a second determination module 550 .

[0096] The extraction module 510 is used to input the data to be mined of each object in the initial object group into the prediction model, and output the prediction value of each object, wherein the prediction model is determined based on the data type to which the data to be mined belongs. In one embodiment, the extraction module 510 can be used to perform the operation S210 described above, which will not be repeated here.

[0097] The first determination module 520 is used to determine the mining object group from the initial object group based on the predicted value of each object and the importance of the to-be-mined data in the mining task. In one embodiment, the first determination module 520 can be used to perform the operation S220 described above, which will not be repeated here.

[0098] The classification module 530 is used to classify the mining object groups based on the preset classification index to obtain mining object groups of different levels. In one embodiment, the classification module 530 can be used to perform the operation S230 described above, which will not be described in detail here.

[0099] The push module 540 is used to push content to mining object groups of different levels based on push strategies matching mining object groups of different levels, so as to obtain feedback from mining object groups of different levels on the content. In one embodiment, the push module 540 can be used to perform the operation S240 described above, which will not be repeated here.

[0100] The second determination module 550 is used to determine the target object from the mining object group based on the feedback situation. In one embodiment, the second determination module 550 can be used to perform the operation S250 described above, which will not be repeated here.

[0101] According to an embodiment of the present disclosure, the data to be mined includes at least basic attribute data, behavior data and social media data, and the data types include at least discrete, nonlinear and sequence types with time dependency. The target object determination device 500 may also include a classification module, a matching module and a third determination module.

[0102] The classification module is used to classify basic attribute data, behavior data and social media data based on discrete, nonlinear and sequential types to obtain classification results. The matching module is used to match the classification results with multiple pre-trained models to obtain matching results, wherein the functions of the multiple pre-trained models are different and the multiple pre-trained models are obtained by joint training. The third determination module is used to determine the prediction model based on the matching results.

[0103] According to an embodiment of the present disclosure, any multiple modules of the extraction module 510, the first determination module 520, the division module 530, the push module 540, and the second determination module 550 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the extraction module 510, the first determination module 520, the division module 530, the push module 540, and the second determination module 550 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the extraction module 510, the first determination module 520, the division module 530, the push module 540 and the second determination module 550 may be at least partially implemented as a computer program module, which may perform a corresponding function when executed.

[0104] Figure 6 A block diagram of an electronic device suitable for implementing a target object determination method according to an embodiment of the present disclosure is schematically shown.

[0105] like Figure 6 As shown, the electronic device 600 according to an embodiment of the present disclosure includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 to a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0106] In RAM 603, various programs and data required for the operation of electronic device 600 are stored. Processor 601, ROM 602 and RAM 603 are connected to each other via bus 604. Processor 601 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in one or more memories.

[0107] According to an embodiment of the present disclosure, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0108] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0109] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 602 and / or RAM 603 described above and / or one or more memories other than ROM 602 and RAM 603.

[0110] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the method provided by the embodiment of the present disclosure.

[0111] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 601. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0112] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0113] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0114] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0115] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0116] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0117] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for determining a target object, characterized in that: The method comprises: Inputting the data to be mined of each object in the initial object group into the prediction model respectively, and outputting the prediction value of each object, wherein the prediction model is determined based on the data type to which the data to be mined belongs; Determining a mining object group from the initial object group based on the predicted value of each of the objects and the importance of the to-be-mined data in the mining task; Based on a preset grade classification index, the mining object group is graded to obtain mining object groups of different grades; Pushing content to the mining object groups at different levels based on push strategies matching the mining object groups at different levels, so as to obtain feedback from the mining object groups at different levels on the content; and Based on the feedback, a target object is determined from the mining object group.

2. The method according to claim 1, characterized in that The data to be mined includes at least basic attribute data, behavior data and social media data, and the data types include at least discrete, nonlinear and sequential types with time dependency; The method further comprises: Based on the discrete type, the nonlinear type and the sequence type, the basic attribute data, the behavior data and the social media data are classified to obtain a classification result; Matching the classification result with a plurality of pre-trained models to obtain a matching result, wherein the plurality of pre-trained models have different functions and are obtained by joint training; and The prediction model is determined according to the matching result.

3. The method according to claim 2, characterized in that The plurality of pre-trained models include at least a first prediction model, a second prediction model, and a third prediction model; The joint training method includes: Classifying the sample mining data of each sample object in the sample object group to obtain discrete sample data, nonlinear sample data and sequence sample data corresponding to each sample object; For each of the sample objects: Inputting the discrete sample data into the first prediction model and outputting a first prediction value of the sample; Inputting the nonlinear sample data into the second prediction model and outputting a second prediction value of the sample; Input the sequence type sample data into the third prediction model and output the sample third prediction value; The first sample prediction value, the second sample prediction value and the third sample prediction value are weighted and summed with the importance of the discrete sample data, the nonlinear sample data and the sequence sample data in the mining task to obtain a sample evaluation value; Determining an evaluation error based on the sample evaluation values ​​corresponding to the sample objects and the labels matching the sample objects; and Based on the evaluation error, the model parameters of the first prediction model, the second prediction model and the third prediction model are adjusted to obtain a plurality of the pre-trained models.

4. The method according to any one of claims 1 to 3, characterized in that The step of determining a target object from the mining object group based on the feedback situation includes: Determining an object behavior indicator according to the feedback, wherein the object behavior indicator is used to indicate the acceptance degree of the mining object group to the content; Adjusting the importance based on the object behavior indicator to obtain an adjusted importance; and The target object is determined from the mining object group based on the adjusted importance and the predicted value.

5. The method according to claim 4, characterized in that The step of adjusting the importance based on the object behavior indicator to obtain the adjusted importance includes: Inputting the object behavior indicator into a reinforcement learning model and outputting an adjustment action for the importance level, wherein the reinforcement learning model is pre-trained with sample object behavior indicators as input and sample adjustment actions for sample importance levels as output; and Based on the adjustment action, the importance is adjusted to obtain the adjusted importance.

6. The method according to claim 2, characterized in that The data type, the predicted value, and the importance have a mapping relationship; The step of determining a mining object group from the initial object group based on the predicted value of each object and the importance of the to-be-mined data in the mining task comprises: For each of said objects: Respectively multiplying the predicted values ​​and the importance levels corresponding to the discrete type, the nonlinear type, and the sequential type, to obtain sub-evaluation values ​​corresponding to the discrete type, the nonlinear type, and the sequential type; Sum the sub-evaluation values ​​corresponding to the discrete type, the nonlinear type, and the sequential type to obtain an evaluation value; and The mining object group is determined from the initial object group based on the evaluation value corresponding to each of the objects.

7. The method according to claim 6, characterized in that The preset grading index includes the evaluation value; The step of classifying the mining object groups based on the preset class classification index to obtain mining object groups of different classes includes: Classifying the evaluation values ​​corresponding to the objects in the mining object group into different levels; and Based on the evaluation values ​​of the different levels, the mining object groups of the different levels are determined.

8. A target object determination device, characterized in that: The device comprises: An extraction module, used to input the data to be mined of each object in the initial object group into the prediction model, and output the prediction value of each object, wherein the prediction model is determined based on the data type to which the data to be mined belongs; A first determination module, configured to determine a mining object group from the initial object group based on the predicted value of each object and the importance of the to-be-mined data in the mining task; A classification module, used to classify the mining object group based on a preset classification index to obtain mining object groups of different levels; a push module, configured to push content to the mining object groups at different levels based on push strategies matching the mining object groups at different levels, so as to obtain feedback from the mining object groups at different levels on the content; and The second determination module is used to determine the target object from the mining object group based on the feedback situation.

9. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images