CN108804387A - Target user determines method and device - Google Patents
Target user determines method and device Download PDFInfo
- Publication number
- CN108804387A CN108804387A CN201710287454.5A CN201710287454A CN108804387A CN 108804387 A CN108804387 A CN 108804387A CN 201710287454 A CN201710287454 A CN 201710287454A CN 108804387 A CN108804387 A CN 108804387A
- Authority
- CN
- China
- Prior art keywords
- user
- sequence
- main body
- convolution
- target user
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of target users to determine method and device.This method includes：Generate operation frequency sequence, wavelet decomposition is carried out to the operation frequency sequence, main body sequence is obtained, convolution algorithm is carried out by increasing sequence successively to each element in the main body sequence, when the result of the convolution algorithm meets preset condition, determine that the user is target user.Operation frequency sequence of the user on the specified type page is handled by wavelet decomposition, noise can more accurately be removed, and decomposite the variation profile (i.e. main running signal) that can accurately indicate the operation frequency of the user in the specified type page, result by the way that main running signal and increasing sequence to be carried out to convolution algorithm can judge whether user has the tendency that operating the frequency obviously rises or falls, to judge whether the user is target user, the accuracy of target user's detection is improved.
Description
Technical field
The present invention relates to net application technology field, more particularly to a kind of target user determines method and device.
Background technology
With the continuous development of net application technology, many Internet service providers can be by analyzing user in Webpage
Operation behavior determine whether user is to the interested target object of a certain things, targetedly to be pushed away into row information
It send.
In the related art, the server of Internet service provider can count in nearest a period of time user in a certain things pair
The number of browse operation in the page answered, when the number of browse operation of the user in the corresponding page of the things has reached
When certain numerical value, it may be determined that the user is the corresponding target user of the things.For example, by taking the page of automotive media as an example,
Server can be to browse the number of the page of automotive media, when the user is within a nearest week in a counting user nearest week
When the number of the page of browsing automotive media reaches 10 times, it can be confirmed that the user is the target user for having purchase vehicle intention.
It is above-mentioned in the related technology, server is only corresponding to a certain things within nearest a period of time by counting user
The number of the browse operation of the page determines whether the user is target user, and the accuracy that target user determines is relatively low.
Invention content
In order to solve in the prior art only by counting user nearest interior to the corresponding page of a certain things for a period of time
The number of browse operation determine whether the user is target user, cause the accuracy that target user determines is lower to ask
Topic, an embodiment of the present invention provides a kind of target users to determine that method and device, technical solution are as follows：
In a first aspect, providing a kind of target user determines method, the method includes：
Operation frequency sequence, the operation frequency sequence are generated to the historical operation record of the specified type page according to user
In each element indicate respectively the user in each unit interval to the operation frequency of the specified type page, and
Each element being ranked sequentially after arriving first according to the corresponding time；
Wavelet decomposition is carried out to the operation frequency sequence, obtains main body sequence, the main body sequence is and the operation
The corresponding sequence of low frequency minor element in frequency sequence；
Convolution algorithm is carried out successively to each element in the main body sequence by increasing sequence；
When the result of the convolution algorithm meets preset condition, determine that the user is target user.
Second aspect, provides a kind of target user's determining device, and described device includes：
Generation module, for generating operation frequency sequence, institute to the historical operation record of the specified type page according to user
The each element stated in operation frequency sequence indicates respectively the user in each unit interval to the specified type page
The operation frequency in face, and each element according to the corresponding time being ranked sequentially after arriving first；
Decomposing module obtains main body sequence, the main body sequence for carrying out wavelet decomposition to the operation frequency sequence
It is sequence corresponding with the low frequency element in the operation frequency sequence；
Convolution module, for carrying out convolution algorithm successively to each element in the main body sequence by increasing sequence；
Determining module, for when the result of the convolution algorithm meets preset condition, determining that the user uses for target
Family.
Technical solution provided in an embodiment of the present invention can include the following benefits：
Operation frequency sequence of the user on the specified type page is handled by wavelet decomposition, it can be more accurate
Noise is removed, and decomposites variation profile (the i.e. main body that can accurately indicate the operation frequency of the user in the specified type page
Signal), the result by the way that main running signal and increasing sequence to be carried out to convolution algorithm can judge whether user has the operation frequency
The trend obviously risen or fallen improves the accuracy of target user's detection to judge whether the user is target user.
It should be understood that above general description and following detailed description is only exemplary and explanatory, not
It can the limitation present invention.
Description of the drawings
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the present invention
Example, and be used to explain the principle of the present invention together with specification.
Fig. 1 is the structural schematic diagram that a kind of target user shown according to an exemplary embodiment determines system；
Fig. 2 is the flow chart that a kind of target user shown according to an exemplary embodiment determines method；
Fig. 3 is a kind of schematic diagram for wavelet decomposition that embodiment shown in Fig. 2 is related to；
Fig. 4 is a kind of schematic diagram of the realization process of determining target user；
Fig. 5 is a kind of block diagram of target user's determining device shown according to an exemplary embodiment；
Fig. 6 is a kind of structural schematic diagram of server shown according to an exemplary embodiment.
Specific implementation mode
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent and the consistent all embodiments of the present invention.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects being described in detail in claims, of the invention.
Fig. 1 is the structural schematic diagram that a kind of target user shown in an of the invention exemplary embodiment determines system.This is
System includes：Several user terminals 120 and server cluster 140.
User terminal 120 can be mobile phone, tablet computer, E-book reader, MP3 player (Moving Picture
Experts Group Audio Layer III, dynamic image expert's compression standard audio level 3), MP4 (Moving
Picture Experts Group Audio Layer IV, dynamic image expert's compression standard audio level 4) player, knee
Mo(u)ld top half pocket computer and desktop computer etc..
It is connected by communication network between user terminal 120 and server cluster 140.Optionally, communication network is wired
Network or wireless network.
Server cluster 140 is a server, or by several servers or a virtual platform, or
Person is a cloud computing service center.Optionally, server cluster 140 may include determining platform for realizing target user
142 server, optionally, server cluster 140 further include for realizing the server of page management platform 144；Optionally,
Server cluster 140 further includes user operation records management server 146.
Optionally, page management platform 144 includes：For push and safeguard Webpage server, for manage and
Store the server of each user account.By logical between page management platform 144 and user operation records management server 146
Communication network is connected.
Optionally, user operation records management server 146 includes：For operation row of the counting user in each page
For server, the server for storing operation behavior of the user in each page.
Optionally, user operation records management server 146, can be from page management under the premise of user's vesting assent
Platform 144, alternatively, operation behavior data of the user in each page are obtained from other associated page management platforms, and
According to historical operation record of the operation note counting user got in each page.
Optionally, which can also include management equipment 160, lead between the management equipment 160 and server cluster 140
Communication network is crossed to be connected.Optionally, communication network is cable network or wireless network.
Optionally, above-mentioned wireless network or cable network use standard communication techniques and/or agreement.Network be usually because
Special net, it may also be any network, including but not limited to LAN (Local Area Network, LAN), Metropolitan Area Network (MAN)
(Metropolitan Area Network, MAN), wide area network (Wide Area Network, WAN), mobile, wired or nothing
Any combinations of gauze network, dedicated network or Virtual Private Network).In some embodiments, using including hypertext markup
Language (Hyper Text Mark-up Language, HTML), extensible markup language (Extensible Markup
Language, XML) etc. technology and/or format represent the data by network exchange.It can additionally use such as safe
Socket layer (Secure Socket Layer, SSL), Transport Layer Security (Trassport Layer Security, TLS), void
Quasi- dedicated network (Virtual Private Network, VPN), Internet Protocol Security (Internet Protocol
Security, IPsec) etc. conventional encryption techniques encrypt all or some links.In further embodiments, can also make
Replace or supplement above-mentioned data communication technology with customization and/or the exclusive data communication technology.
Wavelet decomposition is a kind of method carrying out signal processing, by flexible shift operations to signal (or being function)
Multi-scale refinement is gradually carried out, high frequency treatment time subdivision is finally reached, frequency is segmented at low frequency, can adapt to time frequency signal point automatically
The requirement of analysis is well suited for analyzing non-stationary signal (signal that i.e. regularity of distribution changes over time).
When user generates interest to a certain part things, it will usually search for and browse and the relevant webpage of the things in a network
The page, to promote understanding to the things.Also, the continuous intensification with user to the interest of the things, user browsing
The frequency of the relevant page of the things also can be in apparent ascendant trend.Correspondingly, emerging to something or other sense when if user just starts
Interest gradually lost interest to the things later, then the frequency that user browses the relevant page of the things can be in apparent declines
Trend.
Such as, it is to be understood that the user of automobile related news, it will usually browse oneself on the page that automotive media provides
The information needed, wherein have purchase vehicle intention the browsing frequency of the user in the page that automotive media provides would generally it is gradual on
It rises.I.e. when a user just begins preparing for buying car, the browsing frequency in the page that automotive media provides can be relatively low, with
The wish for user's buying car is more and more stronger, and the browsing frequency in the page that automotive media provides generally also can be increasingly
It is high.Conversely, if user has just purchased vehicle completion, the browsing frequency in the page that automotive media provides can be relatively high, with purchase
Time passage after vehicle, the browsing frequency of the user in the page that automotive media provides generally also can be lower and lower.And this
The basic thought for inventing each embodiment is exactly the frequency by analyzing operation behavior of the user in the relevant page of a certain things
It is secondary, and the frequency of the operation behavior in the relevant page of a certain things is determined as in the user of rising or downward trend to this
The corresponding target user of things, to improve the accuracy of target user's determination.
Due to the individual internet behavior difference of different user, it is relevant which is browsed to the interested user of a certain things
The frequency of the page is also different in the concrete condition of rising or downward trend.For example, for having intention the user of purchase vehicle, some
The browsing frequency for the page that " fast heat " user may provide automotive media in one week has obviously ascendant trend；And have
" slow heat " user may then be gradually increasing the browsing frequency for the page that automotive media provides within one or two months a bit.In addition, with
Family within some shorter time in the navigation process of the relevant page of a certain things, may suddenly improve in the things phase
The operation frequency in the page of pass then reduces the operation frequency in the relevant page of the things in a long time again, this
The operation frequency that kind improves suddenly is not representative for the analysis of the operation behavior of user, belongs to user's operation row
For the interference information (also referred to as noise) in analytic process.
If can be seen that user by the analysis of the above-mentioned operation frequency to user in the relevant page of something or other
In each unit interval, the operation frequency in the relevant page of something or other is arranged sequentially in time, is obtained
Sequence can regard a non-stationary signal with two dimensions of time and frequency as, and may be mingled in the signal
Therefore noise handles above-mentioned sequence by wavelet decomposition, can more accurately remove noise, and decompositing can be accurate
User is really indicated in the variation profile (i.e. main running signal) to the operation frequency in the relevant page of something or other, by believing main body
Number analysis can detect whether the user is the corresponding target user of the things.For example, when analyzing main running signal,
Convolution algorithm can be carried out using convolution mask to main running signal, and judge whether the user is target according to convolution algorithm result
User.
Fig. 2 is the flow chart that a kind of target user shown according to an exemplary embodiment determines method, to be applied to such as
For server cluster in system shown in FIG. 1, which determines that method may include the following steps：
Step 201, operation frequency sequence is generated to the historical operation record of the specified type page according to user.
Wherein, each element in the operation frequency sequence indicates respectively user in each unit interval to specifying class
The operation frequency of the type page, and each element according to the corresponding time being ranked sequentially after arriving first.
The above-mentioned specified type page can be the relevant page of some things, which can be the single page, can also
Including multiple pages, also, the page can be HTML (the Hyper Text Markup opened by browser
Language, hypertext markup language) page, alternatively, the page can also be the built-in Web page opened by application program
Face.
In embodiments of the present invention, server cluster can obtain user and remember to the historical operation of the above-mentioned specified type page
Record, historical operation record may include operating time of the user to all previous operation behavior of the specified type page, server
Cluster can generate the corresponding operation frequency sequence of the user according to the historical operation record got.
By taking unit interval is 1 day as an example, aforesaid operations frequency sequence can be user in continuous several days, right respectively
Sequence made of the number of the operation of the specified type page arranges sequentially in time.For example, user is in nearest 2*2*N days
Operation frequency sequence for, which can be expressed as：
X=[x1, x2, x3, x4..., x2*2*N-1, x2*2*N]。
Wherein, x1In be user in nearest 2*2*N days the 1st day, to the frequency of the operation of the specified type page；x2
In be user in nearest 2*2*N days the 2nd day, to the frequency of the operation of the specified type page；And so on, x2*2*NFor with
Family is in last 1 day in nearest 2*2*N days, to the frequency of the operation of the specified type page.
In practical applications, there are many kinds of the type possibility of operation of the user in the page, for example, user can be in the page
Middle execution is clicked browsing (including click browsing text, picture or video cartoon etc.), is compared (such as to two or more
The attribute of object is compared), make comments and contact the operations such as customer service.In embodiments of the present invention, server cluster can
To do same processing to different types of user's operation, for example, server cluster in counting user in some unit interval
When to the frequency of the operation of the specified type page, ignore the style differences between different operations, by each operation point of user
It Tong Ji not be 1 frequency.
Specifically, assuming in some unit interval, user clicks in the specified type page and has browsed 50 times, compares
3 times, delivered 2 comments, 1 customer service of taken in conjunction, then server cluster count the user in the unit interval
The frequency of operation is 50+3+2+1=56 times.
In alternatively possible realization method, server cluster, can be to not when generating aforesaid operations frequency sequence
The user's operation of same type does differentiation processing, for example, the historical operation that server cluster is got is recorded in addition to including user
Further include corresponding action type except the operating time of the operation behavior of the specified type page, server cluster can root
It is recorded according to the historical operation and counts the user in each unit interval, the number of operations of each corresponding action type；Service
Device cluster obtains the weight coefficient of each action type, will for each unit interval in above-mentioned each unit interval
The number of operations of each action type and the sum of products of the weight coefficient of each action type are determined as user in the unit
Between the operation frequency in section, finally the operation frequency by user in each unit interval according to the corresponding time after arriving first
Be ranked sequentially, you can to obtain aforesaid operations frequency sequence.
For example, assuming that the weight coefficient for clicking browsing is 1, the weight coefficient of comparison is 2, and the weight coefficient made comments is
5, the weight coefficient for contacting customer service is 10, and in some unit interval, user clicks in the specified type page and browses
50 times, than right 3 times, 2 comments are delivered, 1 customer service of taken in conjunction, then server cluster can be by the user in the unit
The frequency statistics of operation in period are 50*1+3*2+2*5+1*10=76 times.
Wherein, the weight coefficient of each above-mentioned action type can be first passed through management equipment in advance in server by administrative staff
It is arranged in cluster and preserves, in use, administrative staff can also repaiies above-mentioned weight coefficient by management equipment
Change update.
Step 202, wavelet decomposition is carried out to the operation frequency sequence, obtains main body sequence, which is and the behaviour
Make the corresponding sequence of low frequency element in frequency sequence.
Referring to FIG. 3, it illustrates a kind of schematic diagrames of wavelet decomposition.It is to carry out 3 wavelet decompositions to original signal
Example, as shown in figure 3, in the 1st wavelet decomposition, cA1 and cD1 is decomposed by original signal, wherein cA1 corresponding is original
Low frequency signal in signal, it is high-frequency signal in original signal that cD1 is corresponding.In the 2nd wavelet decomposition, cA1 is decomposed
For cA2 and cD2, wherein it is low frequency signal in cA1 that cA2 is corresponding, and it is high-frequency signal in cA1 that cD2 is corresponding.At the 3rd time
When wavelet decomposition, cA2 is decomposed into cA3 and cD3, wherein it is low frequency signal in cA2 that cA3 is corresponding, and cD3 corresponding is cA2
In high-frequency signal.After 3 wavelet decompositions, the decomposition result of acquisition is C and L, wherein C is to decompose 4 letters obtained
Number, respectively cA3, cD3, cD2 and cD1, L are 4 respective signal lengths of signal that above-mentioned decomposition obtains.
Wherein, above-mentioned signal cA3 represents the profile and general picture of original signal, and cD3, cD2, cD1 respectively represent original signal
Detailed information, the frequency highest of wherein cD1, cD2 and cD3 take second place, and the frequency of cA3 is minimum.Under normal conditions, work as original signal
In contain noise when, after wavelet decomposition, noise is concentrated in high frequency section.
It in embodiments of the present invention, can be using the operation frequency sequence X that step 201 obtains as original signal, to the behaviour
Make frequency sequence X and carry out wavelet decomposition, you can to obtain at least two groups sequence, wherein the low frequency in respective operations frequency sequence
The sequence of minor element is main body sequence, is considered as noise if the accidental number of visits of user fluctuated, after wavelet decomposition, behaviour
Main body sequence will be separated by making the noise in frequency sequence X, at this point, the main body sequence can accurately indicate user to referring to
Determine the variation tendency of the operation frequency of type page.
It optionally, in embodiments of the present invention, can be according to the behaviour when carrying out wavelet decomposition to operation frequency sequence X
The number for making the element in frequency sequence X determines decomposition number, and according to the determining decomposition number to the operation frequency sequence
Carry out wavelet decomposition.
Fig. 3 is illustrated for carrying out 3 wavelet decompositions to original signal, in embodiments of the present invention, to operation frequency
When secondary sequence X carries out wavelet decomposition, 3 wavelet decompositions can be carried out to operation frequency sequence X, alternatively, can also be to operation frequency
Secondary sequence X carries out the fewer or more wavelet decompositions of number, and the low frequency minor element that last 1 wavelet decomposition is obtained is corresponding
Sequence based on retrieval.Specifically, in order to reduce the complexity subsequently calculated, the length for the main body sequence that wavelet decomposition obtains
Spending (number of the element i.e. in main body sequence) cannot be too long, that is, decomposing number cannot be very little；Meanwhile in order to ensure main body sequence
It can accurately indicate variation tendency of the user to the operation frequency of the specified type page, the length for the main body sequence that wavelet decomposition obtains
Degree can not be too short, that is, decomposing number cannot be too many；It therefore, in embodiments of the present invention, can be according in operation frequency sequence X
The element number length of frequency sequence X (operate) determine decomposition number, the element number operated in frequency sequence X gets over
More, decomposition number is more, correspondingly, the element number in operation frequency sequence X is fewer, it is also fewer to decompose number.
Specifically, the correspondence decomposed between number and element number section can be pre-set in server cluster,
When carrying out wavelet decomposition to operation frequency sequence X, the element section where the element number of operation frequency sequence X is determined, and
The decomposition number inquired the decomposition number corresponding to determining element section, and obtained according to inquiry to operation frequency sequence X into
Row decomposes.
For example, can be pre-set in server cluster decompose correspondence between number and element number section can be with
As shown in table 1：
| Decompose number | 1 | 2 | 3 | … |
| Element number section | (30,90] | (90,360] | (360,1080] | … |
Table 1
As shown in table 1, when the number for operating the element in frequency sequence X be in section (30,90] when, server cluster is true
The fixed number that wavelet decomposition is carried out to operation frequency sequence X is 1 time, when the number for operating the element in frequency sequence X is in area
Between (90,360] when, server cluster determines that the number that wavelet decomposition is carried out to operation frequency sequence X is 2 times, when the operation frequency
The number of element in sequence X be in section (360,1080] when, server cluster determine to operation frequency sequence X carry out small echo
The number of decomposition is 3 times, and so on.
For example, for decomposing number and be 2 times, server cluster is to operating frequency sequence X=[x1, x2, x3, x4...,
x2*2*N-1, x2*2*N] carry out 2 times after decomposing, the decomposition result of acquisition is：
CA2=[a1, a2, a3..., aN]；
CD2=[d21, d22, d23..., d2N]；
DD1=[d11, d12, d13..., d12*N]。
Wherein, cA2 is the main body sequence decomposed and obtained.
Step 203, convolution algorithm is carried out by increasing sequence successively to each element in the main body sequence.
In embodiments of the present invention, server cluster can do convolution algorithm respectively to each element in main body sequence,
Specifically, for i-th of element in the main body sequence, server cluster can be calculated by the following formula i-th of element
Corresponding convolution results：
Wherein, F (i) is the corresponding convolution results of i-th of element, and M is the number of the element in the increasing sequence, ai+h
For the elements of the i-th+h in the main body sequence,For (M+1+2h)/2 element in the increasing sequence, M >=3, and M
For odd number, (M-1)/2+1≤i≤N- (M-1)/2, N is the number of the element in the main body sequence, i+h≤N, and i and N are
Positive integer.
Optionally, in the scheme shown in the embodiment of the present invention, server cluster can be selected according to actual applicable cases
Suitable increasing sequence is selected, for example, server can obtain the increasing sequence according to the number of the element in the main body sequence, and
Convolution algorithm is carried out to the element in the main body sequence by the increasing sequence got.
Specifically, several groups increasing sequence can be set in server cluster, the number of the element in every group of increasing sequence
Difference, for example, respectively the number of the element in increasing sequence can be 3,5,7 respectively or more, meanwhile, server
The correspondence between increasing sequence and element number section can be pre-set in cluster, server cluster does main body sequence
When convolution algorithm, it is first determined the element section where the element number of main body sequence, and it is right to inquire determining element section institute
The increasing sequence answered, and convolution algorithm is done to main body sequence according to the increasing sequence inquired.
Wherein, above-mentioned increasing sequence is the arithmetic sequence of the element centered on 0.For example, can be arranged in server cluster
Two groups of increasing sequences, respectively：
5 dimensions (i.e. element number is 5) sequence：[- 2, -1,0,1,2]；
And 7 dimension (i.e. element number be 7) sequence：[- 3, -2, -1,0,1,2,3].
For example, to main body sequence cA2=[a1, a2, a3..., aN] carry out convolution algorithm when, with the increasing sequence of selection
For above-mentioned 5 dimension sequence, i.e., the numerical value of above-mentioned M is 5, then 3≤i≤N-2, and server cluster is from a3Start, in cA2
Each element carries out convolution algorithm respectively, until for a in cA2N-2Operation is carried out to terminate.Specifically：
For a3When carrying out convolution algorithm (i=3), F (3)=- 2a1-a2+0*a3+a4+2a5；
For a4When carrying out convolution algorithm (i=4), F (4)=- 2a2-a3+0*a4+a5+2a6；
And so on, for aN-2When carrying out convolution algorithm (i=N-2), F (N-2)=- 2aN-4-aN-3+0*aN-2+aN-1+
2aN。
Step 204, when the result of the convolution algorithm meets preset condition, target user is determined that the user is.
In above-mentioned steps 203, when server cluster carries out convolution algorithm every time, by a in main body sequenceiIt is front and back several
Element carries out convolution algorithm with increasing sequence, if aiSeveral front and back elements are incremental (i.e. aiBefore corresponding unit interval
Afterwards in a period of time, the operation frequency of user gradually increases), then convolution algorithm result can be bigger at this time, if for example, progressive sequence
Row are the arithmetic sequences of the element centered on 0, then convolution algorithm result at this time can be a larger positive value；If aiIt is front and back
Several elements difference less (i.e. aiBefore and after corresponding unit interval in a period of time, the operation frequency of user keeps certain
Quantity and fluctuation it is smaller), then convolution algorithm result can be moderate at this time, if for example, increasing sequence be centered on 0 element etc.
Difference sequence, then convolution algorithm result at this time can be close to 0；If aiSeveral front and back elements are (the i.e. a to successively decreaseiCorresponding list
Before and after the period of position in a period of time, the operation frequency of user gradually decreases), then convolution algorithm result can be smaller at this time, for example,
If increasing sequence is the arithmetic sequence of the element centered on 0, convolution algorithm result at this time can be a smaller negative value.
Based on the feature of above-mentioned convolution algorithm result, in embodiments of the present invention, server cluster is carrying out convolution algorithm
When, in each element of aforementioned body sequence, the corresponding convolution results of either element meet preset condition, you can are somebody's turn to do with determining
User is target user.
Specifically, by taking above-mentioned either element is i-th of element in main body sequence as an example, when target user is to specifying class
When the interest of the corresponding object of type page user on the rise, server cluster calculates i-th of element in main body sequence
After corresponding convolution results, if calculating the convolution results F (i) obtained is more than the first predetermined threshold value, then it is assumed that user is above-mentioned i-th
Before and after the corresponding unit interval of a element in a period of time, have on apparent to the interest of the corresponding object of the specified type page
The trend of liter, at this point it is possible to determine that the user is target user.
In practical applications, user to the interest of a certain things such as time fluctuation, for example, whithin a period of time,
User is interested in a certain things, and in a period of time later, user may lose interest again to the things, after a period of time
Afterwards, user may will produce interest again to the things, therefore, if the corresponding time gap current time of i-th of element is longer,
F (i) to user currently whether indicative significance interested in the things or losing interest can be smaller, if conversely, i-th yuan
The corresponding time gap current time of element is closer, then whether F (i) is currently interested in the things or lose interest in user
Indicative significance can be larger.Therefore, in alternatively possible realization method, when target user is corresponding to the specified type page
When the interest of object user on the rise, server cluster calculates the corresponding convolution knot of i-th of element in main body sequence
After fruit, if calculate obtain convolution results F (i) be more than the second predetermined threshold value, and the corresponding unit interval of i-th of element away from
Duration from current time is less than the first preset duration, then it is assumed that user is right within a period of time closer apart from current time
The interest of the corresponding object of the specified type page has apparent ascendant trend, at this point it is possible to determine that the user is target user.
Alternatively, when target user is that have the user of downward trend to the interest of the corresponding object of the specified type page, if
It calculates the convolution results F (i) obtained and is less than third predetermined threshold value, then it is assumed that user is in the corresponding unit of above-mentioned i-th of element
Between before and after section in a period of time, significantly decrease trend to the interest of the corresponding object of the specified type page, at this point, can also
Determine that the user is target user.Alternatively, when target user is that have decline to become the interest of the corresponding object of the specified type page
When the user of gesture, if convolution results F (i) is less than four predetermined threshold values, and the corresponding unit interval segment distance of i-th of element
The duration at current time is less than the second preset duration, then it is assumed that user is within a period of time closer apart from current time, to referring to
The interest for determining the corresponding object of type page significantly decreases trend, at this point, target user can also be determined that the user is.
Wherein, above-mentioned first predetermined threshold value, the second predetermined threshold value, third predetermined threshold value, the 4th predetermined threshold value, first are preset
Duration and the second preset duration can be the ginsengs that administrative staff are manually arranged in server cluster in advance by management equipment
Number, also, administrative staff can also modify or update to above-mentioned parameter during operation.
Optionally, server cluster can first calculate subject sequence successively when executing above-mentioned steps 203 and step 204
The corresponding convolution results of each element in row, then further according to the corresponding convolution results of each element determine the user whether be
Target user.
In alternatively possible realization method, server cluster can also be in one in calculating main body sequence every time
After the corresponding convolution results of element, i.e., judge whether the user is target user by the corresponding convolution results of the element, if so,
Then stop follow-up convolutional calculation process, otherwise, continues to calculate the corresponding convolution results of next element.
For example, if target user is the user on the rise to the interest of the corresponding object of the specified type page, alternatively,
Target user is the user for having downward trend to the interest of the corresponding object of the specified type page, then server cluster is to subject sequence
When each element in row carries out convolutional calculation successively, a convolution results are often calculated, i.e., being determined according to the convolution results should
Whether user is target user, if so, stopping subsequent convolutional calculation process, otherwise continues to execute subsequent convolutional calculation.
If desired two class target users are determined simultaneously, for example, having rising to the interest of the corresponding object of the specified type page
The user of trend is first kind target user, has the user of downward trend to be the interest of the corresponding object of the specified type page
Second Type target user, then server cluster during carrying out convolutional calculation successively to each element in main body sequence,
Calculating a convolution results, and according to the convolution results determine the user be in two class target users a kind of target user it
Afterwards, it is also necessary to continue to execute subsequent convolutional calculation, the user is in two class target users if being determined in subsequent calculating process
Another kind of target user, then server can stop subsequent convolutional calculation process.
Optionally, the scheme above the embodiment of the present invention is with the corresponding convolution results of either element in main body sequence
Meet and illustrate for preset condition, in practical applications, the result by convolution algorithm come determine user whether be
When target user, other determining methods can also be used, for example, when corresponding convolution algorithm result meets the member of preset condition
The quantity of element is not less than the first amount threshold, alternatively, when corresponding convolution algorithm result meets preset condition and adjacent element
Quantity be not less than the second amount threshold when, it may be determined that the user is target user.
Optionally, server cluster determines active user as that after target user, can be provided targetedly to the user
Service, for example, targetedly pushing the corresponding message (such as advertisement information etc.) of the specified type page to the user.
In conclusion target user provided in an embodiment of the present invention determines method, by wavelet decomposition to user specified
Operation frequency sequence in type page is handled, and noise can be more accurately removed, and is decomposited and can accurately be indicated to use
The variation profile (i.e. main running signal) of the operation frequency of the family in the specified type page, by by main running signal and increasing sequence into
The result of row convolution algorithm can judge whether user has the tendency that operating the frequency obviously rises or falls, to judge the use
Whether family is target user, improves the accuracy of target user's detection.
Specifically, referring to FIG. 4, it illustrates a kind of schematic diagrames of the realization process of determining target user.With above-mentioned reality
Target user in the server cluster that now process can be as shown in Figure 1 determines platform 142, page management platform 144 and uses
The target user that family operation note management server 146 is realized, and determined is to have the user of purchase vehicle intention or purchased vehicle to complete
User for, as shown in figure 4, page management platform 144 be automotive media website platform, safeguard in the automotive media website
Each page, and record operation behavior and operating time of each user in each page of the automotive media website,
And the operation behavior of record and operating time are sent to user operation records management server 146.User operation records pipe
Manage server 146 to operation behavior of each user in each page of the automotive media website and operating time respectively into
Row statistics obtains the corresponding historical operation record of each user, the corresponding user's account of historical operation record that statistics is obtained
It number is stored.
When determining target user, target user determines user account of the platform 142 by some user, from user's operation
The historical operation record of the user is extracted in record management server 146, and statistics life is recorded according to the historical operation of the user
At by nearest several days, the operation frequency of the user daily in each page of automotive media website is arranged sequentially in time
The operation frequency sequence of row.Target user determines platform 142 according to the operation frequency sequence of generation determination decomposition number, and according to
Determining decomposition number carries out wavelet decomposition to operation frequency sequence according to preset decomposition algorithm, and last time is decomposed
It obtains, sequence based on the retrieval of corresponding low frequency minor element.
Target user determines that platform 142 selects a preset increasing sequence according to the length of the main body sequence, and will pass
Increasing sequence does convolution algorithm respectively with each element in main body sequence, obtains convolution algorithm as a result, and according to convolution algorithm knot
Magnitude relationship between fruit and pre-set threshold value determines whether the user is target user.
Specifically, when determining the target user for having purchase vehicle intention, if i-th of element in aforementioned body sequence is corresponding
Convolution algorithm result is more than pre-set threshold value A, and the corresponding time gap current time of i-th of element is closer, then may be used
To think that the user has purchase vehicle intention, which can be determined as to the target user for having purchase vehicle intention.
Alternatively, when determining the target user for having purchased vehicle completion, if the corresponding volume of i-th of element in aforementioned body sequence
Product operation result is less than pre-set threshold value B, and the corresponding time gap current time of i-th of element is closer, then can be with
Think that the user has just purchased vehicle completion, which can be determined as having purchased the target user of vehicle completion.
After determining that the user is target user, server cluster can be provided to the user and targetedly be serviced, for example,
After determining that the user is the target user of purchase vehicle intention, server cluster can further analyze the page pair of user's browsing
It answers which brand and vehicle, and advertisement that the user may need is matched according to analysis result and is pushed to the user.Or
Person, when determine that the user is purchased vehicle completion target user after, server cluster can to the user push Motor Maintenance or
Insure relevant advertisement.
Fig. 5 is a kind of block diagram of target user's determining device shown according to an exemplary embodiment.The target
User's determining device some or all of can be embodied as in server cluster by way of hardware or soft or hard combination, to hold
All or part of step in row embodiment illustrated in fig. 2.Target user's determining device may include：
Generation module 501, for generating operation frequency sequence to the historical operation record of the specified type page according to user,
Each element in the operation frequency sequence indicates respectively the user in each unit interval to the specified type
The operation frequency of the page, and each element according to the corresponding time being ranked sequentially after arriving first；
Decomposing module 502 obtains main body sequence, the main body for carrying out wavelet decomposition to the operation frequency sequence
Sequence is sequence corresponding with the low frequency element in the operation frequency sequence；
Convolution module 503, for carrying out convolution fortune successively to each element in the main body sequence by increasing sequence
It calculates；
Determining module 504, for when the result of the convolution algorithm meets preset condition, determining that the user is target
User.
Optionally, the convolution module, is used for
For i-th of element in the main body sequence, it is calculated by the following formula the corresponding convolution of i-th of element
As a result：
Wherein, F (i) is the corresponding convolution results of i-th of element, and M is the number of the element in the increasing sequence,
ai+hFor the elements of the i-th+h in the main body sequence,For (M+1+2h)/2 element in the increasing sequence, M
>=3, and it is integer that M, which is odd number, (M-1)/2+1≤i≤N- (M-1)/2, and i,.
Optionally, the increasing sequence is the arithmetic sequence of the element centered on 0.
Optionally, the determining module, for meeting when the corresponding convolution results of either element in each element
When the preset condition, determine that the user is target user.
Optionally, when the target user is on the rise to the interest of the corresponding object of the specified type page
When user, the preset condition includes：
The convolution results are more than the first predetermined threshold value；Alternatively, the convolution results are more than the second predetermined threshold value, and it is described
The duration at either element corresponding unit interval segment distance current time is less than the first preset duration；
Optionally, when the target user is that have downward trend to the interest of the corresponding object of the specified type page
When user, the preset condition includes：
The convolution results are less than third predetermined threshold value；Alternatively, the convolution results are less than the 4th predetermined threshold value, and it is described
The duration at either element corresponding unit interval segment distance current time is less than the second preset duration.
Optionally, the generation module, including：
Acquiring unit is recorded, the historical operation for obtaining the user records, and the historical operation record includes described
Action type and operating time of the user to the operation behavior of the specified type page；
Statistic unit, it is corresponding for counting the user in each unit interval according to historical operation record
The number of operations of each action type；
Coefficient acquiring unit, the weight coefficient for obtaining each action type；
Frequency determination unit, for for each unit interval in each unit interval, will it is described each
The sum of products of the number of operations of action type and the weight coefficient of each action type is determined as the user described
The operation frequency in unit interval；
Arrangement units are used for the operation frequency by the user in each unit interval according to the corresponding time
Being ranked sequentially after arriving first obtains the operation frequency sequence.
Optionally, described device further includes：
Acquisition module, for the convolution module by increasing sequence to each element in the main body sequence successively
Before carrying out convolution algorithm, the increasing sequence is obtained according to the number of the element in the main body sequence.
Optionally, the decomposing module, including：
Number determination unit decomposes number for being determined according to the number of the element in the operation frequency sequence；
Resolving cell, for carrying out wavelet decomposition to the operation frequency sequence according to the determining decomposition number.
In conclusion target user's determining device provided in an embodiment of the present invention, by wavelet decomposition to user specified
Operation frequency sequence in type page is handled, and noise can be more accurately removed, and is decomposited and can accurately be indicated to use
The variation profile (i.e. main running signal) of the operation frequency of the family in the specified type page, by by main running signal and increasing sequence into
The result of row convolution algorithm can judge whether user has the tendency that operating the frequency obviously rises or falls, to judge the use
Whether family is target user, improves the accuracy of target user's detection.
Fig. 6 is a kind of structural schematic diagram of server shown according to an exemplary embodiment.The server 600 includes
Central processing unit (CPU) 601 including the storage of the system of random access memory (RAM) 602 and read-only memory (ROM) 603
Device 604, and connect the system bus 605 of system storage 604 and central processing unit 601.The server 600 further includes
The basic input/output (I/O systems) 606 of information is transmitted between each device in help computer, and for storing
The mass-memory unit 607 of operating system 613, application program 614 and other program modules 615.
The basic input/output 606 includes display 608 for showing information and inputs letter for user
The input equipment 609 of such as mouse, keyboard etc of breath.The wherein described display 608 and input equipment 609 are all by being connected to
The input and output controller 610 of system bus 605 is connected to central processing unit 601.The basic input/output 606
Can also include input and output controller 610 for receive and handle from keyboard, mouse or electronic touch pen etc. it is multiple its
The input of his equipment.Similarly, input and output controller 610 also provides output to display screen, printer or other kinds of defeated
Go out equipment.
The mass-memory unit 607 is by being connected to the bulk memory controller (not shown) of system bus 605
It is connected to central processing unit 601.The mass-memory unit 607 and its associated computer-readable medium are server
600 provide non-volatile memories.That is, the mass-memory unit 607 may include such as hard disk or CD-ROM
The computer-readable medium (not shown) of driver etc.
Without loss of generality, the computer-readable medium may include computer storage media and communication media.Computer
Storage medium includes information such as computer-readable instruction, data structure, program module or other data for storage
The volatile and non-volatile of any method or technique realization, removable and irremovable medium.Computer storage media includes
RAM, ROM, EPROM, EEPROM, flash memory or other solid-state storages its technologies, CD-ROM, DVD or other optical storages, tape
Box, tape, disk storage or other magnetic storage apparatus.Certainly, skilled person will appreciate that the computer storage media
It is not limited to above-mentioned several.Above-mentioned system storage 604 and mass-memory unit 607 may be collectively referred to as memory.
According to various embodiments of the present invention, the server 600 can also be arrived by network connections such as internets
Remote computer operation on network.Namely server 600 can be by the network interface that is connected on the system bus 605
Unit 611 is connected to network 612, in other words, can also be connected to using Network Interface Unit 611 other kinds of network or
Remote computer system (not shown).
The memory further includes that one or more than one program, the one or more programs are stored in
In memory, central processing unit 601 realizes that target user shown in Fig. 2 determines by executing one or more programs
Method.
In the exemplary embodiment, it includes the non-transitorycomputer readable storage medium instructed, example to additionally provide a kind of
Such as include the memory of instruction, above-metioned instruction can be executed by the processor of server to complete shown in each embodiment of the present invention
Target user determines method.For example, the non-transitorycomputer readable storage medium can be ROM, random access memory
(RAM), CD-ROM, tape, floppy disk and optical data storage devices etc..
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to its of the present invention
Its embodiment.This application is intended to cover the present invention any variations, uses, or adaptations, these modifications, purposes or
Person's adaptive change follows the general principle of the present invention and includes undocumented common knowledge in the art of the invention
Or conventional techniques.The description and examples are only to be considered as illustrative, and true scope and spirit of the invention are by following
Claim is pointed out.
It should be understood that the invention is not limited in the precision architectures for being described above and being shown in the accompanying drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present invention is limited only by the attached claims.
Claims (14)
1. a kind of target user determines method, which is characterized in that the method includes：
The historical operation record of the specified type page is generated according to user and operates frequency sequence, in the operation frequency sequence
Each element indicate respectively the user in each unit interval to the operation frequency of the specified type page, and it is described
Each element being ranked sequentially after arriving first according to the corresponding time；
Wavelet decomposition is carried out to the operation frequency sequence, obtains main body sequence, the main body sequence is and the operation frequency
The corresponding sequence of low frequency minor element in sequence；
Convolution algorithm is carried out successively to each element in the main body sequence by increasing sequence；
When the result of the convolution algorithm meets preset condition, determine that the user is target user.
2. according to the method described in claim 1, it is characterized in that, it is described by increasing sequence to each in the main body sequence
A element carries out convolution algorithm successively, including：
For i-th of element in the main body sequence, it is calculated by the following formula the corresponding convolution knot of i-th of element
Fruit：
Wherein, F (i) is the corresponding convolution results of i-th of element, and M is the number of the element in the increasing sequence, ai+h
For the elements of the i-th+h in the main body sequence,For (M+1+2h)/2 element in the increasing sequence, M >=
3, and M is odd number, (M-1)/2+1≤i≤N- (M-1)/2, N is the number of the element in the main body sequence, i+h≤N, and i
It is positive integer with N.
3. according to the method described in claim 1, it is characterized in that, the increasing sequence is the equal difference sequence of the element centered on 0
Row.
4. according to the method described in claim 1, it is characterized in that, described when the result of the convolution algorithm meets preset condition
When, determine that the user is target user, including：
When the corresponding convolution results of either element in each element meet the preset condition, determine that the user is
Target user.
5. according to the method described in claim 4, it is characterized in that,
It is described when the target user is the user on the rise to the interest of the corresponding object of the specified type page
Preset condition includes：
The convolution results are more than the first predetermined threshold value；Alternatively, the convolution results are more than the second predetermined threshold value, and it is described any
The duration at element corresponding unit interval segment distance current time is less than the first preset duration；
It is described when the target user is that have the user of downward trend to the interest of the corresponding object of the specified type page
Preset condition includes：
The convolution results are less than third predetermined threshold value；Alternatively, the convolution results are less than the 4th predetermined threshold value, and it is described any
The duration at element corresponding unit interval segment distance current time is less than the second preset duration.
6. according to the method described in claim 1, it is characterized in that, historical operation record includes the user to specified class
It is the action type of the operation behavior of the type page and operating time, described that the historical operation of the specified type page is recorded according to user
Operation frequency sequence is generated, including：
The user is counted in each unit interval according to historical operation record, the operation of each corresponding action type
Number；
Obtain the weight coefficient of each action type；
For each unit interval in each unit interval, by the number of operations of each action type and institute
The sum of products for stating the weight coefficient of each action type is determined as the operation frequency of the user in the unit interval；
The operation frequency of the user in each unit interval is arranged according to sequence of the corresponding time after arriving first
Row obtain the operation frequency sequence.
7. method according to any one of claims 1 to 6, which is characterized in that by increasing sequence to the main body sequence
In each element carry out convolution algorithm successively before, the method further includes：
The increasing sequence is obtained according to the number of the element in the main body sequence.
8. a kind of target user's determining device, which is characterized in that described device includes：
Generation module, for generating operation frequency sequence, the behaviour to the historical operation record of the specified type page according to user
Making each element in frequency sequence indicates respectively the user in each unit interval to the specified type page
Operate the frequency, and each element according to the corresponding time being ranked sequentially after arriving first；
Decomposing module, for the operation frequency sequence carry out wavelet decomposition, acquisition main body sequence, the main body sequence be with
The corresponding sequence of low frequency element in the operation frequency sequence；
Convolution module, for carrying out convolution algorithm successively to each element in the main body sequence by increasing sequence；
Determining module, for when the result of the convolution algorithm meets preset condition, determining that the user is target user.
9. device according to claim 8, which is characterized in that the convolution module is used for
For i-th of element in the main body sequence, it is calculated by the following formula the corresponding convolution knot of i-th of element
Fruit：
Wherein, F (i) is the corresponding convolution results of i-th of element, and M is the number of the element in the increasing sequence, ai+h
For the elements of the i-th+h in the main body sequence,For (M+1+2h)/2 element in the increasing sequence, M >=
3, and M is odd number, (M-1)/2+1≤i≤N- (M-1)/2, N is the number of the element in the main body sequence, i+h≤N, and i
It is positive integer with N.
10. device according to claim 8, which is characterized in that the increasing sequence is the equal difference sequence of the element centered on 0
Row.
11. device according to claim 8, which is characterized in that
The determining module, for meeting the preset condition when the corresponding convolution results of either element in each element
When, determine that the user is target user.
12. according to the devices described in claim 11, which is characterized in that
It is described when the target user is the user on the rise to the interest of the corresponding object of the specified type page
Preset condition includes：
The convolution results are more than the first predetermined threshold value；Alternatively, the convolution results are more than the second predetermined threshold value, and it is described any
The duration at element corresponding unit interval segment distance current time is less than the first preset duration；
It is described when the target user is that have the user of downward trend to the interest of the corresponding object of the specified type page
Preset condition includes：
The convolution results are less than third predetermined threshold value；Alternatively, the convolution results are less than the 4th predetermined threshold value, and it is described any
The duration at element corresponding unit interval segment distance current time is less than the second preset duration.
13. device according to claim 8, which is characterized in that the historical operation record includes the user to specified
The action type of the operation behavior of type page and operating time, the generation module, including：
Statistic unit corresponds to each for counting the user in each unit interval according to historical operation record
The number of operations of action type；
Coefficient acquiring unit, the weight coefficient for obtaining each action type；
Frequency determination unit, for for each unit interval in each unit interval, described each to be operated
The number of operations of type and the sum of products of the weight coefficient of each action type are determined as the user in the unit
The operation frequency in period；
Arrangement units, for by the operation frequency of the user in each unit interval according to the corresponding time from elder generation
Being ranked sequentially after obtains the operation frequency sequence.
14. according to any device of claim 8 to 13, which is characterized in that described device further includes：
Acquisition module, for being carried out successively to each element in the main body sequence by increasing sequence in the convolution module
Before convolution algorithm, the increasing sequence is obtained according to the number of the element in the main body sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710287454.5A CN108804387B (en) | 2017-04-27 | 2017-04-27 | Target user determination method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710287454.5A CN108804387B (en) | 2017-04-27 | 2017-04-27 | Target user determination method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108804387A true CN108804387A (en) | 2018-11-13 |
| CN108804387B CN108804387B (en) | 2021-07-23 |
Family
ID=64070078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710287454.5A Active CN108804387B (en) | 2017-04-27 | 2017-04-27 | Target user determination method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108804387B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110570217A (en) * | 2019-09-10 | 2019-12-13 | 北京百度网讯科技有限公司 | cheating detection method and device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080033991A1 (en) * | 2006-08-03 | 2008-02-07 | Jayanta Basak | Prediction of future performance of a dbms |
| CN102216941A (en) * | 2008-08-19 | 2011-10-12 | 数字标记公司 | Methods and systems for content processing |
| CN103744929A (en) * | 2013-12-30 | 2014-04-23 | 传神联合（北京）信息技术有限公司 | Target user object determination method |
| US20140236968A1 (en) * | 2011-10-31 | 2014-08-21 | Li-Mei Jiao | Discrete Wavelet Transform Method for Document Structure Similarity |
| CN104090908A (en) * | 2014-06-06 | 2014-10-08 | 微梦创科网络科技（中国）有限公司 | Method and device for counting mean detention time in page group and generalizing content in website |
| CN105335519A (en) * | 2015-11-18 | 2016-02-17 | 百度在线网络技术（北京）有限公司 | Model generation method and device as well as recommendation method and device |
| CN106354822A (en) * | 2016-08-30 | 2017-01-25 | 五八同城信息技术有限公司 | Method and device for acquiring target user |
-
2017
- 2017-04-27 CN CN201710287454.5A patent/CN108804387B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080033991A1 (en) * | 2006-08-03 | 2008-02-07 | Jayanta Basak | Prediction of future performance of a dbms |
| CN102216941A (en) * | 2008-08-19 | 2011-10-12 | 数字标记公司 | Methods and systems for content processing |
| US20140236968A1 (en) * | 2011-10-31 | 2014-08-21 | Li-Mei Jiao | Discrete Wavelet Transform Method for Document Structure Similarity |
| CN103744929A (en) * | 2013-12-30 | 2014-04-23 | 传神联合（北京）信息技术有限公司 | Target user object determination method |
| CN104090908A (en) * | 2014-06-06 | 2014-10-08 | 微梦创科网络科技（中国）有限公司 | Method and device for counting mean detention time in page group and generalizing content in website |
| CN105335519A (en) * | 2015-11-18 | 2016-02-17 | 百度在线网络技术（北京）有限公司 | Model generation method and device as well as recommendation method and device |
| CN106354822A (en) * | 2016-08-30 | 2017-01-25 | 五八同城信息技术有限公司 | Method and device for acquiring target user |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110570217A (en) * | 2019-09-10 | 2019-12-13 | 北京百度网讯科技有限公司 | cheating detection method and device |
| CN110570217B (en) * | 2019-09-10 | 2022-10-14 | 北京百度网讯科技有限公司 | Cheating detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108804387B (en) | 2021-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112148987B (en) | Message pushing method based on target object activity and related equipment | |
| US9043302B1 (en) | Campaign and competitive analysis and data visualization based on search interest data | |
| US20160134934A1 (en) | Estimating audience segment size changes over time | |
| Cheng et al. | An ARL-unbiased design of time-between-events control charts with runs rules | |
| CA2901073C (en) | Predictive model generator | |
| US9911130B1 (en) | Attribution modeling using regression analysis | |
| CN102750320B (en) | Method, device and system for calculating network video real-time attention | |
| US10679137B2 (en) | Systems and methods for determining real-time visitor segments | |
| Chakraborty et al. | A generally weighted moving average chart for time between events | |
| Alevizakos et al. | A double exponentially weighted moving average chart for time between events | |
| US20230017443A1 (en) | Dynamically varying remarketing based on evolving user interests | |
| Rakitzis et al. | The effect of parameter estimation on the performance of one-sided Shewhart control charts for zero-inflated processes | |
| Vasilyev | On some distributions associated to boundary value problems | |
| Song et al. | Convergence of the tamed Euler method for stochastic differential equations with piecewise continuous arguments under non-global Lipschitz continuous coefficients | |
| CN108804387A (en) | Target user determines method and device | |
| Wang | Change-in-mean problem for long memory time series models with applications | |
| CN114746899A (en) | Methods, systems, and apparatus for estimating census-level audience, impressions, and duration across demographic groups | |
| WO2014163993A2 (en) | Inferring attribute and item preferences | |
| BenMabrouk et al. | Wavelet fuzzy hybrid model for physico-financial signals | |
| JP2011227720A (en) | Recommendation system, recommendation method and recommendation program | |
| CN109583922B (en) | Method and device for analyzing purchase demand | |
| Takeuchi | Joint distributions for stochastic functional differential equations | |
| Brown | Monitoring a Poisson process subject to gradual changes in the arrival rates | |
| CN110162697A (en) | User browsing behavior analysis method, device, medium and terminal based on big data | |
| CN113254501B (en) | Control method and device for predicting program TAD through discretization of continuous features |
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 |