CN109992606A - A kind of method for digging of target user, device, electronic equipment and storage medium - Google Patents
A kind of method for digging of target user, device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN109992606A CN109992606A CN201910195363.8A CN201910195363A CN109992606A CN 109992606 A CN109992606 A CN 109992606A CN 201910195363 A CN201910195363 A CN 201910195363A CN 109992606 A CN109992606 A CN 109992606A
- Authority
- CN
- China
- Prior art keywords
- user
- vector
- similarity
- digging
- module
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides the method for digging of target user a kind of, device, electronic equipment and storage medium, this method and device are specially the figure insertion vector that multiple users to be selected are calculated according to vector computation model trained in advance；Calculate the similarity between each figure insertion vector and the reference vector of reference user；Target user will be chosen to be more than user to be selected corresponding to the similarity of preset threshold.This programme carries out usage mining by way for the treatment of and being compared between the figure insertion vector at selection family and reference vector, effectively reduces the data volume of retrieval, to accelerate recall precision, and then effectively increases the efficiency of usage mining.
Description
Technical field
This disclosure relates to Internet technical field more particularly to a kind of method for digging of target user, device, electronic equipment
And storage medium.
Background technique
With the development of network technology, people's lives gradually full Internet and mobile Internet transfer, however I
Enjoy network bring it is convenient while, the problem of internet of very fast development also brings information explosion to us.It is right
For internet platform, if it is possible to if finding suitable target user, so that it may realize phase for respective objects user
Mesh of marketing accordingly may be implemented for example, can push corresponding information to it for potential commercial user in the purpose answered
's.
It is dense to map users to based on the study of extensive discrete depths at present when finding suitable target user
Feature space method, user's frequency feature is converted by a neural network connected entirely, to realize user
It excavates, the complexity of entire calculating process is relatively high, and calculating speed is slower when data volume is big, and since it is desired that arrives user
Entire vector space in retrieved, and number of users is often more than one hundred million ranks so that entire retrieving is very very long,
It is lower so as to cause digging efficiency.
Summary of the invention
To overcome the problems in correlation technique, the disclosure provides the method for digging, device, electronics of a kind of target user
Equipment and storage medium.
In a first aspect, providing the method for digging of target user a kind of, comprising:
Vector is embedded according to the figure that vector computation model trained in advance calculates multiple users to be selected；
Calculate the similarity between each figure insertion vector and the reference vector of reference user；
Target user will be chosen to be more than user to be selected corresponding to the similarity of preset threshold.
Optionally, the similarity calculated between each figure insertion vector and the reference vector of reference user, packet
It includes:
The figure insertion vector of the multiple user to be selected is hashing onto multiple Hash buckets using local sensitivity hash algorithm；
According to the reference vector to calculating in each Hash bucket, the similarity is obtained.
Optionally, the figure insertion vector step of multiple users to be selected is calculated in the vector computation model that the basis is trained in advance
After rapid, further includes:
The Demographics value of the user to be selected is added in corresponding figure insertion vector.
Optionally, further includes:
Behavior based on user constructs user's figure of multiple users, and user's figure includes multiple nodes, each section
Point represents a user；
Selected positive sample and negative sample are schemed according to the user；
Model training is carried out using the positive sample and the negative sample, obtains the vector computation model.
It is optionally, described that selected positive sample and negative sample are schemed according to the user, comprising:
The positive sample is selected using weighting migration method in user's figure；
It is randomly selected from user's figure with the associated other users of specific user, obtains the negative sample.
Second aspect provides the excavating gear of target user a kind of, comprising:
Vector calculation module, the figure for being configured as calculating multiple users to be selected according to vector computation model trained in advance are embedding
Incoming vector；
Similarity calculation module is configured as calculating between each figure insertion vector and the reference vector of reference user
Similarity；
Target chosen module is configured as that target will be chosen to be more than user to be selected corresponding to the similarity of preset threshold
User.
Optionally, the similarity calculation module includes:
First computing unit, be configured as using local sensitivity hash algorithm by the figure of the multiple user to be selected be embedded in
Amount is hashing onto multiple Hash buckets；
Second computing unit is configured as being obtained according to the reference vector to calculating in each Hash bucket
The similarity.
Optionally, further includes:
Vector add-on module is configured as being calculated in the vector calculation module according to vector computation model trained in advance
After the figure insertion vector of multiple users to be selected, the Demographics value of the user to be selected is added to corresponding
In the figure insertion vector.
Optionally, further includes:
Figure building module, the behavior based on user that is configured as construct user's figure of multiple users, and user's figure includes
Multiple nodes, each one user of the node on behalf；
Sample chosen module is configured as scheming selected positive sample and negative sample according to the user；
Model training module is configured as carrying out model training using the positive sample and the negative sample, obtains described
Vector computation model.
Optionally, the sample chosen module includes:
First selected unit is configured as in user's figure selecting the positive sample using weighting migration method；
Second selected unit is configured as selecting at random from user's figure with the associated other users of specific user
It takes, obtains the negative sample.
The third aspect provides a kind of electronic equipment, comprising:
Processor；
Memory for storage processor executable instruction；
Wherein, the processor is configured to executing method for digging as described in relation to the first aspect.
Fourth aspect provides a kind of non-transitorycomputer readable storage medium, when the instruction in the storage medium by
When the processor of mobile terminal executes, so that mobile terminal is able to carry out method for digging as described in relation to the first aspect.
5th aspect, provides a kind of computer program, including method for digging described in first aspect；
The technical scheme provided by this disclosed embodiment can include the following benefits: this programme selects family by treating
Figure insertion vector and reference vector between the mode that is compared carry out usage mining, effectively reduce the data volume of retrieval,
To accelerate recall precision, and then effectively increase the efficiency of usage mining.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
The disclosure can be limited.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows and meets implementation of the invention
Example, and be used to explain the principle of the present invention together with specification.
Fig. 1 is a kind of flow chart of the method for digging of target user shown according to an exemplary embodiment；
Fig. 2 is the flow chart of the method for digging of another target user shown according to an exemplary embodiment；
Fig. 3 is the flow chart of the method for digging of another target user shown according to an exemplary embodiment；
Fig. 4 is a kind of block diagram of the excavating gear of target user shown according to an exemplary embodiment；
Fig. 5 is the block diagram of the excavating gear of another target user shown according to an exemplary embodiment；
Fig. 6 is the block diagram of the excavating gear of another target user shown according to an exemplary embodiment；
Fig. 7 is the block diagram of a kind of electronic equipment shown according to an exemplary embodiment；
Fig. 8 is the block diagram of another electronic equipment shown according to an exemplary embodiment.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all embodiments consistented with the present invention.On the contrary, they be only with it is such as appended
The example of device and method being described in detail in claims, some aspects of the invention are consistent.
Fig. 1 is a kind of flow chart of the method for digging of target user shown according to an exemplary embodiment.
Shown in referring to Fig.1, method for digging provided by the present embodiment is used to excavate target user, example according to corresponding purpose
Potential commercial user is such as excavated from Internet user, is joined so that business occurs according to Result and potential commercial user
System, to realize corresponding commercial object, such as information popularization, advertisement pushing.The method for digging of the target user includes following step
It is rapid:
S101, the figure insertion vector that multiple users to be selected are calculated according to vector computation model.
When the application one of technical solution be applied to corresponding website when, which can regard the website as
All users, including permanent subscriber, temporary subscriber, registration user, Guest User etc..It more or less all can be in view of these users
There are some information for website, therefore obtain the relevant information of multiple users to be selected first here, then using in advance training to
Amount computation model calculates the relevant information of these users, to obtain the figure insertion vector of all users to be selected.
It is worth noting that the vector computation model is to be trained gained based on the user's figure constructed in advance.
S102, the figure for calculating each user to be selected are embedded in the similarity of vector and reference vector.
The basis calculated herein is that website generally has some reference users, and preserve the benchmark of the reference user to
Amount, for example, website can generally accumulate some commercial users, and preserves these commercial users being calculated according to the above method
User vector, the user vector of these commercial users is reference vector here.
In the figure insertion vector for obtaining all users, i.e. obtain multiple figures corresponding with accordingly user to be selected respectively be embedded in
After amount, the similarity between each figure insertion vector and reference vector, and a preset threshold determines according to actual conditions are calculated,
Each similarity is compared by a preset similarity threshold in other words with the preset threshold.
When calculating similarity, the calculating of the similarity is realized by following specific steps:
Firstly, the figure insertion vector of all users to be selected is hashing onto Hash one by one using local sensitivity hash algorithm
In bucket, local sensitivity hash algorithm can guarantee that the higher figure insertion vector of similarity is hashing onto a Hash bucket；Then, needle
Similarity relative to reference vector is calculated to figure insertion vector all in each Hash bucket.It by this method only need to be more
It is found in new Hash bucket, is not required to carry out global searching, so as to reduce the burden of calculating.
S103, the user to be selected that similarity is more than preset threshold is chosen to be target user.
It is more to get arriving after determining the similarity between the figure insertion vector of each user and the reference vector of reference user
It after a similarity, is therefrom chosen, the corresponding user to be selected of figure insertion vector that similarity is higher than the preset threshold is selected
For target user, i.e., similarity is higher than the user to be selected of preset threshold as final Result.
In the case where reference user is commercial user, selected target user is potential commercial user here.And
And for potential commercial user, with the similarity of the reference vector of the commercial user as reference user it is higher to
Family is selected more to be possible to as potential commercial user.At this point, preferential Securities or discount Securities is sent to the potential commercial user, with
It is translated into practical commercial user.
It can be seen from the above technical proposal that this application provides the method for digging of target user a kind of, specially basis
Trained vector computation model calculates the figure insertion vector of multiple users to be selected in advance；Calculate each figure insertion vector and base
Similarity between the reference vector at mutatis mutandis family；Target will be chosen to be more than user to be selected corresponding to the similarity of preset threshold
User.This programme carries out user's digging by way for the treatment of and being compared between the figure insertion vector at selection family and reference vector
Pick, effectively reduces the data volume of retrieval, to accelerate recall precision, and then effectively increases the efficiency of usage mining.
Fig. 2 is the flow chart of the method for digging of another target user shown according to an exemplary embodiment.
Referring to shown in Fig. 2, method for digging provided by the present embodiment includes the following steps:
S201, the figure insertion vector that multiple users to be selected are calculated according to vector computation model.
When the application one of technical solution be applied to corresponding website when, which can regard the website as
All users, including permanent subscriber, temporary subscriber, registration user, Guest User etc..It more or less all can be in view of these users
There are some information for website, therefore obtain the relevant information of multiple users to be selected first here, then using in advance training to
Amount computation model calculates the relevant information of these users, to obtain the figure insertion vector of all users to be selected.
S202, the other information of user to be selected is added to corresponding figure insertion vector.
After the figure for obtaining each user to be selected is embedded in vector, the Demographics value of each user to be selected is added
Into corresponding figure insertion vector, such as by discretization or normalized age, gender, regional feature etc., to guarantee corresponding figure
It is embedded in the constant of vector.
S203, the figure for calculating each user to be selected are embedded in the similarity of vector and reference vector.
In the figure insertion vector for obtaining all users, i.e. obtain multiple figures corresponding with accordingly user to be selected respectively be embedded in
After amount, the similarity between each figure insertion vector and reference vector, and a preset threshold determines according to actual conditions are calculated,
Each similarity is compared by a preset similarity threshold in other words with the preset threshold.
S204, the user to be selected that similarity is more than preset threshold is chosen to be target user.
It is more to get arriving after determining the similarity between the figure insertion vector of each user and the reference vector of reference user
It after a similarity, is therefrom chosen, the corresponding user to be selected of figure insertion vector that similarity is higher than the preset threshold is selected
For target user, i.e., similarity is higher than the user to be selected of preset threshold as final Result.
It can be seen from the above technical proposal that this application provides the method for digging of target user a kind of, specially basis
Trained vector computation model calculates the figure insertion vector of multiple users to be selected in advance；And population system will be added in figure insertion vector
Meter learns feature；Calculate the similarity between each figure insertion vector and the reference vector of reference user；It will be more than default threshold
User to be selected corresponding to the similarity of value is chosen to be target user.This programme, which passes through, treats the figure insertion vector for selecting family and base
The mode being compared between quasi- vector carries out usage mining, effectively reduces the data volume of retrieval, to accelerate retrieval effect
Rate, and then effectively increase the efficiency of usage mining., relative to a upper embodiment, the population by splicing some users is united for it
Meter learns feature, thereby may be ensured that the constant of corresponding figure insertion vector.
Fig. 3 is the flow chart of the method for digging of another target user shown according to an exemplary embodiment.
Referring to shown in Fig. 3, method for digging provided by the present embodiment includes the following steps:
S301, the behavior based on user establish user's figure of user.
One user of each node on behalf in user's figure, and the side of relationship, i.e. figure between user is connected by works
Knot, user has viewing to the works of other users, concern, likes that behaviors is waited to be considered as having side between the two nodes.It is different
Behavior have different weights, the weight highest liked, concern is taken second place, viewing power it is lightest.G=(V, E, W), V represents node,
E representative edge, the weight of W representative edge.
S302, selected positive sample and negative sample are schemed according to user.
The positive sample is selected using weighting migration method specially in user's figure；Negative sample does not have from specific user then
It is randomly selected in the other users of connection.
S303, model training is carried out according to positive sample and negative sample.
After obtaining positive sample and negative sample, model training is carried out using above-mentioned sample, to obtain accordingly to meter
Model is calculated, which is used to calculate the figure insertion vector of each user to be selected.
It is divided into 2 parts for carrying out the neural network of model training in the present embodiment, lower layer is the volume solution in machine translation
Code device, is unsupervised learning, this part is the autocoder, x for extracting global informationi=[xi1..., xij..., xin] it is in G
The input vector of node i is expressed, xijWhat is represented is the normalized weight on side between node i and node j, and not connecting side is 0.So
xiThe interaction scenario of user Yu other user are represented, so the embedding characterization by autocoder out is to use
The context mechanism information at family.The forward direction study of this part is input node xiEmbedding is obtained by dnn encoder,
Embedding is reduced by the decoder of dnnTherefore the loss function of this part isUnder
The parameter of layer part each encoder be it is shared, the parameter of same decoder is also shared.
Upper layer is the neural network connected entirely, this part is supervised learning, the user formed before by weighting migration and use
Pair pairs of family, as positive sample, then again in the user's stochastical sampling not being connected with user i as negative sample.Here
Input is that the embedding for two users that lower layer learns out is stitched together, and then extracts feature by hidden layer, eventually passes through
Sigmoid function is converted into probability, and wherein sigmoid function calculation formula is as follows:
σ (a)=1/ (1+exp (- a))
The value range of sigmoid function is label of the similarity further according to sample that two users are obtained between (0,1)
(belonging to weighting migration to generate pair pairs is positive sample, is otherwise negative sample), uses LogLoss loss function, loss function
It is as follows:
The probability estimated is pt=σ (wt·xt), σ is sigmoid function, yt∈ { 0,1 } is the label of sample.Or more
What layer network mainly characterized is the similarity (whether being connected) of user couple, and extraction is local message in figure.
The loss of whole network is made of this two-part loss, it may be assumed that
ltotal=ls+λ1lf+λ2lreg
Wherein lregThat represent is regular terms, λ1And λ2The hyper parameter coefficient of representative, we are to excavate commercial user herein,
It is considered that global interactive information is more important, so λ1< 1.And lose lfPart can effectively evade unsupervised model without
The problem of method offline evaluation modelling effect, also bring certain information gain.Loss is minimized using stochastic gradient descent method
Function, and the gradient of loss function is solved, the parameter of network is then successively updated, this two parts updates embedding simultaneously,
So it includes global contextual information that embedding, which is, and include partial structurtes information, thus can more comprehensively characterize
The role of user in the entire network.
S304, the figure insertion vector that multiple users to be selected are calculated according to vector computation model.
When the application one of technical solution be applied to corresponding website when, which can regard the website as
All users, including permanent subscriber, temporary subscriber, registration user, Guest User etc..It more or less all can be in view of these users
There are some information for website, therefore obtain the relevant information of multiple users to be selected first here, then using in advance training to
Amount computation model calculates the relevant information of these users, to obtain the figure insertion vector of all users to be selected.
S305, the figure for calculating each user to be selected are embedded in the similarity of vector and reference vector.
In the figure insertion vector for obtaining all users, i.e. obtain multiple figures corresponding with accordingly user to be selected respectively be embedded in
After amount, the similarity between each figure insertion vector and reference vector, and a preset threshold determines according to actual conditions are calculated,
Each similarity is compared by a preset similarity threshold in other words with the preset threshold.
S306, the user to be selected that similarity is more than preset threshold is chosen to be target user.
It is more to get arriving after determining the similarity between the figure insertion vector of each user and the reference vector of reference user
It after a similarity, is therefrom chosen, the corresponding user to be selected of figure insertion vector that similarity is higher than the preset threshold is selected
For target user, i.e., similarity is higher than the user to be selected of preset threshold as final Result.
It can be seen from the above technical proposal that this application provides the method for digging of target user a kind of, specially first
Then training vector computation model calculates the figure insertion vector of multiple users to be selected using the vector computation model；And it will scheme embedding
Demographics is added in incoming vector；Calculate the phase between each figure insertion vector and the reference vector of reference user
Like degree；Target user will be chosen to be more than user to be selected corresponding to the similarity of preset threshold.This programme is by treating selection
The mode being compared between the figure insertion vector and reference vector at family carries out usage mining, effectively reduces the data of retrieval
Amount, to accelerate recall precision, and then effectively increases the efficiency of usage mining.Relative to above embodiment, here
Loss function is minimized using stochastic gradient descent method, and solves the gradient of loss function, then successively updates the ginseng of network
Number, thus can more comprehensively characterize the role of user in the entire network.
Fig. 4 is a kind of block diagram of the excavating gear of target user shown according to an exemplary embodiment.
Referring to shown in Fig. 4, excavating gear provided by the present embodiment is used to excavate target user, example according to corresponding purpose
Potential commercial user is such as excavated from Internet user, is joined so that business occurs according to Result and potential commercial user
System, to realize corresponding commercial object, such as information popularization, advertisement pushing.The excavating gear of the target user includes to meter
Calculate module 10, similarity calculation module 20 and target chosen module 30.
Vector calculation module is configured as calculating the figure insertion vector of multiple users to be selected according to vector computation model.
When the application one of technical solution be applied to corresponding website when, which can regard the website as
All users, including permanent subscriber, temporary subscriber, registration user, Guest User etc..It more or less all can be in view of these users
There are some information for website, therefore obtain the relevant information of multiple users to be selected first here, then using in advance training to
Amount computation model calculates the relevant information of these users, to obtain the figure insertion vector of all users to be selected.
It is worth noting that the vector computation model is to be trained gained based on the user's figure constructed in advance.
Similarity calculation module is configured as calculating the similarity of figure the insertion vector and reference vector of each user to be selected.
The basis calculated herein is that website generally has some reference users, and preserve the benchmark of the reference user to
Amount, for example, website can generally accumulate some commercial users, and preserves these commercial users being calculated according to the above method
User vector, the user vector of these commercial users is reference vector here.
In the figure insertion vector for obtaining all users, i.e. obtain multiple figures corresponding with accordingly user to be selected respectively be embedded in
After amount, the similarity between each figure insertion vector and reference vector, and a preset threshold determines according to actual conditions are calculated,
Each similarity is compared by a preset similarity threshold in other words with the preset threshold.
The similarity calculation module specifically includes the first computing unit and the second computing unit.
First calculates and is applied alone in the figure of all users to be selected insertion vector is hashing onto one using local sensitivity hash algorithm
In each and every one Hash bucket, local sensitivity hash algorithm can guarantee that the higher figure insertion vector of similarity is hashing onto a Hash bucket
In；Second computing unit is used to calculate for figure insertion vector all in each Hash bucket relative to the similar of reference vector
Degree.It need to only be found in the Hash bucket of update by this method, be not required to carry out global searching, so as to reduce the negative of calculating
Load.
Target chosen module is configured as the user to be selected that similarity is more than preset threshold being chosen to be target user.
It is more to get arriving after determining the similarity between the figure insertion vector of each user and the reference vector of reference user
It after a similarity, is therefrom chosen, the corresponding user to be selected of figure insertion vector that similarity is higher than the preset threshold is selected
For target user, i.e., similarity is higher than the user to be selected of preset threshold as final Result.
In the case where reference user is commercial user, selected target user is potential commercial user here.And
And for potential commercial user, with the similarity of the reference vector of the commercial user as reference user it is higher to
Family is selected more to be possible to as potential commercial user.At this point, preferential Securities or discount Securities is sent to the potential commercial user, with
It is translated into practical commercial user.
It can be seen from the above technical proposal that this application provides the excavating gear of target user a kind of, specially basis
Trained vector computation model calculates the figure insertion vector of multiple users to be selected in advance；Calculate each figure insertion vector and base
Similarity between the reference vector at mutatis mutandis family；Target will be chosen to be more than user to be selected corresponding to the similarity of preset threshold
User.This programme carries out user's digging by way for the treatment of and being compared between the figure insertion vector at selection family and reference vector
Pick, effectively reduces the data volume of retrieval, to accelerate recall precision, and then effectively increases the efficiency of usage mining.
In addition, further including vector add-on module 40 in the present embodiment, as shown in figure 5, the vector add-on module is configured as
After vector calculation module calculates the figure insertion vector of each user to be selected, the other information of user to be selected is added to correspondence
Figure insertion vector in.
Specially the Demographics value of each user to be selected is added in corresponding figure insertion vector, such as will
Discretization or normalized age, gender, regional feature etc., to guarantee the constant of corresponding figure insertion vector.
In addition, as shown in fig. 6, further including figure building module 50, sample chosen module 60 and model training in the present embodiment
Module 70.
Figure building module is configured as user's figure that user is established in the behavior based on user.
One user of each node on behalf in user's figure, and the side of relationship, i.e. figure between user is connected by works
Knot, user has viewing to the works of other users, concern, likes that behaviors is waited to be considered as having side between the two nodes.It is different
Behavior have different weights, the weight highest liked, concern is taken second place, viewing power it is lightest.G=(V, E, W), V represents node,
E representative edge, the weight of W representative edge.
Sample chosen module is not configured to scheme selected positive sample and negative sample according to user.
The module specifically includes the first selected unit and the second selected unit, and the first selected unit in user's figure for adopting
The positive sample is selected with weighting migration method；For selecting negative sample, negative sample does not have second selected unit from specific user
It is randomly selected in the other users of connection.
Model training module is configured as carrying out model training according to positive sample and negative sample.
After obtaining positive sample and negative sample, model training is carried out using above-mentioned sample, to obtain accordingly to meter
Model is calculated, which is used to calculate the figure insertion vector of each user to be selected.
Here loss function is minimized using stochastic gradient descent method, and solves the gradient of loss function, then successively
The parameter for updating network, thus can more comprehensively characterize the role of user in the entire network.
The present embodiment also provides a kind of computer program, and the computer program is as shown in Figure 1, Figure 2 or shown in Fig. 3 for executing
The method for digging of target user.
Fig. 7 is the block diagram of a kind of electronic equipment shown according to an exemplary embodiment.
For example, electronic equipment 700 may be provided as a server, including processing component 722, it further comprise one
Or multiple processors, and the memory resource as representated by memory 732, it can be by the execution of processing component 722 for storing
Instruction, such as application program.
The application program stored in memory 732 may include it is one or more each correspond to one group of instruction
Module.In addition, processing component 722 is configured as executing instruction, to execute target user's as shown in Figure 1, Figure 2 or shown in Fig. 3
Method for digging.
Electronic equipment 700 can also include that a power supply module 726 is configured as executing the power supply pipe of electronic equipment 700
Reason, a wired or wireless network interface 750 are configured as electronic equipment 700 being connected to network and an input and output (I/
O) interface 758.Electronic equipment 700 can be operated based on the operating system for being stored in memory 732, such as Windows
ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM or similar.
Fig. 8 is the block diagram of another electronic equipment shown according to an exemplary embodiment.
For example, electronic equipment 800 can be mobile phone, computer, digital broadcasting terminal, messaging device, game
Console, tablet device, Medical Devices, body-building equipment, the mobile terminals such as personal digital assistant.
Electronic equipment 800 may include following one or more components: processing component 802, memory 804, electric power assembly
806, multimedia component 809, audio component 810, the interface 812 of input/output (I/O), sensor module 814, and communication
Component 816.
The integrated operation of the usual controlling electronic devices 800 of processing component 802, such as with display, call, data are logical
Letter, camera operation and record operate associated operation.Processing component 802 may include one or more processors 820 to hold
Row instruction, to perform all or part of the steps of the methods described above.In addition, processing component 802 may include one or more moulds
Block, convenient for the interaction between processing component 802 and other assemblies.For example, processing component 802 may include multi-media module, with
Facilitate the interaction between multimedia component 809 and processing component 802.
Memory 804 is configured as storing various types of data to support the operation in equipment 800.These data are shown
Example includes the instruction of any application or method for operating on electronic equipment 800, contact data, telephone directory number
According to, message, picture, video etc..Memory 804 can by any kind of volatibility or non-volatile memory device or they
Combination realize, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable
Programmable read only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, quick flashing
Memory, disk or CD.
Power supply module 806 provides electric power for the various assemblies of electronic equipment 800.Power supply module 806 may include power supply pipe
Reason system, one or more power supplys and other with for electronic equipment 800 generate, manage, and distribute the associated component of electric power.
Multimedia component 809 includes the screen of one output interface of offer between the electronic equipment 800 and user.
In some embodiments, screen may include liquid crystal display (LCD) and touch panel (TP).If screen includes touch surface
Plate, screen may be implemented as touch screen, to receive input signal from the user.Touch panel includes one or more touches
Sensor is to sense the gesture on touch, slide, and touch panel.The touch sensor can not only sense touch or sliding
The boundary of movement, but also detect duration and pressure associated with the touch or slide operation.In some embodiments,
Multimedia component 809 includes a front camera and/or rear camera.When equipment 800 is in operation mode, as shot mould
When formula or video mode, front camera and/or rear camera can receive external multi-medium data.Each preposition camera shooting
Head and rear camera can be a fixed optical lens system or have focusing and optical zoom capabilities.
Audio component 810 is configured as output and/or input audio signal.For example, audio component 810 includes a Mike
Wind (MIC), when electronic equipment 800 is in operation mode, when such as call mode, recording mode, and voice recognition mode, microphone
It is configured as receiving external audio signal.The received audio signal can be further stored in memory 804 or via logical
Believe that component 816 is sent.In some embodiments, audio component 810 further includes a loudspeaker, is used for output audio signal.
I/O interface 812 provides interface between processing component 802 and peripheral interface module, and above-mentioned peripheral interface module can
To be keyboard, click wheel, button etc..These buttons may include, but are not limited to: home button, volume button, start button and lock
Determine button.
Sensor module 814 includes one or more sensors, for providing the state of various aspects for electronic equipment 800
Assessment.For example, sensor module 814 can detecte the state that opens/closes of equipment 800, the relative positioning of component, such as institute
The display and keypad that component is electronic equipment 800 are stated, sensor module 814 can also detect electronic equipment 800 or electronics
The position change of 800 1 components of equipment, the existence or non-existence that user contacts with electronic equipment 800,800 orientation of electronic equipment
Or the temperature change of acceleration/deceleration and electronic equipment 800.Sensor module 814 may include proximity sensor, be configured to
It detects the presence of nearby objects without any physical contact.Sensor module 814 can also include optical sensor, such as
CMOS or ccd image sensor, for being used in imaging applications.In some embodiments, which can be with
Including acceleration transducer, gyro sensor, Magnetic Sensor, pressure sensor or temperature sensor.
Communication component 816 is configured to facilitate the communication of wired or wireless way between electronic equipment 800 and other equipment.
Electronic equipment 800 can access the wireless network based on communication standard, such as WiFi, carrier network (such as 2G, 3G, 4G or 5G),
Or their combination.In one exemplary embodiment, communication component 816 receives via broadcast channel and comes from external broadcasting management
The broadcast singal or broadcast related information of system.In one exemplary embodiment, the communication component 816 further includes that near field is logical
(NFC) module is believed, to promote short range communication.For example, radio frequency identification (RFID) technology, infrared data association can be based in NFC module
Meeting (IrDA) technology, ultra wide band (UWB) technology, bluetooth (BT) technology and other technologies are realized.
In the exemplary embodiment, electronic equipment 800 can be by one or more application specific integrated circuit (ASIC), number
Word signal processor (DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for executing as shown in Figure 1, Figure 2 or shown in Fig. 3
Target user method for digging.
In the exemplary embodiment, a kind of non-transitorycomputer readable storage medium including instruction, example are additionally provided
It such as include the memory 804 of instruction, above-metioned instruction can be executed by the processor 820 of electronic equipment 800 to complete the above method.Example
Such as, the non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-ROM, tape, soft
Disk and optical data storage devices etc..
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to of the invention its
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the invention, these modifications, purposes or
Person's adaptive change follows general principle of the invention and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and true scope and spirit of the invention are by following
Claim is pointed out.
It should be understood that the present invention is not limited to the precise structure already described above and shown in the accompanying drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present invention is limited only by the attached claims.
Claims (10)
1. a kind of method for digging of target user characterized by comprising
Vector is embedded according to the figure that vector computation model trained in advance calculates multiple users to be selected；
Calculate the similarity between each figure insertion vector and the reference vector of reference user；
Target user will be chosen to be more than user to be selected corresponding to the similarity of preset threshold.
2. method for digging as described in claim 1, which is characterized in that described to calculate each figure insertion vector and benchmark use
Similarity between the reference vector at family, comprising:
The figure insertion vector of the multiple user to be selected is hashing onto multiple Hash buckets using local sensitivity hash algorithm；
According to the reference vector to calculating in each Hash bucket, the similarity is obtained.
3. method for digging as claimed in claim 1 or 2, which is characterized in that calculate mould in the vector that the basis is trained in advance
Type calculates after the figure insertion vector step of multiple users to be selected, further includes:
The Demographics value of the user to be selected is added in corresponding figure insertion vector.
4. method for digging as claimed in claim 1 or 2, which is characterized in that further include:
Behavior based on user constructs user's figure of multiple users, and user's figure includes multiple nodes, each node generation
One user of table；
Selected positive sample and negative sample are schemed according to the user；
Model training is carried out using the positive sample and the negative sample, obtains the vector computation model.
5. method for digging as claimed in claim 4, which is characterized in that described to scheme selected positive sample and negative sample according to the user
This, comprising:
The positive sample is selected using weighting migration method in user's figure；
It is randomly selected from user's figure with the associated other users of specific user, obtains the negative sample.
6. a kind of excavating gear of target user characterized by comprising
Vector calculation module, be configured as being embedded according to the figure that vector computation model trained in advance calculates multiple users to be selected to
Amount；
Similarity calculation module is configured as calculating the phase between each figure insertion vector and the reference vector of reference user
Like degree；
Target chosen module, target will be chosen to be more than user to be selected corresponding to the similarity of preset threshold by, which being configured as, uses
Family.
7. excavating gear as claimed in claim 6, which is characterized in that further include:
Vector add-on module is configured as being calculated in the vector calculation module according to vector computation model trained in advance multiple
User to be selected figure insertion vector after, by the Demographics value of the user to be selected be added to it is corresponding described in
In figure insertion vector.
8. excavating gear as claimed in claim 6, which is characterized in that further include:
Figure building module, the behavior based on user that is configured as construct user's figure of multiple users, and user's figure includes multiple
Node, each one user of the node on behalf；
Sample chosen module is configured as scheming selected positive sample and negative sample according to the user；
Model training module is configured as carrying out model training using the positive sample and the negative sample, obtains the vector
Computation model.
9. a kind of electronic equipment characterized by comprising
Processor；
Memory for storage processor executable instruction；
Wherein, the processor is configured to executing method for digging as claimed in any one of claims 1 to 5.
10. a kind of non-transitorycomputer readable storage medium, when the instruction in the storage medium is by the processing of mobile terminal
When device executes, so that mobile terminal is able to carry out method for digging as claimed in any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910195363.8A CN109992606A (en) | 2019-03-14 | 2019-03-14 | A kind of method for digging of target user, device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910195363.8A CN109992606A (en) | 2019-03-14 | 2019-03-14 | A kind of method for digging of target user, device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109992606A true CN109992606A (en) | 2019-07-09 |
Family
ID=67129599
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910195363.8A Pending CN109992606A (en) | 2019-03-14 | 2019-03-14 | A kind of method for digging of target user, device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109992606A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111667181A (en) * | 2020-06-08 | 2020-09-15 | 拉扎斯网络科技（上海）有限公司 | Task processing method and device, electronic equipment and computer readable storage medium |
| CN112395490A (en) * | 2019-08-16 | 2021-02-23 | 京东数字科技控股有限公司 | Method and apparatus for generating information |
| CN112597363A (en) * | 2020-12-22 | 2021-04-02 | 南京领行科技股份有限公司 | User processing method and device, electronic equipment and storage medium |
| CN113110887A (en) * | 2021-03-31 | 2021-07-13 | 联想(北京)有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN113269323A (en) * | 2020-02-17 | 2021-08-17 | 北京达佳互联信息技术有限公司 | Data processing method, processing device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106874509A (en) * | 2017-03-01 | 2017-06-20 | 广州大学 | Resource recommendation method and device based on middle granularity user grouping |
| CN107103057A (en) * | 2017-04-13 | 2017-08-29 | 腾讯科技（深圳）有限公司 | A kind of resource supplying method and device |
| CN107729290A (en) * | 2017-09-21 | 2018-02-23 | 北京大学深圳研究生院 | A kind of expression learning method of ultra-large figure using the optimization of local sensitivity Hash |
| US20180349501A1 (en) * | 2017-06-05 | 2018-12-06 | Linkedln Corporation | Embedding user categories using graphs for enhancing searches based on similarities |
-
2019
- 2019-03-14 CN CN201910195363.8A patent/CN109992606A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106874509A (en) * | 2017-03-01 | 2017-06-20 | 广州大学 | Resource recommendation method and device based on middle granularity user grouping |
| CN107103057A (en) * | 2017-04-13 | 2017-08-29 | 腾讯科技（深圳）有限公司 | A kind of resource supplying method and device |
| US20180349501A1 (en) * | 2017-06-05 | 2018-12-06 | Linkedln Corporation | Embedding user categories using graphs for enhancing searches based on similarities |
| CN107729290A (en) * | 2017-09-21 | 2018-02-23 | 北京大学深圳研究生院 | A kind of expression learning method of ultra-large figure using the optimization of local sensitivity Hash |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112395490A (en) * | 2019-08-16 | 2021-02-23 | 京东数字科技控股有限公司 | Method and apparatus for generating information |
| CN112395490B (en) * | 2019-08-16 | 2023-09-29 | 京东科技控股股份有限公司 | Method and device for generating information |
| CN113269323A (en) * | 2020-02-17 | 2021-08-17 | 北京达佳互联信息技术有限公司 | Data processing method, processing device, electronic equipment and storage medium |
| CN113269323B (en) * | 2020-02-17 | 2024-03-12 | 北京达佳互联信息技术有限公司 | Data processing method, processing device, electronic equipment and storage medium |
| CN111667181A (en) * | 2020-06-08 | 2020-09-15 | 拉扎斯网络科技（上海）有限公司 | Task processing method and device, electronic equipment and computer readable storage medium |
| CN112597363A (en) * | 2020-12-22 | 2021-04-02 | 南京领行科技股份有限公司 | User processing method and device, electronic equipment and storage medium |
| CN113110887A (en) * | 2021-03-31 | 2021-07-13 | 联想(北京)有限公司 | Information processing method and device, electronic equipment and storage medium |
| CN113110887B (en) * | 2021-03-31 | 2023-07-21 | 联想(北京)有限公司 | Information processing method, device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109800325B (en) | Video recommendation method and device and computer-readable storage medium | |
| WO2021196401A1 (en) | Image reconstruction method and apparatus, electronic device and storage medium | |
| CN109992606A (en) | A kind of method for digging of target user, device, electronic equipment and storage medium | |
| CN109089133B (en) | Video processing method and device, electronic equipment and storage medium | |
| KR20210053122A (en) | Method, apparatus and storage medium for training of neural network | |
| CN111524521B (en) | Voiceprint extraction model training method, voiceprint recognition method, voiceprint extraction model training device and voiceprint recognition device | |
| CN107102746B (en) | Candidate word generation method and device and candidate word generation device | |
| WO2021031645A1 (en) | Image processing method and apparatus, electronic device and storage medium | |
| WO2021027343A1 (en) | Human face image recognition method and apparatus, electronic device, and storage medium | |
| CN111160448B (en) | Training method and device for image classification model | |
| WO2021036382A1 (en) | Image processing method and apparatus, electronic device and storage medium | |
| CN108073303B (en) | Input method and device and electronic equipment | |
| CN110659690B (en) | Neural network construction method and device, electronic equipment and storage medium | |
| CN112101238A (en) | Clustering method and device, electronic equipment and storage medium | |
| CN109670077B (en) | Video recommendation method and device and computer-readable storage medium | |
| CN109117874A (en) | Operation behavior prediction technique and device | |
| CN112926310B (en) | Keyword extraction method and device | |
| CN109685041B (en) | Image analysis method and device, electronic equipment and storage medium | |
| CN111242303A (en) | Network training method and device, and image processing method and device | |
| CN112906484B (en) | Video frame processing method and device, electronic equipment and storage medium | |
| CN112307281A (en) | Entity recommendation method and device | |
| TW202133030A (en) | Image processing method and apparatus, and electronic device, and storage medium | |
| CN111553464A (en) | Image processing method and device based on hyper network and intelligent equipment | |
| CN109886211A (en) | Data mask method, device, electronic equipment and storage medium | |
| CN114154068A (en) | Media content recommendation method and device, electronic equipment and storage medium |
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 |