CN116861085A - Application ordering method, device, computer equipment and storage medium - Google Patents

Application ordering method, device, computer equipment and storage medium Download PDF

Info

Publication number
CN116861085A
CN116861085A CN202310825969.1A CN202310825969A CN116861085A CN 116861085 A CN116861085 A CN 116861085A CN 202310825969 A CN202310825969 A CN 202310825969A CN 116861085 A CN116861085 A CN 116861085A
Authority
CN
China
Prior art keywords
application
target
sample
user
characteristic 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
Application number
CN202310825969.1A
Other languages
Chinese (zh)
Inventor
王润泽
陈雪峰
徐赟佳
刘梓豪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Industrial and Commercial Bank of China Ltd ICBC
Original Assignee
Industrial and Commercial Bank of China Ltd ICBC
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Industrial and Commercial Bank of China Ltd ICBC filed Critical Industrial and Commercial Bank of China Ltd ICBC
Priority to CN202310825969.1A priority Critical patent/CN116861085A/en
Publication of CN116861085A publication Critical patent/CN116861085A/en
Pending legal-status Critical Current

Links

Classifications

    • 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
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
    • G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
    • G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
    • G06F21/6254—Protecting personal data, e.g. for financial or medical purposes by anonymising data, e.g. decorrelating personal data from the owner's identification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
    • G06F3/0481—Interaction techniques based on graphical user interfaces [GUI] based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance
    • G06F3/04817—Interaction techniques based on graphical user interfaces [GUI] based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance using icons
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
    • G06F3/0481—Interaction techniques based on graphical user interfaces [GUI] based on specific properties of the displayed interaction object or a metaphor-based environment, e.g. interaction with desktop elements like windows or icons, or assisted by a cursor's changing behaviour or appearance
    • G06F3/0482—Interaction with lists of selectable items, e.g. menus
    • 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
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Human Computer Interaction (AREA)
  • Bioethics (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Computer Security & Cryptography (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present application relates to the field of artificial intelligence, and in particular, to an application ranking method, apparatus, computer device, and storage medium. The method comprises the following steps: acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user; inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user; and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user. The application can meet the sorting requirements of different user applications.

Description

Application ordering method, device, computer equipment and storage medium
Technical Field
The present application relates to the field of artificial intelligence, and in particular, to an application ranking method, apparatus, computer device, and storage medium.
Background
At present, for more deeply decoupling each functional module of mobile terminal office mobile phone application, many office applications decouple functions into a mobile platform with a trunk and a plurality of application modules with different functions, each application module completes the internal functional logic of the mobile terminal office platform, and meanwhile, the mobile terminal office platform is used as an office portal to provide a quick call entry for each accessed application module, so that the coupling degree between application functions is lower.
In the conventional technology, when the background server of the mobile terminal office platform sorts the icons of a plurality of applications, the icons are generally sorted according to the total number of clicks counted by the previous buried points, or the sorting order is comprehensively considered according to the sequence of accessing each application by the mobile terminal office platform, after the icon order is generated, the background server sends a unified icon sorting sequence to each mobile terminal, so that the application icons on the mobile terminal are sorted according to the icon sorting sequence.
However, when different people use the mobile terminal, the use authority and the use frequency of different people for each application are different, so that the unified icon ordering sequence cannot meet the use requirement of each user; therefore, improvements are needed.
Disclosure of Invention
In view of the foregoing, it is desirable to provide an application ranking method, apparatus, computer device and storage medium that can meet the ranking needs of different users.
In a first aspect, the present application provides an application ranking method, the method comprising:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
Inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user.
In one embodiment, inputting the target feature information into the ranking model to obtain a target ranking result for each application installed in a terminal held by a target user, including:
encoding the target characteristic information to obtain an application click sequence of the target user;
inputting the application click sequence into the ordering model to obtain probability values of all applications installed in the terminal held by the target user;
and generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
In one embodiment, the ordering model includes an embedded layer, a convolutional network layer, and a fully-connected layer; inputting the application click sequence into the ordering model to obtain probability values of each application installed in the terminal held by the target user, wherein the method comprises the following steps:
inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence;
Inputting the initial feature representation into a convolution network layer for depth feature extraction to obtain a depth feature representation;
and inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In one embodiment, the method further comprises:
acquiring a training sample set from a data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user;
and training the initial model by adopting a training sample set to obtain a sequencing model.
In one embodiment, obtaining a training sample set from a data lake corresponding to a buried point acquisition server includes:
candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets;
and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
In one embodiment, training the initial model with a training sample set to obtain a ranking model includes:
determining sample characteristic information corresponding to the application with the application use record times larger than a preset threshold value in the training sample set as a positive sample;
determining sample characteristic information corresponding to the application of which the application use record frequency is less than or equal to a preset threshold value in the training sample set as a negative sample;
and training the initial model by adopting a Focal loss function and positive and negative samples in a training sample set to obtain a sorting model.
In a second aspect, the present application also provides an application ranking apparatus, which includes:
the acquisition module is used for acquiring target characteristic information of a target user sent by the application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
the sequencing module is used for inputting the target characteristic information into the sequencing model to obtain a target sequencing result of each application installed in the terminal held by the target user;
and the feedback module is used for sending a return notice comprising the target sorting result to the application management server, and the return notice is used for indicating the application management server to forward the target sorting result to the terminal held by the target user. .
In a third aspect, the present application also provides a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user. .
In a fourth aspect, the present application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
Inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user.
In a fifth aspect, the application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user.
The application ordering method, the device, the computer equipment and the storage medium are characterized in that an application management server is used as a front-end server and is communicated with a user terminal; the data processing server is used as a back-end server and is only used for data processing of the ordering model, is not directly communicated with the user terminal, and the data security is improved by the protection effect of the ordering model in the data processing server through the isolation of the application management server; the application management server is used as a transfer party of the data flow, and after the target characteristic information of the target user is obtained, the data processing server generates a targeted application ordering result for the target user through an ordering model, so that a personalized application ordering mode can be provided for each user.
Drawings
FIG. 1 is an application environment diagram of an application ordering method in one embodiment;
FIG. 2 is a flow diagram of an application ordering method in one embodiment;
FIG. 3 is a flow diagram of training a ranking model in one embodiment;
FIG. 4 is a flow diagram of processing feature information of a target user in one embodiment;
FIG. 5 is a flow diagram of a ranking model process flow in one embodiment;
FIG. 6 is a flow chart of an application ordering method in another embodiment;
FIG. 7 is a block diagram of an embodiment of an application ordering apparatus;
fig. 8 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
At present, for more deeply decoupling each functional module of mobile terminal office mobile phone application, many office applications decouple functions into a mobile platform with a trunk and a plurality of application modules with different functions, each application module completes the internal functional logic of the mobile terminal office platform, and meanwhile, the mobile terminal office platform is used as an office portal to provide a quick call entry for each accessed application module, so that the coupling degree between application functions is lower. In the conventional technology, when a background server of a mobile terminal office platform sorts icons of a plurality of applications, the icons are generally sorted according to the total number of clicks counted by previous buried points, or are comprehensively considered in modes such as access sequence and the like. However, the access application icon sequence of each user is relatively fixed, and cannot meet the use requirement of each user, so improvement is needed.
The application ordering method provided by the embodiment of the application is realized based on an application ordering system, and as shown in fig. 1, the application ordering system buries a point acquisition server, a data lake, a data processing server, a database and an application management server. The embedded point acquisition server and the application management server are used as front-end servers and are communicated with the user terminal; the data processing server acts as a back-end server and does not communicate directly with the user terminal. The embedded point acquisition server is used for acquiring embedded point data of each user terminal in real time, writing the embedded point data into the data lake in batches, wherein the embedded point data is characteristic information generated when a user uses (clicks) each application; the data lake is used for storing buried point data; the database is used for storing buried point data; the above steps complete the data flow of (1), (2) and (3) in the above diagram.
The data processing server is used for acquiring buried point data in the data lake and training the initial model based on the buried point data to obtain the ordering model. After the training of the ranking model is completed, when the user terminal needs to make recommendation of application ranking, a request is initiated to the application management server (fig. 1 (4)), and the application management server obtains the required feature information (namely the dimension trained above) through the database query related data. After the processing is completed, the characteristic information is used as an input parameter to be provided to the data processing server, the data processing server generates a sequencing result by calling a sequencing model, the sequencing result is a sequencing result of each application installed on the user terminal, and after the data processing service returns the sequencing result (a sequencing sequence corresponding to each application ID) to the application management server, the user terminal performs sequencing according to the sequencing result. The embedded point acquisition server, the data processing server and the application management server can be realized by independent servers or a server cluster formed by a plurality of servers.
In one embodiment, as shown in fig. 2, an application ordering method is provided, and the method is applied to the data processing server in fig. 1 for illustration, and includes the following steps:
s201, obtaining target characteristic information of a target user sent by an application management server.
The target feature information is acquired by the application management server based on the application ordering request of the target user.
Specifically, when the target user sends an application ordering request through the terminal, the application management server receives the application ordering request ((4) in fig. 1), the application management server queries related data through the database to obtain target feature information of the target user, and then the application management server sends the target feature information of the target user to the data processing server.
It will be appreciated that the target feature information may be related data generated by a single click of any application by the target user, including fields such as a user ID, a click application ID, and a click time.
S202, inputting the target characteristic information into the sequencing model to obtain a target sequencing result of each application installed in the terminal held by the target user.
The ordering model can be a neural network model, and the data processing server is used for training the initial model to obtain the ordering model; and uses the ranking model when application ranking is required.
Specifically, the ranking model processes the target feature information and outputs a target ranking result for each application of the target user, so that the terminal corresponding to the target user ranks the icons of each application according to the target ranking result.
S203, sending a return notification comprising the target ordering result to the application management server.
The return notice is used for indicating the application management server to forward the target sorting result to the terminal held by the target user. And transmitting the target sorting result to the application management server, so that the application management server forwards the target sorting result to a terminal held by the target user.
Specifically, when the target sorting result is returned, the application management server continues to serve as a transfer party, that is, the data processing server sends the target sorting result to the application management server, and then the application management server forwards the target sorting result to a terminal held by the target user.
In the application ordering method, an application management server is used as a front-end server and is communicated with a user terminal; the data processing server is used as a back-end server and is only used for data processing of the ordering model, is not directly communicated with the user terminal, and the data security is improved by the protection effect of the ordering model in the data processing server through the isolation of the application management server; the application management server is used as a transfer party of the data flow, and after the target characteristic information of the target user is obtained, the data processing server generates a targeted application ordering result for the target user through an ordering model, so that a personalized application ordering mode can be provided for each user.
As shown in fig. 3, the application ranking method further includes: s301, acquiring a training sample set from a data lake corresponding to the buried point acquisition server.
The training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user.
It may be appreciated that the sample characteristic information of the sample user includes relevant data generated by the sample user during a history period when clicking on each application, including information such as user ID, click application ID, click time, etc.
Further, the sample characteristic information of the sample user is obtained by desensitizing the original characteristic information of the sample user by the buried point acquisition server, and the sample characteristic information at least comprises desensitized identity information of the sample user and application use records of the sample user; the application use record comprises information such as a clicking application ID, clicking time and the like.
Specifically, the desensitization processing is performed on the original characteristic information of the sample user to obtain desensitization identity information of the sample user, which comprises the following steps: the user ID is converted into user category information, and the user ID data is not reserved, so that the personal privacy of the user is protected. By way of example, the user category information may be a department to which the user belongs, an age group to which the user belongs, and so forth.
In this embodiment, the application usage records of the users in the same category are analyzed to obtain the application ranking result corresponding to the user category as a whole, so that the ranking result has a more comprehensive coverage.
It can be appreciated that after the desensitized non-secret related data is sorted, the desensitized non-secret related data is converted into a line of data and is pushed into a data lake in near real time, and the click application ID, the click time and the like generated by each click correspond to the line of data in the data lake.
Meanwhile, the embedded point acquisition server converts the embedded point acquisition server into a line of data and stores the line of data into a database through quasi-real-time pushing, in this case, after receiving an application ordering request, the application server analyzes the user ID sending the application ordering request, and queries the user identity characteristic information corresponding to the user after desensitization processing from the database based on the user ID.
Optionally, obtaining a training sample set from a data lake corresponding to the buried point acquisition server includes: candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets; and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
Specifically, the life cycle of the data lake-entering data can be set according to the requirements (such as average access frequency), and the data can be cleaned regularly; the time when each row of data is added to the data lake is taken as a candidate sample set of application usage records with updated time falling into a set period, and the candidate sample set is taken as a training sample set. For example, the set period may be a period of a preset length of time before the current time, such as 1 month before the current time.
S302, training an initial model by using a training sample set to obtain a sorting model.
Specifically, desensitization identity information of a sample user and application use records of the sample user are input into an initial model, the initial model is trained until a loss function of the initial model reaches a corresponding threshold value, so that training of the initial model is completed, and a sequencing model is obtained.
In one embodiment, since part of the applications are only opened to some authorized users, when training the initial model by using a training sample set of random sampling, there may be a problem of sample imbalance; for example, a certain application has only a small number of users, the amount of the historical user data generated correspondingly is also greatly reduced compared with that of other applications, and for example, the application with part of later access but higher use frequency is ranked later, so that the application may not be displayed in a limited page.
It will be understood that the sample (class) sample Imbalance (class-immalance) refers to a situation that the number of training samples of different classes in the classification task is very different, and in general, the sample class Ratio (immalance Ratio) (most classes vs. minority class) is significantly greater than 1:1 (e.g. 4:1) can be classified as a problem of sample Imbalance. The imbalance solution can be summarized as: samples of different classes are made relatively uniform for Loss (or gradient) contributions in model learning by some method. The core idea of Focal loss is to add different weights of categories and weights of difficult (high-loss) samples based on a cross entropy loss function (CE) to improve model learning effect. Therefore, in order to reduce the influence of sample imbalance, the loss function of the initial recommended network model in the application is Focal loss, and the following formula (1) is shown:
FL(p t )＝-α t (1-p t ) γ log(p t ) (1)
by a coefficient alpha t When label=1, α t =α; when label=0, α t =1- α; wherein, alpha ranges from 0 to 1, and the contribution of positive and negative samples to loss is controlled by setting alpha; (1-p) t ) γ Is the modulation factor. When p is t Towards 0, the modulation factor towards 1, the positive and negative samples contribute significantly to the total loss. When p is t Towards 1, the modulation factor towards 0 contributes little to the total loss. Modulation factor change is achieved by adjusting γ. The positive samples in this embodiment may be sample feature information corresponding to applications whose number of uses is greater than the corresponding threshold, and the negative samples may be sample feature information corresponding to applications whose number of uses is less than or equal to the corresponding threshold.
Specifically, sample characteristic information corresponding to applications with the application use record times greater than a preset threshold value in a training sample set is determined to be a positive sample; determining sample characteristic information corresponding to the application of which the application use record frequency is less than or equal to a preset threshold value in the training sample set as a negative sample; and training the initial model by adopting a Focal loss function and positive and negative samples in a training sample set to obtain a sorting model.
It can be appreciated that two important properties of Focal loss: (1) When a sample is misclassified, pt is small and therefore the modulation factor tends to be 1, i.e. there is little change from the original loss. When pt tends to be 1 (where the classification is correct and is a sample easy to classify), the modulation factor tends to be 0, i.e. the contribution to the total loss is small; (2) When γ=0, focal loss is a conventional cross entropy loss, and when γ increases, the modulation factor increases. Thus, focal loss is a suitable function to measure the contribution of difficult-to-classify and easy-to-classify samples to total loss.
In this embodiment, the protection of the user identity is achieved by desensitizing the sample characteristic information of the sample user.
As shown in fig. 4, this embodiment provides an alternative way to input the target feature information into the ranking model to obtain the target ranking result of each application installed in the terminal held by the target user, that is, a way to refine S202. The specific implementation process can comprise the following steps:
s401, encoding the target characteristic information to obtain an application click sequence of the target user.
Specifically, the IDs of the icons of the clicked applications in the target feature information are ordered according to the click time, and an application click sequence of the target user is generated, for example, N time points correspond to N single click data, and each single click data corresponds to an icon ID of an application;
then, the application icon IDs are coded and represented by using a single-hot coding technology, and for each application icon ID, the corresponding position of the application icon ID is marked as 1, and the rest positions are marked as 0. For example, in this embodiment, there are 250 applications, and the click vector corresponding to each single click data is a 250-dimensional vector, that is, the generated application click sequence of the target user is N250-dimensional vectors.
S402, inputting the application click sequence into the sorting model to obtain probability values of all applications installed in the terminal held by the target user.
The ordering model comprises an embedded layer, a convolution network layer and a full connection layer.
Specifically, as shown in fig. 5, inputting the application click sequence into the ranking model to obtain probability values of each application installed in the terminal held by the target user, including:
s501, inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence.
It will be appreciated that since the one-hot encoding results in extremely high and sparse training data dimensions, a dimension reduction operation is required, and the embedded layer in this embodiment includes a weight matrix of 250×80 dimensions.
Specifically, the embedding layer uses a weight matrix to perform dimension reduction, is used for mapping a high-dimension sparse vector to a low-dimension dense vector, and uses a nonlinear activation function tanh to learn a nonlinear relation between features, so as to extract feature representation information of an application click sequence of a target user. The embedded layer may be viewed as dimension-down encoding the target user's application click sequence, with the encoded rules automatically generated by training weights in the network model.
S502, inputting the initial feature representation into a convolution network layer to perform depth feature extraction, and obtaining a depth feature representation.
The convolution network layer is a one-dimensional convolution network, the one-dimensional convolution network carries out convolution calculation by adopting 20 groups of one-dimensional convolution kernels with the convolution kernel lengths of 1,2,3 and 4, the activation function is tanh, and a 20-dimensional output sequence is generated after convolution; and then splicing the output results of different convolution kernels by using a splicing layer to generate an 80-dimensional time sequence, wherein the length of the output sequence is the length of the input sequence.
Specifically, the convolution network layer comprises 20 groups of convolution kernels with the length of 1, 20 groups of convolution kernels with the length of 2, 20 groups of convolution kernels with the length of 3 and 20 groups of convolution kernels with the length of 4; specifically, after the conversion sequence is input to the convolutional network layer, a plurality of unit sequences are obtained, including: convolving the conversion sequence with 20 groups of convolution kernels with the convolution kernel length of 1 to obtain N first unit sequences with 20 dimensions; convolving the conversion sequence with 20 groups of convolution kernels with the length of 2 to obtain N second unit sequences with 20 dimensions; convolving the conversion sequence with 20 groups of convolution kernels with the convolution kernel length of 3 to obtain N third unit sequences with 20 dimensions; convolving the conversion sequence with 20 groups of convolution kernels with the convolution kernel length of 4 to obtain N fourth unit sequences with 20 dimensions; and splicing the unit sequences according to the splicing layer to obtain N80-dimensional splicing sequences.
And S503, inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In this embodiment, by designing a full connection layer to restore the output to the access application ID code of the access application dimension number, the activation function is softmax, and the recommended access application list is generated for performing access application recommendation ordering.
Specifically, the activation function of the fully-connected layer uses softmax, which is generally used in a task of performing multiple classifications, and can output a plurality of neurons respectively, wherein the neurons are mapped to intervals of (0, 1) respectively, the sum of all the outputs is 1, and the property of probability is met, so that each output can be regarded as the probability of corresponding classification, the total classification number of the fully-connected layer is the total number of application icon IDs, namely 250 classes, and each classification corresponds to a probability value respectively.
S403, generating a target ordering result of each application installed in the terminal held by the target user according to the probability value of each application.
Specifically, the scalar result is generated in the order from high to low according to the respective probability values of the 250 classifications.
In the embodiment, the coding mode is adopted, so that the calculated amount of the sequencing model is reduced, and meanwhile, the model processing precision is ensured.
Illustratively, on the basis of the above embodiments, this embodiment provides an alternative example of an application ranking method. As shown in fig. 6, the specific implementation process includes:
s601, obtaining target characteristic information of a target user sent by an application management server.
The target feature information is acquired by the application management server based on the application ordering request of the target user.
S602, encoding the target characteristic information to obtain an application click sequence of the target user.
S603, inputting the application click sequence into an embedded layer of the sequencing model for dimension reduction processing, and obtaining an initial characteristic representation corresponding to the application click sequence.
Specifically, a training sample set is obtained from a data lake corresponding to a buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user; and training the initial model by adopting a training sample set to obtain a sequencing model.
Further, a candidate sample set of the application use record, of which the updating moment in the data lake corresponding to the buried point acquisition server meets the set period, is used as a training sample set; and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
S604, inputting the initial feature representation into a convolutional network layer of the sequencing model to extract depth features, and obtaining the depth feature representation.
S605, the depth characteristic representation is input to a full connection layer of the sequencing model, and probability values of all applications installed in a terminal held by a target user are obtained.
S606, generating a target ordering result of each application installed in the terminal held by the target user according to the probability value of each application.
S607, a return notification including the target sort result is sent to the application management server.
The return notice is used for indicating the application management server to forward the target sorting result to the terminal held by the target user. .
The specific process of S601 to S607 may refer to the description of the above method embodiment, and its implementation principle and technical effect are similar, and will not be described herein.
In the embodiment, the application ordering method and the corresponding application ordering system solve the problems that the mobile terminal office portal multi-access application single page has limited displayable icons, the user has different use frequencies for different access applications, and the different access applications have different open ranges. The icon can be displayed according to the more scientific calculation that the user is likely to use more frequently according to the use condition of the user. The insufficient space in other icon pages can not be displayed and hidden. And further, the user experience is improved in the limited page space. And the usability of the mobile terminal product is improved.
It should be understood that, although the steps in the flowcharts related to the above embodiments are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides an application ordering device for realizing the above related application ordering method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in one or more embodiments of the application ranking device provided below may refer to the limitation of the application ranking method described above, and will not be repeated here.
In one embodiment, as shown in fig. 7, there is provided an application ranking apparatus 1, comprising: an acquisition module 11, a ranking module 12 and a feedback module 13, wherein:
an obtaining module 11, configured to obtain target feature information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
the sorting module 12 is configured to input target feature information into the sorting model, and obtain a target sorting result for each application installed in a terminal held by a target user;
and the feedback module 13 is used for sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to the terminal held by the target user.
In one embodiment, the ranking module 12 includes:
the coding sub-module is used for coding the target characteristic information to obtain an application click sequence of the target user;
the probability calculation sub-module is used for inputting the application click sequence into the ordering model to obtain probability values of each application installed in the terminal held by the target user;
and the sequencing sub-module is used for generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
In one embodiment, the ordering model includes an embedded layer, a convolutional network layer, and a fully-connected layer; the probability calculation sub-module is further used for: inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence;
inputting the initial feature representation into a convolution network layer for depth feature extraction to obtain a depth feature representation;
and inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In one embodiment, the application ranking model further comprises a training module comprising:
the desensitization sub-module is used for acquiring a training sample set from the data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user;
and the training sub-module is used for training the initial model by adopting the training sample set to obtain the ordering model.
In one embodiment, the desensitizing sub-module is further configured to:
candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets;
and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
In one embodiment, the training sub-module is further configured to: determining sample characteristic information corresponding to the application with the application use record times larger than a preset threshold value in the training sample set as a positive sample; determining sample characteristic information corresponding to the application of which the application use record frequency is less than or equal to a preset threshold value in the training sample set as a negative sample; and training the initial model by adopting a Focal loss function and positive and negative samples in a training sample set to obtain a sorting model.
The various modules in the application ordering apparatus described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 8. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is for storing data for applying the ranking method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an application ordering method.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
Inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user.
In one embodiment, when the processor executes the logic for inputting the target feature information into the ranking model to obtain the target ranking result for each application installed in the terminal held by the target user, the following steps are specifically implemented: encoding the target characteristic information to obtain an application click sequence of the target user; inputting the application click sequence into the ordering model to obtain probability values of all applications installed in the terminal held by the target user; and generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
In one embodiment, the ordering model includes an embedded layer, a convolutional network layer, and a fully-connected layer; when the processor executes logic that the computer program inputs the application click sequence into the sorting model to obtain probability values of applications installed in the terminal held by the target user, the following steps are specifically realized: inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence; inputting the initial feature representation into a convolution network layer for depth feature extraction to obtain a depth feature representation; and inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In one embodiment, the processor when executing the computer program further performs the steps of: acquiring a training sample set from a data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user; and training the initial model by adopting a training sample set to obtain a sequencing model.
In one embodiment, when the processor executes the logic of the computer program to obtain the training sample set from the data lake corresponding to the buried point acquisition server, the following steps are specifically implemented: candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets; and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
In one embodiment, the computer program trains the initial model with a training sample set, and the logic to obtain the ranking model, when executed by the processor, performs the steps of: determining sample characteristic information corresponding to the application with the application use record times larger than a preset threshold value in the training sample set as a positive sample; determining sample characteristic information corresponding to the application of which the application use record frequency is less than or equal to a preset threshold value in the training sample set as a negative sample; and training the initial model by adopting a Focal loss function and positive and negative samples in a training sample set to obtain a sorting model.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user. .
In one embodiment, the computer program inputs the target feature information into the ranking model, and when the logic for obtaining the target ranking result for each application installed in the terminal held by the target user is executed by the processor, the following steps are specifically implemented: encoding the target characteristic information to obtain an application click sequence of the target user; inputting the application click sequence into the ordering model to obtain probability values of all applications installed in the terminal held by the target user; and generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
In one embodiment, the ordering model includes an embedded layer, a convolutional network layer, and a fully-connected layer; the computer program inputs the application click sequence into the ordering model, and when logic for obtaining probability values of each application installed in the terminal held by the target user is executed by the processor, the following steps are specifically implemented: inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence; inputting the initial feature representation into a convolution network layer for depth feature extraction to obtain a depth feature representation; and inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a training sample set from a data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user; and training the initial model by adopting a training sample set to obtain a sequencing model.
In one embodiment, the logic of the computer program for obtaining the training sample set from the data lake corresponding to the buried point acquisition server is executed by the processor, and specifically implements the following steps: candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets; and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
In one embodiment, the computer program trains the initial model with a training sample set, and when the logic to obtain the ordered model is executed by the processor, the following steps are specifically implemented: determining sample characteristic information corresponding to the application with the application use record times larger than a preset threshold value in the training sample set as a positive sample;
determining sample characteristic information corresponding to the application of which the application use record frequency is less than or equal to a preset threshold value in the training sample set as a negative sample; and training the initial model by adopting a Focal loss function and positive and negative samples in a training sample set to obtain a sorting model.
In one embodiment, a computer program product is provided comprising a computer program which, when executed by a processor, performs the steps of:
Acquiring target characteristic information of a target user sent by an application management server; the application management server obtains the target feature information based on the application ordering request of the target user;
inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by a target user;
and sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for instructing the application management server to forward the target sorting result to a terminal held by a target user.
In one embodiment, the computer program inputs the target feature information into the ranking model, and when the logic for obtaining the target ranking result for each application installed in the terminal held by the target user is executed by the processor, the following steps are specifically implemented: encoding the target characteristic information to obtain an application click sequence of the target user; inputting the application click sequence into the ordering model to obtain probability values of all applications installed in the terminal held by the target user; and generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
In one embodiment, the ordering model includes an embedded layer, a convolutional network layer, and a fully-connected layer; the computer program inputs the application click sequence into the ordering model, and when logic for obtaining probability values of each application installed in the terminal held by the target user is executed by the processor, the following steps are specifically implemented: inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence; inputting the initial feature representation into a convolution network layer for depth feature extraction to obtain a depth feature representation; and inputting the depth characteristic representation into the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a training sample set from a data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user by a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user; and training the initial model by adopting a training sample set to obtain a sequencing model.
In one embodiment, the logic of the computer program for obtaining the training sample set from the data lake corresponding to the buried point acquisition server is executed by the processor, and specifically implements the following steps: candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets; and constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
In one embodiment, the computer program trains the initial model with a training sample set, and when the logic to obtain the ordered model is executed by the processor, the following steps are specifically implemented: and training the initial model by adopting a Focal loss function and a training sample set to obtain a sorting model.
The user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.

Claims (10)

1. An application ranking method, the method comprising:
acquiring target characteristic information of a target user sent by an application management server; the target feature information is acquired by the application management server based on an application ordering request of the target user;
inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in a terminal held by the target user;
And sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for indicating the application management server to forward the target sorting result to a terminal held by the target user.
2. The method according to claim 1, wherein the inputting the target feature information into the ranking model to obtain a target ranking result for each application installed in the terminal held by the target user includes:
encoding the target characteristic information to obtain an application click sequence of the target user;
inputting the application click sequence into a sequencing model to obtain probability values of all applications installed in a terminal held by the target user;
and generating a target sequencing result of each application installed in the terminal held by the target user according to the probability value of each application.
3. The method of claim 2, wherein the ordering model comprises an embedded layer, a convolutional network layer, and a fully-connected layer; the step of inputting the application click sequence into a ranking model to obtain probability values of applications installed in a terminal held by the target user, wherein the method comprises the following steps:
inputting the application click sequence into the embedded layer for dimension reduction processing to obtain an initial characteristic representation corresponding to the application click sequence;
Inputting the initial characteristic representation into the convolutional network layer for depth characteristic extraction to obtain a depth characteristic representation;
and inputting the depth characteristic representation to the full connection layer to obtain probability values of all applications installed in the terminal held by the target user.
4. The method according to claim 1, wherein the method further comprises:
acquiring a training sample set from a data lake corresponding to the buried point acquisition server; the training sample set comprises sample characteristic information of a sample user and sample sequencing results corresponding to the sample user, wherein the sample characteristic information of the sample user is obtained by desensitizing original characteristic information of the sample user through a buried point acquisition server, and the sample characteristic information at least comprises desensitization identity information of the sample user and application use records of the sample user;
and training the initial model by adopting the training sample set to obtain a sequencing model.
5. The method of claim 4, wherein the obtaining a training sample set from a data lake corresponding to a point-of-burial collection server comprises:
candidate sample sets of application use records, of which the updating moment in the data lake corresponds to the buried point acquisition server meets a set period, are used as training sample sets;
And constructing a training sample set according to the sample characteristic information of the sample user and the sample sequencing result corresponding to the sample user.
6. The method of claim 4, wherein training the initial model using the training sample set to obtain the ranking model comprises:
determining sample characteristic information corresponding to the application, of which the application use record frequency is larger than a preset threshold value, in the training sample set as a positive sample;
determining sample characteristic information corresponding to the application of which the frequency of application use records in the training sample set is less than or equal to the preset threshold value as a negative sample;
and training the initial model by adopting a Focalloss loss function and positive and negative samples in the training sample set to obtain a sequencing model.
7. An application ranking apparatus, the apparatus comprising:
the acquisition module is used for acquiring target characteristic information of a target user sent by the application management server; the target feature information is acquired by the application management server based on an application ordering request of the target user;
the sequencing module is used for inputting the target characteristic information into a sequencing model to obtain a target sequencing result of each application installed in the terminal held by the target user;
And the feedback module is used for sending a return notice comprising the target sorting result to the application management server, wherein the return notice is used for indicating the application management server to forward the target sorting result to a terminal held by the target user.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 6 when the computer program is executed.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
CN202310825969.1A 2023-07-06 2023-07-06 Application ordering method, device, computer equipment and storage medium Pending CN116861085A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310825969.1A CN116861085A (en) 2023-07-06 2023-07-06 Application ordering method, device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310825969.1A CN116861085A (en) 2023-07-06 2023-07-06 Application ordering method, device, computer equipment and storage medium

Publications (1)

Publication Number Publication Date
CN116861085A true CN116861085A (en) 2023-10-10

Family

ID=88224684

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310825969.1A Pending CN116861085A (en) 2023-07-06 2023-07-06 Application ordering method, device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN116861085A (en)

Similar Documents

Publication Publication Date Title
Corchado et al. Ibr retrieval method based on topology preserving mappings
CN109711883B (en) Internet advertisement click rate estimation method based on U-Net network
CN112328741B (en) Intelligent association reply method and device based on artificial intelligence and computer equipment
CN110197207B (en) Method and related device for classifying unclassified user group
CN108182633B (en) Loan data processing method, loan data processing device, loan data processing program, and computer device and storage medium
CN115293919B (en) Social network distribution outward generalization-oriented graph neural network prediction method and system
CN113869420A (en) Text recommendation method based on comparative learning and related equipment
Peng et al. Research on telecom customer churn prediction based on ga-xgboost and shap
CN113656699A (en) User feature vector determination method, related device and medium
CN115270002B (en) Knowledge distillation-based item recommendation method, electronic equipment and storage medium
CN116861085A (en) Application ordering method, device, computer equipment and storage medium
CN111552827B (en) Labeling method and device, behavior willingness prediction model training method and device
CN113935788A (en) Model evaluation method, device, equipment and computer readable storage medium
CN117078441B (en) Method, apparatus, computer device and storage medium for identifying claims fraud
CN112182357B (en) Data recommendation method, device, computer equipment and storage medium
Kaedi et al. Holographic memory-based Bayesian optimization algorithm (HM-BOA) in dynamic environments
CN112800048B (en) Communication network user communication record completion method based on graph representation learning
CN117112724A (en) Search pushing method, apparatus, device, storage medium and computer program product
CN117786208A (en) Agricultural knowledge recommendation method and device, electronic equipment and storage medium
CN116861071A (en) Information pushing method, information pushing device, computer equipment, storage medium and program product
CN117290594A (en) Financial business activity recommendation method, apparatus, computer device and storage medium
CN116150462A (en) Vector construction method and device for target object and computer equipment
Yang et al. Hyperbolic multichannel graph convolutional network for social recommendation
CN116702980A (en) Object loss prediction model training method, object loss prediction method and device
CN116894112A (en) Data classification method, device, computer equipment and storage medium thereof

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination