Information Generation Method and Device

Abstract

Embodiments of the present disclosure disclose an information generation method and apparatus, including: obtaining historical user data, preprocessing the user data and performing label annotation; automatically screening the preprocessed data according to preset screening conditions to obtain valid user data; inputting the labeled valid user data into an information generation model, training a pre-established information generation model with a preset optimization objective as the training objective, and outputting the optimal configuration information of the information generation model under the optimization objective; after the information generation model is trained, if an input of new user data in the current traffic is received, outputting the configuration information corresponding to the new user data. Through the established information generation model, it is realized to output the corresponding configuration information for different user data, thereby solving the problem in the related art that the optimal configuration information based on different user data in the RTA advertising placement service cannot be automatically output.

Description

Technical Field

[0001] The present disclosure relates to the technical field of data processing, and particularly to an information generation method and apparatus. Background Art

[0002] In related technologies, technologies such as RTA are usually used to find users to whom advertisements are to be delivered.

[0003] In the business process of optimizing advertisement delivery, it is usually impossible to automatically obtain the corresponding configuration information for user data in the current traffic. Summary of the Invention

[0004] The main purpose of the present disclosure is to provide an information generation method and apparatus.

[0005] To achieve the above object, according to the first aspect of the present disclosure, an information generation method is provided, including: obtaining historical user data, preprocessing the user data and performing label annotation; automatically screening the preprocessed data according to preset screening conditions to obtain valid user data; inputting the labeled valid user data into an information generation model, training the pre-established information generation model with a preset optimization target as the training target, and outputting the optimal configuration information of the information generation model under the optimization target; after the information generation model is trained, if an input of new user data in the current traffic is received, outputting the optimal configuration information corresponding to the new user data.

[0006] Optionally, the information generation model includes: the advertisement delivery service is pre-segmented into multiple service nodes, and the information generation model predicts the conversion rate of the advertisement delivery service for each node based on the user data; the information generation model estimates the size of user value based on the user data to obtain the size of user value; the information generation model also estimates the competing bid based on the size of user value.

[0007] Optionally, training the pre-established information generation model with a preset optimization target as the training target includes: presetting a first target for the conversion rate of each node in the service; when using the preset first target as the optimization target, determining the configuration information in the conversion rate prediction process.

[0008] Optionally, training the pre-established information generation model with a preset optimization target as the training target includes: presetting a second target for the service information used to evaluate the size of user value; when using the preset second target as the optimization target, determining the configuration information in the user value estimation process.

[0009] Optionally, training the pre-established information generation model with a preset optimization target includes: presetting a third target for the participation competition cost; when using the preset third target as the optimization target, simultaneously determining the configuration information of the conversion rate and the configuration information of the user value.

[0010] Optionally, when using the preset optimization target as the training target, the configuration information of the preset screening conditions is also output.

[0011] According to a second aspect of the present disclosure, an information generation device is provided. The device is used in an advertising placement service implemented based on RTA. The device includes:

[0012] A labeling unit configured to obtain historical user data, preprocess the user data, and perform label annotation;

[0013] A data screening unit configured to automatically screen the preprocessed data according to preset screening conditions to obtain valid user data;

[0014] A training unit configured to input the labeled valid user data into an information generation model, use a preset optimization target as the training target, train the pre-established information generation model, and output the configuration information of the information generation model under the optimization target;

[0015] An output unit configured to, after the information generation model is trained, if receiving the input of new user data in the current traffic, output the optimal configuration information corresponding to the new user data.

[0016] According to a third aspect of the present disclosure, a computer-readable storage medium is provided, storing computer instructions for causing a computer to execute the information generation method according to any implementation manner of the first aspect.

[0017] According to a fourth aspect of the present disclosure, an electronic device is provided, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to cause the at least one processor to execute the information generation method according to any implementation manner of the first aspect.

[0018] In the information generation method and apparatus according to the embodiments of the present disclosure, it includes: obtaining historical user data, preprocessing the user data and performing label annotation; automatically screening the preprocessed data according to preset screening conditions to obtain valid user data; inputting the labeled valid user data into an information generation model, training the pre-established information generation model with a preset optimization objective as the training target, and outputting the optimal configuration information of the information generation model under the optimization objective; after the information generation model is trained, if an input of new user data in the current traffic is received, outputting the configuration information corresponding to the new user data. Through the established information generation model, it is realized to output the corresponding configuration information for different user data, thereby solving the problem in the related art that it is impossible to automatically output the optimal configuration information based on different user data in the RTA advertising placement service. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present disclosure. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0020] Figure 1 is a flowchart of the information generation method according to the embodiments of the present disclosure;

[0021] Figure 2 is an application scenario diagram of the information generation method according to the embodiments of the present disclosure;

[0022] Figure 3 is a schematic diagram of an electronic device according to the embodiments of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In order to enable those skilled in the art to better understand the solutions of the present disclosure, the following will clearly and completely describe the technical solutions in the embodiments of the present disclosure with reference to the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only some embodiments of the present disclosure, rather than all embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present disclosure.

[0024] It should be noted that the terms "first", "second", etc. in the description, claims and above-mentioned drawings of the present disclosure are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so as to implement the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0025] It should be noted that, without conflict, the embodiments in the present disclosure and the features in the embodiments can be combined with each other. The present disclosure will be described in detail below with reference to the drawings and in combination with the embodiments.

[0026] According to an embodiment of the present disclosure, an information generation method is provided, as Figure 1 shown, the method includes the following steps 101 to 104:

[0027] Step 101: Obtain historical user data, preprocess the user data and perform label annotation.

[0028] In this embodiment, the historical user data can be used as a training sample, and the content representing the user characteristics in the sample is labeled, including the user's device information (for example, the model of the device, etc.), the user's attribute information (for example, the user's age, etc.), the user's historical behavior information, the information of the delivery channel, the variables such as the frequency, maximum, minimum, variance, etc. derived by processing. Before labeling the tags, the method can also preprocess the data, including missing value filling, one-hot encoding, mean, variance, etc. The implementation methods are as follows:

[0029] df_cat = data.describe(include=['object'])

[0030] df_num = data.describe(include=['number'])

[0031] cat_features_3 = ['county_id', 'app_classify_type_max_use', 'app_classify_type_sec_use',

[0032] 'app_name_max_use', 'app_name_sec_use']

[0033] df_dumm = alldata

[0034] for each in cat_features_3:

[0035] dummies = pd.get_dummies(df_dumm.loc[:, each], prefix = each)

[0036] df_dumm = pd.concat([df_dumm, dummies], axis = 1)

[0037] data = df_dumm.drop(cat_features_3, axis = 1)

[0038] Step 102: Automatically screen the pre - processed data according to preset screening conditions to obtain valid user data.

[0039] In this embodiment, screening the user data to determine valid user data includes: eliminating user data that does not meet the preset qualifications; and for preset nodes in the advertising placement business, eliminating user data that does not meet the conversion rules at the preset nodes.

[0040] In this embodiment, when traffic is obtained, user data can be screened. The screening method may include eliminating user data that does not match the preset qualification information, which may include the scope of business based on the product outline indicated in the advertisement to be placed. Based on this, data of target customer groups that do not operate, user data in the dynamic blacklist, registered user data, and user data in the overdue list shared by the Internet Finance Association can be excluded. In addition, for user data that does not meet the preset conversion rules at the preset nodes of the business to be placed, the rule may be N exposures but not M conversions, and M / N does not meet the preset value. Such user data is filtered to exclude user data with ineffective multiple exposures.

[0041] Step 103: Input the labeled valid user data into the information generation model, use the preset optimization objective as the training objective, train the pre - established information generation model, and output the optimal configuration information of the information generation model under the optimization objective.

[0042] In this embodiment, by training the pre - established information generation model, the trained model can differentially output the optimal configuration information.

[0043] Step 104: After the information generation model is trained, if an input of new user data in the current traffic is received, output the optimal configuration information corresponding to the new user data.

[0044] In this embodiment, after receiving the user data in the current traffic, the user data can be automatically screened to obtain the current valid user data, and the optimal configuration information can be visually output from the valid user data through the information generation model.

[0045] As an optional implementation manner of this embodiment, the information generation model includes: the advertising placement service is pre-segmented into multiple service nodes, and the information generation model predicts the conversion rate of the advertising placement service for each node based on the user data; the information generation model estimates the user value based on the user data to obtain the size of the user value; the information generation model also estimates the competing bid based on the size of the user value.

[0046] In this optional implementation manner, the information generation model can be used to predict the conversion rate of any user data at each service node, estimate the value of the user data, and the model can also determine the bidding price based on the user value.

[0047] Furthermore, the service of implementing advertising placement based on RTA can be segmented into multiple nodes, including but not limited to click node, registration node, first login node, service application node (for example, loan service application node), node where the applied service is implemented (for example, node where loan operation is performed after applying for the loan service), evaluation node (for example, credit limit evaluation node), usage node (for example, usage rate of the limit), charging node (for example, CPS, CPA, CPCL, CPC, or CPM, etc.) and other nodes.

[0048] When predicting the conversion rate, a conversion rate prediction model can be established and trained in advance, and this model can predict the conversion rate of each node for the user data. During training, multi-objective optimization can be performed on the prediction model to achieve the optimal prediction model for the conversion rate of the entire service link.

[0049] When predicting the user value, the user value can be obtained through the user value estimation model, and this value can be reflected in numerical form. The evaluation of this value can be based on the credit limit, total transaction amount, number of transactions, usage rate of the limit, etc. of the user data. The value estimation model can be obtained through training. During training, the relevant information of the user ID can be used as the dependent variable (label), and the credit limit, total transaction amount, number of transactions, usage rate of the limit, etc. of the user data can be used as the target variables to train the model. The dependent variable can include tags processed from device information and APP list, internal historical information flow placement conversion information, external third-party information, etc.

[0050] As an alternative implementation of this embodiment, training the pre-established information generation model with a preset optimization target includes: making a first target preset for the conversion rates of each node in the business; when using the preset first target as the optimization target, determining the optimal configuration information in the conversion rate prediction process.

[0051] In this alternative implementation, through this training process, the conversion rates of each node corresponding to user data with different characteristics can be determined. When setting the conversion rate (such as the conversion rate of the entire link) as the optimization target, the conversion rates corresponding to each business node are determined by adjusting the parameters of the model. When determining the optimal configuration information, the optimal conversion rate of the entire link can be adjusted and set, and then the conversion rates corresponding to each node under this target can be determined, and this conversion rate can be used as the optimal configuration information. The conversion rates of different nodes can belong to different competing strategies, as shown in Table 1 below:

[0052] Table 1

[0053]

[0054] Furthermore, after receiving new user data, the trained information generation model can output the conversion rates of each business node corresponding to the current user data and the corresponding competing strategies.

[0055] As an alternative implementation of this embodiment, training the pre-established information generation model with a preset optimization target includes: making a second target preset for the business information used to evaluate the value size; when using the preset second target as the optimization target, determining the optimal configuration information in the user value estimation process.

[0056] In this alternative implementation, through this training process, the user value sizes corresponding to user data with different characteristics can be determined. Different values can correspond to different bidding coefficients, and the competitive bidding price of the advertisement can be determined through this bidding coefficient. When setting the user value threshold as the optimization target, the parameters of the value estimation model in the training process are adjusted to determine the preset information of the users in the user data under different value thresholds, and this preset information can include the content in Table 2.

[0057] When determining the optimal configuration information, the optimal threshold can be adjusted and determined, and then the optimal configuration information can be determined. Different preset information can correspond to different competing strategies, as shown in Table 2 below:

[0058] Table 2

[0059]

[0060] It can be understood that after obtaining the configuration information of different optimization objectives, the corresponding policy content can be determined, including the strong parameter competition policy, the strong non-competition policy, or others.

[0061] Furthermore, after receiving new user data, the trained information generation model can output the preset information corresponding to the current user data, as well as the competition strategy that the preset information can correspond to.

[0062] As an optional implementation manner of this embodiment, training the pre-established information generation model with the preset optimization objective includes: presetting a third objective for the competition cost; when using the preset third objective as the optimization objective, simultaneously determining the configuration information of the conversion rate and the configuration information of the user value.

[0063] In this optional implementation manner, when using the competition cost as the optimization objective, the conversion rate prediction model and the value estimation model can be optimized simultaneously, and after setting the competition cost, the conversion rate of each node and the preset information of the user can be determined. When new user data is received, by setting different competition costs, the conversion rate of each node, the preset information of the user, and their corresponding policy content can be directly determined.

[0064] As an optional implementation manner of this embodiment, when using the preset optimization objective as the training objective, the configuration information of the preset screening condition is also output.

[0065] In this optional implementation manner, the configuration information of the preset screening condition can be the threshold allowing the user to correspond to the multiple head count. The multiple head count of the user can be the number of services the user conducts simultaneously. For example, the multiple head count is the number of borrowing channels the user uses simultaneously. For instance, when borrowing from A, B, and C simultaneously, the multiple head count is 3. In this optional implementation manner, the multiple head count can be estimated by establishing a multiple head count estimation model, and the estimation model is trained to determine the model that can correspond to the optimal multiple head count for different user feature data. After training to obtain this model, when new user data is received, the multiple head count corresponding to the new user data can be obtained.

[0066] Furthermore, the accuracy of the multiple head count estimation can be improved by combining the multiple head count estimation model with different third-party data (for example, the data details of users and multiple head counts provided by different third parties). Therefore, during the training process, the combination method of the multiple head data estimation model and different third parties can be adjusted (such as, multiple head data estimation model + A, or multiple head data + B) to determine the optimal estimation model, and finally, through training, the optimal combination method and the corresponding multiple head count for different user data can be learned.

[0067] After receiving new user data, the information generation model that has completed training can output the number of heads corresponding to the current user data and the combination method for this new data. Finally, after visual output, the number of heads corresponding to different user data can be intuitively obtained.

[0068] Reference Figure 2 , Figure 2 shows an application schematic diagram of the information generation method. After training, it realizes automatic screening, automated strategy learning, and obtains the automatically generated optimal RTA marketing strategy after output.

[0069] The module for data preprocessing automatically processes various types of data mentioned above, such as missing value filling, one-hot encoding, mean, variance, etc. The automated algorithm optimization module performs strategy learning through the processed data and clear optimization target parameters.

[0070] By automatically analyzing and optimizing the strategy effect, it realizes visual and full-process evaluation of the strategy effect and automatically determines the optimal strategy.

[0071] The automatically generated strategy report includes the content mentioned in the above embodiments, such as the content of the optimal strategy, the variables and thresholds used in the strategy, the combined logical relationship of the strategy, the weights and ratios affected by the strategy, the estimated business performance of strategy optimization, etc.

[0072] This embodiment realizes the automatic generation of a strategy analysis report for different user data, which is convenient for business personnel to use, reduces the requirements for the business experience of business personnel, and thus reduces the labor cost.

[0073] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. And although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0074] According to an embodiment of the present disclosure, there is also provided an apparatus for implementing the above information generation. The apparatus is used in an advertising placement service implemented based on RTA. The apparatus includes: a labeling unit configured to obtain historical user data, preprocess the user data, and perform label annotation; a data screening unit configured to automatically screen the preprocessed data according to preset screening conditions to obtain valid user data; a training unit configured to input the labeled valid user data into an information generation model, use a preset optimization objective as the training objective, train the pre-established information generation model, and output the configuration information of the information generation model under the optimization objective; and an output unit configured to, after the information generation model is trained, if receiving an input of new user data in the current traffic, output the optimal configuration information corresponding to the new user data.

[0075] As an optional implementation manner of this embodiment, the information generation model includes: the advertising placement service is pre-segmented into multiple service nodes, and the information generation model predicts the conversion rate of the advertising placement service for each node based on the user data; the information generation model estimates the user value size based on the user data to obtain the size of the user value; and the information generation model also estimates the competing bid based on the size of the user value.

[0076] An embodiment of the present disclosure provides an electronic device, as Figure 3 shown. The electronic device includes one or more processors 31 and a memory 32. Figure 3 Taking one processor 31 as an example in

[0077] The controller may further include: an input device 33 and an output device 34.

[0078] The processor 31, the memory 32, the input device 33, and the output device 34 may be connected through a bus or other means. Figure 3 Taking connection through a bus as an example in

[0079] The processor 31 may be a central processing unit (CPU). The processor 31 may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. chips, or a combination of the above types of chips. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0080] The memory 32, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as the program instructions / modules corresponding to the control method in the embodiments of the present disclosure. The processor 31 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions, and modules stored in the memory 32, that is, implements the information generation method in the above method embodiments.

[0081] The memory 32 may include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the processing device of the server operation, etc. In addition, the memory 32 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory 32 may optionally include a memory remotely set relative to the processor 31, and these remote memories can be connected to the network connection device through a network. Examples of the above networks include, but are not limited to, the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0082] The input device 33 can receive input digital or character information, and generate key signal inputs related to the user settings and function controls of the processing device of the server. The output device 34 may include a display device such as a display screen.

[0083] One or more modules are stored in the memory 32, and when executed by one or more processors 31, they execute as Figure 1 shown in the method.

[0084] Those skilled in the art can understand that to implement all or part of the processes in the above method embodiments, it can be completed by instructing relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the above embodiments of each motor control method. Among them, the storage medium can be a magnetic disk, an optical disc, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated: HDD), or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above types of memories.

[0085] Although embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations fall within the scope defined by the appended claims.

Claims

1. An information generation method, characterized in that, The method is used in the advertising placement service implemented based on RTA, and the method includes: Obtain historical user data, preprocess the user data and perform label annotation; Automatically screen the preprocessed data according to preset screening conditions to obtain valid user data; Input the labeled valid user data into the information generation model, use the preset optimization goal as the training goal, train the pre-established information generation model, and output the optimal configuration information of the information generation model under the optimization goal; After the information generation model is trained, if the input of new user data in the current traffic is received, output the optimal configuration information corresponding to the new user data; Among them, the information generation model includes: The advertising placement service is pre-segmented into multiple service nodes, and the information generation model predicts the conversion rate of the advertising placement service for each node based on the user data; The information generation model estimates the size of user value based on the user data to obtain the size of user value; The information generation model also estimates the competing bid based on the size of user value.

2. The information generation method according to claim 1, wherein Training the pre-established information generation model with the preset optimization goal as the training goal includes: Perform a first target preset on the conversion rate of each node in the service, including setting the conversion rate as the optimization goal; When using the preset first target as the optimization goal, determine the configuration information in the conversion rate prediction process.

3. The information generation method according to claim 2, wherein Training the pre-established information generation model with the preset optimization goal as the training goal includes: Perform a second target preset on the service information for evaluating the size of user value, including setting the user value threshold as the optimization goal; When using the preset second target as the optimization goal, determine the configuration information in the user value estimation process.

4. The information generation method according to claim 1, wherein Training the pre-established information generation model with the preset optimization goal as the training goal includes: Perform a third target preset on the competing cost, including using the competing cost as the optimization goal; When using the preset third target as the optimization goal, simultaneously determine the configuration information of the conversion rate and the configuration information of user value.

5. The information generation method according to claim 1, wherein When using the preset optimization goal as the training goal, also output the configuration information of the preset screening conditions.

6. An information generation device, characterized in that, The device is used in the advertising placement service implemented based on RTA, and the device includes: A labeling unit configured to obtain historical user data, preprocess the user data and perform label annotation; A data screening unit configured to automatically screen the preprocessed data according to preset screening conditions to obtain valid user data; A training unit configured to input the labeled valid user data into the information generation model, use the preset optimization goal as the training goal, train the pre-established information generation model, and output the configuration information of the information generation model under the optimization goal; An output unit configured to, after the information generation model is trained, if the input of new user data in the current traffic is received, output the optimal configuration information corresponding to the new user data; Among them, the information generation model includes: The advertising placement service is pre-segmented into multiple service nodes, and the information generation model predicts the conversion rate of the advertising placement service for each node based on the user data; The information generation model estimates the magnitude of user value based on the user data to obtain the magnitude of user value; The information generation model also estimates the competing bid based on the magnitude of user value.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the information generation method according to any one of claims 1-5.

8. An electronic device, characterized in that, Comprising: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to cause the at least one processor to execute the information generation method according to any one of claims 1-5.

Images