CN110175438A - Share account detection method and relevant device - Google Patents
Share account detection method and relevant device Download PDFInfo
- Publication number
- CN110175438A CN110175438A CN201910445651.4A CN201910445651A CN110175438A CN 110175438 A CN110175438 A CN 110175438A CN 201910445651 A CN201910445651 A CN 201910445651A CN 110175438 A CN110175438 A CN 110175438A
- Authority
- CN
- China
- Prior art keywords
- account
- value
- characteristic
- sample
- item
- 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
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/30—Authentication, i.e. establishing the identity or authorisation of security principals
- G06F21/31—User authentication
- G06F21/316—User authentication by observing the pattern of computer usage, e.g. typical user behaviour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L63/00—Network architectures or network communication protocols for network security
- H04L63/08—Network architectures or network communication protocols for network security for authentication of entities
Abstract
The present invention provides a kind of sharing account detection method, this method obtains account to be detected in the characteristic value of several preset behavioural characteristic items;Obtain account Rating Model trained in advance;Wherein the parameter of account Rating Model includes each behavioural characteristic item and each corresponding weighted value of behavioural characteristic item, and weighted value is to be trained by neural network model training algorithm to having markd account sample;The characteristic value of account to be detected is input in account Rating Model, to obtain the target score value of account to be detected;Whether meet default sharing condition according to target score value, to determine whether account to be detected is to share account.The method provided through the invention, it can be determined that going out account to be detected is to share account or non-sharing account.The present invention also provides the relevant devices for sharing account detection, to guarantee the application and realization of the method in practice.
Description
Technical field
The present invention relates to account detection technique fields, more particularly sharing account detection method and relevant device.
Background technique
Currently, many website platforms provide different access authority for different types of access user.Such as access user
Type include member user and two kinds of ordinary user, can be accessed from website platform compared to member user for ordinary user
To more resources, the more service functions provided using website platform etc..
The type of user is accessed generally by the type classification of login account.Different types of access user has not
The register account number of same type, website platform log in the type of used register account number according to access user to determine that the access is used
Family is what type of access user, and then provides access authority corresponding with its type for access user.
For website platform, management needs guarantee access user and register account number are one-to-one for convenience,
But at present for various reasons, some user will use the register account number of other users to access website platform, especially general
General family may use the register account number of member user to scheme to enjoy more access authority.Such case is known as account point
Enjoy, in other words, account sharing refer to a register account number for multiple users using come the case where accessing website platform, simultaneously
The register account number, which is referred to as, shares account.
Therefore, how whether the account of test access website platform is the problem of sharing account, be industry urgent need to resolve.
Summary of the invention
In view of this, the present invention provides a kind of sharing account detection method, to realize the detection for sharing account.Separately
Outside, the present invention also provides it is a kind of sharing account detection relevant device, to guarantee the method in practice application and
It realizes.
In order to achieve the object, technical solution provided by the invention is as follows:
In a first aspect, the present invention provides a kind of sharing account detection methods, comprising:
Account to be detected is obtained in the characteristic value of preset behavioural characteristic item;
Obtain account Rating Model trained in advance;Wherein the parameter of the account Rating Model includes each behavior
Characteristic item and each corresponding weighted value of behavioural characteristic item, and the weighted value is calculated by neural network model training
Method is trained the markd account sample of tool;
The characteristic value of the account to be detected is input in the account Rating Model, to obtain the account to be detected
Target score value;
Whether meet default sharing condition according to the target score value, to determine whether the account to be detected is sharing
Account.
Second aspect, the present invention provides a kind of sharing account detection devices, comprising:
Characteristic value acquisition module, for obtaining account to be detected in the characteristic value of preset behavioural characteristic item;
Rating Model obtains module, for obtaining account Rating Model trained in advance;The wherein account Rating Model
Parameter include each behavioural characteristic item and each corresponding weighted value of behavioural characteristic item, and the weighted value
It is to be trained by neural network model training algorithm to having markd account sample;
Feature-value-score module, for the characteristic value of the account to be detected to be input in the account Rating Model,
To obtain the target score value of the account to be detected;
Share account detection module, for whether meeting default sharing condition according to the target score value, to determine
State whether account to be detected is to share account.
The third aspect, the present invention provides a kind of sharing account detection device, including processor and memory, the processing
Software program, calling storage data in the memory of the device by operation storage in the memory, at least execute
Following steps:
Account to be detected is obtained in the characteristic value of preset behavioural characteristic item;
Obtain account Rating Model trained in advance;Wherein the parameter of the account Rating Model includes each behavior
Characteristic item and each corresponding weighted value of behavioural characteristic item, and the weighted value is calculated by neural network model training
Method is trained the markd account sample of tool;
The characteristic value of the account to be detected is input in the account Rating Model, to obtain the account to be detected
Target score value;
Whether meet default sharing condition according to the target score value, to determine whether the account to be detected is sharing
Account.
Fourth aspect, the present invention provides a kind of storage mediums, are stored thereon with computer program, the computer program
When being executed by processor, above-mentioned sharing account detection method is realized.
Based on above technical scheme as can be seen that the present invention provides a kind of sharing account detection method, this method is obtained
The account to be detected account Rating Model trained in advance in the characteristic value of behavioural characteristic item and acquisition, by account to be detected in behavior
The characteristic value of characteristic item is input in account Rating Model, to obtain the score value of account to be detected, and then according to score value is
It is no to meet preset sharing condition, to determine whether account to be detected is to share account.As it can be seen that sharing account provided by the invention
Detection method can be used to whether a unknown access account is to share account.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow chart for sharing account detection method provided by the invention;
Fig. 2 is a kind of structural block diagram for sharing account detection device provided by the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The embodiment of the present invention provides a kind of method of sharing account detection, and this method can be applied in network platform side,
It is detected for the account to the access network platform, whether specific test access account is to share account.Specifically, referring to figure
1, which specifically includes following steps S101~S104.
S101, account to be detected is obtained in the characteristic value of several preset behavioural characteristic items.
Specifically, the account to access to the network platform can be obtained, using the account as account to be detected.In addition pre-
One or some behavioural characteristic items are first set, these behavioural characteristic items can be used to the characteristics of portraying an access account, these
Feature can be used for determining accessing whether account is to share account.That is, preset behavioural characteristic item is to be able to reflect point
Enjoy the behavioural characteristic item of feature.
In practical applications, website platform can be specially to provide the platform of Video service, then preset behavioural characteristic item
It can specifically include: any one of registration behavioural characteristic, login behavioural characteristic, viewing behavioural characteristic, payment behavior feature
Or a variety of combination.More specifically:
Registering behavioural characteristic may include one of following feature or a variety of combinations: what same time (range) was registered
Account number, the ip number of registration of same time (range), on the same day the account number of identical ip registration, on the same day identical user_agent
Register account number number, on the same day the register account number number of the same user_agent of same ip, the same place same time (range) note
The wind of duration (number of days), registration behavior that the account number of volume, registered place, registration ip, registion time, register account number have used
Dangerous grade (judgement from air control system), registration mailbox, the accounting of registration mailbox suffix, registration user_agent, registration
The accounting of user_agent, registered place with it is the last log in whether place consistent, account registered place access times,
Whether account is the common site of user, the cell-phone number of registration, registration cell-phone number using duration, registered place registered place
Ownership place, registration cell-phone number ownership place it is whether consistent etc. with the registration place ip;
Logging in behavioural characteristic may include one of following feature or a variety of combinations: the number of user's logon attempt,
Mode number that number that user successfully logs in, user log in (password login, short message log in, mailbox log in etc.), user's success
The ground points of number of devices, user's login that the probability of login, user log in, the risk class of user's login behavior (come from air control
The judgement of system), user app version number, user_agent (user agent) quantity for using of user for logging in etc., log in ip,
Login time logs in ip, logs in user_agent;
Viewing behavioural characteristic may include one of following feature or a variety of combinations: user requests time of vip video
Number, user request the vip view of the Annual distribution (quantity for initiating vip video request for such as each hour) of vip video, user's request
User_agent quantity that ip quantity that the quantity of frequency, user request vip video to use, user request vip video to use, user
Ground points, user while the online number of devices, user for requesting the number of devices of vip video, user to request vip video are detected
More equipment number online simultaneously etc., viewing time, viewing place, viewing ip, viewing user_agent;
Payment behavior feature may include one of following feature or a variety of combinations: the source of user member's equity
Payment behavior number, the user's note of (nominal price purchase, activation code, activity reward etc.), the membership grade of user, user's registration so far
Membership type number that volume payment amount number so far, user bought, the user place of payment and log in whether place consistent, user
The account number of place of payment time (range) payment whether consistent with viewing place, same, the same user_agent branch of same ip
Account number, the place of payment, payment ip, time of payment, the payment user_agent of the account number, the payment of same place paid.
Certainly, if the network platform is the other kinds of network platform, behavioural characteristic item can be other types, and the present invention is simultaneously
It is not specifically limited.
After this step determines account to be detected, need to obtain account to be detected in the characteristic value of the above behavioural characteristic item.It needs
It is noted that it is subsequent characteristic value is handled when, the data format of characteristic value may be required, so if being obtained
The characteristic value obtained meets the processing requirement, then directly subsequent step can be carried out to characteristic value, if characteristic value obtained is simultaneously
The processing requirement is not met, then needs to pre-process characteristic value.Specific pre-treatment step is as follows.
Several behavioural characteristic items in account to be detected are converted into the identifiable format of subsequent processes.For example,
Preset behavioural characteristic item includes the date, in the value on date, it is understood that there may be " 20180901 ", or " on September 1st, 2018 ",
Perhaps the multiple formats such as date value of " Friday on the 1st of September in 2018 " or inaccuracy, the date that subsequent processes require
Format is the first format, then the characteristic value that extended formatting will be present is converted, and the characteristic value after conversion is knowing for unification
The first other date format.
It is further to note that the characteristic value of certain behavioural characteristic items is text formatting, in order to which subsequent calculation processing needs
It is converted into specific value, if behavioural characteristic item includes viewing place, the corresponding numerical value in each viewing place is preset, such as exists
Viewing numerical value in Beijing is 1, is 2 in Shanghai viewing numerical value, is 3 in Shenzhen viewing numerical value, with the methods of this by behavioural characteristic item
The characteristic value of text type is converted to the characteristic value of value type.
It should be noted that the characteristic value of behavioural characteristic item, which can be, carries out treated characteristic value to primitive character value.
It specifically, include the characteristic value of behavioural characteristic item in the behavioural characteristic data of account to be detected, this feature value is original spy
Value indicative, it is thus necessary to determine that Discrete Eigenvalue corresponding to the primitive character value.
The method of determination of Discrete Eigenvalue is to obtain generated discrete by the behavioural characteristic data of a large amount of account sample
Characteristic value has corresponding relationship between these Discrete Eigenvalues and primitive character value, can will be to be detected according to the corresponding relationship
The primitive character value of account is converted to Discrete Eigenvalue.The Discrete Eigenvalue scores as account is input in subsequent step S103
Characteristic value in model.
It should be noted that the behavioural characteristic data Discrete Eigenvalue generated of account sample, can be evidence weight
It is worth (weight of evidence, abbreviation WOE), the generating mode of weight evidence weight values may refer to following the description, this time not
It repeats.
S102, account Rating Model trained in advance is obtained;Wherein the parameter of account Rating Model includes that each behavior is special
Item and each corresponding weighted value of behavioural characteristic item are levied, and weighted value is by neural network model training algorithm to mark
What the account sample of note was trained.
Wherein, the characteristic value of account Rating Model usage behavior characteristic item and the weight of behavior characteristic item are come to be detected
Account scores.It should be noted that the characteristic value of behavioural characteristic item is unknown-value in account Rating Model, which is waited for
Detection account is detected, and the characteristic value of the behavioural characteristic item of which account to be detected is inputted.But in account Rating Model,
The weighted value of behavioural characteristic item is given value, and weighted value is to be obtained by neural network model to having markd account sample training
, it is labeled as black account sample or white account sample, black account sample is the sample for being identified as sharing account, white account sample
It is not identified as sharing the sample of account.
Each corresponding weighted value of behavioural characteristic item is the foundation of scoring.Account is in the characteristic value of behavioural characteristic item and this
The product of the corresponding weight of behavioural characteristic item is score of the account in behavior characteristic item.All rows for the account that adds up
It is characterized the scoring event of item, the final score value of the account can be obtained.Based on the scoring thought, account can be constructed and commented
Sub-model.
S103, the characteristic value of account to be detected is input in account Rating Model, to obtain the target of account to be detected
Score value.
Specifically, the available account Rating Model of step S102, the available account to be detected of step S101 is multiple
The characteristic value of behavioural characteristic item, by account to be detected in the characteristic value input account Rating Model of behavioural characteristic item.
The behavioural characteristic item of each account to be detected has corresponding weight evidence weight values in account Rating Model, by behavior spy
The characteristic value of sign item is multiplied with the weight evidence weight values of behavioural characteristic item, the score of each behavioural characteristic item is obtained, by all behaviors
The score of characteristic item is added, and can obtain account to be detected in the total score of all behavioural characteristic items.It for ease of description, can be with
The gross score is known as target score value.
For example, a specific example of account Rating Model are as follows: score=w0+w1x1+w2x2+w3x3, wherein score be
Target score value, what 3 x were indicated is three behavioural characteristic items, and what 4 w values indicated is the weighted value of behavioural characteristic item.It needs
Bright, each behavioural characteristic item has corresponding behavioural characteristic value, w0It is that neural network model training algorithm can be trained
The special weighted value arrived, i.e., a weighted value of any one not corresponding behavioural characteristic item.
S104, whether default sharing condition is met according to target score value, to determine whether account to be detected is to share account
Number.
Specifically, sharing condition can be preset, shares condition and is used for by the evaluation to target score value, to determine
Whether account to be detected is to share account.Sharing condition is according to known sharing account and the non-score value for sharing account
What statistical conditions were provided, i.e., the score value that statistics shares account is any situation, what the non-score value for sharing account is
The particular content of sharing condition is arranged according to these situations for situation.The sharing condition of such setting, can be to unknown to be checked
The score value for surveying account is evaluated.
In a kind of specific implementation, sharing condition includes score value threshold value, by the target for determining account to be detected
The relationship of score value and score value threshold value, to determine whether account to be detected is to share account.
For example, the score value threshold value for presetting sharing account is 500, target score value is greater than or equal to the score value threshold
The account to be detected of value is to share account, is otherwise non-sharing account.If account Rating Model exports some account to be detected
Target score value be 200, at this time output valve be less than preset value, the account to be detected be not belonging to share account;If account scores
The target score value of model output is 1000, and output valve is greater than preset value at this time, then the account to be detected belongs to sharing account.
In another specific implementation, sharing grade can be set, by realizing pair to the judgement for sharing grade
The judgement of the sharing situation of account to be detected.
Specifically, it presets sharing condition to be specifically used for determining score value corresponding account sharing grade, then this step
Suddenly it can specifically include: obtain pre-set multiple accounts and share grade, wherein different accounts shares the corresponding difference of grade
Score value section;Determine the corresponding target account in target score value section and target score value section of target score value ownership
Number share grade;Share whether grade meets default sharing condition according to target account, to determine whether account to be detected is point
Enjoy account.
For example, presetting high, medium and low three accounts shares grade, wherein the corresponding score value section of high sharing grade
It is that score value is greater than 2000, the corresponding score value section of middle sharing grade is [1000,2000], the corresponding scoring of low sharing grade
[0,1000) value section is.It is assumed that the target score value that account Rating Model exports certain account to be detected is 200, can determine
The account to be detected belong to [0,1000) this score value section, and the score value area can be determined according to above-mentioned setting condition
Between it is corresponding be low sharing grade.Further, it is assumed that belong to the low account to be detected for sharing grade and not share account, ownership
In and the high account to be detected for sharing grade is to share account, then can be determined that the non-sharing account of above-mentioned detection account.
From the above technical scheme, the present invention provides a kind of sharing account detection method, this method obtains to be detected
The account account Rating Model trained in advance in the characteristic value of behavioural characteristic item and acquisition, by account to be checked in behavioural characteristic item
Whether characteristic value is input in account Rating Model, to obtain the score value of account to be detected, and then met according to score value pre-
If sharing condition, come determine account to be detected whether be share account.As it can be seen that the sharing account detection provided through the invention
The detection to account is shared may be implemented in method.
It should be noted that current account detection method can be by manual analysis data, and according to different sharing rows
It is characterized the corresponding detected rule of customization, after the behavioural characteristic for accessing account changes, needs not only but also manually reanalyses access
The related data of account reformulates detected rule.It is clear that this mode needs labor intensive cost, and flexibility compared with
It is low.However, the present invention can obtain account Rating Model, the weighted value in account Rating Model is instructed by neural network model
It practises, as long as the splitting glass opaque feature that discovery is new, is continued using neural network model training method to these splitting glass opaques spy
Sign is trained, so as to reduce cost of labor, and the flexibility that can detecte.
Described above is the application process of account Rating Model, it may be assumed that using account Rating Model to unknown to be detected
The process that account is detected.The training process of detailed description below account Rating Model, specifically in account Rating Model
The training process of weighted value.
Specifically, a kind of specific training method of the weighted value in step S102 account Rating Model obtained includes such as
Lower step A1~A3.
A1, the behavioural characteristic data for obtaining a plurality of account sample;Wherein the behavioural characteristic data of all account samples are wrapped
Several identical behavioural characteristic items are included, and every account sample all has characteristic value in each behavioural characteristic item.
Specifically, the behavioural characteristic data of a plurality of account sample are obtained, there are several behaviors in each account sample
Characteristic item, and the behavioural characteristic item of all account samples is identical.
It should be noted that the characteristic value that this step obtains is the original value of the behavioural characteristic item of account sample, therefore can
It is subsequent primitive character value to be handled with referred to as primitive character value, it is carried out with obtaining can be used neural network algorithm
Trained characteristic value.
A2, according to every account sample behavioural characteristic item characteristic value, determine the behavioural characteristic of every account sample to
Amount;Wherein behavioural characteristic vector is used to indicate the behavioural characteristic of account sample.
Specifically, this step is to determine every account sample respectively corresponding behavioural characteristic vector, behavioural characteristic
What vector included is the characteristic value of each behavioural characteristic item after treatment.Behavioural characteristic vector may be considered a characteristic value
Combination, how many behavioural characteristic item just include then how many a characteristic values in the combination of this feature value.It should be noted that feature
Each characteristic value in value combination is the characteristic value obtained after handling primitive character value.
A kind of mode of processing is primitive character value to be carried out sliding-model control, specific sliding-model control process includes:
For every account sample, the corresponding evidence weight of each characteristic value is calculated in the characteristic value of behavioural characteristic item according to account sample
WOE value combines the corresponding weight evidence weight values of each characteristic value of every account sample the behavioural characteristic vector for account sample.
Subsequent processing step is that behavioural characteristic item is divided into discrete behavioural characteristic item and two class of Continuous behavior characteristic item point
Other places reason, it may be assumed that the weight evidence weight values of different type behavioural characteristic item are calculated using different calculating method methods.
If account sample is continuously, to need the distribution situation by characteristic value with section in the characteristic value of behavioural characteristic item
Form divides, then is the corresponding weight evidence weight values of each interval computation, the corresponding weight evidence weight values in each section, according to account
The section conclusion evidence weighted value that the characteristic value of sample is belonged to, behavioural characteristic vector are exactly these determining weight evidence weight values
Combination.If account sample in the characteristic value of behavioural characteristic item be it is discrete, directly take this feature value for calculating weight evidence weight values,
Again by all weight evidence weight values combination of behavior characteristic item to get the behavioural characteristic vector for arriving discrete behavioural characteristic item.
Therefore, first according to the quantity situation of characteristic value, behavioural characteristic item is divided into discrete behavioural characteristic item and continuous
Behavioural characteristic item.
Specifically, behavioural characteristic item is likely to occur various characteristic values in account sample, according to the feature being likely to occur
The quantity situation of value, to determine that behavioural characteristic item is discrete behavioural characteristic item or Continuous behavior characteristic item.It should be noted that
The signified characteristic value being likely to occur is all characteristic values that behavioural characteristic item can occur herein.For example, viewing number this
Behavioural characteristic item, user can with viewing primary, viewing twice, viewing three times, viewing four times, viewing five times ... viewings tens
Secondary, viewing several hundred times, viewing thousands of times etc..
If the quantity of the characteristic value that can occur of behavioural characteristic item reaches certain quantitative requirement, it is determined that the behavior
Characteristic item is Continuous behavior characteristic item, is then determined as discrete behavioural characteristic item on the contrary.Quantitative requirement herein can be according to reality
Demand and be arranged.Based on the criteria for classifying, behavioural characteristic item can be divided into discrete behavioural characteristic item and Continuous behavior feature
?.
It is understood that out according to the above-mentioned criteria for classifying, if the quantity for the characteristic value that can occur is more, the behavior
Characteristic item may be divided into Continuous behavior characteristic item, if the negligible amounts for the characteristic value that can occur on the contrary, the behavior
Characteristic item may be divided into discrete behavioural characteristic item.Such as viewing number this behavioural characteristic item, user can with viewing at
Thousand up to ten thousand times, then behavior characteristic item may be divided into Continuous behavior characteristic item;For another example this behavioural characteristic of viewing place
, the viewing place of user is limited in a region, such as Beijing, Shanghai, the pre-set viewing place in Guangzhou, because
This behavior characteristic item may be confirmed as discrete behavioural characteristic item.
It should be noted that the primitive character value of the behavioural characteristics item such as viewing place and nonumeric, therefore will be non-obtaining
After the primitive character value of value type, need to handle the characteristic value for value type, and then carry out subsequent processing again.For example, right
In this behavioural characteristic item of viewing place, specific value can be set according to region, specifically such as can be set Beijing place be 1,
Shanghai place is 2, Shanghai place is 3 etc..
Individually below to discrete behavioural characteristic item and Continuous behavior characteristic item, illustrate how to calculate corresponding weight evidence
Weight values.
For discrete behavioural characteristic item, the corresponding weight evidence weight values of each characteristic value of discrete behavioural characteristic item are calculated.Tool
Body, according to the characteristic value distribution situation of discrete behavioural characteristic item, obtain characteristic value therein, count corresponding to each characteristic value
Black account sample and white account sample quantity and calculate their accounting;Further according to weight evidence weight values calculation formula:
Obtain weight evidence weight values corresponding to each characteristic value.Wherein black account sample is the account for being marked as sharing account
Number sample, white account sample are the account sample for being not labeled as sharing account.
Such as discrete behavioural characteristic item includes viewing place, by taking this characteristic value of Beijing viewing place as an example, counts Beijing
The quantity of black account sample and white account sample corresponding to viewing place, according to the quantity of black account sample and white account sample
Corresponding accounting is obtained, obtains this characteristic value pair of Beijing viewing place further according to accounting and weight evidence weight values calculation formula
The weight evidence weight values answered.
For Continuous behavior characteristic item, the selected characteristic value section from continuous characteristic value is needed, and calculate characteristic value area
Between corresponding weight evidence weight values.Illustrate individually below how from continuous characteristic value selected characteristic value section, Yi Jiru
What calculates the corresponding weight evidence weight values in characteristic value section.
Wherein, from continuous characteristic value a kind of specific embodiment in selected characteristic value section include step 1.1~
1.3。
1.1, the item number of the corresponding account sample of each characteristic value of Continuous behavior characteristic item is counted, it is each continuous to obtain
The characteristic value distribution situation of behavioural characteristic item.
Specifically, any one behavioural characteristic item existing characteristics value in each account sample, for continuous row
It is characterized item, the corresponding characteristic value of behavior characteristic item is also continuously, to pass through and count behavior feature in each account sample
The characteristic value of item, obtains a distribution map about behavior characteristic item.
For example, be directed to this behavioural characteristic item of viewing quantity, first count viewing quantity in each account sample this
Behavioural characteristic item, to obtain a mapping relations of viewing quantity Yu account sample;Assuming that when viewing quantity is 1, it is corresponding
Account sample have 100, when viewing quantity is 2, corresponding account has 200, so obtain about viewing quantity this
The characteristic value distribution curve of behavioural characteristic item.
1.2, according to the characteristic value distribution situation of Continuous behavior characteristic item, in the characteristic value of Continuous behavior characteristic item, really
The variation for determining account sample strip number meets the object feature value of preset condition;Wherein the number of object feature value is N.
Specifically, it after the characteristic value distribution situation for obtaining behavioural characteristic item, determines and changes apparent characteristic value, by this feature
Value is individually extracted from distribution situation as object feature value.For example, it is assumed that the feature Distribution value according to viewing quantity is bent
Line it can be concluded that, in the characteristic value distribution curve of viewing quantity, account sample strip number is at characteristic value 2 and 3 the two characteristic values
Significant change has occurred, then 2 and 3 be viewing quantity object feature value.
Wherein, hence it is evident that variation can be judged by the way that whether the variation of account sample strip number meets preset condition.Specifically
Judgment mode can be, the absolute value of variation is more than preset absolute value threshold value;Or can be, change rate is more than preset
Change rate threshold value;Or can also refer to skilled artisans appreciate that judgement two values obviously it is changed other
Mode.
It 1.3, is N+1 characteristic value section by the feature value division of Continuous behavior characteristic item according to object feature value.
Specifically, the object feature value determined according to step 1.2, the characteristic value distribution situation of behavioural characteristic item is divided into
Different sections.Such as, premised on step 1.2 example, after obtaining 2,3 two object feature values of object feature value, according to this two
The characteristic value distribution curve of a feature value division viewing quantity, can be by feature value division section less than 2, greater than 2 less than 3
Section, the section greater than 3.
It can be above multiple characteristic value sections by the feature value division of Continuous behavior characteristic item.How is detailed description below
Calculate the corresponding weight evidence weight values in characteristic value section, in practical applications, a kind of specific embodiment include step 2.1~
2.2。
2.1, the quantity of the corresponding black account sample in characteristic value section and the quantity of corresponding white account sample are determined;Its
In black account sample be marked as share account account sample, white account sample be not labeled as share account account
Sample.
Specifically, identical as the mode that above-mentioned discrete behavioural characteristic item calculates corresponding black and white account sample size, only not
Cross is determining entire section herein rather than the corresponding black and white account sample size of a characteristic value.
For example, be directed to this behavioural characteristic item of viewing quantity, viewing quantity characteristic value in the section less than 2, system
Account sample of the meter with black account label and the account sample with white account label;Characteristic value 2 and spy in viewing quantity
In 3 section of value indicative, account sample of the statistics with black account label and the account sample with white account label;In viewing quantity
Characteristic value be greater than 3 sections in, statistics with black account label account sample and with white account label account sample;
To obtain the quantity of black account sample and white account sample in different sections.
2.2, according to the ratio of the quantity of the corresponding black account sample in characteristic value section and the quantity of white account sample, meter
Calculate the weight evidence weight values in each characteristic value section.
Specifically, according to the quantity of black account sample and white account sample in a certain section counted, black account is calculated
Ratio shared by ratio shared by number sample and white account sample, and the section is obtained according to weight evidence weight values calculation formula (1)
Weight evidence weight values.
For example, for for the section greater than 3 in this behavioural characteristic item of viewing quantity, in the section obtained by statistics
Black account sample number be 8, white sample number be 12;Therefore, the accounting of black account sample is 8/20, and the accounting of white account sample is
12/20, and then the weight evidence weight values according to corresponding to the section that weight evidence weight values calculation formula (1) is calculated greater than 3.
In conclusion by above step 1.1-1.3 and step 2.1-2.2, in available Continuous behavior characteristic item
The corresponding weight evidence weight values in each characteristic value section.In addition, more than the present invention also illustrating how to obtain discrete behavioural characteristic item
The corresponding weight evidence weight values of each characteristic value.It therefore, can be by the original spy of account sample for each account sample
Value indicative is converted into corresponding weight evidence weight values.
Specifically, for each account sample, account sample is obtained corresponding to the characteristic value of discrete behavioural characteristic item
Weight evidence weight values, and determine that account sample belongs to (ownership is referred to as corresponding to) in the characteristic value of Continuous behavior characteristic item
Object feature value section, and obtain the corresponding weight evidence weight values in object feature value section.These weight evidence weight values are for forming
The behavioural characteristic vector of account sample.It should be noted that weight evidence weight values are referred to as Discrete Eigenvalue, in step S101
The characteristic value of account to be detected can also according to above two corresponding relationship, by the primitive character value of account to be detected be converted to from
Dissipate characteristic value.
More specifically, the characteristic value in each account sample, has plenty of the characteristic value of Continuous behavior characteristic item, has plenty of
The characteristic value of discrete behavioural characteristic item.If it is the characteristic value of discrete behavioural characteristic item, then this directly is converted by this feature value
The corresponding weight evidence weight values of characteristic value;If it is the characteristic value of Continuous behavior characteristic item, then it needs to be determined that this feature value belongs to
Which characteristic value section obtains the belonged to corresponding weight evidence weight values in this feature value section, then converts this for this feature value
The corresponding weight evidence weight values in characteristic value section.
Weight evidence weight values are used to form the feature vector of account sample, as a specific example of feature vector is
[0.00651,2.29272,1.99425,8.04397,1.60549,1.74118,2.73867], each numerical value in the combination
It is evident feature value.
It should be noted that obtaining this processing mode of behavioural characteristic vector based on evidence weight value, make neural network
Model can handle nonumeric type feature, and reduce the rule for the feature space for needing to search for when neural network model training
Mould reduces calculation amount required for neural network model training, improves training effectiveness.
A3, using neural network model training algorithm, the behavioural characteristic vector of all account samples is trained, with
To each corresponding weighted value of behavioural characteristic item.
Specifically, after the characteristic value of account sample being converted to feature vector, neural network model training can be used
Algorithm is trained these feature vectors.
It should be noted that neural network is a kind of operational model, by mutual between a large amount of node (or neuron)
It connects and composes.A kind of each specific output function of node on behalf, referred to as excitation function.Connection between every two node all represents
One weighted value for passing through the connection signal, referred to as weight, this is equivalent to the memory of artificial neural network.Network it is defeated
Out then with the connection type of network, weighted value can change therewith because of the difference of excitation function.And network itself is usually all to certainly
Right certain algorithm of boundary or function approach, it is also possible to the expression to a kind of logic strategy.
It for the present invention, is by the feature vector of each behavioural characteristic item in input account sample, to these spies
Sign vector is trained, and neural network model training is completed to mean that the weighted value in neural network model is fixed up, this
A little weight, that is, corresponding weighted values of behavioural characteristic item, while these weighted values are also the weighted value in account Rating Model.
It should be noted that neural network model training algorithm used in this application can be specially Logic Regression Models instruction
Practice algorithm.Based on the available corresponding account Rating Model of the Logic Regression Models, it is illustrated below by way of an example.
For example, the formula of Logic Regression Models are as follows:Wherein e is natural constant, p be one between 0 to 1 it
Between probability.
Assuming that Logic Regression Models include 3 behavioural characteristic items, respectively x1, x2, x3, z=w is enabled0+w1x1+w2x2+
w3x3, following formula can be obtained after z is carried out algebraic transformation:
It enablesSo as to obtain following account Rating Model: score=w0+w1x1+w2x2+w3x3。
Wherein x1 in above formula, x2, x3 are the behavioural characteristic vector after conversion, and the form for vector of being write as is [x1, x2, x3],
W0, w1, w2, w3 are the parameters of Logic Regression Models, using the account sample training model, exactly in order to estimate to parameter
Meter, obtains w0, w1, w2, the fixation value of w3.
See Fig. 2, it illustrates a kind of structures for sharing account detection device provided by the present application, can specifically include: special
Value indicative obtains module 201, Rating Model acquisition module 202, feature-value-score module 203 and shares account detection module 204.
Characteristic value acquisition module 201, for obtaining account to be detected in the characteristic value of several preset behavioural characteristic items;
Rating Model obtains module 202, for obtaining account Rating Model trained in advance;The wherein account scoring mould
The parameter of type includes each behavioural characteristic item and each corresponding weighted value of behavioural characteristic item, and the weight
Value is to be trained by neural network model training algorithm to having markd account sample;
Feature-value-score module 203, for the characteristic value of the account to be detected to be input to the account Rating Model
In, to obtain the target score value of the account to be detected;
Share account detection module 204, for whether meeting default sharing condition according to the target score value, to determine
Whether the account to be detected is to share account.
In one example, the default sharing condition is specifically used for sentencing score value corresponding account sharing grade
It is fixed;Then the sharing account detection module 204 can specifically include: share grade classification submodule, sharing grade determines submodule
Block and sharing account detection sub-module.
Share grade classification submodule, shares grade for obtaining pre-set multiple accounts, wherein different accounts
Share grade and corresponds to different score value sections;
Share grade and determine submodule, for determining the target score value section of target score value ownership and described
The corresponding target account in target score value section shares grade;
Share account detection sub-module, for sharing whether grade meets the default sharing item according to the target account
Part, to determine whether the account to be detected is to share account.
In one example, sharing account detection device can also include: weighted value training module, for training weight
Value;Wherein the weighted value training module can specifically include: characteristic acquisition submodule, feature vector determine submodule,
Weighted value trains submodule.
Characteristic acquisition submodule, for obtaining the behavioural characteristic data of a plurality of account sample;Wherein all account samples
This behavioural characteristic data include several identical behavioural characteristic items, and every account sample has in each behavioural characteristic item
There is characteristic value;
Feature vector determines submodule, for, in the characteristic value of behavioural characteristic item, determining every according to every account sample
The behavioural characteristic vector of account sample;Wherein behavioural characteristic vector is used to indicate the behavioural characteristic of account sample;
Weighted value trains submodule, special to the behavior of all account samples for using neural network model training algorithm
Sign vector is trained, to obtain each corresponding weighted value of behavioural characteristic item.
In one example, described eigenvector determines that submodule can specifically include: evidence weight computing unit and
Evidence weight assembled unit.
Evidence weight computing unit, for being directed to every account sample, according to the account sample in behavioural characteristic item
Characteristic value calculates the corresponding weight evidence weight values of each characteristic value;
Evidence weight assembled unit, for being by the corresponding weight evidence weight values combination of each characteristic value of every account sample
The behavioural characteristic vector of the account sample.
In one example, the evidence weight computing unit can specifically include: behavioural characteristic item divide subelement, from
It dissipates feature calculation subelement, continuous feature calculation subelement and evidence weight and obtains subelement.
Behavioural characteristic divides subelement and behavioural characteristic item is divided into discrete lines for the quantity situation according to characteristic value
It is characterized item and Continuous behavior characteristic item;
Discrete features computation subunit calculates each spy of discrete behavioural characteristic item for being directed to discrete behavioural characteristic item
The corresponding weight evidence weight values of value indicative;
Continuous feature calculation subelement, for being directed to Continuous behavior characteristic item, the selected characteristic value from continuous characteristic value
Section, and calculate the corresponding weight evidence weight values in characteristic value section;
Evidence weight obtains subelement, and for being directed to each account sample, it is special in discrete behavior to obtain the account sample
Weight evidence weight values corresponding to the characteristic value of item are levied, and determine the account sample in the characteristic value institute of Continuous behavior characteristic item
The object feature value section of ownership, and obtain the corresponding weight evidence weight values in the object feature value section.
In one example, the continuous feature calculation subelement includes: that subelement and weight calculation are chosen in section
Unit.
Subelement is chosen in section, for being directed to Continuous behavior characteristic item, the selected characteristic value section from continuous characteristic value;
Weight calculation subelement, for calculating the corresponding weight evidence weight values in characteristic value section.
In one example, the section is chosen subelement and can be specifically included: statistics subelement, value subelement and
Divide subelement.
Count subelement, the item of the corresponding account sample of each characteristic value for counting the Continuous behavior characteristic item
Number, to obtain the characteristic value distribution situation of each Continuous behavior characteristic item;
Value subelement, according to the characteristic value distribution situation of the Continuous behavior characteristic item, in the Continuous behavior feature
In the characteristic value of item, determine that the variation of account sample strip number meets the object feature value of preset condition;The wherein target signature
The number of value is N;
Subelement is divided, for being N by the feature value division of the Continuous behavior characteristic item according to the object feature value
+ 1 characteristic value section.
In one example, the weight calculation subelement can specifically include: black and white sample determines subelement and card
According to weight calculation subelement.
Black and white sample determines subelement, for determining the quantity of the corresponding black account sample in characteristic value section and corresponding
The quantity of white account sample;Wherein black account sample be marked as share account account sample, white account sample be not by
Labeled as the account sample for sharing account;
Evidence weight computation subunit, for the quantity and white account according to the corresponding black account sample in characteristic value section
The ratio of the quantity of sample calculates the weight evidence weight values in each characteristic value section.
Present invention also provides a kind of sharing account detection device, which may include processor and memory, described
Software program, calling storage data in the memory of the processor by operation storage in the memory, at least
Execute following steps:
Account to be detected is obtained in the characteristic value of several preset behavioural characteristic items;
Obtain account Rating Model trained in advance;Wherein the parameter of the account Rating Model includes each behavior
Characteristic item and each corresponding weighted value of behavioural characteristic item, and the weighted value is calculated by neural network model training
Method is trained the markd account sample of tool;
The characteristic value of the account to be detected is input in the account Rating Model, to obtain the account to be detected
Target score value;
Whether meet default sharing condition according to the target score value, to determine whether the account to be detected is sharing
Account.
Present invention also provides a kind of storage mediums, are stored thereon with computer program, and the computer program is processed
When device executes, the sharing account detection method that any one above-mentioned embodiment provides is realized.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
It should also be noted that, herein, relational terms such as first and second and the like are used merely to one
Entity or operation are distinguished with another entity or operation, without necessarily requiring or implying between these entities or operation
There are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant are intended to contain
Lid non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including above-mentioned element.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
Claims (10)
1. a kind of sharing account detection method characterized by comprising
Account to be detected is obtained in the characteristic value of preset behavioural characteristic item;
Obtain account Rating Model trained in advance;Wherein the parameter of the account Rating Model includes each behavioural characteristic
Item and each corresponding weighted value of behavioural characteristic item, and the weighted value is by neural network model training algorithm pair
Have what markd account sample was trained;
The characteristic value of the account to be detected is input in the account Rating Model, to obtain the mesh of the account to be detected
Mark score value;
Whether meet default sharing condition according to the target score value, to determine whether the account to be detected is to share account
Number.
2. sharing account detection method according to claim 1, which is characterized in that the default sharing condition is specifically used for
Share grade to the corresponding account of score value to determine;
It is then described that whether default sharing condition is met according to the target score value, to determine whether the account to be detected is point
Enjoying account includes:
It obtains pre-set multiple accounts and shares grade, wherein different accounts, which shares grade, corresponds to different score value areas
Between;
Determine the corresponding target account in target score value section and target score value section of the target score value ownership
Number share grade;
Share whether grade meets default sharing condition according to the target account, to determine whether the account to be detected is point
Enjoy account.
3. sharing account detection method according to claim 1, which is characterized in that the training process packet of the weighted value
It includes:
Obtain the behavioural characteristic data of a plurality of account sample;Wherein the behavioural characteristic data of all account samples include identical
Behavioural characteristic item, and every account sample all has characteristic value on each behavioural characteristic item;
According to every account sample in the characteristic value of behavioural characteristic item, the behavioural characteristic vector of every account sample is determined;Wherein
Behavioural characteristic vector is used to indicate the behavioural characteristic of account sample;
Using neural network model training algorithm, the behavioural characteristic vector of all account samples is trained, it is each to obtain
The corresponding weighted value of behavioural characteristic item.
4. sharing account detection method according to claim 3, which is characterized in that described to be expert at according to every account sample
It is characterized the characteristic value of item, determines the behavioural characteristic vector of every account sample, comprising:
It is corresponding in each characteristic value of characteristic value calculating of behavioural characteristic item according to the account sample for every account sample
Weight evidence weight values;
By the corresponding weight evidence weight values of each characteristic value of every account sample combine the behavioural characteristic for the account sample to
Amount.
5. sharing account detection method according to claim 4, which is characterized in that described to be directed to every account sample, root
The corresponding weight evidence weight values of each characteristic value are calculated in the characteristic value of behavioural characteristic item according to the account sample, comprising:
According to the quantity situation of characteristic value, behavioural characteristic item is divided into discrete behavioural characteristic item and Continuous behavior characteristic item;
For discrete behavioural characteristic item, the corresponding weight evidence weight values of each characteristic value of discrete behavioural characteristic item are calculated;
For Continuous behavior characteristic item, the selected characteristic value section from continuous characteristic value, and it is corresponding to calculate characteristic value section
Weight evidence weight values;
For each account sample, account sample evidence weight corresponding to the characteristic value of discrete behavioural characteristic item is obtained
Value, and determine that the account sample in the object feature value section that the characteristic value of Continuous behavior characteristic item is belonged to, and obtains
The corresponding weight evidence weight values in the object feature value section.
6. sharing account detection method according to claim 5, which is characterized in that described to be chosen from continuous characteristic value
Characteristic value section, comprising:
The item number of the corresponding account sample of each characteristic value of the Continuous behavior characteristic item is counted, to obtain each Continuous behavior
The characteristic value distribution situation of characteristic item;
According to the characteristic value distribution situation of the Continuous behavior characteristic item, in the characteristic value of the Continuous behavior characteristic item, really
The variation for determining account sample strip number meets the object feature value of preset condition;Wherein the number of the object feature value is N;
It is N+1 characteristic value section by the feature value division of the Continuous behavior characteristic item according to the object feature value.
7. sharing account detection method according to claim 5, which is characterized in that the calculating characteristic value section is corresponding
Weight evidence weight values, comprising:
Determine the quantity of the corresponding black account sample in characteristic value section and the quantity of corresponding white account sample;Wherein black account
Sample is the account sample for being marked as sharing account, and white account sample is the account sample for being not labeled as sharing account;
According to the ratio of the quantity of the corresponding black account sample in characteristic value section and the quantity of white account sample, each spy is calculated
The weight evidence weight values in value indicative section.
8. a kind of sharing account detection device characterized by comprising
Characteristic value acquisition module, for obtaining account to be detected in the characteristic value of preset behavioural characteristic item;
Rating Model obtains module, for obtaining account Rating Model trained in advance;The wherein ginseng of the account Rating Model
Number include each behavioural characteristic item and each corresponding weighted value of behavioural characteristic item, and the weighted value be by
Neural network model training algorithm is trained the markd account sample of tool;
Feature-value-score module, for the characteristic value of the account to be detected to be input in the account Rating Model, with
To the target score value of the account to be detected;
Share account detection module, for whether meeting default sharing condition according to the target score value, come determine it is described to
Whether detection account is to share account.
9. a kind of sharing account detection device, which is characterized in that including processor and memory, the processor is deposited by operation
Software program, the data of calling storage in the memory, at least execution following steps of storage in the memory:
Account to be detected is obtained in the characteristic value of preset behavioural characteristic item;
Obtain account Rating Model trained in advance;Wherein the parameter of the account Rating Model includes each behavioural characteristic
Item and each corresponding weighted value of behavioural characteristic item, and the weighted value is by neural network model training algorithm pair
Have what markd account sample was trained;
The characteristic value of the account to be detected is input in the account Rating Model, to obtain the mesh of the account to be detected
Mark score value;
Whether meet default sharing condition according to the target score value, to determine whether the account to be detected is to share account
Number.
10. a kind of storage medium, is stored thereon with computer program, which is characterized in that the computer program is held by processor
When row, sharing account detection method described in claim 1-7 any one is realized.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910445651.4A CN110175438A (en) | 2019-05-27 | 2019-05-27 | Share account detection method and relevant device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910445651.4A CN110175438A (en) | 2019-05-27 | 2019-05-27 | Share account detection method and relevant device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110175438A true CN110175438A (en) | 2019-08-27 |
Family
ID=67695873
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910445651.4A Pending CN110175438A (en) | 2019-05-27 | 2019-05-27 | Share account detection method and relevant device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110175438A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111104979A (en) * | 2019-12-18 | 2020-05-05 | 北京思维造物信息科技股份有限公司 | Method, device and equipment for generating user behavior value evaluation model |
CN111582448A (en) * | 2020-04-30 | 2020-08-25 | 广州市百果园信息技术有限公司 | Weight training method and device, computer equipment and storage medium |
CN111970250A (en) * | 2020-07-27 | 2020-11-20 | 深信服科技股份有限公司 | Method for identifying account sharing, electronic device and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103258039A (en) * | 2013-05-20 | 2013-08-21 | 中国地质大学(武汉) | Method for detecting false information on microblog |
CN103607419A (en) * | 2013-08-23 | 2014-02-26 | 合一网络技术(北京)有限公司 | High-quality user account anti-sharing method and system |
CN104954350A (en) * | 2014-03-31 | 2015-09-30 | 腾讯科技(深圳)有限公司 | Account information protection method and system thereof |
CN105915364A (en) * | 2015-12-08 | 2016-08-31 | 乐视网信息技术(北京)股份有限公司 | Method for preventing malicious sharing of user identity and device thereof |
CN107185243A (en) * | 2017-04-26 | 2017-09-22 | 厦门吉比特网络技术股份有限公司 | A kind of recognition methods of network game external store |
CN109635529A (en) * | 2018-11-13 | 2019-04-16 | 平安科技(深圳)有限公司 | Account shares detection method, device, medium and electronic equipment |
-
2019
- 2019-05-27 CN CN201910445651.4A patent/CN110175438A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103258039A (en) * | 2013-05-20 | 2013-08-21 | 中国地质大学(武汉) | Method for detecting false information on microblog |
CN103607419A (en) * | 2013-08-23 | 2014-02-26 | 合一网络技术(北京)有限公司 | High-quality user account anti-sharing method and system |
CN104954350A (en) * | 2014-03-31 | 2015-09-30 | 腾讯科技(深圳)有限公司 | Account information protection method and system thereof |
CN105915364A (en) * | 2015-12-08 | 2016-08-31 | 乐视网信息技术(北京)股份有限公司 | Method for preventing malicious sharing of user identity and device thereof |
CN107185243A (en) * | 2017-04-26 | 2017-09-22 | 厦门吉比特网络技术股份有限公司 | A kind of recognition methods of network game external store |
CN109635529A (en) * | 2018-11-13 | 2019-04-16 | 平安科技(深圳)有限公司 | Account shares detection method, device, medium and electronic equipment |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111104979A (en) * | 2019-12-18 | 2020-05-05 | 北京思维造物信息科技股份有限公司 | Method, device and equipment for generating user behavior value evaluation model |
CN111582448A (en) * | 2020-04-30 | 2020-08-25 | 广州市百果园信息技术有限公司 | Weight training method and device, computer equipment and storage medium |
CN111970250A (en) * | 2020-07-27 | 2020-11-20 | 深信服科技股份有限公司 | Method for identifying account sharing, electronic device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wang et al. | Truthful incentive mechanism with location privacy-preserving for mobile crowdsourcing systems | |
CN110009174A (en) | Risk identification model training method, device and server | |
CN103412918B (en) | A kind of service trust degree appraisal procedure based on service quality and reputation | |
CN109461070A (en) | A kind of risk measures and procedures for the examination and approval, device, storage medium and server | |
CN108269087B (en) | Method and device for processing position information | |
CN110175438A (en) | Share account detection method and relevant device | |
CN108833458B (en) | Application recommendation method, device, medium and equipment | |
CN106875110A (en) | Operational indicator layered calculation method and device, distributed computing method and system | |
CN108805391A (en) | Determine the method and device of high risk user | |
CN112016788A (en) | Risk control strategy generation and risk control method and device and electronic equipment | |
CN107657357B (en) | Data processing method and device | |
US20110119173A1 (en) | System and Method for Providing Intelligent Market Data Snapshots | |
CN109785167A (en) | Risk information methods of exhibiting, device, terminal and computer readable storage medium | |
CN113222326A (en) | Method and device for evaluating maturity of scientific and technological resource service platform | |
Wang et al. | The truthful evolution and incentive for large-scale mobile crowd sensing networks | |
CN112579910A (en) | Information processing method, information processing apparatus, storage medium, and electronic device | |
Li et al. | Capitalize your data: Optimal selling mechanisms for IoT data exchange | |
CN109075987A (en) | Optimize digital assembly analysis system | |
CN106897198A (en) | A kind of processing method and processing device of daily record data | |
CN110297948B (en) | Relational network construction method and device | |
CN103729369B (en) | The method and device of automatically processing coexisting orders | |
CN110706122A (en) | Method, device and readable medium for improving social security agent efficiency based on big data analysis | |
CN110263551A (en) | A kind of test method and device | |
CN113205391B (en) | Historical order matching degree based order dispatching method, electronic equipment and computer readable medium | |
CN110489394A (en) | Intermediate data processing method and equipment |
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 |