Method and device for adjusting delivery strategy of delivery object, equipment and storage medium

Abstract

The invention belongs to the field of artificial intelligence and financial science and technology, and relates to a delivery object delivery strategy adjustment method, which comprises the following steps: obtaining target users having historical behaviors on all delivery objects according to obtained object index data of a current period, user attribute data and user behavior data of the target user in the current period and object attribute data of the delivery object are obtained, and time sequence features are constructed for the user attribute data, the user behavior data and the object attribute data through feature engineering; and the index prediction result of the next period is predicted through the index prediction model according to the time sequence characteristics. The invention further provides a putting object putting strategy adjusting device and equipment and a storage medium. In addition, the invention also relates to a block chain technology, and the user attribute data and the user behavior data can be stored in a block chain. According to the invention, the delivery effect of the delivery object can be pre-estimated in advance, and accurate delivery is realized.

Description

Technical Field

[0001] The present application relates to the fields of artificial intelligence and financial technology, and in particular to a method, device, equipment and storage medium for adjusting a delivery strategy for a delivery object. Background Art

[0002] Product advertising aims to promote product sales by introducing relevant product information to the target audience, highlighting the characteristics of the product, and attracting the attention of the target audience and potential consumers to the corresponding product. It strives to achieve direct and immediate advertising effects, leaving a good product image in the minds of the target audience, thereby increasing the market share of the product and laying a solid foundation for ultimately achieving the company's goals.

[0003] In the financial field, with the rapid development of the automobile industry, automobile finance business has gradually become an important service field. Various automobile financial products are emerging in an endless stream, and how to make their own financial products known and used by users in a very short time has become the focus of marketing. In order to increase the usage rate of automobile financial products and increase customer stickiness, business personnel will choose to place multiple different marketing advertisements on major platforms. Usually, a financial product has a long marketing stage. It is difficult to evaluate the effects of marketing advertisements on various platforms in a short period of time, and it is impossible to accurately compare the effects of marketing advertisements on various platforms, and it is impossible to provide a relatively scientific adjustment strategy. Summary of the invention

[0004] The purpose of the embodiments of the present application is to propose a method, device, equipment and storage medium for adjusting the delivery strategy of a delivery object, so as to solve the technical problems that it is currently difficult to evaluate the effectiveness of marketing advertisements on various platforms in the short term, it is impossible to accurately compare the effects of marketing advertisements on various platforms, and it is impossible to provide a relatively scientific adjustment strategy.

[0005] In order to solve the above technical problems, the present application embodiment provides a method for adjusting the delivery strategy of a delivery object, which adopts the following technical solution:

[0006] Obtaining target index data of all delivery objects of the target product in the current cycle, and obtaining target users who have historical behaviors on all delivery objects according to the target index data;

[0007] Acquire user attribute data and user behavior data of the target user in the current period, and acquire object attribute data of the delivery object;

[0008] Performing feature selection on the user attribute data, the user behavior data, and the object attribute data through feature engineering to construct time series features;

[0009] Input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next cycle;

[0010] Scoring the indicator prediction results to obtain scoring results;

[0011] The delivery strategy of the delivery object is adjusted based on the scoring result.

[0012] In order to solve the above technical problems, the embodiment of the present application further provides a delivery strategy adjustment device for delivery objects, which adopts the following technical solution:

[0013] A first acquisition module is used to acquire the target index data of all the delivery objects of the target product in the current cycle, and acquire the target users who have historical behaviors on all the delivery objects according to the object index data;

[0014] A second acquisition module is used to acquire user attribute data and user behavior data of the target user in the current period, and acquire object attribute data of the delivery object;

[0015] A feature engineering module, used for performing feature selection on the user attribute data, the user behavior data and the object attribute data through feature engineering to construct time series features;

[0016] A prediction module, used to input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next period;

[0017] A scoring module is used to score the indicator prediction results to obtain a scoring result;

[0018] An adjustment module is used to adjust the delivery strategy of the delivery object based on the scoring result.

[0019] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0020] The computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for adjusting the delivery strategy of the delivery object as described above are implemented.

[0021] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0022] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the method for adjusting the delivery strategy of the delivery object as described above are implemented.

[0023] Compared with the prior art, this application has the following beneficial effects:

[0024] The present application provides a method for adjusting the delivery strategy of a delivery object, which obtains target users who have historical behaviors on all delivery objects through object indicator data of the current cycle, and obtains user attribute data and user behavior data of the target users in the current cycle and object attribute data of the delivery object, constructs time series features for the user attribute data, user behavior data and object attribute data through feature engineering, and predicts the indicator prediction result of the next cycle according to the time series features through an indicator prediction model, which can achieve early estimation of the delivery effect of the delivery object, assist in the operation management of the delivery object and the adjustment of the delivery strategy of the delivery object, so that the delivery object can be accurately delivered and avoid waste of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0027] Figure 2 It is a flow chart of an embodiment of a method for adjusting a delivery strategy of a delivery object according to the present application;

[0028] Figure 3 It is a structural schematic diagram of an embodiment of a delivery strategy adjustment device for delivery objects according to the present application;

[0029] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0031] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0033] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0034] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0035] This application provides a method for adjusting the delivery strategy of a delivery object, which can be applied to Figure 1 In the system architecture 100 shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0036] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0037] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, etc.

[0038] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0039] It should be noted that the delivery strategy adjustment method for delivery objects provided in the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the delivery strategy adjustment device for delivery objects is generally arranged in a server / terminal device.

[0040] 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.

[0041] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for adjusting a delivery strategy of a delivery object according to the present application, comprising the following steps:

[0042] Step S201, obtaining the target index data of all delivery objects of the target product in the current cycle, and obtaining the target users who have historical behaviors on all delivery objects according to the object index data.

[0043] The target product can refer to physical goods or services, including financial services, insurance, investment, travel, etc. The delivery targets are all marketing advertisements delivered by the target product, and all delivery targets are all types of advertisements delivered by the target product on various platforms. For example, when the target product is auto finance business, the delivery targets are auto finance products, including auto advertisements, auto insurance advertisements, auto consumer loan advertisements, auto mortgage advertisements, credit card installment car purchase advertisements, etc.

[0044] The current cycle can be set according to actual conditions, such as in units of years, months, weeks, days, etc. In this embodiment, in order to evaluate the effectiveness of marketing advertisements on each platform in a short period of time, one week is used as a cycle, and the current cycle is the current week.

[0045] Object indicator data are indicators used to measure the effectiveness of advertising, including clicks, visits, reservations, transactions, conversion rates, etc.

[0046] Historical behaviors include but are not limited to clicks, visits, reservations, conversions, sharing, viewing, browsing, comments, likes, following, feedback, complaints, etc.

[0047] In this embodiment, the delivery strategy adjustment method for delivery objects is executed on the electronic device (eg Figure 1 The server / terminal device shown in the figure can obtain the object indicator data through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.

[0048] Step S202, obtaining user attribute data and user behavior data of the target user in the current cycle, and obtaining object attribute data of the delivery object.

[0049] In this embodiment, user attribute data includes gender, age, social media account, commonly used IP address, marital status, etc.; user behavior data includes views, clicks, reservations, transactions, stay time, likes, comments, favorites, viewing time, viewing completion rate, interaction time (i.e. timestamp), etc.; object attribute data includes delivery time, delivery channel, delivery content, delivery budget, delivery platform, object type, object target user, delivery cycle, number of deliveries within the delivery cycle, etc.

[0050] In this embodiment, data embedding can be used to obtain user attribute data and user behavior data. Specifically, a specific code or tool is embedded in the application or website to track and record the user's attribute information and behavior information. Object attribute data can be obtained from a third-party object delivery monitoring platform or from a database corresponding to the delivery object.

[0051] It should be emphasized that in order to further ensure the privacy and security of user attribute data and user behavior data, the above user attribute data and user behavior data can also be stored in a blockchain node.

[0052] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0053] Step S203, performing feature selection on user attribute data, user behavior data and object attribute data through feature engineering to construct time series features.

[0054] Among them, feature engineering is used to select the most useful features (variables or attributes) for the prediction model from user attribute data, user behavior data, and object attribute data.

[0055] In some optional implementations of this embodiment, the step of performing feature selection on user attribute data, user behavior data, and object attribute data through feature engineering to construct time series features includes:

[0056] Preprocessing the user attribute data, the user behavior data and the object attribute data to obtain preprocessed object data;

[0057] Encode the preprocessed object data to obtain an object feature vector;

[0058] The random forest model is used to calculate the feature importance of the object feature vector, and the target object features are selected based on the feature importance;

[0059] The time features in the target object features are extracted, and the time series features of the target object features are constructed according to the time features.

[0060] Preprocessing includes cleaning data, processing missing values ​​and outliers, and filling missing values ​​of data according to multiple interpolation methods. Encoding the preprocessed object data includes embedding encoding or one-hot encoding, where embedding encoding can use word embedding or BERT embedding.

[0061] In this embodiment, the feature importance of the object feature vector is calculated using a random forest model, and the target object features are selected according to the feature importance. When training with a random forest model, each decision tree will randomly select a portion of features for splitting, which helps to reduce the model's dependence on specific features. After the training is completed, the feature importance of each feature is usually evaluated based on the Gini impurity reduced by each feature when building a tree; according to the feature importance, the features are sorted, and G object feature vectors with feature importance greater than a preset threshold are selected as the extracted target object features, and those features that contribute less to model prediction are removed to identify the most influential features; the random forest model is retrained using the selected feature subset to evaluate the effect of feature selection; then the parameters of the random forest model, such as the number of trees, maximum depth, etc., are adjusted according to the effect of feature selection to find the optimal model; after determining the optimal model, the performance of the optimal model is evaluated on an independent test set to ensure that the results of feature engineering can improve the prediction accuracy of the model.

[0062] The time series feature is constructed based on the extracted target object features. The time series is a data column recorded in chronological order with a unified indicator. Specifically, the time feature, i.e., the timestamp, can be extracted from the target object features, and the corresponding time series feature can be constructed based on the timestamp.

[0063] In some optional implementations, based on the timestamp itself, the main feature is about the time sliding window. The features obtained by automatic time series feature engineering include target features based on the time sliding window, target statistical features based on the time sliding window, target trend features based on the time sliding window, important original features based on the time sliding window, statistical features based on the time sliding window, and other features. Among them, for the target features based on the time sliding window, in the time series data, the target generally does not deviate too much from the adjacent time step values, and they have a strong correlation, so the adjacent targets in the past can be used as features first.

[0064] Through feature engineering, data quality can be improved and the expressiveness of features can be enhanced, thereby improving the performance, accuracy and generalization ability of the model.

[0065] Step S204, input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next period.

[0066] The indicator prediction model includes a shared LSTM layer, a specific indicator prediction layer and an output layer. In this embodiment, the step of inputting the time series features into the trained indicator prediction model to obtain the indicator prediction results of the delivery object in the next cycle includes:

[0067] Input the time series features into the shared LSTM layer for sequence data processing and output the shared feature hidden state;

[0068] Extract the shared feature hidden state through the specific indicator prediction layer to obtain the prediction indicator feature;

[0069] The predicted indicator features are input into the output layer for calculation to obtain the corresponding indicator prediction results.

[0070] Among them, the shared LSTM layer can process sequence data and extract time-dependent features, which are used to extract the common features of the delivery objects in the time series features; the output of the LSTM layer is a series of hidden states, which contain the feature representation of the input sequence. The specific indicator prediction layer uses the output of the shared LSTM layer as feature input and passes it to the specific layer of each task, such as the visit volume prediction layer, click volume prediction layer, reservation volume prediction layer, transaction volume visit layer, etc., to extract task-related features, such as features related to visit volume prediction. The specific indicator prediction layer of each task is finally connected to the corresponding output layer to output the final prediction result.

[0071] By predicting indicators through the indicator prediction model, useful feature information can be shared between different tasks, reducing the amount of calculation and preventing overfitting. The strong correlation between the prediction results of different indicators is used to make the model's feature representation ability stronger, thereby ensuring the accuracy of the model's prediction.

[0072] In some optional implementations of this embodiment, the shared LSTM layer includes a forget gate, an input gate, an update gate, and an output gate, wherein the time series feature passes through the forget gate, the input gate, the update gate, and the output gate. The forget gate determines which information in the memory state of the previous moment will be discarded. It generates an output between 0 and 1 by inputting the input of the current moment and the hidden state of the previous moment through a sigmoid activation function, where 0 indicates that the relevant information is completely forgotten and 1 indicates that the relevant information is completely retained. The input gate determines which information input at the current moment will be added to the memory state. It consists of two parts: a sigmoid activation function for controlling the importance of information and a tanh activation function for creating a new candidate value. The update gate combines the information of the forget gate and the input gate for update. The output gate determines the output of the hidden state at the current moment. It generates an output between 0 and 1 by inputting the input of the current moment and the hidden state of the previous moment through a sigmoid activation function. The memory state at the current moment is processed through a tanh activation function and multiplied by the output of the output gate to obtain the final hidden state.

[0073] In this embodiment, the step of inputting the time series features into the shared LSTM layer for sequence data processing and outputting the hidden state of the shared features includes:

[0074] Based on the shared feature hidden state at the previous moment and the current time series features, the forget gate output features are calculated through the forget gate;

[0075] The forget gate output feature, the shared feature hidden state at the previous moment, and the current time series feature are calculated through the input gate to obtain an input gating signal and a candidate memory cell state;

[0076] Calculate the updated memory cell state through the update gate according to the forget gate output feature, the input gating signal and the candidate memory cell state;

[0077] The output gating signal is calculated through the output gate according to the shared feature hidden state at the previous moment and the current time series characteristics, and the shared feature hidden state at the current moment is output based on the output gating signal and the updated memory cell state.

[0078] Among them, the calculation formula of the forget gate is as follows:

[0079] f t =σ(W xf x t +W hf h t-1 +b f );

[0080] Among them, σ is the sigmoid function, W xf , W hf and b f are the weight and bias of the forget gate respectively; x t is the current time series feature, h t-1 is the hidden state of the shared features at the previous moment.

[0081] The calculation formula of the input gate is as follows:

[0082] i t =σ(W xi x t +W hi h t-1 +b i );

[0083] Among them, W xi , W hi and b i are the weight and bias of the input gate respectively; x t is the current time series feature, h t-1 is the hidden state of the shared features at the previous moment.

[0084] The calculation formula for the candidate memory cell state is as follows:

[0085]

[0086] Among them, tanh is the hyperbolic tangent function; W xc , W hc and b c They are the weight and bias for calculating the candidate memory cell state; x t is the current time series feature, h t-1 is the hidden state of the shared features at the previous moment.

[0087] The calculation formula of the memory cell state after the update gate is updated is as follows:

[0088]

[0089] The formula for the shared feature hidden state at the current moment is as follows:

[0090]

[0091] Among them, u t is the output of the update gate; is the candidate hidden state at the current moment.

[0092] The LSTM layer overcomes the problems of gradient vanishing and gradient exploding during model prediction by using a mechanism constructed using input gates, forget gates, update gates, and output gates, and is able to capture long-term dependencies in the sequence and improve prediction accuracy.

[0093] In some optional implementations, before the step of inputting the time series features into the trained indicator prediction model, the following steps are also included:

[0094] Obtain a sample data set of all historical delivery objects corresponding to all historical products in a target historical period, the sample data set includes historical user attribute data, historical user behavior data, and historical object attribute data, and the sample data set carries real indicator data of the next period of the target historical period corresponding to the sample data one by one;

[0095] The pre-built initial indicator prediction model is trained based on the sample data set to obtain the final indicator prediction model.

[0096] During the training process, the mean square error (MSE) can be used as the loss function. The loss function is used to calculate the difference between the model output and the true label to obtain the loss value. The back propagation algorithm is used to calculate the gradient of the loss value relative to the network parameters. The optimizer is used to update the network weights and biases according to the calculated gradient. The iterative training continues until the performance of the model on the validation set no longer improves, or the preset number of iterations is reached, and the final indicator prediction model is output.

[0097] By training the model, the training process optimizes the model parameters to improve the fit to the training data and the prediction accuracy; it can automatically learn useful features in the data and improve the accuracy of the prediction; and the training process allows the model to adapt to different data distributions and changes, improve its robustness on new data, and reduce human bias.

[0098] Step S205, scoring the indicator prediction results to obtain scoring results.

[0099] Specifically, the average indicator data of different delivery objects in the next cycle and the average unit indicator data of each delivery object in the next cycle are calculated according to the indicator prediction results; the score of the delivery object is calculated based on the average indicator data and the average unit indicator data to obtain a scoring result.

[0100] For example, if one week is a cycle, and the indicator prediction result is the click volume data of all delivery objects of the target product, then the average indicator data is the average weekly click volume avg(week) of all delivery objects, and the average unit indicator data is the average daily click volume avg(day) of each delivery object. The number of days required to reach the average weekly click volume avg(week) is calculated and rounded up. The calculation formula is as follows:

[0101] d=ceil(avg(week) / avg(day));

[0102] Based on 7-d, the scores of the delivery objects are adjusted, and the scoring results are:

[0103] s=s+2*(7-d).

[0104] The sample results are shown in Table 1:

[0105] Table 1

[0106] Current score avg(week) avg(day) Number of days Rating results Latest scores 0 1000 110 10 s＝0-2*3 -6 0 1000 400 3 s＝0+2*4 8

[0107] Taking one week as a cycle, the indicator prediction result is the reservation volume data of all delivery objects of the target product. The average weekly reservation volume avg(week) of different advertisements in the current week is counted. The average daily reservation volume avg(day) of each advertisement in the current week is counted. The number of days required to reach the average weekly reservation volume avg(week) is calculated and rounded up.

[0108] d=ceil(avg(week) / avg(day)), then the scoring result is:

[0109] s=s-2^(d-7)(when d≥7)

[0110] s=s+2^(7-d)(when d<7)

[0111] The same calculation method is used for other different indicator data. It should be understood that the importance of different indicator data is different. For example, among click behavior, reservation behavior, and transaction behavior, click has the lowest importance and transaction has the highest importance. Therefore, the final scoring result is weighted.

[0112] By scoring the indicator prediction results, decision makers can be provided with clear and comparable data support, and make more objective and accurate decisions based on the scoring results. Scoring also simplifies complex data into easy-to-understand numerical values, allowing non-professionals to quickly grasp key information.

[0113] Step S206: adjusting the delivery strategy of the delivery object based on the scoring result.

[0114] Specifically, the delivery strategy includes selecting advertising channels, designing advertising content ideas, identifying target audiences, determining delivery time, delivery frequency and continuity, geographic positioning, and multi-channel integrated marketing. The delivery strategy can be re-customized based on the scoring results to maximize the delivery effect.

[0115] This application constructs time series features for user attribute data, user behavior data and object attribute data through feature engineering, and predicts the indicator prediction results of the next period according to the time series features through the indicator prediction model. It can achieve early estimation of the delivery effect of the delivery object, assist in the operation management of the delivery object and adjust the delivery strategy of the delivery object, so that the delivery object can be delivered accurately and avoid waste of resources.

[0116] In some optional implementations, the step of encoding the preprocessed object data to obtain the object feature vector includes:

[0117] Classify the preprocessed object data to obtain classification results, and extract classification labels based on the classification results;

[0118] The preprocessed object data is encoded according to the classification label to obtain the object feature vector.

[0119] In this embodiment, the classification results are one type of feature data and the other type of indicator data's true value. The classification labels are the feature labels and the indicator's true labels. The data are then encoded according to the classification labels, which can optimize data feature representation, improve data consistency, reduce data processing time, and improve the model's understanding of the data's internal structure, thereby improving prediction accuracy.

[0120] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0121] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0122] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0123] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a device for adjusting the delivery strategy of a delivery object. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0124] like Figure 3 As shown, the delivery strategy adjustment device 300 of the delivery object described in this embodiment includes: a first acquisition module 301, a second acquisition module 302, a feature engineering module 303, a prediction module 304, a scoring module 305 and an adjustment module 306. Among them:

[0125] The first acquisition module 301 is used to acquire the target index data of all the delivery objects of the target product in the current cycle, and acquire the target users who have historical behaviors on all the delivery objects according to the object index data;

[0126] The second acquisition module 302 is used to acquire the user attribute data and user behavior data of the target user in the current period, and acquire the object attribute data of the delivery object;

[0127] The feature engineering module 303 is used to perform feature selection on the user attribute data, the user behavior data and the object attribute data through feature engineering to construct time series features;

[0128] The prediction module 304 is used to input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next period;

[0129] The scoring module 305 is used to score the indicator prediction result to obtain a scoring result;

[0130] The adjustment module 306 is used to adjust the delivery strategy of the delivery object based on the scoring result.

[0131] It should be emphasized that in order to further ensure the privacy and security of user attribute data and user behavior data, the above user attribute data and user behavior data can also be stored in a blockchain node.

[0132] Based on the above-mentioned delivery object delivery strategy adjustment device 300, time series features are constructed for user attribute data, user behavior data and object attribute data through feature engineering, and the indicator prediction model is used to predict the indicator prediction result of the next period according to the time series features. This can achieve early estimation of the delivery effect of the delivery object, assist in the operation management of the delivery object and adjust the delivery strategy of the delivery object, so that the delivery object can be delivered accurately and avoid waste of resources.

[0133] In some optional implementations, the feature engineering module 303 includes:

[0134] A preprocessing submodule, used for preprocessing the user attribute data, the user behavior data and the object attribute data to obtain preprocessed object data;

[0135] An encoding submodule, used for encoding the preprocessed object data to obtain an object feature vector;

[0136] A feature selection submodule, used to calculate the feature importance of the object feature vector using a random forest model, and select target object features according to the feature importance;

[0137] The construction submodule is used to extract the time features in the target object features and construct the time series features of the target object features according to the time features.

[0138] Through feature engineering, data quality can be improved and the expressiveness of features can be enhanced, thereby improving the performance, accuracy and generalization ability of the model.

[0139] In some optional implementations of this embodiment, the encoding submodule is further used for:

[0140] Classifying the preprocessed object data to obtain a classification result, and extracting a classification label according to the classification result;

[0141] The preprocessed object data is encoded according to the classification label to obtain an object feature vector.

[0142] By classifying and encoding the preprocessed object data, we can optimize the data feature representation, improve data consistency, reduce data processing time, and improve the model's understanding of the data's internal structure, thereby improving prediction accuracy.

[0143] In some optional implementations, the indicator prediction model includes a shared LSTM layer, a specific indicator prediction layer, and an output layer; the prediction module 304 includes:

[0144] A hidden submodule, used for inputting the time series features into the shared LSTM layer for sequence data processing, and outputting the shared feature hidden state;

[0145] An extraction submodule, used for extracting features of the shared feature hidden state through the specific indicator prediction layer to obtain prediction indicator features;

[0146] The output submodule is used to input the prediction index features into the output layer for calculation to obtain the corresponding index prediction results.

[0147] By predicting indicators through the indicator prediction model, useful feature information can be shared between different tasks, reducing the amount of calculation and preventing overfitting. The strong correlation between the prediction results of different indicators is used to make the model's feature representation ability stronger, thereby ensuring the accuracy of the model's prediction.

[0148] In some optional implementations of this embodiment, the shared LSTM layer includes a forget gate, an input gate, an update gate, and an output gate; the hidden submodule is further used for:

[0149] Based on the shared feature hidden state at the previous moment and the current time series feature, the forget gate output feature is calculated through the forget gate;

[0150] The forget gate output feature, the shared feature hidden state at the previous moment, and the current time series feature are calculated through the input gate to obtain an input gating signal and a candidate memory cell state;

[0151] Calculating an updated memory cell state through the update gate according to the forget gate output feature, the input gating signal and the candidate memory cell state;

[0152] The output gate calculates an output gating signal according to the shared feature hidden state at the previous moment and the current time series feature, and outputs the shared feature hidden state at the current moment based on the output gating signal and the updated memory cell state.

[0153] The LSTM layer overcomes the problems of gradient vanishing and gradient exploding during model prediction by using a mechanism constructed using input gates, forget gates, update gates, and output gates, and is able to capture long-term dependencies in the sequence and improve prediction accuracy.

[0154] In some optional implementations, the delivery strategy adjustment device 300 based on the delivery object further includes a training module for:

[0155] Obtain a sample data set of all historical delivery objects corresponding to all historical products in a target historical period, wherein the sample data set includes historical user attribute data, historical user behavior data, and historical object attribute data, and the sample data set carries real indicator data of the next period of the target historical period corresponding to the sample data one by one;

[0156] The pre-constructed initial indicator prediction model is trained based on the sample data set to obtain a final indicator prediction model.

[0157] By training the model, the training process optimizes the model parameters to improve the fit to the training data and the prediction accuracy; it can automatically learn useful features in the data and improve the accuracy of the prediction; and the training process allows the model to adapt to different data distributions and changes, improve its robustness on new data, and reduce human bias.

[0158] In some optional implementations, the scoring module 305 is further configured to:

[0159] Calculate the average index data of different delivery objects in the next period and the average unit index data of each delivery object in the next period according to the index prediction result;

[0160] The score of the delivery object is calculated based on the average indicator data and the average unit indicator data to obtain a final scoring result.

[0161] By scoring the indicator prediction results, decision makers can be provided with clear and comparable data support, and make more objective and accurate decisions based on the scoring results. Scoring also simplifies complex data into easy-to-understand numerical values, allowing non-professionals to quickly grasp key information.

[0162] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0163] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0164] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0165] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk equipped on the computer device 4, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the delivery strategy adjustment method of the delivery object, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0166] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions or process data stored in the memory 41, such as computer-readable instructions for running the delivery strategy adjustment method of the delivery object.

[0167] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0168] This embodiment implements the steps of the delivery strategy adjustment method for delivery objects as in the above-mentioned embodiment when the processor executes computer-readable instructions stored in the memory, constructs time series features for user attribute data, user behavior data and object attribute data through feature engineering, and predicts the indicator prediction results of the next period according to the time series features through the indicator prediction model, which can achieve early estimation of the delivery effect of the delivery object, assist in the operation management of the delivery object and the adjustment of the delivery strategy of the delivery object, so that the delivery object can be accurately delivered and avoid waste of resources.

[0169] The present application also provides another implementation mode, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor, so that the at least one processor executes the steps of the delivery strategy adjustment method for the delivery object as described above, constructs time series features for user attribute data, user behavior data and object attribute data through feature engineering, and predicts the indicator prediction results of the next period according to the time series features through an indicator prediction model, which can achieve early estimation of the delivery effect of the delivery object, assist in the operation management of the delivery object and the adjustment of the delivery strategy of the delivery object, so that the delivery object can be accurately delivered and avoid waste of resources.

[0170] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0171] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for adjusting a delivery strategy of a delivery object, characterized in that: The steps include: Obtaining target index data of all delivery objects of the target product in the current cycle, and obtaining target users who have historical behaviors on all delivery objects according to the target index data; Acquire user attribute data and user behavior data of the target user in the current period, and acquire object attribute data of the delivery object; Performing feature selection on the user attribute data, the user behavior data, and the object attribute data through feature engineering to construct time series features; Input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next cycle; Scoring the indicator prediction results to obtain scoring results; The delivery strategy of the delivery object is adjusted based on the scoring result.

2. The method for adjusting the delivery strategy of a delivery object according to claim 1, characterized in that: The step of performing feature selection on the user attribute data, the user behavior data and the object attribute data through feature engineering to construct time series features includes: Preprocessing the user attribute data, the user behavior data, and the object attribute data to obtain preprocessed object data; Encoding the preprocessed object data to obtain an object feature vector; Calculating the feature importance of the object feature vector using a random forest model, and selecting target object features according to the feature importance; The time features in the target object features are extracted, and the time series features of the target object features are constructed according to the time features.

3. The method for adjusting the delivery strategy of a delivery object according to claim 2, characterized in that: The step of encoding the preprocessed object data to obtain an object feature vector comprises: Classifying the preprocessed object data to obtain a classification result, and extracting a classification label according to the classification result; The preprocessed object data is encoded according to the classification label to obtain an object feature vector.

4. The method for adjusting the delivery strategy of a delivery object according to claim 1, characterized in that: The indicator prediction model includes a shared LSTM layer, a specific indicator prediction layer and an output layer; the step of inputting the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next cycle includes: Input the time series features into the shared LSTM layer for sequence data processing, and output the shared feature hidden state; Extracting features of the shared feature hidden state through the specific indicator prediction layer to obtain prediction indicator features; The prediction index feature is input into the output layer for calculation to obtain the corresponding index prediction result.

5. The method for adjusting the delivery strategy of a delivery object according to claim 4, characterized in that: The shared LSTM layer includes a forget gate, an input gate, an update gate, and an output gate; the step of inputting the time series features into the shared LSTM layer for sequence data processing and outputting the hidden state of the shared features includes: Based on the shared feature hidden state at the previous moment and the current time series feature, the forget gate output feature is calculated through the forget gate; The forget gate output feature, the shared feature hidden state at the previous moment, and the current time series feature are calculated through the input gate to obtain an input gating signal and a candidate memory cell state; Calculating an updated memory cell state through the update gate according to the forget gate output feature, the input gating signal and the candidate memory cell state; The output gate calculates an output gating signal according to the shared feature hidden state at the previous moment and the current time series feature, and outputs the shared feature hidden state at the current moment based on the output gating signal and the updated memory cell state.

6. The method for adjusting the delivery strategy of a delivery object according to claim 1, characterized in that: Before the step of inputting the time series features into the trained indicator prediction model, the method further includes: Obtain a sample data set of all historical delivery objects corresponding to all historical products in a target historical period, wherein the sample data set includes historical user attribute data, historical user behavior data, and historical object attribute data, and the sample data set carries real indicator data of the next period of the target historical period corresponding to the sample data one by one; The pre-constructed initial indicator prediction model is trained based on the sample data set to obtain a final indicator prediction model.

7. The method for adjusting the delivery strategy of a delivery object according to claim 1, characterized in that: The step of scoring the indicator prediction result to obtain the scoring result comprises: Calculate the average index data of different delivery objects in the next period and the average unit index data of each delivery object in the next period according to the index prediction result; The score of the delivery object is calculated based on the average indicator data and the average unit indicator data to obtain a final scoring result.

8. A device for adjusting the delivery strategy of a delivery object, characterized in that: include: A first acquisition module is used to acquire the target index data of all the delivery objects of the target product in the current cycle, and acquire the target users who have historical behaviors on all the delivery objects according to the object index data; A second acquisition module is used to acquire user attribute data and user behavior data of the target user in the current period, and acquire object attribute data of the delivery object; A feature engineering module, used for performing feature selection on the user attribute data, the user behavior data and the object attribute data through feature engineering to construct time series features; A prediction module, used to input the time series features into the trained indicator prediction model to obtain the indicator prediction result of the delivery object in the next period; A scoring module is used to score the indicator prediction results to obtain a scoring result; An adjustment module is used to adjust the delivery strategy of the delivery object based on the scoring result.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for adjusting the delivery strategy of the delivery object as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the method for adjusting the delivery strategy of a delivery object according to any one of claims 1 to 7 are implemented.

Images