CN108804387B - Target user determination method and device - Google Patents
Target user determination method and device Download PDFInfo
- Publication number
- CN108804387B CN108804387B CN201710287454.5A CN201710287454A CN108804387B CN 108804387 B CN108804387 B CN 108804387B CN 201710287454 A CN201710287454 A CN 201710287454A CN 108804387 B CN108804387 B CN 108804387B
- Authority
- CN
- China
- Prior art keywords
- sequence
- user
- convolution
- target user
- frequency
- 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.)
- Active
Links
Images
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
Abstract
The invention relates to a target user determination method and a target user determination device. The method comprises the following steps: generating an operation frequency sequence, performing wavelet decomposition on the operation frequency sequence to obtain a main body sequence, sequentially performing convolution operation on each element in the main body sequence through an increasing sequence, and determining the user as a target user when the result of the convolution operation meets a preset condition. The wavelet decomposition is used for processing the operation frequency sequence of the user on the specified type page, so that the noise can be removed more accurately, the change outline (namely a main signal) capable of accurately representing the operation frequency of the user in the specified type page is decomposed, and whether the user has the trend that the operation frequency is obviously increased or decreased can be judged by carrying out convolution operation on the main signal and the incremental sequence, so that whether the user is a target user is judged, and the detection accuracy of the target user is improved.
Description
Technical Field
The present invention relates to the field of network application technologies, and in particular, to a method and an apparatus for determining a target user.
Background
With the continuous development of network application technology, many network service providers can determine whether a user is a target object interested in a certain object by analyzing the operation behavior of the user in a webpage so as to carry out targeted information push.
In the related art, the server of the network service provider may count the number of browsing operations of the user in the page corresponding to the certain thing in the recent period of time, and when the number of browsing operations of the user in the page corresponding to the thing reaches a certain value, it may be determined that the user is the target user corresponding to the thing. For example, taking the page of the car media as an example, the server may count the number of times that the user browses the page of the car media in the last week, and when the number of times that the user browses the page of the car media in the last week reaches 10 times, may confirm that the user is the target user with the intention of purchasing cars.
In the related art, the server determines whether the user is the target user only by counting the number of browsing operations of the user on the page corresponding to the certain event in the latest period of time, and the accuracy of determining the target user is low.
Disclosure of Invention
In order to solve the problem that in the prior art, whether a user is a target user is determined only by counting the number of browsing operations of the user on a page corresponding to an event within a recent period of time, which results in low accuracy of target user determination, embodiments of the present invention provide a method and an apparatus for determining a target user, where the technical scheme is as follows:
in a first aspect, a target user determination method is provided, where the method includes:
generating an operation frequency sequence according to a historical operation record of a user on a page of an appointed type, wherein each element in the operation frequency sequence indicates the operation frequency of the user on the page of the appointed type in each unit time period respectively, and the elements are arranged according to the sequence of corresponding time from first to last;
performing wavelet decomposition on the operation frequency sequence to obtain a main body sequence, wherein the main body sequence is a sequence corresponding to low-frequency elements in the operation frequency sequence;
sequentially performing convolution operation on each element in the main body sequence through an increasing sequence;
and when the result of the convolution operation meets a preset condition, determining the user as a target user.
In a second aspect, a target user determination apparatus is provided, the apparatus comprising:
the generating module is used for generating an operation frequency sequence according to a historical operation record of a user on a specified type page, wherein each element in the operation frequency sequence indicates the operation frequency of the user on the specified type page in each unit time period respectively, and the elements are arranged according to the sequence of corresponding time from first to last;
the decomposition module is used for performing wavelet decomposition on the operation frequency sequence to obtain a main body sequence, wherein the main body sequence is a sequence corresponding to low-frequency elements in the operation frequency sequence;
the convolution module is used for sequentially carrying out convolution operation on each element in the main body sequence through the incremental sequence;
and the determining module is used for determining the user as a target user when the result of the convolution operation meets a preset condition.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
the wavelet decomposition is used for processing the operation frequency sequence of the user on the specified type page, so that the noise can be removed more accurately, the change outline (namely a main signal) capable of accurately representing the operation frequency of the user in the specified type page is decomposed, and whether the user has the trend that the operation frequency is obviously increased or decreased can be judged by carrying out convolution operation on the main signal and the incremental sequence, so that whether the user is a target user is judged, and the detection accuracy of the target user is improved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a block diagram illustrating a target user determination system in accordance with an exemplary embodiment;
FIG. 2 is a flow diagram illustrating a method of target user determination in accordance with an exemplary embodiment;
FIG. 3 is a schematic diagram of a wavelet decomposition to which the embodiment shown in FIG. 2 relates;
FIG. 4 is a schematic illustration of an implementation process for determining a target user;
FIG. 5 is a block diagram illustrating the structure of a target user determination device in accordance with an exemplary embodiment;
fig. 6 is a schematic diagram illustrating a configuration of a server according to an example embodiment.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the invention, as detailed in the appended claims.
Fig. 1 is a schematic structural diagram of a target user determination system according to an exemplary embodiment of the present invention. The system comprises: a number of user terminals 120 and a server cluster 140.
The user terminal 120 may be a mobile phone, a tablet computer, an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III, mpeg compression standard Audio Layer 3), an MP4 player (Moving Picture Experts Group Audio Layer IV, mpeg compression standard Audio Layer 4), a laptop, a desktop computer, or the like.
The user terminal 120 and the server cluster 140 are connected via a communication network. Optionally, the communication network is a wired network or a wireless network.
The server cluster 140 is a server, or a plurality of servers, or a virtualization platform, or a cloud computing service center. Optionally, the server cluster 140 may include a server for implementing the target user determination platform 142, and optionally, the server cluster 140 further includes a server for implementing the page management platform 144; optionally, the server cluster 140 further includes a user operation record management server 146.
Optionally, the page management platform 144 includes: the system comprises a server for pushing and maintaining webpage pages and a server for managing and storing various user accounts. The page management platform 144 and the user operation record management server 146 are connected through a communication network.
Optionally, the user operation record management server 146 includes: the server is used for counting the operation behaviors of the user in each page, and the server is used for storing the operation behaviors of the user in each page.
Optionally, on the premise that the user authorizes and approves, the user operation record management server 146 may obtain operation behavior data of the user in each page from the page management platform 144, or from other associated page management platforms, and count the historical operation record of the user in each page according to the obtained operation record.
Optionally, the system may further include a management device 160, and the management device 160 is connected to the server cluster 140 through a communication network. Optionally, the communication network is a wired network or a wireless network.
Optionally, the wireless network or wired network described above uses standard communication techniques and/or protocols. The Network is typically the Internet, but may be any Network including, but not limited to, a Local Area Network (LAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), a mobile, wireline or wireless Network, a private Network, or any combination of virtual private networks. In some embodiments, data exchanged over a network is represented using techniques and/or formats including Hypertext Mark-up Language (HTML), Extensible Markup Language (XML), and the like. All or some of the links may also be encrypted using conventional encryption techniques such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec). In other embodiments, custom and/or dedicated data communication techniques may also be used in place of, or in addition to, the data communication techniques described above.
Wavelet decomposition is a method for processing signals, which gradually carries out multi-scale refinement on signals (or called as functions) through telescopic translation operation, finally achieves time subdivision at high frequency and frequency subdivision at low frequency, can automatically adapt to the requirement of time-frequency signal analysis, and is very suitable for analyzing non-stationary signals (namely, signals with distribution rules changing along with time).
When a user is interested in a certain object, the user usually searches and browses the web pages related to the object in the network to improve the understanding of the object. Moreover, as the interest of the user in the object increases, the frequency of browsing the page related to the object by the user also tends to increase obviously. Accordingly, if the user is interested in a certain object at the beginning and gradually loses interest in the object later, the frequency of browsing the page related to the object by the user is in a significantly decreasing trend.
For example, a user who needs to know about car related messages usually browses information needed by the user on a page provided by a car media, wherein the browsing frequency of the user who intends to buy a car on the page provided by the car media is usually gradually increased. That is, when a user just starts to buy a car, the frequency of browsing the pages provided by the car media will be low, and as the user's will get stronger and stronger, the frequency of browsing the pages provided by the car media will be higher and higher. On the contrary, if the user just purchases the car, the browsing frequency of the user in the page provided by the car media is higher, and the browsing frequency of the user in the page provided by the car media is generally lower and lower as the time after the car is purchased. The basic idea of each embodiment of the present invention is to determine a user with an ascending or descending frequency of the operation behavior in the page related to the certain object as a target user corresponding to the certain object by analyzing the frequency of the operation behavior of the user in the page related to the certain object, so as to improve the accuracy of the determination of the target user.
Due to the difference of the individual internet surfing behaviors of different users, the specific situations that the frequency of browsing the page related to a certain thing by a user interested in the thing is in an ascending or descending trend are different. For example, taking a user who intends to purchase a car as an example, some "fast-heating" users may have a very obvious upward trend in browsing frequency of pages provided by car media within a week; while some "slow-hot" users may gradually increase their frequency of browsing pages provided by automotive media within a month or two. In addition, during browsing a certain event-related page by a user, the operation frequency in the event-related page may be suddenly increased in a short time, and then decreased in a longer time, and such suddenly increased operation frequency is not representative for the operation behavior analysis of the user and belongs to the interference information (also referred to as noise) during the operation behavior analysis of the user.
It can be seen from the above analysis of the operation frequency of the user in the page related to some object that, if the operation frequency of the user in the page related to some object is arranged according to the time sequence in each unit time period, the obtained sequence can be regarded as a non-stationary signal with two dimensions of time and frequency, and noise may be included in the signal, so that the sequence is processed through wavelet decomposition, noise can be more accurately removed, a change profile (i.e., a main signal) capable of accurately representing the operation frequency of the user in the page related to some object can be decomposed, and whether the user is a target user corresponding to the object can be detected through analysis of the main signal. For example, when analyzing the main signal, the convolution template may be used to perform convolution operation on the main signal, and whether the user is the target user may be determined according to the convolution operation result.
Fig. 2 is a flowchart illustrating a target user determination method according to an exemplary embodiment, which is taken as an example of a server cluster applied in the system shown in fig. 1, and the target user determination method may include the following steps:
Each element in the operation frequency sequence respectively indicates the operation frequency of the user on the specified type page in each unit time period, and the elements are arranged according to the sequence of the corresponding time from first to last.
The page of the specified type may be a page related to some thing, the page may be a single page or may include a plurality of pages, and the page may be an HTML (hypertext Markup Language) page opened by a browser, or the page may be a built-in Web page opened by an application program.
In the embodiment of the present invention, the server cluster may obtain a historical operation record of the user on the page of the specified type, where the historical operation record may include operation time of the user on a past operation behavior of the page of the specified type, and the server cluster may generate an operation frequency sequence corresponding to the user according to the obtained historical operation record.
Taking a unit time period of 1 day as an example, the operation frequency sequence may be a sequence in which the times of operations on the specified type of page by the user are arranged in time order in several consecutive days. For example, the operation frequency sequence of the user in the last 2 × N days is an example, and the operation frequency sequence may be represented as:
X＝[x1，x2，x3，x4，…，x2*2*N-1，x2*2*N]。
wherein x is1The frequency of operations on the specified type pages within day 1 of the last 2 x N days for the user; x is the number of2Frequency of operations on the specified type pages within the 2 nd day of the last 2 x N days for the user; by analogy, x2*2*NThe frequency of operations on the specified type pages within the last 1 day of the last 2 x N days is given to the user.
In practical applications, the types of operations performed by the user in the page may be many, for example, the user may perform operations of click browsing (including clicking browsing text, pictures, or video animations), comparing (for example, comparing attributes of two or more objects), posting comments, and contacting customer service in the page. In the embodiment of the present invention, the server cluster may perform equal processing on different types of user operations, for example, when the server cluster counts the frequency of operations on a specified type page by a user in a certain unit time period, the server cluster ignores the type difference between different operations, and counts each operation of the user as 1 frequency respectively.
Specifically, if the user clicks and browses 50 times in the specified type page in a certain unit time period, compares 3 times, issues 2 comments, and contacts 1 customer service, the server cluster counts the frequency of the user's operation in the unit time period as 50+3+2+ 1-56 times.
In another possible implementation manner, when the server cluster generates the operation frequency sequence, differentiation processing may be performed on different types of user operations, for example, a historical operation record obtained by the server cluster includes, in addition to operation time of a user on an operation behavior of a specified type page, a corresponding operation type, and the server cluster may count, according to the historical operation record, the number of times of operations of the user corresponding to each operation type in each unit time period; the server cluster obtains the weight coefficient of each operation type, determines the sum of the product of the operation times of each operation type and the weight coefficient of each operation type as the operation frequency of the user in the unit time period for each unit time period in each unit time period, and finally arranges the operation frequency of the user in each unit time period according to the sequence of the corresponding time from first to last, namely, the operation frequency sequence can be obtained.
For example, assuming that the weight coefficient of click browsing is 1, the weight coefficient of comparison is 2, the weight coefficient of posting a comment is 5, the weight coefficient of contacting customer service is 10, and in a certain unit time period, the user clicks and browses 50 times in a specified type page, compares 3 times, posts 2 comments, and contacts 1 customer service, the server cluster may count the frequency of operations of the user in the unit time period as 50+ 1+ 3+2+ 5+ 1+ 10 times as 76 times.
The weighting coefficient of each operation type can be set and stored in the server cluster by a manager through the management device in advance, and the manager can modify and update the weighting coefficient through the management device in the using process.
Referring to fig. 3, a diagram of a wavelet decomposition is shown. Taking 3-time wavelet decomposition of the original signal as an example, as shown in fig. 3, at the time of 1 st wavelet decomposition, the original signal is decomposed into cA1 and cD1, where cA1 corresponds to the low frequency signal in the original signal, and cD1 corresponds to the high frequency signal in the original signal. At the time of wavelet decomposition 2, cA1 is decomposed into cA2 and cD2, where cA2 corresponds to a low-frequency signal in cA1 and cD2 corresponds to a high-frequency signal in cA 1. At the time of wavelet decomposition 3, cA2 is decomposed into cA3 and cD3, where cA3 corresponds to a low-frequency signal in cA2 and cD3 corresponds to a high-frequency signal in cA 2. After 3 times of wavelet decomposition, the obtained decomposition results are C and L, wherein C is 4 signals obtained by decomposition, namely cA3, cD3, cD2 and cD1, and L is the signal length of each of the 4 signals obtained by the decomposition.
The signal cA3 represents the outline and the overview of the original signal, and cD3, cD2 and cD1 represent the detailed information of the original signal, wherein the frequency of cD1 is the highest, the frequency of cD2 is the second of cD3, and the frequency of cA3 is the lowest. In general, when noise is contained in an original signal, the noise is concentrated in a high frequency portion after wavelet decomposition.
In the embodiment of the present invention, the operation frequency sequence X obtained in step 201 may be used as an original signal, and wavelet decomposition may be performed on the operation frequency sequence X, so as to obtain at least two groups of sequences, where a sequence corresponding to a low-frequency element in the operation frequency sequence is a main sequence, and if accidental browsing frequency fluctuation of a user is regarded as noise, after the wavelet decomposition, the noise in the operation frequency sequence X is separated into the main sequence, and at this time, the main sequence may accurately represent a variation trend of the operation frequency of the user on a specified type of page.
Optionally, in the embodiment of the present invention, when performing wavelet decomposition on the operation frequency sequence X, the number of decomposition times may be determined according to the number of elements in the operation frequency sequence X, and the wavelet decomposition may be performed on the operation frequency sequence according to the determined number of decomposition times.
Fig. 3 illustrates an example of performing wavelet decomposition 3 times on an original signal, in the embodiment of the present invention, when performing wavelet decomposition on the operation frequency sequence X, the operation frequency sequence X may be subjected to wavelet decomposition 3 times, or wavelet decomposition with fewer or more times may also be performed on the operation frequency sequence X, and a sequence corresponding to a low-frequency element obtained by wavelet decomposition 1 time last is acquired as a main sequence. Specifically, in order to reduce the complexity of subsequent calculation, the length of the subject sequence obtained by wavelet decomposition (i.e., the number of elements in the subject sequence) cannot be too long, i.e., the number of decomposition times cannot be too small; meanwhile, in order to ensure that the main body sequence can accurately represent the variation trend of the operation frequency of the user on the specified type page, the length of the main body sequence obtained by wavelet decomposition cannot be too short, namely the decomposition frequency cannot be too many; therefore, in the embodiment of the present invention, the decomposition times may be determined according to the number of elements in the operation frequency sequence X (i.e. the length of the operation frequency sequence X), and the more the number of elements in the operation frequency sequence X is, the more the decomposition times are, and correspondingly, the less the number of elements in the operation frequency sequence X is, the less the decomposition times are.
Specifically, the server cluster may preset a corresponding relationship between the number of decomposition times and the element number interval, and when performing wavelet decomposition on the operation frequency sequence X, determine the element interval in which the element number of the operation frequency sequence X is located, query the number of decomposition times corresponding to the determined element interval, and decompose the operation frequency sequence X according to the queried number of decomposition times.
For example, the correspondence between the decomposition times and the element number intervals may be preset in the server cluster as shown in table 1:
| number of decompositions | 1 | 2 | 3 | … |
| Interval of element number | (30，90] | (90，360] | (360，1080] | … |
TABLE 1
As shown in table 1, when the number of elements in the operation frequency sequence X is in the section (30, 90), the server cluster determines that the number of times of wavelet decomposition on the operation frequency sequence X is 1, when the number of elements in the operation frequency sequence X is in the section (90, 360), the server cluster determines that the number of times of wavelet decomposition on the operation frequency sequence X is 2, when the number of elements in the operation frequency sequence X is in the section (360, 1080), the server cluster determines that the number of times of wavelet decomposition on the operation frequency sequence X is 3, and so on.
For example, taking the number of times of decomposition as 2 as an example, the server cluster pair operation frequency sequence X ═ X1，x2，x3，x4，…，x2*2*N-1，x2*2*N]After 2 decompositions, the decomposition results obtained were:
cA2＝[a1，a2，a3，…，aN]；
cD2＝[d21，d22，d23，…，d2N]；
dD1＝[d11，d12，d13，…，d12*N]。
wherein, cA2 is the main sequence obtained by decomposition.
In step 203, the convolution operation is performed on each element in the main body sequence sequentially through the incremental sequence.
In this embodiment of the present invention, the server cluster may perform convolution operation on each element in the main sequence, specifically, for the ith element in the main sequence, the server cluster may calculate a convolution result corresponding to the ith element by using the following formula:
wherein F (i) is the convolution result corresponding to the ith element, M is the number of elements in the increasing sequence, ai+hIs the i + h th element in the subject sequence,
is the (M +1+2h)/2 th element in the increasing sequence, M is more than or equal to 3, and M is an odd number, (M-1)/2+1 is more than or equal to i and less than or equal to N- (M-1)/2, N is the number of the elements in the main body sequence, i + h is more than or equal to N, and i and N are positive integers.
Optionally, in the scheme shown in the embodiment of the present invention, the server cluster may select a suitable increment sequence according to an actual application situation, for example, the server may obtain the increment sequence according to the number of elements in the main body sequence, and perform convolution operation on the elements in the main body sequence through the obtained increment sequence.
Specifically, a plurality of groups of increasing sequences may be set in the server cluster, the number of elements in each group of increasing sequences is different, for example, the number of elements in each increasing sequence may be 3, 5, 7, or more, meanwhile, a corresponding relationship between the increasing sequence and an element number interval may be set in advance in the server cluster, and when the server cluster performs convolution operation on the main sequence, the element interval in which the number of elements of the main sequence is located is determined first, the increasing sequence corresponding to the determined element interval is queried, and convolution operation is performed on the main sequence according to the queried increasing sequence.
Wherein the increasing sequence is an arithmetic sequence having 0 as a central element. For example, two sets of incrementing sequences may be set in the server cluster, which are:
5-dimensional (i.e., number of elements is 5) sequence: [ -2, -1, 0, 1, 2 ];
and a 7-dimensional (i.e., 7 element number) sequence: [ -3, -2, -1,0,1,2,3].
For example, in the case of the subject sequence cA2 ═ a1，a2，a3，…，aN]When convolution operation is carried out, the selected increasing sequence is taken as the 5-dimensional sequence as an example, namely the numerical value of M is 5, i is more than or equal to 3 and less than or equal to N-2, and the server cluster is connected with the server cluster from a3Initially, convolution operations are performed for each element in cA2 until a in cA2 is performedN-2And finishing the operation. Specifically, the method comprises the following steps:
for a3When convolution operation (i is 3), F (3) is-2 a1-a2+0*a3+a4+2a5；
For a4When convolution operation (i is 4), F (4) is-2 a2-a3+0*a4+a5+2a6；
By analogy, for aN-2When convolution operation is performed (i-N-2), F (N-2) ═ 2aN-4-aN-3+0*aN-2+aN-1+2aN。
And step 204, when the result of the convolution operation meets a preset condition, determining that the user is a target user.
In step 203, the server cluster performs a convolution operation each time a in the body sequence is determinediPerforming convolution operation on a plurality of front and back elements and the increasing sequence if the aiThe front and back elements being incremental (i.e. a)iThe operation frequency of the user gradually increases in a period of time before and after the corresponding unit time period), the convolution operation result is larger, for example, if the incremental sequence is an arithmetic sequence with 0 as a central element, the convolution operation result is a larger positive value; if the a isiThe front and back elements are not greatly different (i.e. a)iThe operation frequency of the user keeps a certain amount and the fluctuation is small in a period of time before and after the corresponding unit time period), the convolution operation result is moderate, for example, if the incremental sequence is an arithmetic sequence with 0 as a central element, the convolution operation result is close to 0; if the a isiThe front and rear elements being decreasing (i.e. a)iCorresponding unitThe operation frequency of the user gradually decreases in a period of time before and after the time period), the convolution operation result is smaller, for example, if the incremental sequence is an arithmetic sequence with 0 as a central element, the convolution operation result is a smaller negative value.
Based on the characteristics of the convolution operation result, in the embodiment of the present invention, when the server cluster performs convolution operation, when the convolution result corresponding to any element in each element of the main sequence meets a preset condition, it may be determined that the user is a target user.
Specifically, for example, when the target user is a user whose interest in the object corresponding to the specified type page has an ascending trend, after the server cluster calculates the convolution result corresponding to the i-th element in the main body sequence, if the calculated convolution result f (i) is greater than a first preset threshold, it is determined that the interest in the object corresponding to the specified type page of the user has an obvious ascending trend within a period of time before and after the unit time period corresponding to the i-th element, and at this time, it may be determined that the user is the target user.
In practical applications, the interest of the user in a certain event may fluctuate over time, for example, the user is interested in a certain event for a period of time, then the user may lose interest in the event for a period of time, and then the user may be interested in the event for a period of time, so that if the time corresponding to the ith element is longer than the current time, f (i) indicates that the user is currently interested in or loses interest, and conversely, if the time corresponding to the ith element is closer to the current time, f (i) indicates that the user is currently interested in or loses interest, and then the time corresponding to the ith element is greater. Therefore, in another possible implementation manner, when the target user is a user whose interest in the object corresponding to the specified type page has an increasing trend, after the server cluster calculates the convolution result corresponding to the ith element in the main body sequence, if the convolution result f (i) obtained by the calculation is greater than the second preset threshold and the time length of the unit time period corresponding to the ith element from the current time is less than the first preset time length, it is considered that the interest in the object corresponding to the specified type page has an obvious increasing trend within a period of time closer to the current time, and at this time, it may be determined that the user is the target user.
Or, when the target user is a user whose interest in the object corresponding to the specified type page is in a downward trend, if the calculated convolution result f (i) is smaller than the third preset threshold, it is considered that the user has an obvious downward trend in the interest in the object corresponding to the specified type page in a period of time before and after the unit time period corresponding to the ith element, and at this time, the user may also be determined to be the target user. Or, when the target user is a user whose interest in the object corresponding to the specified type page is in a downward trend, if the convolution result f (i) is less than the fourth preset threshold and the duration of the unit time period corresponding to the ith element from the current time is less than the second preset duration, it is considered that the interest in the object corresponding to the specified type page is in an obvious downward trend within a period of time closer to the current time by the user, and at this time, the user may also be determined to be the target user.
The first preset threshold, the second preset threshold, the third preset threshold, the fourth preset threshold, the first preset duration and the second preset duration may be parameters manually set in the server cluster by a manager in advance through a management device, and the manager may modify or update the parameters in the operation process.
Optionally, when the server cluster executes the steps 203 and 204, it may first sequentially calculate a convolution result corresponding to each element in the main sequence, and then determine whether the user is the target user according to the convolution result corresponding to each element.
In another possible implementation manner, after the convolution result corresponding to one element in the main sequence is calculated each time, the server cluster may also determine whether the user is the target user according to the convolution result corresponding to the element, if so, stop the subsequent convolution calculation process, and otherwise, continue to calculate the convolution result corresponding to the next element.
For example, if the target user is a user whose interest in the object corresponding to the specified type page is in an ascending trend, or the target user is a user whose interest in the object corresponding to the specified type page is in a descending trend, when the server cluster sequentially performs convolution calculation on each element in the main body sequence, each convolution result is calculated, that is, whether the user is the target user is determined according to the convolution result, if yes, the subsequent convolution calculation process is stopped, otherwise, the subsequent convolution calculation is continuously performed.
If two types of target users need to be determined simultaneously, for example, a user with an upward interest trend of an object corresponding to a specified type page is a first type of target user, and a user with a downward interest trend of the object corresponding to the specified type page is a second type of target user, the server cluster needs to continue to perform subsequent convolution calculation after calculating a convolution result and determining that the user is one of the two types of target users according to the convolution result in the process of sequentially performing convolution calculation on each element in the main body sequence, and the server can stop the subsequent convolution calculation process if determining that the user is another of the two types of target users in the subsequent calculation process.
Optionally, in the above scheme of the embodiment of the present invention, it is described by taking an example that a convolution result corresponding to any element in a main sequence meets a preset condition, in practical application, when determining whether a user is a target user according to a result of convolution operation, other determining methods may also be used, for example, when the number of elements whose corresponding result of convolution operation meets the preset condition is not less than a first number threshold, or when the corresponding result of convolution operation meets the preset condition and the number of adjacent elements is not less than a second number threshold, it may be determined that the user is the target user.
Optionally, after determining that the current user is the target user, the server cluster may provide a targeted service to the user, for example, push a message (such as an advertisement message, etc.) corresponding to a specified type of page to the user in a targeted manner.
In summary, the target user determining method provided in the embodiment of the present invention processes the operation frequency sequence of the user on the specified type page through wavelet decomposition, so as to remove noise more accurately, and decompose a change profile (i.e., a main signal) that can accurately represent the operation frequency of the user on the specified type page, and determine whether the user has a trend that the operation frequency significantly increases or decreases by performing a convolution operation on the main signal and the incremental sequence, thereby determining whether the user is the target user, and improving the accuracy of target user detection.
Specifically, please refer to fig. 4, which shows a schematic diagram of an implementation process of determining a target user. Taking the implementation process that can be implemented by the target user determination platform 142, the page management platform 144, and the user operation record management server 146 in the server cluster shown in fig. 1, and the determined target user is a user with a car purchasing intention or a user who has already purchased a car, as shown in fig. 4, the page management platform 144 is a platform of an automobile media website, maintains each page in the automobile media website, records the operation behavior and the operation time of each user in each page of the automobile media website, and sends the recorded operation behavior and the recorded operation time to the user operation record management server 146. The user operation record management server 146 respectively counts the operation behaviors and the operation times of each user in each page of the automobile media website to obtain the historical operation records respectively corresponding to each user, and stores the historical operation records obtained by counting corresponding to the user account.
When determining the target user, the target user determination platform 142 extracts the historical operation record of the user from the user operation record management server 146 through the user account of the user, and generates an operation frequency sequence in which the operation frequency of the user in each page of the automobile media website in the last several days is arranged according to the time sequence according to the historical operation record statistics of the user. The target user determination platform 142 determines the decomposition times according to the generated operation frequency sequence, performs wavelet decomposition on the operation frequency sequence according to a preset decomposition algorithm according to the determined decomposition times, and acquires a sequence corresponding to a low-frequency secondary element, which is obtained by the last decomposition, as a main sequence.
The target user determination platform 142 selects a preset incremental sequence according to the length of the main sequence, performs convolution operation on the incremental sequence and each element in the main sequence respectively to obtain a convolution operation result, and determines whether the user is the target user according to the size relationship between the convolution operation result and a preset threshold value.
Specifically, when a target user with an intention to purchase a vehicle is determined, if the convolution operation result corresponding to the ith element in the main sequence is greater than a preset threshold a and the time corresponding to the ith element is closer to the current time, the user may be considered to have the intention to purchase the vehicle, and the user may be determined as the target user with the intention to purchase the vehicle.
Or, when determining the target user who has purchased the vehicle, if the convolution operation result corresponding to the ith element in the main sequence is smaller than the preset threshold B, and the time corresponding to the ith element is closer to the current time, the user may be considered to have purchased the vehicle just now, and the user may be determined as the target user who has purchased the vehicle.
After determining that the user is a target user, the server cluster may provide a targeted service to the user, for example, after determining that the user is a target user with a car-buying intention, the server cluster may further analyze which brands and vehicle types correspond to pages browsed by the user, match advertisements that may be needed by the user according to an analysis result, and push the advertisements to the user. Alternatively, after determining that the user is a target user who has purchased a car, the server cluster may push an advertisement related to car maintenance or insurance to the user.
Fig. 5 is a block diagram illustrating a structure of a target user determination apparatus according to an exemplary embodiment. The target user determination device may be implemented as part or all of a server cluster in hardware or a combination of hardware and software to perform all or part of the steps in the embodiment shown in fig. 2. The target user determination means may comprise:
a generating module 501, configured to generate an operation frequency sequence according to a historical operation record of a user on a specified type page, where each element in the operation frequency sequence indicates an operation frequency of the user on the specified type page in each unit time period, and the elements are arranged in a sequence from first to last according to corresponding time;
a decomposition module 502, configured to perform wavelet decomposition on the operation frequency sequence to obtain a main body sequence, where the main body sequence is a sequence corresponding to a low-frequency element in the operation frequency sequence;
a convolution module 503, configured to perform convolution operation on each element in the main sequence sequentially through an incremental sequence;
a determining module 504, configured to determine that the user is a target user when a result of the convolution operation satisfies a preset condition.
Optionally, the convolution module is used for
For the ith element in the main body sequence, calculating a convolution result corresponding to the ith element by the following formula:
wherein F (i) is the convolution result corresponding to the ith element, M is the number of elements in the increasing sequence, ai+hIs the i + h element in the subject sequence,
is the (M +1+2h)/2 th element in the increasing sequence, M is more than or equal to 3, and M is an odd number, (M-1)/2+1 is more than or equal to i and less than or equal to N- (M-1)/2, and i is an integer.
Optionally, the increment sequence is an arithmetic sequence with 0 as a central element.
Optionally, the determining module is configured to determine that the user is a target user when a convolution result corresponding to any one of the elements satisfies the preset condition.
Optionally, when the target user is a user whose interest in the object corresponding to the specified type page is trending upward, the preset condition includes:
the convolution result is larger than a first preset threshold value; or, the convolution result is greater than a second preset threshold, and the time length of the unit time period corresponding to any element from the current time is less than a first preset time length;
optionally, when the target user is a user whose interest in the object corresponding to the specified type page is in a downward trend, the preset condition includes:
the convolution result is smaller than a third preset threshold value; or, the convolution result is smaller than a fourth preset threshold, and the time length of the unit time period corresponding to any element from the current time is smaller than a second preset time length.
Optionally, the generating module includes:
the record acquisition unit is used for acquiring a historical operation record of the user, wherein the historical operation record comprises an operation type and an operation time of an operation behavior of the user on a specified type page;
the counting unit is used for counting the operation times of the user corresponding to each operation type in each unit time period according to the historical operation record;
a coefficient acquisition unit configured to acquire a weight coefficient for each operation type;
a frequency determining unit configured to determine, for each of the unit time periods, a sum of products of the number of operations of each operation type and a weight coefficient of each operation type as an operation frequency of the user in the unit time period;
and the arranging unit is used for arranging the operation frequency of the user in each unit time period according to the sequence of the corresponding time from first to last to obtain the operation frequency sequence.
Optionally, the apparatus further comprises:
and the obtaining module is used for obtaining the incremental sequence according to the number of the elements in the main body sequence before the convolution module carries out convolution operation on each element in the main body sequence sequentially through the incremental sequence.
Optionally, the decomposition module includes:
the frequency determining unit is used for determining the decomposition frequency according to the number of the elements in the operation frequency sequence;
and the decomposition unit is used for performing wavelet decomposition on the operation frequency sequence according to the determined decomposition times.
In summary, the target user determining apparatus provided in the embodiment of the present invention processes the operation frequency sequence of the user on the specified type page through wavelet decomposition, so as to remove noise more accurately, and decompose a change profile (i.e., a main signal) that can accurately represent the operation frequency of the user on the specified type page, and determine whether the user has a trend that the operation frequency significantly increases or decreases by performing a convolution operation on the main signal and the incremental sequence, thereby determining whether the user is the target user, and improving the accuracy of target user detection.
Fig. 6 is a schematic diagram illustrating a configuration of a server according to an example embodiment. The server 600 includes a Central Processing Unit (CPU)601, a system memory 604 including a Random Access Memory (RAM)602 and a Read Only Memory (ROM)603, and a system bus 605 connecting the system memory 604 and the central processing unit 601. The server 600 also includes a basic input/output system (I/O system) 606, which facilitates the transfer of information between devices within the computer, and a mass storage device 607, which stores an operating system 613, application programs 614, and other program modules 615.
The basic input/output system 606 includes a display 608 for displaying information and an input device 609 such as a mouse, keyboard, etc. for a user to input information. Wherein the display 608 and the input device 609 are connected to the central processing unit 601 through an input output controller 610 connected to the system bus 605. The basic input/output system 606 may also include an input/output controller 610 for receiving and processing input from a number of other devices, such as a keyboard, mouse, or electronic stylus. Similarly, input/output controller 610 may also provide output to a display screen, a printer, or other type of output device.
The mass storage device 607 is connected to the central processing unit 601 through a mass storage controller (not shown) connected to the system bus 605. The mass storage device 607 and its associated computer-readable media provide non-volatile storage for the server 600. That is, the mass storage device 607 may include a computer-readable medium (not shown) such as a hard disk or CD-ROM drive.
Without loss of generality, the computer-readable media may comprise computer storage media and communication media. Computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes RAM, ROM, EPROM, EEPROM, flash memory or other solid state memory technology, CD-ROM, DVD, or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices. Of course, those skilled in the art will appreciate that the computer storage media is not limited to the foregoing. The system memory 604 and mass storage device 607 described above may be collectively referred to as memory.
The server 600 may also operate as a remote computer connected to a network via a network, such as the internet, in accordance with various embodiments of the present invention. That is, the server 600 may be connected to the network 612 through the network interface unit 611 connected to the system bus 605, or may be connected to other types of networks or remote computer systems (not shown) using the network interface unit 611.
The memory further includes one or more programs, the one or more programs are stored in the memory, and the central processing unit 601 implements the target user determination method shown in fig. 2 by executing the one or more programs.
In an exemplary embodiment, a non-transitory computer-readable storage medium, such as a memory, including instructions executable by a processor of a server to perform the targeted user determination methods illustrated by the various embodiments of the present invention is also provided. For example, the non-transitory computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
It will be understood that the invention is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.
Claims (9)
1. A method for target user determination, the method comprising:
generating an operation frequency sequence according to a historical operation record of a user on a page of an appointed type, wherein each element in the operation frequency sequence indicates the operation frequency of the user on the page of the appointed type in each unit time period respectively, and the elements are arranged according to the sequence of corresponding time from first to last;
performing wavelet decomposition on the operation frequency sequence to obtain a main body sequence, wherein the main body sequence is a sequence corresponding to low-frequency elements in the operation frequency sequence;
sequentially performing convolution operation on each element in the main body sequence through an increasing sequence;
for the ith element in the main body sequence, calculating a convolution result corresponding to the ith element by the following formula:
wherein F (i) is the convolution result corresponding to the ith element, M is the number of elements in the increasing sequence, ai+hIs the i + h element in the subject sequence,
is (M +1+2h)/2 th element in the increasing sequence, M is more than or equal to 3, and M is an odd number, (M-1)/2+1 is more than or equal to i and less than or equal to N- (M-1)/2, N is the number of the elements in the main body sequence, i + h is more than or equal to N, and i and N are positive integers; the increasing sequence is an arithmetic sequence which is obtained according to the number of elements in the main body sequence and takes 0 as a central element;
and when the result of the convolution operation meets a preset condition, determining the user as a target user.
2. The method according to claim 1, wherein the determining that the user is a target user when the result of the convolution operation satisfies a preset condition comprises:
and when the convolution result corresponding to any element in the elements meets the preset condition, determining that the user is the target user.
3. The method of claim 2,
when the target user is a user with an increasing trend of interest in the object corresponding to the specified type page, the preset condition includes:
the convolution result is larger than a first preset threshold value; or, the convolution result is greater than a second preset threshold, and the time length of the unit time period corresponding to any element from the current time is less than a first preset time length;
when the target user is a user with a downward trend in the interest of the object corresponding to the specified type page, the preset condition includes:
the convolution result is smaller than a third preset threshold value; or, the convolution result is smaller than a fourth preset threshold, and the time length of the unit time period corresponding to any element from the current time is smaller than a second preset time length.
4. The method according to claim 1, wherein the historical operation record includes operation types and operation times of operation behaviors of the user on the specified type of page, and the generating of the operation frequency sequence according to the historical operation record of the user on the specified type of page comprises:
counting the operation times of the user corresponding to each operation type in each unit time period according to the historical operation records;
acquiring a weight coefficient of each operation type;
for each unit time period in the various unit time periods, determining the sum of the products of the times of operation of each operation type and the weight coefficient of each operation type as the frequency of operation of the user in the unit time period;
and arranging the operation frequency of the user in each unit time period according to the sequence of the corresponding time from first to last to obtain the operation frequency sequence.
5. A target user determination apparatus, the apparatus comprising:
the generating module is used for generating an operation frequency sequence according to a historical operation record of a user on a specified type page, wherein each element in the operation frequency sequence indicates the operation frequency of the user on the specified type page in each unit time period respectively, and the elements are arranged according to the sequence of corresponding time from first to last;
the decomposition module is used for performing wavelet decomposition on the operation frequency sequence to obtain a main body sequence, wherein the main body sequence is a sequence corresponding to low-frequency elements in the operation frequency sequence;
the convolution module is used for sequentially carrying out convolution operation on each element in the main body sequence through the incremental sequence;
for the ith element in the main body sequence, calculating a convolution result corresponding to the ith element by the following formula:
wherein F (i) is the convolution result corresponding to the ith element, M is the number of elements in the increasing sequence, ai+hIs the i + h element in the subject sequence,
is (M +1+2h)/2 th element in the increasing sequence, M is more than or equal to 3, and M is an odd number, (M-1)/2+1 is more than or equal to i and less than or equal to N- (M-1)/2, N is the number of the elements in the main body sequence, i + h is more than or equal to N, and i and N are positive integers; the increasing sequence is an arithmetic sequence which is obtained according to the number of elements in the main body sequence and takes 0 as a central element;
and the determining module is used for determining the user as a target user when the result of the convolution operation meets a preset condition.
6. The apparatus of claim 5,
the determining module is configured to determine that the user is a target user when a convolution result corresponding to any one of the elements satisfies the preset condition.
7. The apparatus of claim 6,
when the target user is a user with an increasing trend of interest in the object corresponding to the specified type page, the preset condition includes:
the convolution result is larger than a first preset threshold value; or, the convolution result is greater than a second preset threshold, and the time length of the unit time period corresponding to any element from the current time is less than a first preset time length;
when the target user is a user with a downward trend in the interest of the object corresponding to the specified type page, the preset condition includes:
the convolution result is smaller than a third preset threshold value; or, the convolution result is smaller than a fourth preset threshold, and the time length of the unit time period corresponding to any element from the current time is smaller than a second preset time length.
8. The apparatus of claim 5, wherein the historical operation record comprises operation types and operation times of operation behaviors of the user on pages of specified types, and the generating module comprises:
the counting unit is used for counting the operation times of the user corresponding to each operation type in each unit time period according to the historical operation record;
a coefficient acquisition unit configured to acquire a weight coefficient for each operation type;
a frequency determining unit configured to determine, for each of the unit time periods, a sum of products of the number of operations of each operation type and a weight coefficient of each operation type as an operation frequency of the user in the unit time period;
and the arranging unit is used for arranging the operation frequency of the user in each unit time period according to the sequence of the corresponding time from first to last to obtain the operation frequency sequence.
9. A computer-readable storage medium having stored therein at least one instruction, which is loaded and executed by a processor to implement the target user determination method as claimed in any one of claims 1 to 4.
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 CN108804387A (en) | 2018-11-13 |
| CN108804387B true 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) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
Family Cites Families (2)
| 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 |
| WO2013063734A1 (en) * | 2011-10-31 | 2013-05-10 | Hewlett-Packard Development Company, L.P. | Determining document structure similarity using discrete wavelet transformation |
-
2017
- 2017-04-27 CN CN201710287454.5A patent/CN108804387B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108804387A (en) | 2018-11-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190138643A1 (en) | Extracting seasonal, level, and spike components from a time series of metrics data | |
| CN105677767B (en) | Equipment configuration recommendation method and device | |
| CN112148987B (en) | Message pushing method based on target object activity and related equipment | |
| US20160134934A1 (en) | Estimating audience segment size changes over time | |
| US20110231256A1 (en) | Automated building of a model for behavioral targeting | |
| US20160285672A1 (en) | Method and system for processing network media information | |
| CN110225366B (en) | Video data processing and advertisement space determining method, device, medium and electronic equipment | |
| CN109522190B (en) | Abnormal user behavior identification method and device, electronic equipment and storage medium | |
| CN104216884B (en) | Collaborative filtering system and method based on time attenuation | |
| US10599981B2 (en) | System and method for estimating audience interest | |
| US8898808B1 (en) | System and method for assessing effectiveness of online advertising | |
| CN110889725B (en) | Online advertisement CTR estimation method, device, equipment and storage medium | |
| US20170061484A1 (en) | Method for determining next purchase interval for customer and system thereof | |
| US20180047048A1 (en) | Digital advertising campaign monitoring system, apparatus, and method | |
| US20170140023A1 (en) | Techniques for Determining Whether to Associate New User Information with an Existing User | |
| WO2021108441A1 (en) | Methods, systems and apparatus to estimate census-level audience size and total impression durations across demographics | |
| US20130091009A1 (en) | Identifying users likely to perform for a specific advertiser's campaign goals | |
| CN107741967A (en) | Method, apparatus and electronic equipment for behavioral data processing | |
| EP3414724A1 (en) | Increase choice shares with personalized incentives using social media data | |
| CN108694174B (en) | Content delivery data analysis method and device | |
| CN108804387B (en) | Target user determination method and device | |
| US10990883B2 (en) | Systems and methods for estimating and/or improving user engagement in social media content | |
| CN107357847B (en) | Data processing method and device | |
| CN112070564B (en) | Advertisement pulling method, device and system and electronic equipment | |
| JP6982404B2 (en) | Information processing equipment, information processing methods and information processing programs |
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 |