CN117390540A - Target operation execution method and device, storage medium and electronic equipment - Google Patents
Target operation execution method and device, storage medium and electronic equipment Download PDFInfo
- Publication number
- CN117390540A CN117390540A CN202311323401.6A CN202311323401A CN117390540A CN 117390540 A CN117390540 A CN 117390540A CN 202311323401 A CN202311323401 A CN 202311323401A CN 117390540 A CN117390540 A CN 117390540A
- Authority
- CN
- China
- Prior art keywords
- target
- prediction
- account
- parameters
- attribute information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000003860 storage Methods 0.000 title claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 77
- 238000012545 processing Methods 0.000 claims description 39
- 238000012512 characterization method Methods 0.000 claims description 37
- 238000012417 linear regression Methods 0.000 claims description 19
- 238000004590 computer program Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 11
- 230000009467 reduction Effects 0.000 claims description 5
- 239000000523 sample Substances 0.000 claims 19
- 230000008569 process Effects 0.000 abstract description 14
- 230000000694 effects Effects 0.000 description 31
- 238000004458 analytical method Methods 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000001965 increasing effect Effects 0.000 description 6
- 238000013136 deep learning model Methods 0.000 description 5
- 230000006399 behavior Effects 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000010200 validation analysis Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000002427 irreversible effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000011480 coordinate descent method Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000005304 joining Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- ZLIBICFPKPWGIZ-UHFFFAOYSA-N pyrimethanil Chemical compound CC1=CC(C)=NC(NC=2C=CC=CC=2)=N1 ZLIBICFPKPWGIZ-UHFFFAOYSA-N 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000008685 targeting Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0253—During e-commerce, i.e. online transactions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/2866—Architectures; Arrangements
- H04L67/30—Profiles
- H04L67/306—User profiles
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/55—Push-based network services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a target operation execution method and device, a storage medium and electronic equipment. The method comprises the following steps: acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing N dimension parameters generated by the target account executing a target type event in a first historical time; inputting account attribute information and object attribute information into a target result prediction model which is trained in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object; inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights; and executing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights. The method and the device solve the technical problem of low accuracy in the execution process of the target operation.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method and apparatus for executing a target operation, a storage medium, and an electronic device.
Background
With the development of internet technology, online shopping is becoming more and more popular through network devices such as computers and mobile phones, and in order to increase the number of commodity transactions, commodity preference information of a shopping platform, such as marketing information including market full-subtracting activities and commodity discount information, is usually pushed to users.
In the related art, marketing-related data is collected mainly depending on database query and batch processing techniques, and then the data is learned and predicted using a conventional machine learning model, for example, linear regression, decision tree, etc., and then marketing strategies are adjusted according to the prediction results.
The marketing strategy determined by the prediction result is to take the whole consumer group as a target object, acquire data such as browsing records, purchasing times and the like of the target object on commodities in a historical time period, predict the predicted sales volume of each commodity in a next period of time according to a machine learning model, assign the commodity marketing strategy, such as commodity advertisement information, according to the predicted sales volume, and send the same commodity preferential activity to each consumer, thereby increasing the commodity trading success rate.
However, in an actual shopping scenario, the consumption capability of each consumer is different, and the focus of attention of each consumer is different when viewing the commodity preference information, in this case, since the commodity preference information seen by the consumer is the same, that means that the consumer may not obtain the key marketing information (such as commodity discount, sales volume, etc.) affecting the self-ordering, the commodity transaction times will be seriously affected, thereby causing a technical problem of lower accuracy in the execution process of the target operation.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the application provides a target operation execution method and device, a storage medium and electronic equipment, so as to at least solve the technical problem of low accuracy in the execution process of the target operation.
According to an aspect of the embodiments of the present application, there is provided a method for performing a target operation, including: acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing N dimension parameters generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing M dimension parameters generated by the target object executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1; inputting account attribute information and object attribute information into a target result prediction model obtained through training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2; inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with the parameters of the P dimensions, and the P prediction weights are used for representing the association between the parameters of the P dimensions and the target object execution target type event; and executing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
Optionally, inputting the account attribute information and the object attribute information into a target result prediction model trained in advance to obtain P prediction results, including: inputting account attribute information and object attribute information into a target result prediction model to obtain p 1 P are predicted results 2 P is the predicted result 1 The prediction results are used for a tableShowing p related to the target account generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 For positive integers greater than or equal to 1, the P predictors include P 1 P are predicted results 2 P is equal to P as a result of prediction 1 And p is as follows 2 And (3) summing.
Optionally, inputting account attribute information and object attribute information into the target result prediction model to obtain p 1 P are predicted results 2 A plurality of prediction results, comprising: inputting account attribute information and object attribute information into a target result prediction model to obtain a first prediction result and a second prediction result, wherein the first prediction result is used for representing the probability that a target account executes a target type event on a target object, and the second prediction result is used for representing the number of the target objects of the executed target type event, and p 1 The prediction results comprise a first prediction result, p 2 The prediction results include a second prediction result.
Optionally, inputting account attribute information and object attribute information into the target result prediction model to obtain p 1 P are predicted results 2 A plurality of prediction results, comprising: respectively converting account attribute information and object attribute information into account characterization features and object characterization features; splicing the account characterization features and the object characterization features to obtain spliced characterization features; sequentially passing the spliced characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain p 1 P are predicted results 2 Predicting results; or the account attribute information and the object attribute information are converted into target characterization features; sequentially passing the target characterization features through a plurality of hidden layers and output layers in a target result prediction model to obtain p 1 P are predicted results 2 And predicting the result.
Optionally, the P prediction results are input into a target weight prediction model obtained by pre-training,obtaining P prediction weights comprises: will p 1 P are predicted results 2 Inputting the prediction results into a target weight prediction model to obtain p 1 Sum of prediction weights p 2 A number of predictive weights, wherein the number P of predictive weights includes P 1 Sum of prediction weights p 2 Each prediction weight, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the individual dimensions have a one-to-one correspondence.
Optionally, inputting the P prediction results into a target weight prediction model trained in advance to obtain P prediction weights, including: and inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the target weight prediction model is a linear regression model, and the P prediction weights are optimal solutions of the target function in the linear regression model under the condition that the input of the linear regression model is the P prediction results.
Optionally, the performing an operation corresponding to a parameter of a part of the parameters of the P dimensions according to the P prediction weights includes: executing an operation corresponding to an ith dimension parameter in the P dimension parameters under the condition that the ith prediction weight in the P prediction weights is the largest, wherein the ith prediction weight corresponds to the ith dimension parameter, and the importance degree of the ith dimension parameter in the P dimension parameters is the highest, i is a positive integer which is greater than or equal to 1 and less than or equal to P; or P is included in the P prediction weights 1 Sum of prediction weights p 2 Predicted weights, and at p 1 The jth prediction weight of the prediction weights is the largest, at p 2 Execution and p in case the kth prediction weight of the prediction weights is the largest 1 Operation corresponding to the parameter of the j-th dimension of the parameters of the dimensions and executing the operation corresponding to p 2 Operations corresponding to parameters of a kth dimension among parameters of the dimensions, wherein the P prediction results comprise P 1 P are predicted results 2 P are predicted by 1 The prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 Is a positive integer greater than or equal to 1, P is equal to P 1 And p is as follows 2 Sum, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the dimensions have a one-to-one correspondence, at p 1 The j-th dimension parameter has the highest importance among the parameters of the dimensions, and is p 2 The parameter of the kth dimension among the parameters of the dimensions has the highest importance, j is greater than or equal to 1 and less than or equal to p 1 K is a positive integer greater than or equal to 1 and less than or equal to p 2 Is a positive integer of (a).
Optionally, the performing an operation corresponding to the parameter of the ith dimension among the parameters of the P dimensions includes one of: pushing media resource information to the target account under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object; controlling the quantity of target resources to be reduced under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the quantity of the target resources is the quantity of resources which are required to be transferred out from a resource set of the target account by the target account for executing the target type event on the target object; in the case where the parameter of the i-th dimension is a parameter corresponding to the target object, the number of control target resources increases.
Optionally, the above-mentioned executing and p 1 The operation corresponding to the parameter of the j-th dimension in the parameters of the dimensions comprises the following steps: pushing media resource information to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object; and/or controlling the reduction of the number of the target resources, wherein the number of the target resources is the number of resources which are required to be transferred out of the resource set of the target account by the target account for executing the target type event on the target object; execution and p 2 Personal dimensionThe operation corresponding to the parameter of the kth dimension in the parameters of the degree comprises the following steps: the number of control target resources increases.
Optionally, the target account executing the target type event on the target object refers to: the target account acquires the target object and transfers the resources with the target resource quantity out of the resource set of the target account.
Optionally, the method further comprises: obtaining a training sample set, wherein each training sample in the training sample set comprises sample account attribute information of a sample account and sample object attribute information of a sample object, the sample account attribute information is used for representing sample parameters of N dimensions generated by executing a target type event on the sample account in a first sample history time, and the sample object attribute information is used for representing sample parameters of M dimensions generated by executing the target type event on the sample object in a second sample history time; and using a training sample set to obtain a result prediction model to be trained, ending training until a loss value corresponding to the result prediction model meets a preset training ending condition, and determining the result prediction model when ending training as a target result prediction model, wherein the loss value is a loss value between P sample prediction results determined according to input training samples and a predetermined P sample actual result.
According to another aspect of the embodiments of the present application, there is also provided an apparatus for performing a target operation, including: the system comprises a first acquisition unit, a second acquisition unit and a first control unit, wherein the first acquisition unit is used for acquiring account attribute information of a target account and object attribute information of a target object, the account attribute information is used for representing parameters of N dimensions generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing parameters of M dimensions generated by the target account executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1; the first processing unit is used for inputting account attribute information and object attribute information into a target result prediction model which is obtained through training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer which is more than or equal to 2; the second processing unit inputs the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used for representing the association degree between the P dimension parameters and the target object execution target type event; and the third processing unit is used for executing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
According to still another aspect of the embodiments of the present application, there is also provided a computer-readable storage medium having a computer program stored therein, wherein the computer program is configured to execute the execution method of the above-described target operation when running.
According to yet another aspect of embodiments of the present application, there is also provided a computer program product comprising a computer program/instruction which, when executed by a processor, implements the steps of the above method.
According to still another aspect of the embodiments of the present application, there is also provided an electronic device including a memory in which a computer program is stored, and a processor configured to execute the execution method of the target operation by the computer program.
According to the embodiment provided by the application, account attribute information and object information are input into a target result prediction model obtained through pre-training, P prediction results are obtained through automatic learning of complex features of data in multiple dimensions, then the P prediction results are input into a target weight prediction model obtained through pre-training, P prediction weights are obtained, and operations corresponding to parameters in partial dimensions of the account attribute information are executed according to the P prediction weights. In other words, by automatically learning complex features in multiple dimensions, the impact weights of the features in each dimension on the target account to perform the target type event are predicted, and the target operation is determined. Therefore, different types of operations are executed according to different prediction weights of the accounts, and the technical effect of improving the accuracy of the executed target operations is achieved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application.
FIG. 1 is a schematic illustration of an application scenario of an alternative method of performing a target operation according to an embodiment of the present application;
FIG. 2 is a flow chart of an alternative method of performing a target operation according to an embodiment of the present application;
FIG. 3 is a schematic diagram of an alternative method of performing a target operation according to an embodiment of the present application;
FIG. 4 is a schematic diagram of another alternative method of performing a target operation according to an embodiment of the present application;
FIG. 5 is a schematic diagram of an alternative determination of P predictions according to an embodiment of the present application;
FIG. 6 is an alternative determination of p in accordance with an embodiment of the present application 1 P are predicted results 2 Schematic representation of individual prediction results;
FIG. 7 is a schematic diagram of an alternative execution of an operation corresponding to one of the P parameters in accordance with an embodiment of the present application;
FIG. 8 is a schematic diagram of an alternative implementation of operations corresponding to two of the P parameters according to an embodiment of the present application;
FIG. 9 is a schematic structural view of an alternative target operation performing device according to an embodiment of the present application;
fig. 10 is a schematic structural diagram of an alternative electronic device according to an embodiment of the present application.
Detailed Description
In order to make the present application solution better understood by those skilled in the art, the following description will be made in detail and with reference to the accompanying drawings in the embodiments of the present application, it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that embodiments of the present application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The technical scheme in the embodiment of the application can follow legal rules in the implementation process, and when the operation is executed according to the technical scheme in the embodiment, the used data cannot relate to user privacy, so that the safety of the data is ensured while the operation process is a compliance method.
Noun interpretation:
1. sensor network: an artificial neural network based on a neuron model simulates the working mode of human brain neurons and can perform nonlinear conversion on input data.
2. Elastic network: the regression method combines the characteristics of ridge regression and Lasso regression, and can perform feature selection to prevent overfitting.
3. Marketing in real time: by collecting and analyzing the real-time data, the effect of the marketing campaign is predicted, and the marketing strategy is dynamically adjusted according to the predicted result.
According to one aspect of the embodiments of the present application, there is provided a method for performing a target operation. As an alternative embodiment, the method for performing the above-mentioned target operation may be, but not limited to, applied to the application scenario shown in fig. 1. In an application scenario as shown in fig. 1, terminal device 102 may be, but is not limited to being, in communication with server 106 via network 104, and server 106 may be, but is not limited to being, performing operations on database 108, such as, for example, write data operations or read data operations. The terminal device 102 may include, but is not limited to, a man-machine interaction screen, a processor, and a memory. The man-machine interaction screen described above may be, but is not limited to, account attribute information of a target account for display on the terminal device 102, an operation corresponding to a parameter of a part of the parameters of the P dimensions, and the like. The processor may be, but is not limited to being, configured to perform a corresponding operation in response to the man-machine interaction operation, or generate a corresponding instruction and send the generated instruction to the server 106. The memory is used for storing related processing data, such as the attribute of a target object, namely Sydney, parameters of P dimensions, P prediction results and the like.
As an alternative, the following steps in the execution method of the target operation may be executed on the server 106: step S102, acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing N dimension parameters generated by the target account executing a target type event in a first history duration, the object attribute information is used for representing M dimension parameters generated by the target account executing the target type event in a second history duration, and N and M are positive integers which are larger than or equal to 1; step S104, inputting account attribute information and object attribute information into a target result prediction model obtained by training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2; step S106, inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used for representing the association degree between the P dimension parameters and the target object execution target type event; step S108, according to the P prediction weights, performing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions.
By adopting the mode, the account attribute information and the object attribute information are input into the target result prediction model obtained by pre-training, P prediction results are obtained by automatically learning complex features of data in multiple dimensions, then the P prediction results are input into the target weight prediction model obtained by pre-training, P prediction weights are obtained, and the operation corresponding to the parameters in part of the dimensions in the account attribute information is executed according to the P prediction weights. In other words, by automatically learning complex features in multiple dimensions, the impact weights of the features in each dimension on the target account to perform the target type event are predicted, and the target operation is determined. Therefore, different types of operations are executed according to different prediction weights of the accounts, and the technical effect of improving the accuracy of the executed target operations is achieved.
In order to solve the problem of low accuracy in the execution process of the target operation, an execution method of the target operation is provided in the embodiment of the present application, and fig. 2 is a flowchart of the execution method of the target operation according to the embodiment of the present application, where the flowchart includes the following steps S202 to S210.
Step S202, acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing parameters of N dimensions generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing parameters of M dimensions generated by the target account executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1;
Step S204, inputting account attribute information and object attribute information into a target result prediction model obtained by training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2;
step S206, inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used for representing the association degree between the P dimension parameters and the target object execution target type event;
the larger the value of the prediction weight is, the larger the relevance between the parameter representing the corresponding dimension and the target object execution target type event is; the smaller the value of the prediction weight, the smaller the association between the parameter representing the corresponding dimension and the target object execution target type event.
For example, assuming that p=3 and the values of the 3 predictive weights are 0.9, 0.5, and 0.3, respectively, the correlation between the parameter in the 1 st dimension corresponding to 0.9 (for example, the purchase probability) and the user's purchase of the commodity is the largest, the correlation between the parameter in the 2 nd dimension corresponding to 0.5 (for example, the user liveness) and the user's purchase of the commodity is moderate, and the correlation between the parameter in the 3 rd dimension corresponding to 0.3 (for example, the sales volume of the commodity) and the user's purchase of the commodity is the smallest.
Step S208, according to the P prediction weights, performing operations corresponding to parameters of partial dimensions in the parameters of the P dimensions.
The overall framework of the execution method of the target operation is shown in fig. 3, and account attribute information in N dimensions and object attribute information in M dimensions are input into a target result prediction model to obtain P prediction results; inputting the P prediction results into a target weight prediction model to obtain P prediction weights corresponding to the parameters in the P dimensions one by one; and then performing operations corresponding to the parameters in the P dimensions according to the P prediction weights.
In order to facilitate understanding of the execution method of the above-described target operation, in the following embodiments, the formulation of the marketing strategy is explained as an example.
It is readily understood that in the course of making marketing strategies, factors affecting the marketing strategy include, but are not limited to, consumer consumption data and merchandise data, such as consumer browsing records, purchase records, user liveness, etc., merchandise prices, merchandise sales, etc., over a historical period. Common marketing strategies include enhancing merchandise discounts, personalized advertisement pushes, issuing coupons, and the like.
It should be noted that the first history duration and the second history duration in the step S202 may be the same or different, which is not limited in the embodiment of the present application.
As shown in fig. 4, 3 prediction results of the probability of purchase, the user liveness, the sales amount of the commodity, and the like in the next period of time are obtained by inputting account attribute information in n=3 dimensions (record of browsing the commodity, the number of times of purchasing the commodity, and average price of purchasing) and object attribute information in m=2 dimensions (commodity price and sales amount) into the target result prediction model.
Wherein, the user liveness includes, but is not limited to, the online time length of the user of the online products (such as the commodities in shopping malls) and the frequency of login and interaction, which reflect the preference and acceptance of the user to the products, and the higher the user liveness, the higher the understanding and acceptance of the products or commodities; conversely, the lower the user liveness, the weaker the instruction will understand and accept the product or commodity.
The 3 predicted results are input into a target weight prediction model to obtain a predicted weight vector (0.9,0.5,0.3), which means that the purchase probability of the user has the greatest influence on the marketing effect, and then the activity of the user is the least influence on the marketing effect by the sales amount of the commodity.
The P prediction weights are used to represent the association degree of the P-dimension parameters and the target type event, i.e. the degree to which the P-dimension parameters affect the marketing effect.
Typically, more media resources are devoted to consumers with high purchase probabilities, e.g., higher discounts, more personalized recommendations, etc.; for commodities with higher sales volume, profit can be increased by increasing the price thereof; for users with low liveness, the liveness of the users can be tried to be improved by means of pushing activities, coupons and the like.
If an operation corresponding to one of the 3-dimensional parameters is performed according to the weight vector (0.9,0.5,0.3), for example, an operation corresponding to the highest weighted purchase probability is performed, then the operation is to provide a higher discount to the target account. If an operation corresponding to 2 or 3 of the 3-dimensional parameters is performed, the operation may be a push of a more personalized advertisement, a commodity price reduction campaign, etc., the determination of the performed targeting operation will be described in detail below in connection with particular embodiments.
As an optional example, the above target account performing a target type event on a target object refers to: the target account acquires the target object and transfers the resources with the target resource quantity out of the resource set of the target account.
For example, the target amount is paid by the target account number, thereby purchasing the target commodity.
In the embodiment of the application, the account attribute information of the target account and the object attribute information of the target object may be acquired, but are not limited to, by the following ways:
s11, creating a Flink application program:
an Apache Flink application is created. This typically involves creating a Java or Scala project and introducing a related library of flank into it. In an application, one StreamExecutionEnvironment is created, which is the entry point for all the flank streaming applications.
S12, configuring a data source;
the Flink provides rich data source interfaces such as Kafka, rabbitMQ, files, sockets, etc. Kafka is used to receive real-time marketing-related data from various interfaces and protocols. A KafkaConsumer is set and added to the StreamExecutionEnvironment.
S13, data processing;
after collecting the data, a series of processes are required on the data, including data cleansing, conversion and integration. In Flink, these operations are accomplished by creating a series of transformations. For example, a mapsection is created to clean the data, a flatmapsection is created to convert the data, and a KeyedStream and WindowFunction are created to integrate the data.
S14, outputting data;
after the data is processed, the data needs to be output to a downstream multi-layer sensor network model. The Flink provides a variety of data output interfaces, such as Kafka, rabbitMQ, files, databases, etc. The processed data was sent into Kafka using Kafka producer and then consumed by downstream models.
S15, starting an application program: and calling an execution method of StreamExecutionEnvironment to start the Flink application program. The Flink application will run in the cluster, collecting and processing data in real-time.
That is, various marketing-related data are collected in real time through various interfaces and protocols, and then the data are cleaned, converted and integrated through a real-time data stream processing technology, such as Apache link or Apache storage, so as to obtain the attribute information of the target account and the object attribute information of the target object.
By the method, the related marketing data can be obtained in real time, so that the marketing data is processed in real time, the quick response to market change is ensured, and the technical effect of improving the accuracy of the prediction result is further realized.
According to the embodiment provided by the application, account attribute information and object information are input into a target result prediction model obtained through pre-training, P prediction results are obtained through automatic learning of complex features of data in multiple dimensions, then the P prediction results are input into a target weight prediction model obtained through pre-training, P prediction weights are obtained, and operations corresponding to parameters in partial dimensions of the account attribute information are executed according to the P prediction weights. In other words, by automatically learning complex features in multiple dimensions, the impact weights of the features in each dimension on the target account to perform the target type event are predicted, and the target operation is determined. Therefore, different types of operations are executed according to different prediction weights of the accounts, and the technical effect of improving the accuracy of the executed target operations is achieved.
As an optional example, inputting the account attribute information and the object attribute information into the target result prediction model trained in advance to obtain P prediction results, where the P prediction results include:
inputting account attribute information and object attribute information into a target result prediction model to obtain p 1 P are predicted results 2 P is the predicted result 1 The prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 For positive integers greater than or equal to 1, the P predictors include P 1 P are predicted results 2 P is equal to P as a result of prediction 1 And p is as follows 2 And (3) summing.
As shown in fig. 5, it is assumed that account attribute information includes parameters of n=3 dimensions, such as a record of browsing goods, the number of times of purchasing goods, and an average price of purchasing in a history period; the object attribute information includes parameters in m=2 dimensions, such as commodity price, sales amount of commodity, and the like.
Wherein p is 1 The prediction results comprise purchase probability and user liveness, wherein the purchase probability represents the prediction probability that the target account will purchase the target commodity, and the user liveness represents the prediction value of the attention degree of the target account to the event of purchasing the target commodity.
P 2 The predicted result includes sales of the commodity, which represents sales of the commodity in a subsequent period of time predicted from the commodity price and sales of the commodity in the historic period of time.
As an optional implementation manner, the account attribute information and the object attribute information are input into the target result prediction model to obtain p 1 P are predicted results 2 A plurality of prediction results, comprising:
inputting account attribute information and object attribute information into a target result prediction model to obtain a first prediction result and a second prediction result, wherein the first prediction result is used for representing the probability that a target account executes a target type event on a target object, and the second prediction result is used for representing an executed target classNumber of target objects of type event, p 1 The prediction results comprise a first prediction result, p 2 The prediction results include a second prediction result.
The account attribute information and the object attribute information are input into a target result prediction model, and 3 prediction results are obtained, for example, the 3 prediction results represent the predicted purchase probability of the user, the predicted liveness of the user and the predicted sales amount of the commodity in a next period of time.
That is, the predicted purchase probability of the user is used to represent a predicted value of the probability of the target account to purchase the commodity, and the activity of the user is used to represent a predicted value of the attention of the target account to purchase the commodity in a future period of time; the predicted sales amount of the commodity is used to represent a predicted value of the sales amount of the commodity by the target account number for a future period of time when the commodity is purchased.
After the P prediction results are obtained, inputting the P prediction results into a target weight prediction model to obtain P prediction weights, which specifically includes:
will p 1 P are predicted results 2 Inputting the prediction results into a target weight prediction model to obtain p 1 Sum of prediction weights p 2 A number of predictive weights, wherein the number P of predictive weights includes P 1 Sum of prediction weights p 2 Each prediction weight, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the individual dimensions have a one-to-one correspondence.
For example, as shown in fig. 7, the prediction weight of 0.5 corresponds to the number of times the target account purchases goods in the history period, and the prediction weight of 0.6 corresponds to the activity of the target account in the history period; the predictive weight of 0.3 corresponds to the sales of the target account number for the commodity in the historical time period.
As an optional implementation manner, the account attribute information and the object attribute information are input into the target result prediction model to obtain p 1 P are predicted results 2 A plurality of prediction results, comprising:
dividing account attribute information and object attribute informationRespectively converting the account characterization features and the object characterization features; splicing the account characterization features and the object characterization features to obtain spliced characterization features; sequentially passing the spliced characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain p 1 P are predicted results 2 Predicting results; or alternatively
Commonly converting account attribute information and object attribute information into target characterization features; sequentially passing the target characterization features through a plurality of hidden layers and output layers in a target result prediction model to obtain p 1 P are predicted results 2 And predicting the result.
As shown in fig. 6 (a), before inputting the account attribute information and the object attribute information into the target result prediction model, respectively performing feature conversion on the account attribute information in each dimension to obtain account characterization features; and performing feature conversion on the object attribute information in each dimension to obtain object characterization features.
Then splicing the account representation feature and the object representation feature to obtain a spliced representation feature; sequentially passing the spliced characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain p 1 P are predicted results 2 And predicting the result.
Or as shown in fig. 6 (b), before inputting account attribute information and object attribute information into the target result prediction model, performing feature conversion on the account attribute information and the object attribute information together to obtain target characterization features, and then sequentially passing the target characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain p 1 P are predicted results 2 And predicting the result.
In this embodiment of the present application, the target result prediction model may, but is not limited to, use a deep learning multi-layer perceptron model to predict marketing data collected and processed in real time, where the predicted target is the effect of a marketing campaign, for example, sales, click rate, conversion rate, etc., and specifically includes the following steps:
s21, data preparation: real-time data of the Flink process is used as input data, including user behavior, merchandise sales data, and market environment data, etc. These data need further processing, such as normalization, missing value filling, etc., to be input into the deep learning model (which in turn can be understood as the target outcome prediction model);
s22, model construction: a multi-layer perceptron network model is constructed, which is a fully connected neural network model. A deep learning framework such as TensorFlow or pyrerch is used. The model comprises a plurality of hidden layers and an output layer, wherein each hidden layer uses an activation function, such as ReLU or sigmoid, to increase the nonlinearity of the model;
s23, model training: model training was performed using a loss function shown in the following formula (1):
Where w is the weight of the model, N is the number of data, t i And y i The target value of the i-th sample and the predicted value of the model, respectively.
The specific training process comprises the following steps:
obtaining a training sample set, wherein each training sample in the training sample set comprises sample account attribute information of a sample account and sample object attribute information of a sample object, the sample account attribute information is used for representing sample parameters of P dimensions generated by executing a target type event on the sample account in a first sample history time, and the sample object attribute information is used for representing sample parameters of M dimensions generated by executing the target type event on the sample object in a second sample history time;
and using a training sample set to obtain a result prediction model to be trained, ending training until a loss value corresponding to the result prediction model meets a preset training ending condition, and determining the result prediction model when ending training as a target result prediction model, wherein the loss value is a loss value between P sample prediction results determined according to input training samples and a predetermined P sample actual result.
The loss function is minimized using an optimization algorithm, such as random gradient descent (SGD). During the training process, the model weights are updated continuously so that the value of the loss function is reduced continuously.
S24, model verification and adjustment: during training, the validation set data is continually used to verify the performance of the model to prevent overfitting. Training is stopped when performance on the validation set is no longer improving through Early stop (Early stop) or other strategies. If the performance is not ideal, model parameters such as the number of hidden layers, the number of neurons, the learning rate and the like may need to be adjusted;
s25, model prediction: after model training is completed, new data collected in real time are input into a target result prediction model to be predicted.
The predicted result is fed into the target weight prediction model for analysis and analysis, and is used for dynamically adjusting the marketing strategy.
As an optional implementation manner, the inputting the P prediction results into the target weight prediction model obtained by training in advance to obtain P prediction weights includes:
and inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the target weight prediction model is a linear regression model, and the P prediction weights are optimal solutions of the target function in the linear regression model under the condition that the input of the linear regression model is the P prediction results.
The target weight prediction model relates to a linear regression model, and mainly solves the linear regression problem from the angles of gradient descent and least square, and corresponds to the ridge regression and the Lasso regression respectively for solving the situation that the overfitting and the matrix are irreversible by two norms commonly used in the linear regression.
The objective function of the ridge regression is to add a regularization term on the basis of the general linear regression, so that parameters are as simple as possible while ensuring the best fitting error, and the generalization capability of the model is strong (i.e. knowledge learned from training data is not excessively believed). The regularization generally adopts one norm and two norms, so that the model has generalization, and meanwhile, the irreversible condition in linear regression can be solved. Lasso regression is constrained by a norm that minimizes the parameter non-zero.
The difference between the ridge regression and the Lasso regression is that in the optimization process, the optimal solution is the intersection of the functional contour and the constraint space, and the regular term can be regarded as the constraint space. Where the constraint space of two ranges is a sphere and the constraint space of one range is a square, i.e. two Fan Huide to values where many parameters are close to 0, and where one range is the smallest possible non-zero parameters.
In the embodiment of the application, a target weight prediction model (elastic network model) is used to analyze and analyze the predicted result of the target result prediction model, and the specific steps are as follows:
s31, data preparation: the input data of the target weight prediction model is a prediction result of a target result prediction model (deep learning model), including a purchase probability of the user, sales amount of the commodity, and the like. These data may require further processing, e.g., normalization, for input into the elastic network model;
S32, model construction: the elastic network is an extension of a linear regression model, and can simultaneously realize ridge regression (L2 regularization) and Lasso regression (L1 regularization) to construct an elastic network model.
S33, model training: model training was performed using a loss function as shown in the following equation (2):
where w is the weight of the model, N is the number of data, t i And y i The target value of the i-th sample and the predicted value of the model, alpha is the regularization coefficient and, ρ is a mixing parameter of the regularization term, i w i 1 And w 2 The L1 norm and the L2 norm, respectively.
The loss function is minimized using an optimization algorithm, such as a coordinate descent method.
S34, model verification and adjustment: during training, the performance of the model is continuously checked using the validation set data to prevent overfitting;
if the performance is not ideal, parameters of the elastic network model, such as regularization coefficients and mixing parameters, may need to be adjusted.
S35, model analysis: after model training is completed, the predicted weight output by the elastic network model is analyzed, and key factors influencing the marketing effect are found out, and the factors are used for dynamically adjusting the marketing strategy.
The output of the elastic network model is a weight vector that is a quantized representation of the importance of the input features (i.e., the predicted results of the deep-learning model). These weights can help us understand the impact of various features on marketing effectiveness and dynamically adjust marketing strategies accordingly.
For example, assume that the input features include a user's purchase probability (predicted by the deep learning model), sales of goods, user's liveness, and the like. After the elastic network model training is completed, the obtained weight vector may be [0.5,0.3,0.2]. This means that the user's purchase probability has the greatest effect on the marketing effect, and secondly, the sales of the commodity, and the user's liveness has the least effect on the marketing effect.
Specific analyses and analyses, including:
executing an operation corresponding to an ith dimension parameter in the P dimension parameters under the condition that the ith prediction weight in the P prediction weights is the largest, wherein the ith prediction weight corresponds to the ith dimension parameter, and the importance degree of the ith dimension parameter in the P dimension parameters is the highest, i is a positive integer which is greater than or equal to 1 and less than or equal to P; or alternatively
Where the P prediction weights include P 1 Sum of prediction weights p 2 Predicted weights, and at p 1 The jth prediction weight of the prediction weights is the largest, at p 2 Execution and p in case the kth prediction weight of the prediction weights is the largest 1 Operation corresponding to the parameter of the j-th dimension of the parameters of the dimensions and executing the operation corresponding to p 2 Operations corresponding to parameters of a kth dimension among parameters of the dimensions, wherein the P prediction results comprise P 1 Individual predictive knotsFruit and p 2 P are predicted by 1 The prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 Is a positive integer greater than or equal to 1, P is equal to P 1 And p is as follows 2 Sum, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the dimensions have a one-to-one correspondence, at p 1 The j-th dimension parameter has the highest importance among the parameters of the dimensions, and is p 2 The parameter of the kth dimension among the parameters of the dimensions has the highest importance, j is greater than or equal to 1 and less than or equal to p 1 K is a positive integer greater than or equal to 1 and less than or equal to p 2 Is a positive integer of (a).
As shown in fig. 7, P predictive weights are 0.9, 0.3 and 0.2, respectively, where 1 st predictive weight is 0.9 maximum, indicating that the number of purchases of the commodity of the target account in the history duration has the greatest effect on the marketing effect, then an operation corresponding to the number of purchases in the parameters of P dimensions will be performed, for example, providing a higher commodity discount.
As another example, as shown in FIG. 8, p 1 The prediction weights comprise 0.5 and 0.6, wherein the prediction weight corresponding to the activity of the target account in the history duration is the largest, and p 2 If the prediction weight is 0.7, operations corresponding to the user's liveness and the sales volume of the commodity, such as pushing a preferential activity to the user and raising the commodity price, are performed.
As an optional implementation manner, the performing an operation corresponding to the parameter of the ith dimension in the parameters of the P dimensions includes one of the following:
pushing media resource information to the target account under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object;
controlling the quantity of target resources to be reduced under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the quantity of the target resources is the quantity of resources which are required to be transferred out from a resource set of the target account by the target account for executing the target type event on the target object;
in the case where the parameter of the i-th dimension is a parameter corresponding to the target object, the number of control target resources increases.
And dynamically adjusting marketing strategies according to the analysis result of the target weight prediction model, including but not limited to adjusting commodity prices, preferential strategies, popularization modes and the like.
For example, more resources may be devoted to those users with high purchase probabilities, higher discounts offered, more personalized recommendations, etc.; for commodities with higher sales volume, profit can be increased by increasing the price thereof; for users with low liveness, an attempt may be made to increase their liveness by pushing campaigns, coupons, etc.
The dynamic adjustment strategy based on the model result not only can improve the marketing effect, but also can reflect market change in real time, so that the marketing strategy is more flexible and efficient.
For example, we find that the user's purchase probability has the greatest impact on the marketing campaign, then it is deeply analyzed: the purchase probability of a user refers to the likelihood that the user purchases a certain good or service within a certain period of time. The probability is predicted by the deep learning model according to the behavior data of the user, the historical purchase record and other factors. For any business company, understanding and predicting the purchasing behavior of the user is a central part of its marketing strategy.
If the user's probability of purchase has the greatest impact on the marketing effect, this means that the marketing strategy should be adjusted and optimized based primarily on the user's probability of purchase.
For example, assume that there are three users a, B and C, and the purchase probabilities of 3 users are 0.9,0.4 and 0.2, respectively. This means that the user a has the greatest possibility of purchasing a certain commodity and the user C has the least possibility of purchasing. Then user a may be more focused in formulating the marketing strategy to provide more attractive offers to ensure or increase his probability of purchase. Meanwhile, the user B can be properly concerned, and the purchase probability of the user B is attempted to be improved. For the user C with lower purchase probability, the user C can choose not to do too many marketing activities temporarily so as to save resources. By the method, the service can be helped to more accurately perform resource allocation, the marketing strategy is optimized, and the marketing effect is improved.
As an alternative example, the above-described execution and p 1 The operation corresponding to the parameter of the j-th dimension in the parameters of the dimensions comprises the following steps: pushing media resource information to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object; and/or controlling the reduction of the number of the target resources, wherein the number of the target resources is the number of resources which are required to be transferred out of the resource set of the target account by the target account for executing the target type event on the target object;
The above-mentioned execution and p 2 The operation corresponding to the parameter of the kth dimension in the parameters of the dimensions comprises the following steps: the number of control target resources increases.
As shown in fig. 7, P predicted weights are 0.9, 0.3 and 0.2, respectively, wherein the 1 st predicted weight is 0.9 maximum, which indicates that the number of purchases of the commodity of the target account in the history duration has the greatest influence on the marketing effect, then at least one of the following operations is performed:
(1) Operations corresponding to the number of purchases in the parameters of the P dimensions, such as pushing an advertisement containing merchandise preference information to the target account;
(2) Offering a higher discount (e.g., common member is 9-fold, target account may enjoy 7-fold offers);
(3) The advertisement containing commodity preferential information is pushed to the target account, and simultaneously, higher discounts are provided.
As another example, as shown in FIG. 8, p 2 The predicted weight includes 0.7, then an operation corresponding to the sales amount of the commodity will be performed, for example,and the commodity price is improved, and the profit is increased by improving the commodity price.
According to the description of the embodiments, the account attribute information of the target account and the object attribute Sydney of the target object are automatically collected and processed in real time, the purchasing behavior of the user and the sales condition of the commodity are predicted in real time, and the marketing strategy is adjusted in real time, so that the marketing effect is improved.
It should be noted that, for simplicity of description, the foregoing method embodiments are all expressed as a series of action combinations, but it should be understood by those skilled in the art that the present application is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present application. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required in the present application.
According to another aspect of the embodiments of the present application, there is also provided an apparatus for performing a target operation as shown in fig. 9, the apparatus including:
a first obtaining unit 902, configured to obtain account attribute information of a target account and object attribute information of a target object, where the account attribute information is used to represent parameters of N dimensions generated by the target account executing a target type event in a first history duration, and the object attribute information is used to represent parameters of M dimensions generated by the target object executing the target type event in a second history duration, and N and M are positive integers greater than or equal to 1;
the first processing unit 904 is configured to input account attribute information and object attribute information into a target result prediction model that is trained in advance, to obtain P prediction results, where the P prediction results are used to represent prediction values of P dimensions of parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2;
The second processing unit 906 inputs the P prediction results into a target weight prediction model obtained by training in advance to obtain P prediction weights, where the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used to represent a degree of association between the P dimension parameters and a target object execution target type event;
the third processing unit 908 is configured to perform an operation corresponding to a parameter of a part of the parameters of the P dimensions according to the P prediction weights.
Optionally, the first processing unit 904 includes:
a first processing module for inputting account attribute information and object attribute information into a target result prediction model to obtain p 1 P are predicted results 2 P is the predicted result 1 The prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 For positive integers greater than or equal to 1, the P predictors include P 1 P are predicted results 2 P is equal to P as a result of prediction 1 And p is as follows 2 And (3) summing.
Optionally, the first processing module includes:
a first processing sub-module, configured to input account attribute information and object attribute information into a target result prediction model, to obtain a first prediction result and a second prediction result, where the first prediction result is used to represent a probability that the target account executes a target type event on the target object, and the second prediction result is used to represent the number of target objects of the executed target type event, p 1 The prediction results comprise a first prediction result, p 2 The prediction results include a second prediction result.
Optionally, the first processing module includes:
the second processing sub-module is used for converting the account attribute information and the object attribute information into account characterization features and object characterization features respectively; splicing the account characterization features and the object characterization features to obtain a spliced characterization featureSign of the disease; sequentially passing the spliced characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain p 1 P are predicted results 2 Predicting results; or alternatively
Commonly converting account attribute information and object attribute information into target characterization features; sequentially passing the target characterization features through a plurality of hidden layers and output layers in a target result prediction model to obtain p 1 P are predicted results 2 And predicting the result.
Optionally, the second processing unit 906 includes:
a second processing module for converting p 1 P are predicted results 2 Inputting the prediction results into a target weight prediction model to obtain p 1 Sum of prediction weights p 2 A number of predictive weights, wherein the number P of predictive weights includes P 1 Sum of prediction weights p 2 Each prediction weight, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the individual dimensions have a one-to-one correspondence.
Optionally, the second processing unit 906 includes:
and the third processing module is used for inputting the P prediction results into the target weight prediction model obtained through pre-training to obtain P prediction weights, wherein the target weight prediction model is a linear regression model, and the P prediction weights are optimal solutions of the target function in the linear regression model under the condition that the input of the linear regression model is the P prediction results.
Optionally, the third processing unit 908 includes:
a fourth processing module, configured to perform an operation corresponding to an i-th dimension parameter among the P-th dimension parameters, where the i-th prediction weight corresponds to the i-th dimension parameter, and where the i is a positive integer greater than or equal to 1 and less than or equal to P, where the i-th prediction weight is the highest importance of the i-th dimension parameter; or alternatively
Where the P prediction weights include P 1 Individual prediction weightsAnd p 2 Predicted weights, and at p 1 The jth prediction weight of the prediction weights is the largest, at p 2 Execution and p in case the kth prediction weight of the prediction weights is the largest 1 Operation corresponding to the parameter of the j-th dimension of the parameters of the dimensions and executing the operation corresponding to p 2 Operations corresponding to parameters of a kth dimension among parameters of the dimensions, wherein the P prediction results comprise P 1 P are predicted results 2 P are predicted by 1 The prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of individual dimensions, p 2 The prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 Is a positive integer greater than or equal to 1, P is equal to P 1 And p is as follows 2 Sum, p 1 Prediction weights and p 1 Parameters of each dimension have a one-to-one correspondence, p 2 Prediction weights and p 2 The parameters of the dimensions have a one-to-one correspondence, at p 1 The j-th dimension parameter has the highest importance among the parameters of the dimensions, and is p 2 The parameter of the kth dimension among the parameters of the dimensions has the highest importance, j is greater than or equal to 1 and less than or equal to p 1 K is a positive integer greater than or equal to 1 and less than or equal to p 2 Is a positive integer of (a).
Optionally, the fourth processing module includes:
a third processing sub-module for performing one of:
pushing media resource information to the target account under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object;
controlling the quantity of target resources to be reduced under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the quantity of the target resources is the quantity of resources which are required to be transferred out from a resource set of the target account by the target account for executing the target type event on the target object;
in the case where the parameter of the i-th dimension is a parameter corresponding to the target object, the number of control target resources increases.
Optionally, the fourth processing module includes:
the pushing sub-module is used for pushing media resource information to the target account, wherein the media resource information is used for guiding the target account to execute a target type event on a target object; and/or controlling the reduction of the number of the target resources, wherein the number of the target resources is the number of resources which are required to be transferred out of the resource set of the target account by the target account for executing the target type event on the target object;
And the fourth processing submodule is used for controlling the increase of the number of target resources.
Optionally, the target account executing the target type event on the target object refers to: the target account acquires the target object and transfers the resources with the target resource quantity out of the resource set of the target account.
Optionally, the apparatus further includes:
the second acquisition unit is used for acquiring a training sample set, wherein each training sample in the training sample set comprises sample account attribute information of a sample account and sample object attribute information of a sample object, the sample account attribute information is used for representing sample parameters of N dimensions generated by the sample account executing a target type event in a first sample history time, and the sample object attribute information is used for representing sample parameters of M dimensions generated by the sample object executing the target type event in a second sample history time;
and the fourth processing unit is used for using the training sample set to obtain a result prediction model to be trained, ending training until a loss value corresponding to the result prediction model meets a preset training ending condition, and determining the result prediction model when ending training as a target result prediction model, wherein the loss value is a loss value between P sample prediction results determined according to the input training samples and the predetermined P sample actual results.
The device is applied to a target result prediction model obtained by inputting account attribute information and object attribute information into a pre-training mode, P prediction results are obtained by automatically learning complex features of data in multiple dimensions, then the P prediction results are input into a target weight prediction model obtained by the pre-training mode, P prediction weights are obtained, and operations corresponding to parameters in partial dimensions in the account attribute information are executed according to the P prediction weights. In other words, by automatically learning complex features in multiple dimensions, the impact weights of the features in each dimension on the target account to perform the target type event are predicted, and the target operation is determined. Therefore, different types of operations are executed according to different prediction weights of the accounts, and the technical effect of improving the accuracy of the executed target operations is achieved.
It should be noted that, the embodiment of the target operation execution device may refer to the embodiment of the target operation execution method, which is not described herein.
According to still another aspect of the embodiments of the present application, there is also provided an electronic device for implementing the execution method of the above-mentioned object operation, which may be the terminal device shown in fig. 10. The present embodiment is described taking the electronic device as a background device as an example. As shown in fig. 10, the electronic device comprises a memory 1002 and a processor 1004, the memory 1002 having stored therein a computer program, the processor 1004 being arranged to perform the steps of any of the method embodiments described above by means of the computer program.
Alternatively, in this embodiment, the electronic device may be located in at least one network device of a plurality of network devices of the computer network.
Alternatively, in the present embodiment, the above-described processor may be configured to execute the following steps by a computer program:
s1, acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing N dimension parameters generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing M dimension parameters generated by the target object executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1;
s2, inputting account attribute information and object attribute information into a target result prediction model obtained through pre-training to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2;
s3, inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used for representing the association degree between the P dimension parameters and the target object execution target type event;
And S4, executing the operation corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
Alternatively, as will be appreciated by those skilled in the art, the structure shown in fig. 10 is merely illustrative, and the electronic device may be a smart phone (such as an Android mobile phone, an iOS mobile phone, etc.), a tablet computer, a palmtop computer, a mobile internet device (Mobile Internet Devices, MID), a PAD, or a target terminal. Fig. 10 is not limited to the structure of the electronic device and the electronic apparatus described above. For example, the electronics may also include more or fewer components (e.g., network interfaces, etc.) than shown in FIG. 10, or have a different configuration than shown in FIG. 10.
The memory 1002 may be configured to store software programs and modules, such as program instructions/modules corresponding to the methods and apparatuses for performing the target operations in the embodiments of the present application, and the processor 1004 executes the software programs and modules stored in the memory 1002 to perform various functional applications and data processing, that is, implement the methods for performing the target operations. The memory 1002 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, the memory 1002 may further include memory located remotely from the processor 1004, which may be connected to the terminal via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof. The memory 1002 may be used for storing account attribute information, parameters in N dimensions, and target type events, among others. As an example, as shown in fig. 10, the memory 1002 may be, but is not limited to, a first acquiring unit 902, a first processing unit 904, a second processing unit 906, and a third processing unit 908 in an executing apparatus including the above-described target operation. In addition, other module units in the target operation execution device may be included, but are not limited to, and are not described in detail in this example.
Optionally, the transmission device 1006 is configured to receive or transmit data via a network. Specific examples of the network described above may include wired networks and wireless networks. In one example, the transmission means 1006 includes a network adapter (Network Interface Controller, NIC) that can be connected to other network devices and routers via a network cable to communicate with the internet or a local area network. In one example, the transmission device 1006 is a Radio Frequency (RF) module for communicating with the internet wirelessly.
In addition, the electronic device further includes: a display 1008 for displaying the azimuth indication information of the target sound; and a connection bus 1010 for connecting the respective module parts in the above-described electronic apparatus.
In other embodiments, the target terminal or the server may be a node in a distributed system, where the distributed system may be a blockchain system, and the blockchain system may be a distributed system formed by connecting the plurality of nodes through a network communication. Among them, the nodes may form a Peer-To-Peer (Peer To Peer) network, and any type of computing device, such as a server, a terminal, etc., may become a node in the blockchain system by joining the Peer-To-Peer network.
According to yet another aspect of the present application, a computer program product or computer program is provided, comprising computer instructions stored in a computer readable storage medium. The computer instructions are read from a computer readable storage medium by a processor of a computer device, and executed by the processor, to cause the computer device to perform a method of performing a target operation provided in various alternative implementations of the server verification process described above, wherein the computer program is configured to perform the steps of any of the method embodiments described above when run.
Alternatively, in the present embodiment, the above-described computer-readable storage medium may be configured to store a computer program for executing the steps of:
s1, acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing N dimension parameters generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing M dimension parameters generated by the target object executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1;
S2, inputting account attribute information and object attribute information into a target result prediction model obtained through pre-training to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing a target type event on a target object, and P is a positive integer greater than or equal to 2;
s3, inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with P dimension parameters, and the P prediction weights are used for representing the association degree between the P dimension parameters and the target object execution target type event;
and S4, executing the operation corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
Alternatively, in this embodiment, it will be understood by those skilled in the art that all or part of the steps in the methods of the above embodiments may be performed by a program for instructing the target terminal related hardware, and the program may be stored in a computer readable storage medium, where the storage medium may include: flash disk, read-Only Memory (ROM), random-access Memory (Random Access Memory, RAM), magnetic or optical disk, and the like.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.
The integrated units in the above embodiments may be stored in the above-described computer-readable storage medium if implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions to cause one or more computer devices (which may be personal computers, servers or network devices, etc.) to perform all or part of the steps of the methods of the various embodiments of the present application.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In several embodiments provided in the present application, it should be understood that the disclosed client may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and are merely a logical functional division, and there may be other manners of dividing the apparatus in actual implementation, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The foregoing is merely a preferred embodiment of the present application and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present application and are intended to be comprehended within the scope of the present application.
Claims (15)
1. A method of performing a target operation, comprising:
acquiring account attribute information of a target account and object attribute information of a target object, wherein the account attribute information is used for representing parameters of N dimensions generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing parameters of M dimensions generated by the target object executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1;
Inputting the account attribute information and the object attribute information into a target result prediction model obtained by training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of parameters of P dimensions generated by the target account executing the target type event on the target object, and P is a positive integer greater than or equal to 2;
inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with the parameters of the P dimensions, and the P prediction weights are used for representing the association degree between the parameters of the P dimensions and the target object executing the target type event;
and executing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
2. The method according to claim 1, wherein said inputting the account attribute information and the object attribute information into a pre-trained target result prediction model to obtain P prediction results includes:
inputting the account attribute information and the object attribute information into the target result prediction model to obtain p 1 P are predicted results 2 A prediction result, wherein the p 1 A prediction result is used for representing p related to the target account, which is generated by the target account executing the target type event on the target object 1 Predicted values of parameters of the individual dimensions, p 2 A prediction result is used for representing p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 Being a positive integer greater than or equal to 1, the P predictors including the P 1 The prediction result and the p 2 P is equal to P as a result of prediction 1 And p is as follows 2 And (3) summing.
3. The method of claim 2, wherein the inputting the account attribute information and the object attribute information into the target result prediction model yields p 1 P are predicted results 2 A plurality of prediction results, comprising:
inputting the account attribute information and the object attribute information into the target result prediction model to obtain a first prediction result and a second prediction result, wherein the first prediction result is used for representing the probability that the target account executes the target type event on the target object, and the second prediction result is used for representing the probability that the target account executes the target type event on the target object Results are used to represent the number of the target objects that are to be subjected to the target type event, the p 1 The prediction results comprise the first prediction result, the p 2 The prediction results include the second prediction result.
4. The method of claim 2, wherein the inputting the account attribute information and the object attribute information into the target result prediction model yields p 1 P are predicted results 2 A plurality of prediction results, comprising:
converting the account attribute information and the object attribute information into account characterization features and object characterization features respectively; splicing the account characterization features and the object characterization features to obtain spliced characterization features; sequentially passing the spliced characterization features through a plurality of hidden layers and output layers in the target result prediction model to obtain the p 1 The prediction result and the p 2 Predicting results; or alternatively
Jointly converting the account attribute information and the object attribute information into target characterization features; sequentially passing the target characterization features through the plurality of hidden layers and the output layer in the target result prediction model to obtain the p 1 The prediction result and the p 2 And predicting the result.
5. The method according to claim 2, wherein inputting the P prediction results into a pre-trained target weight prediction model to obtain P prediction weights comprises:
the p is set 1 The prediction result and the p 2 Inputting the predicted results into the target weight prediction model to obtain p 1 Sum of prediction weights p 2 A number of predictive weights, wherein the number P of predictive weights includes the number P of 1 Individual prediction weights and the p 2 A predictive weight, p 1 A prediction weight and the p 1 The parameters of the dimensions have a one-to-one correspondence, the p 2 A prediction weight and the p 2 Of dimensions ofThe parameters have a one-to-one correspondence.
6. The method according to claim 1, wherein inputting the P prediction results into a pre-trained target weight prediction model to obtain P prediction weights comprises:
and inputting the P prediction results into a target weight prediction model obtained by pre-training to obtain P prediction weights, wherein the target weight prediction model is a linear regression model, and the P prediction weights are optimal solutions of target functions in the linear regression model under the condition that the input of the linear regression model is the P prediction results.
7. The method of claim 1, wherein the performing an operation corresponding to a parameter of a partial dimension among the parameters of the P dimensions according to the P prediction weights comprises:
executing an operation corresponding to an ith dimension parameter in the P dimension parameters under the condition that the ith prediction weight in the P dimension parameters is the largest, wherein the ith prediction weight corresponds to the ith dimension parameter, and the importance degree of the ith dimension parameter in the P dimension parameters is the highest, i is a positive integer which is greater than or equal to 1 and less than or equal to P; or alternatively
Where the P prediction weights include P 1 Sum of prediction weights p 2 Predicted weights, and at p 1 The j-th prediction weight of the prediction weights is the largest, at the p 2 Execution and p in case the kth prediction weight of the prediction weights is the largest 1 Operation corresponding to the parameter of the j-th dimension of the parameters of the dimensions and executing the operation corresponding to p 2 Operations corresponding to parameters of a kth dimension among parameters of a plurality of dimensions, wherein the P prediction results comprise P 1 P are predicted results 2 A prediction result, p 1 A prediction result is used for representing the target account related to the target account generated by the target account executing the target type event on the target object p 1 Predicted values of parameters of the individual dimensions, p 2 A prediction result is used for representing the p related to the target object generated by the target account executing the target type event on the target object 2 Predicted values of parameters of individual dimensions, p 1 And p 2 Is a positive integer greater than or equal to 1, P is equal to P 1 And p is as follows 2 Sum of said p 1 A prediction weight and the p 1 The parameters of the dimensions have a one-to-one correspondence, the p 2 A prediction weight and the p 2 The parameters of the dimensions have a one-to-one correspondence, and the parameters of the dimensions have a one-to-one correspondence with the parameters of the dimensions 1 The parameter of the j-th dimension has the highest importance among the parameters of the p-th dimension 2 The parameter of the kth dimension has the highest importance, j is greater than or equal to 1 and less than or equal to p 1 K is a positive integer greater than or equal to 1 and less than or equal to p 2 Is a positive integer of (a).
8. The method of claim 7, wherein the performing an operation corresponding to the i-th dimension of the P-dimension parameters comprises one of:
pushing media resource information to the target account under the condition that the parameter of the ith dimension is a parameter corresponding to the target account, wherein the media resource information is used for guiding the target account to execute the target type event on the target object;
Controlling a target resource quantity to be reduced when the parameter of the ith dimension is a parameter corresponding to the target account, wherein the target resource quantity is the quantity of resources which are required to be transferred out from a resource set of the target account by the target account for executing the target type event on the target object;
and controlling the number of the target resources to increase in the case that the parameter of the ith dimension is a parameter corresponding to the target object.
9. The method of claim 7, wherein the step of determining the position of the probe is performed,
the execution and p 1 The operation corresponding to the parameter of the j-th dimension in the parameters of the dimensions comprises the following steps: pushing media resource information to the target account, wherein the media resource information is used for guiding the target account to execute the target type event on the target object; and/or controlling a reduction in a target resource quantity, wherein the target resource quantity is a quantity of resources transferred out from a resource set of the target account required by the target account to execute the target type event on the target object;
the execution and p 2 The operation corresponding to the parameter of the kth dimension in the parameters of the dimensions comprises the following steps: and controlling the target resource quantity to increase.
10. The method of any one of claims 1 to 9, wherein the target account performing the target type event on the target object means: and the target account acquires the target object and transfers the resources with the target resource quantity out of the resource set of the target account.
11. The method according to any one of claims 1 to 9, further comprising:
obtaining a training sample set, wherein each training sample in the training sample set comprises sample account attribute information of a sample account and sample object attribute information of a sample object, the sample account attribute information is used for representing sample parameters of the N dimensions generated by the sample account executing the target type event in a first sample history time period, and the sample object attribute information is used for representing sample parameters of the M dimensions generated by the sample object executing the target type event in a second sample history time period;
and using the training sample set to obtain a result prediction model to be trained until a loss value corresponding to the result prediction model meets a preset training ending condition, ending training, and determining the result prediction model when the training is ended as the target result prediction model, wherein the loss value is a loss value between P sample prediction results determined according to the input training samples and P sample actual results determined in advance.
12. An execution apparatus of a target operation, comprising:
the system comprises a first acquisition unit, a second acquisition unit and a first processing unit, wherein the first acquisition unit is used for acquiring account attribute information of a target account and object attribute information of a target object, the account attribute information is used for representing parameters of N dimensions generated by the target account executing a target type event in a first historical time, the object attribute information is used for representing parameters of M dimensions generated by the target object executing the target type event in a second historical time, and N and M are positive integers which are larger than or equal to 1;
the first processing unit is used for inputting the account attribute information and the object attribute information into a target result prediction model which is obtained through training in advance to obtain P prediction results, wherein the P prediction results are used for representing the prediction values of P dimension parameters generated by the target account executing the target type event on the target object, and P is a positive integer which is more than or equal to 2;
the second processing unit inputs the P prediction results into a target weight prediction model obtained through pre-training to obtain P prediction weights, wherein the P prediction weights have a one-to-one correspondence with the parameters of the P dimensions, and the P prediction weights are used for representing the association degree between the parameters of the P dimensions and the target object executing the target type event;
And the third processing unit is used for executing operations corresponding to the parameters of partial dimensions in the parameters of the P dimensions according to the P prediction weights.
13. A computer-readable storage medium, characterized in that the computer-readable storage medium comprises a stored program, wherein the program is executable by a terminal device or a computer to perform the method of any one of claims 1 to 11.
14. A computer program product comprising computer programs/instructions which, when executed by a processor, implement the steps of the method of any one of claims 1 to 11.
15. An electronic device comprising a memory and a processor, characterized in that the memory has stored therein a computer program, the processor being arranged to execute the method according to any of the claims 1 to 11 by means of the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311323401.6A CN117390540A (en) | 2023-10-12 | 2023-10-12 | Target operation execution method and device, storage medium and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311323401.6A CN117390540A (en) | 2023-10-12 | 2023-10-12 | Target operation execution method and device, storage medium and electronic equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117390540A true CN117390540A (en) | 2024-01-12 |
Family
ID=89462349
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311323401.6A Pending CN117390540A (en) | 2023-10-12 | 2023-10-12 | Target operation execution method and device, storage medium and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117390540A (en) |
-
2023
- 2023-10-12 CN CN202311323401.6A patent/CN117390540A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11551239B2 (en) | Characterizing and modifying user experience of computing environments based on behavior logs | |
US20190043081A1 (en) | Management of the display of online ad content consistent with one or more performance objectives for a webpage and/or website | |
EP4181026A1 (en) | Recommendation model training method and apparatus, recommendation method and apparatus, and computer-readable medium | |
US10354184B1 (en) | Joint modeling of user behavior | |
CN110008973B (en) | Model training method, method and device for determining target user based on model | |
KR102049777B1 (en) | Item recommendation method and apparatus based on user behavior | |
CN111242310A (en) | Feature validity evaluation method and device, electronic equipment and storage medium | |
US20220108334A1 (en) | Inferring unobserved event probabilities | |
Xu et al. | Entrepreneurial learning and disincentives in crowdfunding markets | |
CN110598120A (en) | Behavior data based financing recommendation method, device and equipment | |
CN114417174B (en) | Content recommendation method, device, equipment and computer storage medium | |
US20230351435A1 (en) | Training an artificial intelligence engine for most appropriate actions | |
CN113344647B (en) | Information recommendation method and device | |
KR102311107B1 (en) | Customer exit prevention method that provides a solution to prevent customer from leaving the deep learning solution platform that automatically creates a deep learning model, perfomred by a system to avoid leaving the customer | |
CN112269943B (en) | Information recommendation system and method | |
CN117853158A (en) | Enterprise operation data prediction method and device based on dynamic quantity benefit analysis | |
WO2002079901A2 (en) | Customer preference system | |
JP2022534721A (en) | Predictive modeling and analytics to process and distribute data traffic | |
US11983743B2 (en) | Training an artificial intelligence engine for generating models to provide targeted actions | |
CN109636432A (en) | The project selection method and device that computer executes | |
CN117390540A (en) | Target operation execution method and device, storage medium and electronic equipment | |
CN114429384A (en) | Intelligent product recommendation method and system based on e-commerce platform | |
Chiong et al. | Split-second decision-making in the field: Response times in mobile advertising | |
CN115222177A (en) | Service data processing method and device, computer equipment and storage medium | |
Kumar et al. | Uplift modeling: Predicting incremental gains |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication |