CN110210540A - Across social media method for identifying ID and system based on attention mechanism - Google Patents
Across social media method for identifying ID and system based on attention mechanism Download PDFInfo
- Publication number
- CN110210540A CN110210540A CN201910431115.9A CN201910431115A CN110210540A CN 110210540 A CN110210540 A CN 110210540A CN 201910431115 A CN201910431115 A CN 201910431115A CN 110210540 A CN110210540 A CN 110210540A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- different
- similarity
- social media
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000007246 mechanism Effects 0.000 title claims abstract description 29
- 230000008447 perception Effects 0.000 claims abstract description 11
- 230000006870 function Effects 0.000 claims abstract description 9
- 238000005457 optimization Methods 0.000 claims abstract description 8
- 230000015654 memory Effects 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 3
- 241001269238 Data Species 0.000 claims 1
- 230000007774 longterm Effects 0.000 claims 1
- 230000003542 behavioural effect Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 239000013598 vector Substances 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000012358 sourcing Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of across social media method for identifying ID and method based on attention mechanism, the described method comprises the following steps: obtaining the data of the different modalities of multiple users in different social medias as training data;For the data of different modalities, the potential expression of different model learning data is respectively adopted, training user's identification model: in conjunction with the confidence level of sequential relationship and different modalities data, calculating the similarity of data between user in different social medias;Similarity between user is mapped to probability space using multi-layer perception (MLP), obtains the probability that user in different social medias is directed toward same user subject;Objective function is constructed using cross entropy, Optimization Solution is iterated to model parameter;The model is used to determine whether to be directed toward same user for the different modalities data in different social medias.The present invention considers the difference of different modalities data transmitting data, and the accuracy of user identity identification is higher.
Description
Technical field
The present invention relates to user identity identification technical field more particularly to it is a kind of based on attention mechanism across social media
Method for identifying ID and system.
Background technique
It reaches its maturity in current this across social media, the multi-sourcing epoch are gradually presented in user generated data, user's
Multi-source heterogeneous data tend to reflect their daily life from different angles, reflect that their attribute is special in terms of different
Sign.Organic combination user is dispersed in the behavioral data on multiple Social Medias, is deep understanding user behavior, dissects user spy
Sign, comprehensively brings possibility to user modeling.Substantially, the user identity identification across social media is subsequent to integrate user
Premise, therefore cause the attention of many researchers.However existing technology depends on user configuration information (user name, life
Day, gender) and social network structure, more abundant user generated data is had ignored, so that the interpretation of model is poor.
Summary of the invention
To overcome above-mentioned the deficiencies in the prior art, the present invention provides a kind of across social media use based on attention mechanism
Family personal identification method and system, it is contemplated that different modalities data transmit the difference of information, and the accuracy of user identity identification is more
It is high.
To achieve the above object, one or more embodiments of the invention provides following technical solution:
A kind of across social media method for identifying ID based on attention mechanism, comprising the following steps:
The data of the different modalities of multiple users in different social medias are obtained as training data;
For the data of different modalities, the potential expression of different model learning data, training user's identity is respectively adopted
Identification model:
In conjunction with the confidence level of sequential relationship and different modalities data, the phase of data between user in different social medias is calculated
Like degree;
Similarity between user is mapped to probability space using multi-layer perception (MLP), user in different social medias is obtained and refers to
To the probability of same user subject;
Objective function is constructed using cross entropy, Optimization Solution is iterated to model parameter;
The model is used to determine whether to be directed toward same user for the different modalities data in different social medias.
One or more embodiments provide a kind of across social media user identity identification system based on attention mechanism,
Include:
Data acquisition module grabs the data of the different modalities of user in different social medias;
The potential expression of different model learning data is respectively adopted for the data of different modalities in data representation module;
Model training module, user's similarity calculation module, in conjunction with the confidence level of sequential relationship and different modalities data, meter
Calculate the similarity of data between user in different social medias;Probability evaluation entity, using multi-layer perception (MLP) by phase between user
It is mapped to probability space like degree, user is directed toward the probability of same user subject in different social medias;Mesh is constructed using cross entropy
Scalar functions are iterated Optimization Solution to model parameter;
User identification module receives the data of the different modalities in different social medias, judges whether to be directed toward same
User.
A kind of electronic equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, the processor realize across the social media user identity based on attention mechanism when executing described program
Recognition methods.
One or more embodiments provide a kind of computer readable storage medium, are stored thereon with computer program, should
Across the social media method for identifying ID based on attention mechanism is executed when program is executed by processor.
The above one or more technical solution there are following the utility model has the advantages that
Identify in different social medias whether be same user based on user configuration information different from existing, the present invention is logical
The similitude of user generated data in different social medias (issue text, image information) is crossed to be identified.It can be compared with
Good solution user configuration information is inconsistent, and confidence level is lower, and social network structure obtains difficulty, the excessive equal limitation of data volume
Property, while according to user generated data, the potential behavioral characteristic of user can be preferably analyzed, is preferably realized across social media
User matching.
The present invention considers in section at the same time, and user usually issues in different social medias similar or even identical
Content.The phase between time attenuation parameter study user-generated content is thus introduced on the basis of based on content similarity
Like property, calculated result facilitates subsequent user identity and more accurately identifies with more explanatory.
User generated data is usually directed to multiple mode, such as text, picture and video.The data of different modalities for across
Social media user identity identification usually has different confidence levels.Present invention introduces attention mechanism different moulds may be implemented
The automatic distribution of state confidence level, solving different modalities in user generated data in across social media user identity identification has difference to set
Believe horizontal problem, improves the interpretation of performance of modeling and model across social media user identity identification.
Detailed description of the invention
The Figure of description for constituting a part of the invention is used to provide further understanding of the present invention, and of the invention shows
Examples and descriptions thereof are used to explain the present invention for meaning property, does not constitute improper limitations of the present invention.
Fig. 1 is a kind of across social media user identity knowledge based on attention mechanism in the one or more embodiments of the present invention
Other method overview flow chart.
The meter of Fig. 2 similarity between user generated data in different social medias in the one or more embodiments of the present invention
Calculate method flow diagram.
Specific embodiment
It is noted that described further below be all exemplary, it is intended to provide further instruction to the present invention.Unless another
It indicates, all technical and scientific terms used herein has usual with general technical staff of the technical field of the invention
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to exemplary embodiments of the present invention.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
In the absence of conflict, the feature in the embodiment and embodiment in the present invention can be combined with each other.
Technical term is explained:
Attention mechanism: attention mechanism is from the habit regularity summarization of mankind's environment of observation, and the mankind are observing
When environment, brain often only focuses on certain several especially important part, obtains the information of needs, construct about environment certain
Description, attention mechanism generate respective weights, forming more accurate data indicates by learning different local importance.
Embodiment one
To achieve the goals above, the present embodiment passes through the Precise Representation of the study multi-source heterogeneous data of user, in conjunction with user
The sequential relationship inside data is generated, realizes across social media user identity identification.Due to different modalities in user generated data
There are different confidence levels in across social media user identity identification, introduces attention mechanism and realize the automatic of different modalities confidence level
Distribution, to improve the interpretation of performance of modeling and model across social media user identity identification.Specifically, this implementation
Example provides a kind of across social media method for identifying ID based on attention Time Perception user modeling, including following step
It is rapid:
S1: grabbing the data of the different modalities of user in different social medias, for the data of different modalities, is respectively adopted
The potential expression of different model learning data;
S2: based on the sequential relationship inside user generated data, the modeling of user's similitude is carried out;
S3: the expression of the multi-source heterogeneous data based on S1 study and user's similitude of S2 model, and capture different modalities number
According to the confidence level different for across social media user identity identification, the generalization ability of model is improved.
The step S1 data indicate the process to be formed further include:
S11: the text data in two social medias is modeled using different networks respectively, with O1In text
Data instance, it is assumed that O1In userThe kth data of publication(being abbreviated as t) contains M word t={ x1, x2...,
xM, for each word xZ, term vector e is mapped it onto using Global VectorsZ.Pass through two-way shot and long term memory network
(Bi-directionalLong Short-Term Memory, BiLSTM) models data, wherein z-th of word is just
To hiding layer stateIt indicates are as follows:
Wherein uzAnd rzIt is to update door and resetting door, m respectivelyzIt is memory cell state.It is last moment hidden layer
State, σ (x) are sigmoid functions.Wu, Wr, Wm, bu, brAnd bmIt is model parameter.Similarly, available z-th of word is reversed
Hide layer stateTherefore, the expression f of available z-th of wordZ, it is as follows:
Finally, our available data t it is potential be expressed as follows shown in:
Therefore, we can obtainPotential expressionSimilarly, O is obtained by another BiLSTM network2Middle userText information in the g data of publicationPotential expression
S12: it is mentioned using the good residual error neural network of pre-training (Residual Neural Network, ResNet) network
Picture feature is taken, for O2Middle userPictorial information in the g data of publicationIt is sent to ResNet net first
Network, then as follows by fully-connected network:
Wherein, WhAnd bhIt is full link model parameter, h represents ResNet network, ΘrIt is the parameter in network h.
The process of the step S2 similitude modeling further include:
S21: the potential expression of data generated based on S1WithIt (saves respectively slightly), it uses
Cosine method calculates separately O2Different modalities data and O1The similitude of data, to carry out user's similitude modeling:
Wherein,WithIt respectively refers toWithWithBetween similarity.
S22: since in section at the same time, user is usually issued in different social medias in similar or even identical
Hold.Sequential relationship inside user generated data is combined with the modeling of user's similitude, as follows:
Wherein, rK, aRepresent time attenuation parameter, pkAnd qgIt is respectivelyWithTimestamp,WithIt ties respectively
After conjunction sequential relationshipWithWithBetween similarity.
The process of the step S3 user identity identification model construction further include:
S31: it is generated based on S2WithAvailable userG dataWith userThe overall situation
Visual similarity distributionIt (is abbreviated as) and global text similarity distributionIt (saves slightly).In view of the data of different modalities are for across social media user
Identification usually has different confidence levels, introduces the automatic distribution that attention mechanism realizes different modalities confidence level, as follows:
[αv, αc]=softmax (aTcon(hv, hc))
Wherein Wv, Wc, bvAnd bcIt is model parameter.Con () represents cascade operation, αvAnd αcRespectively representing model is vision
With the different confidence levels of text modality distribution.A represents problem, and " this given user issues in content, vision and text mould
Which transmitting information of state is more? " it may finally obtain user'sWithSimilarity:
Therefore, available userEveryWithSimilarity
S32: the similitude of G data is integrated to obtain and uses corpse by the global similitude obtained based on S31WithPhase
Like degree d=[d1, d2..., dG], whereinIt indicatesAverage pond.This reality
It applies example and similarity between user is mapped to by probability space by using multi-layer perception (MLP):
Wherein,Represent userWithIt is directed toward the probability of unification user entity, w and b are model parameters.It finally, can be with
Obtain the target equation of model.The target equation realized using cross entropy, prediction result for providing to model and true
Label value between gap measured:
Wherein, yiRepresent userWithWhether the label of unification user entity is directed toward.
S33: by repetitive exercise until model is restrained, initial value is the random value of normal distribution, automatic using backpropagation
Parameter in learning model carries out more wheel training, and target equation gradually decreases, and training to target equation is stablized, and user identity is saved
The parameter of identification model can be used to export the result across social media user identity identification.
Embodiment two
The purpose of the present embodiment is to provide a kind of across social media user identity identification system.
To achieve the goals above, a kind of across social media user identity based on attention mechanism is present embodiments provided
Identifying system, comprising:
Data acquisition module grabs the data of the different modalities of user in different social medias;
The potential expression of different model learning data is respectively adopted for the data of different modalities in data representation module;
Model training module, user's similarity calculation module, in conjunction with the confidence level of sequential relationship and different modalities data, meter
Calculate the similarity of data between user in different social medias;Probability evaluation entity, using multi-layer perception (MLP) by phase between user
It is mapped to probability space like degree, user is directed toward the probability of same user subject in different social medias;Mesh is constructed using cross entropy
Scalar functions are iterated Optimization Solution to model parameter;
User identification module receives the data of the different modalities in different social medias, judges whether to be directed toward same
User.
Embodiment three
The purpose of the present embodiment is to provide a kind of electronic equipment.
To achieve the goals above, it present embodiments provides a kind of electronic equipment, including memory, processor and is stored in
On memory and the computer program that can run on a processor, the processor are realized when executing described program:
For the data of different modalities, the potential expression of different model learning data, training user's identity is respectively adopted
Identification model:
In conjunction with the confidence level of sequential relationship and different modalities data, the phase of data between user in different social medias is calculated
Like degree;
Similarity between user is mapped to probability space using multi-layer perception (MLP), user in different social medias is obtained and refers to
To the probability of same user subject;
Objective function is constructed using cross entropy, Optimization Solution is iterated to model parameter;
The model is used to determine whether to be directed toward same user for the different modalities data in different social medias.
Example IV
The purpose of the present embodiment is to provide a kind of computer readable storage medium.
A kind of computer readable storage medium, is stored thereon with computer program, execution when which is executed by processor
Following steps:
For the data of different modalities, the potential expression of different model learning data, training user's identity is respectively adopted
Identification model:
In conjunction with the confidence level of sequential relationship and different modalities data, the phase of data between user in different social medias is calculated
Like degree;
Similarity between user is mapped to probability space using multi-layer perception (MLP), user in different social medias is obtained and refers to
To the probability of same user subject;
Objective function is constructed using cross entropy, Optimization Solution is iterated to model parameter;
The model is used to determine whether to be directed toward same user for the different modalities data in different social medias.
Each step involved in above embodiments two, three and four is corresponding with embodiment of the method one, and specific embodiment can
Referring to the related description part of embodiment one.Term " computer readable storage medium " is construed as including that one or more refers to
Enable the single medium or multiple media of collection;It should also be understood as including any medium, any medium can be stored, be encoded
Or it carries instruction set for being executed by processor and processor is made either to execute in the present invention method.
The above one or more embodiment has following technical effect that
Identify in different social medias whether be same user based on user configuration information different from existing, the present invention is logical
The similitude of user generated data in different social medias (issue text, image information) is crossed to be identified.It can be compared with
Good solution user configuration information is inconsistent, and confidence level is lower, and social network structure obtains difficulty, the excessive equal limitation of data volume
Property, while according to user generated data, the potential behavioral characteristic of user can be preferably analyzed, is preferably realized across social media
User matching.
The present invention considers in section at the same time, and user usually issues in different social medias similar or even identical
Content.The phase between time attenuation parameter study user-generated content is thus introduced on the basis of based on content similarity
Like property, calculated result facilitates subsequent user identity and more accurately identifies with more explanatory.
User generated data is usually directed to multiple mode, such as text, picture and video.The data of different modalities for across
Social media user identity identification usually has different confidence levels.Present invention introduces attention mechanism different moulds may be implemented
The automatic distribution of state confidence level, solving different modalities in user generated data in across social media user identity identification has difference to set
Believe horizontal problem, improves the interpretation of performance of modeling and model across social media user identity identification.
It will be understood by those skilled in the art that each module or each step of aforementioned present invention can be filled with general computer
It sets to realize, optionally, they can be realized with the program code that computing device can perform, it is thus possible to which they are stored
Be performed by computing device in the storage device, perhaps they are fabricated to each integrated circuit modules or by they
In multiple modules or step be fabricated to single integrated circuit module to realize.The present invention is not limited to any specific hardware and
The combination of software.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Above-mentioned, although the foregoing specific embodiments of the present invention is described with reference to the accompanying drawings, not protects model to the present invention
The limitation enclosed, those skilled in the art should understand that, based on the technical solutions of the present invention, those skilled in the art are not
Need to make the creative labor the various modifications or changes that can be made still within protection scope of the present invention.
Claims (10)
1. a kind of across social media method for identifying ID based on attention mechanism, which comprises the following steps:
The data of the different modalities of multiple users in different social medias are obtained as training data;
For the data of different modalities, the potential expression of different model learning data, training user's identification is respectively adopted
Model:
In conjunction with the confidence level of sequential relationship and different modalities data, the similar of data between user is calculated in different social medias
Degree;
Similarity between user is mapped to probability space using multi-layer perception (MLP), user in different social medias is obtained and is directed toward together
The probability of one user subject;
Objective function is constructed using cross entropy, Optimization Solution is iterated to model parameter;
The model is used to determine whether to be directed toward same user for the different modalities data in different social medias.
2. a kind of across social media method for identifying ID based on attention mechanism as described in claim 1, feature
It is, the data of the different modalities are text data and image data;For text data, net is remembered by two-way shot and long term
The potential expression of network learning data;For image data, using the potential of residual error neural network and fully-connected network learning data
It indicates.
3. a kind of across social media method for identifying ID based on attention mechanism as described in claim 1, feature
It is, the similarity for calculating data between user in different social medias includes:
Calculate the similarity of different modalities data between user in different social medias;
The time attenuation parameter in different social medias between user data is calculated in conjunction with sequential relationship, corrects above-mentioned similitude;
Confidence level is distributed based on the data that attention mechanism is different modalities, the Similarity-Weighted between different modalities data is obtained
To comprehensive similarity.
4. a kind of across social media method for identifying ID based on attention mechanism as claimed in claim 3, feature
It is, by social media O1In userThe kth data of publication is expressed asBy O2Middle userThe g articles number of publication
Text information in is expressed asPictorial information in g data is expressed asThe potential expression obtained through overfitting
It is expressed asIn different social medias between user different modalities data similarity calculating method are as follows:
Wherein,WithIt respectively refers toWithWithBetween similarity.
5. a kind of across social media method for identifying ID based on attention mechanism as claimed in claim 4, feature
It is, time attenuation parameter calculation method are as follows:
Wherein, rK, gRepresent time attenuation parameter, pkAnd qgIt is respectivelyWithTimestamp;
Similitude is corrected according to time attenuation parameter are as follows:
Wherein,WithAfter sequential relationshipWithWithBetween similarity.
6. a kind of across social media method for identifying ID based on attention mechanism as claimed in claim 5, feature
It is, the calculation method of different modalities confidence level are as follows:
[αv, αc]=softmax (aTcon(hv, hc)) wherein, αvAnd αcRespectively representing model is that vision and text modality are distributed
Different confidence levels;WithRespectively
UserG dataWith userThe distribution of overall Vision similitude and the distribution of global text similarity;Wv, Wc, bv
And bcIt is model parameter, con () represents cascade operation;
UserEveryWithSimilarity it is as follows:
7. a kind of across social media method for identifying ID based on attention mechanism as claimed in claim 6, feature
It is,
Integrate user in a period of timeAll G datas and userSimilarity d=[d1, d2..., dG];
Similarity between user is mapped to probability space using multi-layer perception (MLP):
Wherein,Represent userWithIt is directed toward the probability of unification user entity, w and b are model parameters;
Construct target equation:
Wherein, yiRepresent userWithWhether the label of unification user entity is directed toward.
8. a kind of across social media user identity identification system based on attention mechanism characterized by comprising
Data acquisition module grabs the data of the different modalities of user in different social medias;
The potential expression of different model learning data is respectively adopted for the data of different modalities in data representation module;
Model training module, user's similarity calculation module calculate not in conjunction with the confidence level of sequential relationship and different modalities data
With the similarity of data between user in social media;Probability evaluation entity, using multi-layer perception (MLP) by similarity between user
It is mapped to probability space, obtains the probability that user in different social medias is directed toward same user subject;Mesh is constructed using cross entropy
Scalar functions are iterated Optimization Solution to model parameter;
User identification module receives the data of the different modalities in different social medias, judges whether to be directed toward same user.
9. a kind of electronic equipment including memory, processor and stores the calculating that can be run on a memory and on a processor
Machine program, which is characterized in that the processor is realized when executing described program as claim 1-7 is described in any item based on note
Across the social media method for identifying ID for power mechanism of anticipating.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
Such as described in any item across the social media user identity identification sides based on attention mechanism claim 1-7 are executed when execution
Method.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910431115.9A CN110210540B (en) | 2019-05-22 | 2019-05-22 | Cross-social media user identity recognition method and system based on attention mechanism |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910431115.9A CN110210540B (en) | 2019-05-22 | 2019-05-22 | Cross-social media user identity recognition method and system based on attention mechanism |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110210540A true CN110210540A (en) | 2019-09-06 |
CN110210540B CN110210540B (en) | 2021-02-26 |
Family
ID=67788116
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910431115.9A Active CN110210540B (en) | 2019-05-22 | 2019-05-22 | Cross-social media user identity recognition method and system based on attention mechanism |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110210540B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111046166A (en) * | 2019-12-10 | 2020-04-21 | 中山大学 | Semi-implicit multi-modal recommendation method based on similarity correction |
CN111274491A (en) * | 2020-01-15 | 2020-06-12 | 杭州电子科技大学 | Social robot identification method based on graph attention network |
CN113297397A (en) * | 2021-05-12 | 2021-08-24 | 山东大学 | Information matching method and system based on hierarchical multi-mode information fusion |
CN113779520A (en) * | 2021-09-07 | 2021-12-10 | 中国船舶重工集团公司第七0九研究所 | Cross-space target virtual identity correlation method based on multilayer attribute analysis |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105677830A (en) * | 2016-01-04 | 2016-06-15 | 北京大学 | Heterogeneous media similarity computing method and retrieving method based on entity mapping |
CN107256271A (en) * | 2017-06-27 | 2017-10-17 | 鲁东大学 | Cross-module state Hash search method based on mapping dictionary learning |
CN107562812A (en) * | 2017-08-11 | 2018-01-09 | 北京大学 | A kind of cross-module state similarity-based learning method based on the modeling of modality-specific semantic space |
CN108090857A (en) * | 2017-12-29 | 2018-05-29 | 复旦大学 | A kind of multi-modal student classroom behavior analysis system and method |
CN108805009A (en) * | 2018-04-20 | 2018-11-13 | 华中师范大学 | Classroom learning state monitoring method based on multimodal information fusion and system |
CN109753602A (en) * | 2018-12-04 | 2019-05-14 | 中国科学院计算技术研究所 | A kind of across social network user personal identification method and system based on machine learning |
-
2019
- 2019-05-22 CN CN201910431115.9A patent/CN110210540B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105677830A (en) * | 2016-01-04 | 2016-06-15 | 北京大学 | Heterogeneous media similarity computing method and retrieving method based on entity mapping |
CN107256271A (en) * | 2017-06-27 | 2017-10-17 | 鲁东大学 | Cross-module state Hash search method based on mapping dictionary learning |
CN107562812A (en) * | 2017-08-11 | 2018-01-09 | 北京大学 | A kind of cross-module state similarity-based learning method based on the modeling of modality-specific semantic space |
CN108090857A (en) * | 2017-12-29 | 2018-05-29 | 复旦大学 | A kind of multi-modal student classroom behavior analysis system and method |
CN108805009A (en) * | 2018-04-20 | 2018-11-13 | 华中师范大学 | Classroom learning state monitoring method based on multimodal information fusion and system |
CN109753602A (en) * | 2018-12-04 | 2019-05-14 | 中国科学院计算技术研究所 | A kind of across social network user personal identification method and system based on machine learning |
Non-Patent Citations (2)
Title |
---|
YUXIN PENG ET AL.: "Modality-Specific Cross-Modal Similarity Measurement With Recurrent Attention Network", 《IEEE》 * |
李源鑫 等: "时间衰减制导的协同过滤相似性计算", 《计算机系统应用》 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111046166A (en) * | 2019-12-10 | 2020-04-21 | 中山大学 | Semi-implicit multi-modal recommendation method based on similarity correction |
CN111046166B (en) * | 2019-12-10 | 2022-10-11 | 中山大学 | Semi-implicit multi-modal recommendation method based on similarity correction |
CN111274491A (en) * | 2020-01-15 | 2020-06-12 | 杭州电子科技大学 | Social robot identification method based on graph attention network |
CN113297397A (en) * | 2021-05-12 | 2021-08-24 | 山东大学 | Information matching method and system based on hierarchical multi-mode information fusion |
CN113297397B (en) * | 2021-05-12 | 2022-08-09 | 山东大学 | Information matching method and system based on hierarchical multi-mode information fusion |
CN113779520A (en) * | 2021-09-07 | 2021-12-10 | 中国船舶重工集团公司第七0九研究所 | Cross-space target virtual identity correlation method based on multilayer attribute analysis |
Also Published As
Publication number | Publication date |
---|---|
CN110210540B (en) | 2021-02-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111931062B (en) | Training method and related device of information recommendation model | |
CN110210540A (en) | Across social media method for identifying ID and system based on attention mechanism | |
CN109902222A (en) | Recommendation method and device | |
CN110532571A (en) | Text handling method and relevant apparatus | |
CN110347932B (en) | Cross-network user alignment method based on deep learning | |
CN110383299A (en) | The generation time model of memory-enhancing effect | |
CN109583322A (en) | A kind of recognition of face depth network training method and system | |
CN110826453A (en) | Behavior identification method by extracting coordinates of human body joint points | |
CN106407889A (en) | Video human body interaction motion identification method based on optical flow graph depth learning model | |
CN112052948B (en) | Network model compression method and device, storage medium and electronic equipment | |
CN110516095A (en) | Weakly supervised depth Hash social activity image search method and system based on semanteme migration | |
CN109992676B (en) | Cross-media resource retrieval method and retrieval system | |
CN114332545B (en) | Image data classification method and device based on low-bit pulse neural network | |
CN109978175A (en) | Parallelization coordinate descent for machine learning model | |
CN115222896B (en) | Three-dimensional reconstruction method, three-dimensional reconstruction device, electronic equipment and computer readable storage medium | |
CN112131261A (en) | Community query method and device based on community network and computer equipment | |
WO2024067779A1 (en) | Data processing method and related apparatus | |
Uddin et al. | Proposing logical table constructs for enhanced machine learning process | |
CN115795025A (en) | Abstract generation method and related equipment thereof | |
CN113568983B (en) | Scene graph generation method and device, computer readable medium and electronic equipment | |
Liu | Restricted Boltzmann machine collaborative filtering recommendation algorithm based on project tag improvement | |
Matsuda et al. | Benchmark for Personalized Federated Learning | |
Sun et al. | A Deep Learning Method for Intelligent Analysis of Sports Training Postures | |
Shi et al. | [Retracted] Design of Personalized Recommendation System for Swimming Teaching Based on Deep Learning | |
CN111259138A (en) | Tax field short text emotion classification method and device |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |