CN110189167A - A kind of moving advertising fraud detection method based on the insertion of isomery figure - Google Patents
A kind of moving advertising fraud detection method based on the insertion of isomery figure Download PDFInfo
- Publication number
- CN110189167A CN110189167A CN201910417284.7A CN201910417284A CN110189167A CN 110189167 A CN110189167 A CN 110189167A CN 201910417284 A CN201910417284 A CN 201910417284A CN 110189167 A CN110189167 A CN 110189167A
- Authority
- CN
- China
- Prior art keywords
- node
- isomery
- fraud detection
- moving advertising
- model
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0248—Avoiding fraud
Abstract
The invention discloses a kind of moving advertising fraud detection methods based on the insertion of isomery figure, comprising steps of 1) obtaining moving advertising daily record data and pre-processing to data；2) user, application and advertisement three's incidence relation data are extracted, isomery figure of having the right is constructed；3) first path is defined, the migration number and longest step-length of each node are set, traverses isomery node of graph of having the right, constructs node member path random walk sequence；4) language model is used, the dense vector of lower dimensional space for constructing isomery figure interior joint of having the right indicates；5) label is defined, subject data are constituted；6) moving advertising fraud detection model is constructed；7) the mobile application subject data of training part are input to moving advertising fraud detection model training, obtain moving advertising fraud detection model；8) fraud detection is carried out to mobile application using moving advertising fraud detection model.The present invention effectively detects the mobile application of fraud using the entity associated relationship in moving advertising system.
Description
Technical field
The present invention relates to the technical fields of mobile application cheating in advertisement, refer in particular to a kind of movement based on the insertion of isomery figure
Cheating in advertisement detection method.
Background technique
Moving advertising has accurate as a kind of new-type marketing for relying on intelligent terminal compared with traditional media
The features such as property, interactivity, flexibility and personalization.However ever-increasing cheating in advertisement behavior is brought to moving advertising market
Serious threat identifies that the fraud of mobile application is extremely difficult, and cheating in advertisement detection has become mobile Internet advertisement
Hot issue urgently to be resolved in the ecosystem.Due to having good expression ability to structural data and there is robustness, base
Exception and fraud detection are applied in the figure analysis method of graph structure data.
The efficiency in Large Scale Graphs is lower based on the analysis method of graph structure for tradition, and existing deep learning etc. has efficacious prescriptions
Case is difficult to directly apply to the analysis of graph structure data, and figure embedding grammar is that the node in figure learns having in a lower dimensional space
Imitating vector indicates, so that subsequent diagram data preferably be supported to analyze.It is how sharp for moving advertising fraudulent mean complicated and changeable
Efficient detection is made to fraud mobile application with the method based on figure insertion, is a problem to be solved.
Summary of the invention
The purpose of the present invention is to overcome the shortcomings of the existing technology and deficiency, proposes a kind of shifting based on the insertion of isomery figure
Dynamic cheating in advertisement detection method, can be improved the accuracy of mobile application cheating in advertisement detection.
To achieve the above object, a kind of technical solution provided by the present invention are as follows: moving advertising based on the insertion of isomery figure
Fraud detection method, comprising the following steps:
1) moving advertising daily record data is obtained, data are pre-processed；
2) the incidence relation data for extracting user, application and advertisement in the moving advertising ecosystem, construct isomery figure of having the right,
Its corresponding meta template WithThe relationship classification collection on classification collection and side is respectively represented, and is met
3) first path is definedThe migration frequency n and longest step-length l of each node are set, traversal is had the right in isomery figure G
The n item of node, building node v is had the right random walk path Sv={ Sv1,Sv2,...,Svn, it finally obtains and has the right isomery figure G's
First path random walk sequence S；
4) language model is constructed, study is had the right the dense vector expression X of d dimension space of P mobile application node in isomery figure G
∈RP×d, constitute input feature value；
5) mobile application of training part is manually marked, each shifting is set according to whether the information for fraud application
The label value of dynamic application；The label of fraud application is set as 1, and the label of non-fraud application is set as 0, obtains PtrainA label
DataPtrain< P, PtrainIt is less than 1 for training certain applications sum P* η, η
But it is greater than 0 floating number, is then bonded subject data with the corresponding input feature value in step 4)；
6) fraud detection model is constructed, for detecting the mobile application of fraud；
7) subject data are input in fraud detection model, obtain fraud detection model parameter, obtains moving advertising
Fraud detection model；
8) the mobile application input feature vector not marked is input in moving advertising fraud detection model, carries out fraud and answers
Detection.
In step 1), data prediction includes data cleansing and Missing Data Filling；Moving advertising daily record data includes four
A attribute: a, unique identification attribute: user, application, advertisement unique identifier；B, time attribute: user uses application operating
The specific time that advertisement occurs, it is accurate to the second；C, position attribution: geographical location locating for identity user；D, device attribute: user
Model, the display screen size, operating system of equipment used.
In step 2), the isomery figure G that has the right includes the node of three types, respectively user nodeApplication nodeAnd advertising node
Have the right isomery figure G include three kinds of node relationships, respectively user using applicationUser's operation advertisementIt answers
With displaying advertisementIts corresponding meta templateIndicate that there is a mappings for arbitrary node v ∈ V
FunctionAnd there is mapping functions for arbitrarily connecting side
Having the right in isomery figure G, the weight on side is determined by its corresponding operation information between two neighbor nodes.
In step 3), the random walk sequence S of each node is constructedti, i=1,2 ..., n, the sample mode point of node
For two stages: initial phase and follow-up phase respectively represent migration sequence StiLength of the length in 0 to one half dollar pathBetween migration stage and length existThe migration stage between longest migration step-length l.
Further, the migration probability of the initial phase of the random walk sequence of each node is constructed are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For the category of current node
In Vt+1The neighbor node collection of type,For first path, φ () is node type mapping function.
Further, the migration probability of the follow-up phase of the random walk sequence of each node are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For with present node with it is next
The relationship type of node, wiFor relationshipLast side weight, β is offset,For qualified neighbours section
Point set,For first path,For side Type mapping function.
In step 4), the language model of building is Skip-gram model and is accelerated using negative sample mode, negative to adopt
The negative sample number of sample is fn；
Majorized function such as following formula in the Skip-gram model of building:
Wherein, vtFor the isomery context for giving node vIn node, θ be model parameter, Xv,
For node v and vtThe knot vector expression of corresponding low-dimensional, XaIt is indicated for the vector of the low-dimensional of arbitrary node in figure.
The dense vector in lower dimensional space for obtaining figure interior joint is expressed asWherein d is vector dimension.
In step 6), the fraud detection model of building is sorter model, including conventional machines learning model and depth
Learning model.
Compared with prior art, the present invention have the following advantages that with the utility model has the advantages that
The present invention indicates the multi-section isomery figure of entity associated relationship in moving advertising system by building, captures richer
Structure and semantic relation；Meanwhile for first path random walk of multi-section isomery figure, weighed by increasing in node probability of spreading
Beam is weighed about, so that the relationship between the node of similar behavior is more closely, the vector that figure insertion is obtained is preferably anti-
The behavioural information of node is mirrored, so as to effectively detect the mobile application of fraud.
Detailed description of the invention
Fig. 1 is the specific flow chart of the method for the present invention.
Fig. 2 for institute's structure have the right isomery diagram be intended to.
Specific embodiment
Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited
In this.
As shown in Figure 1, the moving advertising fraud detection method based on the insertion of isomery figure provided by the present embodiment, specific to walk
Suddenly include:
1) moving advertising daily record data is obtained, data are pre-processed.
In the present embodiment, data prediction includes data cleansing, Missing Data Filling；Moving advertising daily record data includes four
A attribute: a, unique identification attribute: the unique identifier of user, application, advertisement etc.；B, time attribute: user uses application behaviour
The specific time for making advertisement generation, it is accurate to the second；C, position attribution: geographical location locating for identity user, such as user the country one belongs to
The IP address etc. that family, city and user use；D, device attribute: the model of equipment used in user, display screen size, operation
System etc..Such as user A has carried out clicking operation to advertisement D on mobile application C using equipment B at some time point.
2) the incidence relation data for extracting user, application and advertisement in the moving advertising ecosystem, construct the isomery figure G that has the right
=< V, E, M >, corresponding meta templateInstitute's structure isomery figure of having the right is as shown in Figure 2.
In the present embodiment, the isomery figure G that has the right includes the node of three types, respectively user nodeUsing
NodeAnd advertising node
Further, have the right isomery figure G include three kinds of node relationships, respectively user using applicationUser
Operate advertisementUsing displaying advertisementIts corresponding meta templateIt indicates for arbitrary
There is a mapping functions by node v ∈ VAnd there is mapping functions for arbitrarily connecting side
Further,WithThe relationship classification collection on classification collection and side is respectively represented, and is met
Further, having the right in isomery figure, the weight on side is true by its corresponding operation information between two neighbor nodes
It is fixed, i.e., the weight on side between two neighbor nodes is determined by the ratio of the number of operation behavior and total degree.
3) it is as follows to define first path:
It is 40 that the migration number of each node, which is set, as 30 and longest step-length, and traversal is had the right the node in isomery figure, building
30 of each node have the right random walk path, to finally obtain the first path random walk sequence for isomery figure of having the right；
In the present embodiment, the random walk sequence S of each node is constructedvi, i=1,2 ..., the sampling side of 30 interior joints
Formula is divided into two stages: initial phase and follow-up phase, respectively represents migration sequence length in the length 2 in 0 to one half dollar path
Between the migration stage and length 2 to the migration stage between longest migration step-length 40；
Further, the migration probability in the starting migration stage of the random walk sequence of each node is constructed are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For the category of current node
In Vt+1The neighbor node collection of type,For first path, φ () is node type mapping function.
Further, the migration probability of the follow-up phase of the random walk sequence of each node are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For with present node with it is next
The relationship type of node, wiFor relationshipLast side weight, β is offset,For qualified neighbours section
Point set,For first path,For side Type mapping function.
4) language model is constructed, the study dense vector of lower dimensional space of a mobile application node in isomery figure of having the right indicates,
Constitute input feature value.
Further, the language model of building is Skip-gram model, and is accelerated using negative sample mode, is born
The negative sample number of sampling is fn, in the present embodiment, the negative sample number for bearing sampling is 5；
Majorized function such as following formula in Skip-gram model:
Wherein, vtFor the isomery context for giving node vIn node, θ be model parameter, Xv,
For node v and vtThe knot vector expression of corresponding low-dimensional, XaIt is indicated for the vector of the low-dimensional of arbitrary node in figure.
The dense vector in lower dimensional space for finally obtaining figure interior joint is expressed asWherein d is vector dimension
Degree.
5) mobile application of training part is manually marked, each shifting is set according to whether the information for fraud application
The label value of dynamic application；The label of fraud application is set as 1, and the label of non-fraud application is set as 0, obtains PtrainA label
DataPtrain< P, PtrainFor training certain applications sum P* η, then with (4)
In correspondence input feature value be bonded subject data；
In the present embodiment, η takes 0.8.
6) fraud detection model is constructed, for detecting the mobile application of fraud.
In the present embodiment, the fraud detection model of building is random forest grader model classifiers model, model
Major parameter is as follows: weak learner number is 150, and the depth capacity of each tree is 5, and non-leaf nodes divides sample, that is, leaf section
The smallest sample number of point is 5, and using score outside bag and stochastic regime is arranged is 10, and Characteristic Number is selected as primitive character
Several square roots, other is model default value.
7) subject data are input in fraud detection model, obtain fraud detection model parameter, obtains moving advertising
Fraud detection model.
8) the mobile application input feature vector not marked is input in moving advertising fraud detection model, carries out fraud and answers
Detection.
In the present embodiment, the input feature vector of target mobile application is input in Random Forest model, obtains one 0
~1 real number py indicates that target mobile application is the probability of fraud application.Threshold tau=0.5, if py > τ, then mesh are set
Marking mobile application is fraud application, is otherwise normal use.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (8)
1. a kind of moving advertising fraud detection method based on the insertion of isomery figure, which comprises the following steps:
1) moving advertising daily record data is obtained, data are pre-processed；
2) the incidence relation data for extracting user, application and advertisement in the moving advertising ecosystem construct isomery figure of having the right, right
The meta template answered WithThe relationship classification collection on classification collection and side is respectively represented, and is met
3) first path is definedThe migration frequency n and longest step-length l of each node are set, the section in the isomery figure G that has the right is traversed
The n item of point, building node v is had the right random walk path Sv={ Sv1,Sv2,...,Svn, finally obtain the member for the isomery figure G that has the right
Path random walk sequence S；
4) language model is constructed, study is had the right the dense vector expression X ∈ R of d dimension space of P mobile application node in isomery figure GP ×d, constitute input feature value；
5) mobile application of training part is manually marked, each movement is set according to whether the information for fraud application and is answered
Label value；The label of fraud application is set as 1, and the label of non-fraud application is set as 0, obtains PtrainA label dataPtrain< P, PtrainIt is less than 1 but big for training certain applications sum P* η, η
In 0 floating number, subject data then are bonded with the corresponding input feature value in step 4)；
6) fraud detection model is constructed, for detecting the mobile application of fraud；
7) subject data are input in fraud detection model, obtain fraud detection model parameter, obtain moving advertising fraud
Detection model；
8) the mobile application input feature vector not marked is input in moving advertising fraud detection model, carries out fraud application
Detection.
2. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 1, which is characterized in that
In step 1), data prediction includes data cleansing and Missing Data Filling；Moving advertising daily record data includes four attributes: a,
Unique identification attribute: user, application, advertisement unique identifier；B, time attribute: user is occurred using application operating advertisement
The specific time, it is accurate to the second；C, position attribution: geographical location locating for identity user；D, device attribute: equipment used in user
Model, display screen size, operating system.
3. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 1, it is characterised in that:
In step 2), the isomery figure G that has the right includes the node of three types, respectively user nodeApplication nodeAnd advertisement
Node
Have the right isomery figure G include three kinds of node relationships, respectively user using applicationUser's operation advertisementUsing exhibition
Show advertisementIts corresponding meta templateIndicate that there is a mapping letters for arbitrary node v ∈ V
NumberAnd there is mapping functions for arbitrarily connecting side
Having the right in isomery figure G, the weight on side is determined by its corresponding operation information between two neighbor nodes.
4. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 1, it is characterised in that:
In step 3), the random walk sequence S of each node is constructedti, i=1,2 ..., n, the sample mode of node is divided into two ranks
Section: initial phase and follow-up phase respectively represent migration sequence StiLength of the length in 0 to one half dollar pathBetween
Migration stage and length existThe migration stage between longest migration step-length l.
5. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 4, it is characterised in that:
Construct the migration probability of the initial phase of the random walk sequence of each node are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For belonging to for current node
Vt+1The neighbor node collection of type,For first path, φ () is node type mapping function.
6. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 4, it is characterised in that:
The migration probability of the follow-up phase of the random walk sequence of each node are as follows:
Wherein,And vi+1∈Vt+1Respectively present node and next node,For with present node and next node
Relationship type, wiFor relationshipLast side weight, β is offset,For qualified neighbor node
Collection,For first path,For side Type mapping function.
7. a kind of moving advertising fraud detection method based on the insertion of isomery figure belonging to according to claim 1, it is characterised in that:
In step 4), the language model of building is Skip-gram model and is accelerated using negative sample mode, bears the negative sample of sampling
This number is fn；
Its majorized function such as following formula in the Skip-gram model of building:
Wherein, vtFor the isomery context N for giving node vt(v),In node；θ is the parameter of model, Xv,For section
Point v and vtThe knot vector of corresponding low-dimensional indicates；XaIt is indicated for the vector of the low-dimensional of arbitrary node in figure；
The dense vector in lower dimensional space for finally obtaining figure interior joint is expressed asWherein d is vector dimension.
8. a kind of moving advertising fraud detection method based on the insertion of isomery figure according to claim 1, it is characterised in that:
In step 6), the fraud detection model of building is sorter model, including conventional machines learning model and deep learning model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910417284.7A CN110189167B (en) | 2019-05-20 | 2019-05-20 | Mobile advertisement fraud detection method based on heterogeneous graph embedding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910417284.7A CN110189167B (en) | 2019-05-20 | 2019-05-20 | Mobile advertisement fraud detection method based on heterogeneous graph embedding |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110189167A true CN110189167A (en) | 2019-08-30 |
| CN110189167B CN110189167B (en) | 2021-06-08 |
Family
ID=67716841
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910417284.7A Active CN110189167B (en) | 2019-05-20 | 2019-05-20 | Mobile advertisement fraud detection method based on heterogeneous graph embedding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110189167B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660019A (en) * | 2019-09-29 | 2020-01-07 | 华北电力大学 | Small data set simplified stroke generation method based on BPL |
| CN110958220A (en) * | 2019-10-24 | 2020-04-03 | 中国科学院信息工程研究所 | Network space security threat detection method and system based on heterogeneous graph embedding |
| CN112232834A (en) * | 2020-09-29 | 2021-01-15 | 中国银联股份有限公司 | Resource account determination method, device, equipment and medium |
| CN112347260A (en) * | 2020-11-24 | 2021-02-09 | 深圳市欢太科技有限公司 | Data processing method and device and electronic equipment |
| CN112767054A (en) * | 2021-01-29 | 2021-05-07 | 北京达佳互联信息技术有限公司 | Data recommendation method, device, server and computer-readable storage medium |
| CN112988501A (en) * | 2019-12-17 | 2021-06-18 | 深信服科技股份有限公司 | Alarm information generation method and device, electronic equipment and storage medium |
| CN112995110A (en) * | 2019-12-17 | 2021-06-18 | 深信服科技股份有限公司 | Method and device for acquiring malicious event information and electronic equipment |
| CN113553446A (en) * | 2021-07-28 | 2021-10-26 | 厦门国际银行股份有限公司 | Financial anti-fraud method and device based on heteromorphic graph deconstruction |
| CN113656797A (en) * | 2021-10-19 | 2021-11-16 | 航天宏康智能科技(北京)有限公司 | Behavior feature extraction method and behavior feature extraction device |
| CN113706279A (en) * | 2021-06-02 | 2021-11-26 | 同盾科技有限公司 | Fraud analysis method and device, electronic equipment and storage medium |
| CN114528479A (en) * | 2022-01-20 | 2022-05-24 | 华南理工大学 | Event detection method based on multi-scale different composition embedding algorithm |
| CN112232834B (en) * | 2020-09-29 | 2024-04-26 | 中国银联股份有限公司 | Resource account determination method, device, equipment and medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120330751A1 (en) * | 2008-06-06 | 2012-12-27 | Oracle International Corporation | Crm system for enhanced retention of customers |
| US20170132636A1 (en) * | 2015-11-11 | 2017-05-11 | Idm Global, Inc. | Systems and methods for detecting relations between unknown merchants and merchants with a known connection to fraud |
| EP3239919A1 (en) * | 2008-03-05 | 2017-11-01 | eBay Inc. | Method and apparatus for image recognition services |
| CN109472626A (en) * | 2018-11-26 | 2019-03-15 | 浙江大学 | A kind of intelligent finance risk control method and system towards mobile phone charter business |
-
2019
- 2019-05-20 CN CN201910417284.7A patent/CN110189167B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3239919A1 (en) * | 2008-03-05 | 2017-11-01 | eBay Inc. | Method and apparatus for image recognition services |
| US20120330751A1 (en) * | 2008-06-06 | 2012-12-27 | Oracle International Corporation | Crm system for enhanced retention of customers |
| US20170132636A1 (en) * | 2015-11-11 | 2017-05-11 | Idm Global, Inc. | Systems and methods for detecting relations between unknown merchants and merchants with a known connection to fraud |
| CN109472626A (en) * | 2018-11-26 | 2019-03-15 | 浙江大学 | A kind of intelligent finance risk control method and system towards mobile phone charter business |
Non-Patent Citations (2)
| Title |
|---|
| FENG YAN-HONG 等: "Domain Named Entity Recognition Method Based on Skip-gram Model", 《2017 FIRST INTERNATIONAL CONFERENCE ON ELECTRONICS INSTRUMENTATION & INFORMATION SYSTEMS (EIIS)》 * |
| 张宝明: "基于Structure2vec算法的网络欺诈", 《软件导刊》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110660019A (en) * | 2019-09-29 | 2020-01-07 | 华北电力大学 | Small data set simplified stroke generation method based on BPL |
| CN110958220A (en) * | 2019-10-24 | 2020-04-03 | 中国科学院信息工程研究所 | Network space security threat detection method and system based on heterogeneous graph embedding |
| CN110958220B (en) * | 2019-10-24 | 2020-12-29 | 中国科学院信息工程研究所 | Network space security threat detection method and system based on heterogeneous graph embedding |
| WO2021077642A1 (en) * | 2019-10-24 | 2021-04-29 | 中国科学院信息工程研究所 | Network space security threat detection method and system based on heterogeneous graph embedding |
| WO2021121244A1 (en) * | 2019-12-17 | 2021-06-24 | 深信服科技股份有限公司 | Alarm information generation method and apparatus, electronic device, and storage medium |
| CN112988501A (en) * | 2019-12-17 | 2021-06-18 | 深信服科技股份有限公司 | Alarm information generation method and device, electronic equipment and storage medium |
| CN112995110A (en) * | 2019-12-17 | 2021-06-18 | 深信服科技股份有限公司 | Method and device for acquiring malicious event information and electronic equipment |
| CN112232834A (en) * | 2020-09-29 | 2021-01-15 | 中国银联股份有限公司 | Resource account determination method, device, equipment and medium |
| CN112232834B (en) * | 2020-09-29 | 2024-04-26 | 中国银联股份有限公司 | Resource account determination method, device, equipment and medium |
| CN112347260A (en) * | 2020-11-24 | 2021-02-09 | 深圳市欢太科技有限公司 | Data processing method and device and electronic equipment |
| CN112767054A (en) * | 2021-01-29 | 2021-05-07 | 北京达佳互联信息技术有限公司 | Data recommendation method, device, server and computer-readable storage medium |
| CN113706279A (en) * | 2021-06-02 | 2021-11-26 | 同盾科技有限公司 | Fraud analysis method and device, electronic equipment and storage medium |
| CN113706279B (en) * | 2021-06-02 | 2024-04-05 | 同盾科技有限公司 | Fraud analysis method, fraud analysis device, electronic equipment and storage medium |
| CN113553446A (en) * | 2021-07-28 | 2021-10-26 | 厦门国际银行股份有限公司 | Financial anti-fraud method and device based on heteromorphic graph deconstruction |
| CN113553446B (en) * | 2021-07-28 | 2022-05-24 | 厦门国际银行股份有限公司 | Financial anti-fraud method and device based on heterograph deconstruction |
| CN113656797A (en) * | 2021-10-19 | 2021-11-16 | 航天宏康智能科技(北京)有限公司 | Behavior feature extraction method and behavior feature extraction device |
| CN113656797B (en) * | 2021-10-19 | 2021-12-21 | 航天宏康智能科技(北京)有限公司 | Behavior feature extraction method and behavior feature extraction device |
| CN114528479A (en) * | 2022-01-20 | 2022-05-24 | 华南理工大学 | Event detection method based on multi-scale different composition embedding algorithm |
| CN114528479B (en) * | 2022-01-20 | 2023-03-21 | 华南理工大学 | Event detection method based on multi-scale heteromorphic image embedding algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110189167B (en) | 2021-06-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110189167A (en) | A kind of moving advertising fraud detection method based on the insertion of isomery figure | |
| CN110287334A (en) | A kind of school's domain knowledge map construction method based on Entity recognition and attribute extraction model | |
| Torralba et al. | Labelme: Online image annotation and applications | |
| CN110059620A (en) | Bone Activity recognition method based on space-time attention | |
| CN107945153A (en) | A kind of road surface crack detection method based on deep learning | |
| CN109918452A (en) | A kind of method, apparatus of data processing, computer storage medium and terminal | |
| CN106326345A (en) | Friend relation mining method based on user behaviors in social network | |
| CN106202362A (en) | Image recommendation method and image recommendation device | |
| CN108960409A (en) | Labeled data generation method, equipment and computer readable storage medium | |
| CN105760439A (en) | Figure cooccurrence relation graph establishing method based on specific behavior cooccurrence network | |
| CN106033435A (en) | Article identification method and apparatus, and indoor map generation method and apparatus | |
| CN110084304A (en) | A kind of object detection method based on generated data collection | |
| CN111008337B (en) | Deep attention rumor identification method and device based on ternary characteristics | |
| CN109815955A (en) | Topic householder method and system | |
| CN110414380A (en) | A kind of students ' behavior detection method based on target detection | |
| CN110223106A (en) | A kind of fraud application detection method based on deep learning | |
| CN109446328A (en) | A kind of text recognition method, device and its storage medium | |
| CN113364748B (en) | Ether house phishing node detection method and system based on transaction subgraph network | |
| CN101271526A (en) | Method for object automatic recognition and three-dimensional reconstruction in image processing | |
| Sloan et al. | Does base map size and imagery matter in sketch mapping? | |
| Barth | Comparative methodologies in the analysis of anthropological data | |
| Câmara et al. | Geocomputation techniques for spatial analysis: are they relevant to health data? | |
| CN108806355A (en) | A kind of calligraphy and painting art interactive education system | |
| Joho et al. | Nonparametric bayesian models for unsupervised scene analysis and reconstruction | |
| CN113469226A (en) | Street view image-based land utilization classification method and system |
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 |