CN109063900A - The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device - Google Patents
The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device Download PDFInfo
- Publication number
- CN109063900A CN109063900A CN201810775857.9A CN201810775857A CN109063900A CN 109063900 A CN109063900 A CN 109063900A CN 201810775857 A CN201810775857 A CN 201810775857A CN 109063900 A CN109063900 A CN 109063900A
- Authority
- CN
- China
- Prior art keywords
- user
- sample
- sequence
- conversion ratio
- behavior sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides training and user's conversion ratio prediction technique and the device of a kind of user's conversion ratio prediction model, it is excavated by the behavior sequence to user, so as to embody the potential behavior pattern of user, and user's conversion ratio prediction model is trained in conjunction with the similarity selection target behavior sequence of user and other users behavior, the model of acquisition can accurately predict user's conversion ratio, improve the predictablity rate of user's conversion ratio.
Description
Technical field
The present invention relates to the training of computer software technical field more particularly to user's conversion ratio prediction model and user to turn
Rate prediction technique and device.
Background technique
User's conversion ratio refers to that user is converted into the probability of specific user by non-user-specific.For example, user is by non-payment
User is converted into the probability of paying customer or user is converted into the probability for registering user by nonregistered user.It is how accurate pre-
User's conversion ratio is surveyed to be of great significance for product operation and UI update etc..However, traditional user's conversion ratio prediction mode
Predictablity rate is lower.
Summary of the invention
In view of this, the present invention provide a kind of user's conversion ratio prediction model training and user's conversion ratio prediction technique and
Device.
Specifically, the present invention is achieved through the following technical solutions:
According to a first aspect of the embodiments of the present invention, a kind of training method of user's conversion ratio prediction model is provided, it is described
Method includes: to be trained according to the goal behavior sequence of multiple user's samples to user's conversion ratio prediction model；Wherein, any
The goal behavior sequence of user's sample is obtained all in accordance with such as under type: obtaining user's sample to the behavior sequence of item objects
After column, the behavior sequence of user's sample is calculated separately according to the registration liveness of user's sample and each other are used
Similarity between the behavior sequence of family sample；It is selected from the behavior sequence of each other users sample according to the similarity
The goal behavior sequence of user's sample.
Optionally, user's conversion ratio prediction model includes being trained according to the goal behavior sequence of first user's sample
First user's conversion ratio prediction model, first user be registration time length parameter be greater than preset duration threshold value, and register live
Jerk is less than or equal to user's sample of default liveness threshold value；Institute is calculated separately according to the registration liveness of user's sample
The step of stating the similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample includes: respectively will
The behavior sequence of each first user sample is converted to vector, and calculate separately each first user sample vector and it is each its
Similarity between the vector of his first user's sample.
Optionally, user's conversion ratio prediction model includes being trained according to the goal behavior sequence of second user sample
Second user conversion ratio prediction model, the second user sample be registration time length parameter be greater than preset duration threshold value, and infuse
Volume liveness is greater than user's sample of default liveness threshold value；It is calculated separately according to the registration liveness of user's sample described
The step of similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample includes: respectively will be each
The behavior sequence of a second user sample is converted to vector, and calculate separately each second user sample vector and it is each other
Similarity between the vector of second user sample；And the behavior sequence of each second user sample is divided into respectively multiple
Each subsequence is respectively converted into subvector by subsequence, generates array according to the subvector, is calculated separately each second and is used
Similarity between the array of family sample and the array of other each second user samples.
Optionally, user's conversion ratio prediction model includes being trained according to the target signature sequence of third user's sample
Third user's conversion ratio prediction model, the third user sample be registration time length parameter be less than or equal to preset duration threshold value
User's sample；Wherein, the target signature sequence is obtained according to such as under type: calculating the feature sequence of the third user sample
Similarity between column and the characteristic sequence of each other thirds user's sample；It is used according to the similarity from other each thirds
The target signature sequence of the third user sample is selected in the characteristic sequence of family sample.
Optionally, the target signature sequence includes fisrt feature sequence and second feature sequence；Wherein, described first is special
Sign sequence is the highest characteristic sequence of similarity with the characteristic sequence of user's sample；The second feature sequence is and institute
State the minimum characteristic sequence of the similarity of the characteristic sequence of user's sample.
Optionally, the use is calculated according to the login liveness of user's sample, behavior liveness and social liveness
The registration liveness of family sample；Wherein, the login liveness logs in frequency for characterizing, and the behavior liveness is for characterizing
The frequency of the historical behavior event is generated, the social activity liveness is used to characterize the frequency for generating history Social behaviors.
Optionally, the registration liveness calculates according to the following formula:In formula, RA is described
Liveness is registered, rd is registration time length, and la is to log in liveness, and ba is behavior liveness, and sa is social liveness, and N is project
Online time.
Optionally, the goal behavior sequence of user's sample includes the first behavior sequence and the second behavior sequence；Wherein,
First behavior sequence is the highest behavior sequence of similarity with the behavior sequence of user's sample；Second behavior
Sequence is the behavior sequence minimum with the similarity of the behavior sequence of user's sample.
According to a second aspect of the embodiments of the present invention, a kind of user's conversion ratio prediction technique is provided, which comprises root
User's conversion ratio of active user is predicted according to user's conversion ratio prediction model trained in advance；User's conversion ratio is pre-
Model is surveyed to be trained according to the training method of user's conversion ratio prediction model of any embodiment.
Optionally, the method also includes: according to user's conversion ratio to user's PUSH message, in the message
The object information of target object is carried, the target object is the maximum object of user's conversion ratio for making active user.
According to a third aspect of the embodiments of the present invention, a kind of training device of user's conversion ratio prediction model is provided, it is special
Sign is that described device includes: training module, pre- to user's conversion ratio for the goal behavior sequence according to multiple user's samples
Model is surveyed to be trained；Wherein, the goal behavior sequence of any user sample is obtained all in accordance with such as under type: first calculates mould
Block, for being enlivened according to the registration of user's sample after obtaining user's sample to the behavior sequence of item objects
Degree calculates separately the similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample；First
Selecting module, for selecting the mesh of user's sample from the behavior sequence of each other users sample according to the similarity
Mark behavior sequence.
According to a fourth aspect of the embodiments of the present invention, a kind of user's conversion ratio prediction meanss are provided, which is characterized in that described
Device includes: prediction module, for user's conversion ratio according to user's conversion ratio prediction model trained in advance to active user
It is predicted；User's conversion ratio prediction model according to the training method of user's conversion ratio prediction model of any embodiment into
Row training.
According to a fifth aspect of the embodiments of the present invention, a kind of computer readable storage medium is provided, calculating is stored thereon with
Machine program, when described program is executed by processor in realization any embodiment the step of method.
According to a sixth aspect of the embodiments of the present invention, a kind of computer equipment, including memory, processor and storage are provided
On a memory and the computer program that can run on a processor, the processor realize any implementation when executing described program
In example the step of method.
It using the embodiment of the present invention, is excavated by the behavior sequence to user, so as to embody user
Potential behavior pattern, and come in conjunction with the similarity selection target behavior sequence of user and other users behavior to user's conversion ratio
Prediction model is trained, and the model of acquisition can accurately predict user's conversion ratio, improves the prediction of user's conversion ratio
Accuracy rate.
It should be understood that above general description and following detailed description be only it is exemplary and explanatory, not
It can the limitation present invention.
Detailed description of the invention
Fig. 1 is the training method flow chart of user's conversion ratio prediction model of one embodiment of the invention.
Fig. 2 is the overview flow chart of the training method of user's conversion ratio prediction model of one embodiment of the invention.
Fig. 3 (a) and Fig. 3 (b) is the schematic diagram of user's division mode of one embodiment of the invention.
Fig. 4 is the schematic diagram of the behavior sequence similarity calculation of one embodiment of the invention.
Fig. 5 is user's conversion ratio prediction technique flow chart of one embodiment of the invention.
Fig. 6 is the block diagram of the training device of user's conversion ratio prediction model of one embodiment of the invention.
Fig. 7 is the block diagram of user's conversion ratio prediction meanss of one embodiment of the invention.
Fig. 8 is the structural representation of the computer equipment for executing present invention method of one embodiment of the invention
Figure.
Specific embodiment
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 all embodiments consistented with the present invention.On the contrary, they be only with it is such as appended
The example of device and method being described in detail in claims, some aspects of the invention are consistent.
It is only to be not intended to limit the invention merely for for the purpose of describing particular embodiments in terminology used in the present invention.
It is also intended in the present invention and the "an" of singular used in the attached claims, " described " and "the" including majority
Form, unless the context clearly indicates other meaning.It is also understood that term "and/or" used herein refers to and wraps
It may be combined containing one or more associated any or all of project listed.
It will be appreciated that though various information, but this may be described using term first, second, third, etc. in the present invention
A little information should not necessarily be limited by these terms.These terms are only used to for same type of information being distinguished from each other out.For example, not departing from
In the case where the scope of the invention, the first information can also be referred to as the second information, and similarly, the second information can also be referred to as
One information.Depending on context, word as used in this " if " can be construed to " ... when " or " when ...
When " or " in response to determination ".
As shown in Figure 1, Fig. 1 is the training method flow chart of user's conversion ratio prediction model of one embodiment of the invention,
The method can include:
Step 101: user's conversion ratio prediction model being trained according to the goal behavior sequence of multiple user's samples；Its
In, the goal behavior sequence of any user sample is obtained all in accordance with such as under type:
Step 102: after obtaining user's sample to the behavior sequence of item objects, according to user's sample
Registration liveness calculates separately the phase between the behavior sequence of user's sample and the behavior sequence of each other users sample
Like degree；
Step 103: user's sample is selected from the behavior sequence of each other users sample according to the similarity
Goal behavior sequence.
User's sample is the user of known users label, and user tag is for characterizing whether user's sample is that conversion is used
Family, if so, 1 can be set as user tag；If it is not, 0 can be set as user tag.User's conversion refers to user by non-spy
Determine user and is converted into specific user.For example, user is converted into paying customer or user by nonregistered user by non-payment user
It is converted into registration user.The behavior sequence of user's sample be according to user's sample to item objects performed by a series of behavior things
Part and the sequence generated.Wherein, project can be application program, and item objects can be the object in application program, for example, can
To be Taobao, Jingdone district etc. for the commodity in the application program of shopping；It is also possible to today's tops, Netease's news etc. for reading
News in the application program of news.Above-mentioned behavior event can be clickthrough in the application, purchase/collection quotient
The behaviors such as comment, forwarding information are delivered/checked to product.
The goal behavior sequence of user's sample be used for characterize user's sample behavior behavior pattern (also referred to as Behavior preference or
Behavioural habits).Goal behavior sequence can be the highest behavior sequence of behavior sequence similarity with user's sample, can also be same
When include with the highest behavior sequence of behavior sequence similarity of user's sample and with the behavior sequence similarity of user's sample
Minimum behavior sequence.
When carrying out model training, the goal behavior sequence of each user's sample can be obtained respectively, then by each mesh
Mark behavior sequence, which is input in user's conversion ratio prediction model, solves model parameter.Assuming that including in user's sample
{ u1, u2 ..., un }, behavior sequence are respectively s1, s2 ..., sn, then can calculate separately s1 and s2 ..., the similarity of sn,
S2 and s1, s3 ..., the similarity ... ... of sn, the similarity of sn and s1 ..., sn-1.Then, according to s2 ..., the phase of sn and s1
The goal behavior sequence for selecting s1 from s2 ..., sn like degree, according to s2 and s1, the similarity of s3 ..., sn is from s1, s3 ..., sn
The goal behavior sequence ... ... of middle selection s2, according to sn and s1 ..., the similarity of sn-1 selects sn's from s1 ..., sn-1
Goal behavior sequence.Finally, according to the goal behavior sequence of s1, the goal behavior sequence ... ... of s2, the goal behavior sequence of sn
User's conversion ratio prediction model is trained.
It using the embodiment of the present invention, is excavated by the behavior sequence to user, so as to embody user
Potential behavior pattern, and come in conjunction with the similarity selection target behavior sequence of user and other users behavior to user's conversion ratio
Prediction model is trained, and the model of acquisition can accurately predict user's conversion ratio, improves the prediction of user's conversion ratio
Accuracy rate.
In one embodiment, user's conversion ratio prediction model includes the goal behavior sequence according to first user's sample
The first user's conversion ratio prediction model trained is arranged, first user is that registration time length parameter is greater than preset duration threshold value,
And register user's sample that liveness is less than or equal to default liveness threshold value；According to the registration liveness of user's sample point
The step of not calculating the similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample packet
Include: the behavior sequence of each first user sample being converted into vector respectively, and calculate separately each first user sample to
Similarity between amount and the vector of other each the first user samples.
Registration time length parameter is used to characterize the time span between user's registration time and current time, can pass through registration
Time (for example, registration number of days, registration year) or other parameters relevant to registion time indicate.Registering liveness can
The time that there is behavior event upon registration for characterizing user's sample accounts for the ratio of registration time length.Liveness threshold value can adopt
Be averaged registration time length with the user in historical time section, can also according to practical application using other parameters as liveness threshold
Value.
First user's conversion ratio prediction model can be used for being greater than registration time length parameter preset duration threshold value, and registers and live
User's conversion ratio that jerk is less than or equal to the first user of default liveness threshold value is predicted.
Whole behavior sequences for the u1 in first user's sample, after its available registration, it is assumed that behavior sequence
For { log in, click advertisement 1, click advertisement 2, read news 1, read news 2, click commodity 1, check comment, buy }, behavior
Each behavior in sequence can correspond to a value, therefore, above-mentioned behavior sequence can be converted to vector (a, c1, c1, r,
R, c2, v, b), a is the corresponding value of login behavior, and c1 is to click the corresponding value of advertisement behavior, and r is to read news behavior pair
The value answered, c2 are to click the corresponding value of commodity behavior, and v is to check the corresponding value of comment behavior, and b is corresponding for buying behavior
Value.For other the first user samples, the mode for obtaining vector is identical as u1, and details are not described herein again.
In one embodiment, user's conversion ratio prediction model includes the goal behavior sequence according to second user sample
The second user conversion ratio prediction model trained is arranged, the second user sample is that registration time length parameter is greater than preset duration threshold
Value, and register user's sample that liveness is greater than default liveness threshold value；Distinguished according to the registration liveness of user's sample
The step of calculating the similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample include:
The behavior sequence of each second user sample is converted into vector respectively, and calculate separately the vector of each second user sample with
Similarity between the vector of other each second user samples；And the behavior sequence of each second user sample is drawn respectively
It is divided into multiple subsequences, each subsequence is respectively converted into subvector, array is generated according to the subvector, calculates separately every
Similarity between the array of a second user sample and the array of other each second user samples.
Second user conversion ratio prediction model can be used for being greater than registration time length parameter preset duration threshold value, and registers and live
User's conversion ratio that jerk is greater than the second user of default liveness threshold value is predicted.
Since the registion time and registration liveness value of second user are larger, the corresponding behavior of second user
Sequence is usually a very long sequence.For ease of calculation, the behavior sequence of second user sample can be divided into more
A subsequence.For the u2 in second user sample, it is assumed that its behavior sequence is { a1, a2 ... ..., an }, then by behavior sequence
Column are divided into m subsequence: { a11, a12 ... ..., a1n }, { a21, a22 ... ..., a2n } ... ..., and am1, am2 ... ...,
Amn }, wherein 1n+2n+ ...+mn=n.For each subsequence, it is respectively converted into subvector, conversion regime and the first user
Conversion regime it is identical, details are not described herein again.Assuming that the corresponding subvector of m subsequence be respectively [x11, x12 ...,
X1n], [x21, x22 ... ..., x2n] ... ..., [xm1, xm2 ... ..., xmn], then the array of each subvector composition can be
Following form:
In addition to this it is possible to by the behavior sequence { a1, a2 ... ..., an } of second user u2 and other each second use
The complete behavior sequence at family calculates separately similarity, calculation and the mode phase for calculating the first user behavior sequence similarity
Together.
Similarity can use the calculation of cosine similarity.For the first user, the first user can be directly calculated
Vector cosine similarity between any two；It, can be using each column of above-mentioned array as vector for the array of second user
An element, then calculate the similarity between array and array.Certainly, the array that subvector is constituted is also possible to above-mentioned number
Then the transposed form of group, then when calculating similarity, can calculate array using every a line of array as an element of vector
Cosine similarity between array.For the vector of second user, the cosine phase of vector between any two can also be directly calculated
Like degree.
It in one embodiment, can be active according to the login liveness, behavior liveness and social activity of user's sample
Degree calculates the registration liveness of user's sample；Wherein, the login liveness logs in frequency for characterizing, and the behavior is living
Jerk generates history Social behaviors for characterizing for characterizing the frequency for generating the historical behavior event, the social activity liveness
Frequency.
Registration time length refers to the time interval between user's registration time interval current time, can be as unit of day.It logs in
Liveness can pass through the ratio calculation of login times and registration time length, wherein user logs in again after publishing every time and calculates one
Secondary login times, that is, as long as user, which does not execute, publishes operation, login times do not add up.Behavior liveness can pass through behavior thing
The ratio of number and registration time length that part occurs calculates, and behavior number here includes all behavior events time that user occurs
Several summations, behavior event include that comment, forwarding are checked in clickthrough in the application, purchase/collecting commodities are delivered/
The behaviors such as information.For that can add the application program of good friend between user, social liveness can by addition good friend's quantity with
The ratio of login time calculates.For it is no addition good friend's function application program, social liveness can by user with
The ratio of interaction (for example, the hair personal letter) numbers of other users and login time calculates, and can also be commented by what user sent
It is calculated by number with the ratio for logging in number of days.Wherein, login time can be as unit of day.
Specifically, the registration liveness calculates according to the following formula:
In formula, RA is the registration liveness, and rd is registration time length, and la is to log in liveness, and ba is behavior liveness, sa
For social liveness, N is project online time.
Logarithm is taken to liveness in above-mentioned formula, liveness can be narrowed down to a fixed range, and logarithm is
It is monotonic increase in its domain, so will not influence the relationship of liveness.Above-mentioned formula can be with the registration day of user
Number is used as benchmark, and using liveness as multiplier factor (enlivening the factor), this enlivens the factor and considers the activity of the user size,
It is reduced the scope using logarithm, attribute when online (for example, application program) of product itself, each user couple is further added
Obtain a RA value.Liveness threshold value can be used user and be averaged registration time length, and all RA values are greater than user and are averaged registration time length
It is the first user less than or equal to user's registration time length that is averaged for second user, the threshold of " user be averaged registration time length " here
Value can flexibly change according to business or operation goal, generally take average user's registration duration preferable.
It in one embodiment, can be by the phase of the behavior sequence with user's sample after calculating similarity
Like highest first behavior sequence of degree, and second behavior sequence minimum with the similarity of the behavior sequence of user's sample
Collectively as the goal behavior sequence of user's sample.
It, can will the corresponding behavior sequence of corresponding with the first user highest vector of vector similarity for the first user
Column, and mesh of the minimum corresponding behavior sequence of vector of vector similarity corresponding with the first user as first user
Mark behavior sequence.It, can will the corresponding behavior of corresponding with the second user highest vector of vector similarity for second user
Sequence, the minimum corresponding behavior sequence of vector of vector similarity corresponding with second user, array corresponding with second user
The corresponding behavior sequence of the highest array of similarity, and array similarity corresponding with second user it is minimum array it is corresponding
Goal behavior sequence of the behavior sequence as the second user.
Further, user's conversion ratio prediction model includes according to the training of the target signature sequence of third user's sample
Third user's conversion ratio prediction model out, the third user sample are that registration time length parameter is less than or equal to preset duration threshold
User's sample of value；Wherein, the target signature sequence is obtained according to such as under type: calculating the feature of the third user sample
Similarity between sequence and the characteristic sequence of each other thirds user's sample；According to the similarity from other each thirds
The target signature sequence of the third user sample is selected in the characteristic sequence of user's sample.
Characteristic sequence can be used for characterizing the user characteristics of user's sample, may include individual's letter from user in characteristic sequence
The feature extracted in breath, such as: one or more of information such as age, gender, residence, hobby, occupation, mailbox.Third
User's conversion ratio prediction model can be used for being less than or equal to registration time length parameter the use of the third user of preset duration threshold value
Family conversion ratio is predicted.Third user may be not present or be only existed and is shorter since its registion time is shorter
Therefore behavior sequence can train third user's conversion ratio prediction model by the characteristic sequence of third user.
It in one embodiment, can be by the phase of the characteristic sequence with user's sample after calculating similarity
Like the highest fisrt feature sequence of degree, and the second feature sequence minimum with the similarity of the characteristic sequence of user's sample
Collectively as the target signature sequence of user's sample.
When being trained to user's conversion ratio prediction model, can turn using behavior sequence or characteristic sequence as user
The input of rate prediction model predicts mould to user's conversion ratio using user tag as the output of user's conversion ratio prediction model
The parameter of type is solved.
The overall procedure of the training method of user's conversion ratio prediction model of one embodiment is as shown in Fig. 2, include as follows
Step:
Step 201: user's sample divides.User's sample is divided into new registration user according to the registion time of user's sample
Sample and old user's sample, new user's sample, that is, registration time length are less than user's sample of preset duration threshold value, and old user's sample is
Registration time length is greater than or equal to user's sample of preset duration threshold value.Old user's sample is divided into further according to registration liveness short
Phase user's sample and long-time users sample, Short-term user sample are to register liveness to be less than or equal to the old of default liveness threshold value
User's sample, long-time users sample are the old user's sample registered liveness and be greater than default liveness threshold value.Wherein, registration is active
Degree can be calculated according to liveness, behavior liveness and social liveness is logged in.Specific user's division mode such as Fig. 3 (a)
With shown in Fig. 3 (b).
Step 202, similarity is calculated.For new user's sample, new user's sample and other each new user's samples are calculated
Similarity between this characteristic sequence vector.For Short-term user sample, the short behavior sequence of the Short-term user sample is calculated
Similarity between vector and the short behavior sequence vector of other Short-term user samples.For long-time users sample, the length is calculated
Similarity between the long behavior sequence vector of phase user's sample and the long behavior sequence vector of other long-time users samples, and
Similarity between the behavior sequence array of the long-time users sample and the behavior sequence array of other long-time users samples.Its
In, characteristic sequence vector is the vector generated according to the characteristic sequence of new user's sample, and short behavior sequence vector is according to short-term
The vector that the behavior sequence of user's sample generates, long behavior sequence vector is generated according to the behavior sequence of long-time users sample
Vector, behavior sequence array are the arrays that the sub- behavior sequence according to long-time users sample in each measurement period generates.Meter
The mode for calculating similarity is as shown in Figure 4.
Wherein, a measurement period can be the period that user uses application program.For example, being opened with login application program
Begin, exiting application program terminates, and a measurement period is used as during this.All behaviors generated in one measurement period can conduct
Behavior in the same behavior sequence.The behavior sequence obtained by way of this division measurement period is more loose, content
Cover more.It can also be started with the homepage for clicking application program, be terminated with payment, from homepage is clicked to payment as a system
Count the period.The behavior sequence obtained by way of this division measurement period is more harsh, can reject some behavior sequences.
In practical applications, it can choose different modes under different operation scenes to divide measurement period.The each system marked off
The short behavior sequence in meter period calculates similarity after may also pass through deduplication operation again.
Step 203: selection target behavior sequence and target signature sequence.It, will be with new user's sample for new user's sample
The minimum and maximum corresponding characteristic sequence of characteristic sequence vector of this characteristic sequence vector similarity is as new user's sample
Target signature sequence.It is for Short-term user sample, the short behavior sequence vector similarity with the Short-term user sample is maximum
Goal behavior sequence of the short behavior sequence corresponding with the smallest short behavior sequence vector as the Short-term user sample.For length
Phase user's sample, by the long behavior sequence vector minimum and maximum with the long behavior sequence vector similarity of the long-time users sample
Corresponding long behavior sequence, and the behavior sequence minimum and maximum with the behavior sequence array similarity of the long-time users sample
The corresponding long behavior sequence of array, collectively as the goal behavior sequence of the long-time users sample.
Step 204: user's conversion ratio prediction model being trained according to goal behavior sequence or target signature sequence.
First user's conversion ratio prediction model is trained according to the target signature sequence of new user's sample, according to Short-term user sample
Goal behavior sequence second user conversion ratio prediction model is trained, according to the goal behavior sequence of long-time users sample
Third user's conversion ratio prediction model is trained.Wherein, first user's conversion ratio prediction model is for predicting new user's
User's conversion ratio, second user conversion ratio prediction model are used to predict user's conversion ratio of Short-term user, third user's conversion ratio
Prediction model is used for user's conversion ratio of predicting long-term user.
As shown in figure 5, being user's conversion ratio prediction technique flow chart of one embodiment of the invention, which comprises
User's conversion ratio of active user is predicted according to user's conversion ratio prediction model trained in advance；Wherein,
User's conversion ratio prediction model is trained according to the training method of user's conversion ratio prediction model of any embodiment.
In one embodiment, for new user, the user characteristics of the new user can be inputted into first user's conversion ratio
Prediction model obtains user's transition probability of the new user.It, can be by the short behavior sequence of the Short-term user for Short-term user
Second user conversion ratio prediction model is inputted, user's transition probability of the Short-term user is obtained.It, can should for long-time users
The long behavior sequence of long-time users inputs third user conversion ratio prediction model, obtains user's transition probability of the long-time users.
User's transition probability can characterize active user to the transition probability of a certain item objects.
It therefore, in one embodiment, can also be according to user's conversion ratio to active user's PUSH message, institute
The object information that target object is carried in message is stated, the target object is to keep user's conversion ratio of active user maximum right
As.In this way, user's transition probability can be improved.
Corresponding with the embodiment of training method of above-mentioned user's conversion ratio prediction model, the present invention also provides a kind of use
The training device of family conversion ratio prediction model, as shown in fig. 6, described device can include:
Training module 601, for according to the goal behavior sequences of multiple user's samples to user's conversion ratio prediction model into
Row training；Wherein, the goal behavior sequence of any user sample is obtained all in accordance with such as under type:
First computing module 602, for after obtaining user's sample to the behavior sequence of item objects, according to institute
The registration liveness for stating user's sample calculates separately the behavior sequence of user's sample and the behavior of each other users sample
Similarity between sequence；
First choice module 603, for being selected from the behavior sequence of each other users sample according to the similarity
The goal behavior sequence of user's sample.
Corresponding with the embodiment of above-mentioned user's conversion ratio prediction technique, the present invention also provides a kind of user's conversion ratio is pre-
Device is surveyed, as shown in fig. 7, described device can include:
Prediction module 701, for being converted according to user of the user's conversion ratio prediction model trained in advance to active user
Rate is predicted；Active user's conversion ratio prediction model is according to the training of user's conversion ratio prediction model of any embodiment
Method is trained.
The function of each unit and the realization process of effect are specifically detailed in above-mentioned corresponding method corresponding step in above-mentioned apparatus
Rapid realization process, details are not described herein.
For device embodiment, since it corresponds essentially to embodiment of the method, so related place is referring to method reality
Apply the part explanation of example.The apparatus embodiments described above are merely exemplary, wherein described be used as separation unit
The unit of explanation may or may not be physically separated, and component shown as a unit can be or can also be with
It is not physical unit, it can it is in one place, or may be distributed over multiple network units.It can be according to actual
The purpose for needing to select some or all of the modules therein to realize the present invention program.Those of ordinary skill in the art are not paying
Out in the case where creative work, it can understand and implement.
In one embodiment, the present invention also provides a kind of computer readable storage mediums, are stored thereon with computer journey
Sequence realizes the method for above-mentioned any one embodiment when described program is executed by processor.
In one embodiment, it the present invention also provides a kind of computer equipment, including memory, processor and is stored in
On reservoir and the computer program that can run on a processor, the processor realize above-mentioned any one reality when executing described program
The method for applying example.
As shown in figure 8, Fig. 8 is shown according to an exemplary embodiment a kind of for executing present invention method
The structural schematic diagram of computer equipment.Referring to Fig. 8, it further comprises one that computer equipment 800, which includes processing component 801,
Or multiple processors, and the memory resource as representated by memory 802, it can be by the execution of processing component 801 for storing
Instruction, such as application program.The application program stored in memory 802 may include that one or more each is right
The module of Ying Yuyi group instruction.In addition, processing component 801 is configured as executing instruction, to execute the above method.
Computer equipment 800 can also include that a power supply module 803 is configured as executing the power supply of computer equipment 800
Management, a wired or wireless network interface 804 is configured as computer equipment 800 being connected to network and an input is defeated
(I/O) interface 805 out.Computer equipment 800 can be operated based on the operating system for being stored in memory 802.Wherein, when described
When instruction in memory 802 is executed by the processing component 801, so that computer equipment 800 is able to carry out any of the above-described reality
The method for applying example.
Theme described in this specification and the embodiment of feature operation can be realized in the following: Fundamental Digital Circuit,
Computer software or firmware, the computer including structure disclosed in this specification and its structural equivalents of tangible embodiment are hard
The combination of part or one or more of which.The embodiment of theme described in this specification can be implemented as one or
Multiple computer programs, i.e. coding are executed by data processing equipment on tangible non-transitory program carrier or are controlled at data
Manage one or more modules in the computer program instructions of the operation of device.Alternatively, or in addition, program instruction can be with
It is coded on manually generated transmitting signal, such as electricity, light or electromagnetic signal that machine generates, the signal are generated will believe
Breath encodes and is transferred to suitable receiver apparatus to be executed by data processing equipment.Computer storage medium can be machine can
Read storage equipment, machine readable storage substrate, random or serial access memory equipment or one or more of which group
It closes.
Processing described in this specification and logic flow can by execute one of one or more computer programs or
Multiple programmable calculators execute, to execute corresponding function by the way that output is operated and generated according to input data.Institute
It states processing and logic flow can also be by dedicated logic circuit-such as FPGA (field programmable gate array) or ASIC (dedicated collection
At circuit) it executes, and device also can be implemented as dedicated logic circuit.
The computer for being suitable for carrying out computer program includes, for example, general and/or special microprocessor or it is any its
The central processing unit of his type.In general, central processing unit will refer to from read-only memory and/or random access memory reception
Order and data.The basic module of computer includes central processing unit for being practiced or carried out instruction and for storing instruction
With one or more memory devices of data.In general, computer will also be including one or more great Rong for storing data
Amount storage equipment, such as disk, magneto-optic disk or CD etc. or computer will be coupled operationally with this mass-memory unit
To receive from it data or have both at the same time to its transmission data or two kinds of situations.However, computer is not required to have in this way
Equipment.In addition, computer can be embedded in another equipment, such as mobile phone, personal digital assistant (PDA), mobile sound
Frequency or video player, game console, global positioning system (GPS) receiver or such as universal serial bus (USB) flash memory
The portable memory apparatus of driver, names just a few.
It is suitable for storing computer program instructions and the computer-readable medium of data including the non-volatile of form of ownership
Memory, medium and memory devices, for example including semiconductor memory devices (such as EPROM, EEPROM and flash memory device),
Disk (such as internal hard drive or removable disk), magneto-optic disk and CD ROM and DVD-ROM disk.Processor and memory can be by special
It is supplemented or is incorporated in dedicated logic circuit with logic circuit.
Although this specification includes many specific implementation details, these are not necessarily to be construed as the model for limiting any invention
It encloses or range claimed, and is primarily used for describing the feature of the specific embodiment of specific invention.In this specification
Certain features described in multiple embodiments can also be combined implementation in a single embodiment.On the other hand, individually implementing
Various features described in example can also be performed separately in various embodiments or be implemented with any suitable sub-portfolio.This
Outside, although feature can work in certain combinations as described above and even initially so be claimed, institute is come from
One or more features in claimed combination can be removed from the combination in some cases, and claimed
Combination can be directed toward the modification of sub-portfolio or sub-portfolio.
Similarly, although depicting operation in the accompanying drawings with particular order, this is understood not to require these behaviour
Make the particular order shown in execute or sequentially carry out or require the operation of all illustrations to be performed, to realize desired knot
Fruit.In some cases, multitask and parallel processing may be advantageous.In addition, the various system modules in above-described embodiment
Separation with component is understood not to be required to such separation in all embodiments, and it is to be understood that described
Program assembly and system can be usually integrated in together in single software product, or be packaged into multiple software product.
The specific embodiment of theme has been described as a result,.Other embodiments are within the scope of the appended claims.?
In some cases, the movement recorded in claims can be executed in different order and still realize desired result.This
Outside, the processing described in attached drawing and it is nonessential shown in particular order or sequential order, to realize desired result.In certain realities
In existing, multitask and parallel processing be may be advantageous.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (14)
1. a kind of training method of user's conversion ratio prediction model, which is characterized in that the described method includes:
User's conversion ratio prediction model is trained according to the goal behavior sequence of multiple user's samples；Wherein, any user
The goal behavior sequence of sample is obtained all in accordance with such as under type:
After obtaining user's sample to the behavior sequence of item objects, according to the registration liveness of user's sample point
The similarity between the behavior sequence of user's sample and the behavior sequence of each other users sample is not calculated；
The goal behavior sequence of user's sample is selected from the behavior sequence of each other users sample according to the similarity
Column.
2. the method according to claim 1, wherein user's conversion ratio prediction model includes using according to first
First user's conversion ratio prediction model that the goal behavior sequence of family sample trains, first user is registration time length parameter
Greater than preset duration threshold value, and register user's sample that liveness is less than or equal to default liveness threshold value；
According to the registration liveness of user's sample calculate separately user's sample behavior sequence and each other users
The step of similarity between the behavior sequence of sample includes:
The behavior sequence of each first user sample is converted into vector respectively, and calculate separately each first user sample to
Similarity between amount and the vector of other each the first user samples.
3. the method according to claim 1, wherein user's conversion ratio prediction model includes using according to second
The second user conversion ratio prediction model that the goal behavior sequence of family sample trains, the second user sample is registration time length
Parameter is greater than preset duration threshold value, and registers user's sample that liveness is greater than default liveness threshold value；
According to the registration liveness of user's sample calculate separately user's sample behavior sequence and each other users
The step of similarity between the behavior sequence of sample includes:
The behavior sequence of each second user sample is converted into vector respectively, and calculate separately each second user sample to
Similarity between amount and the vector of other each second user samples；And
The behavior sequence of each second user sample is divided into multiple subsequences respectively, each subsequence is respectively converted into son
Vector generates array according to the subvector, calculates separately the array and other each second users of each second user sample
Similarity between the array of sample.
4. the method according to claim 1, wherein user's conversion ratio prediction model includes being used according to third
Third user's conversion ratio prediction model that the target signature sequence of family sample trains, the third user sample is registration time length
Parameter is less than or equal to user's sample of preset duration threshold value；
Wherein, the target signature sequence is obtained according to such as under type:
It calculates similar between the characteristic sequence of the third user sample and the characteristic sequence of each other thirds user's sample
Degree；
The mesh of the third user sample is selected from the characteristic sequence of each other thirds user's sample according to the similarity
Mark characteristic sequence.
5. according to the method described in claim 4, it is characterized in that, the target signature sequence includes fisrt feature sequence and
Two characteristic sequences；
Wherein, the fisrt feature sequence is the highest characteristic sequence of similarity with the characteristic sequence of user's sample；
The second feature sequence is the characteristic sequence minimum with the similarity of the characteristic sequence of user's sample.
6. the method according to claim 1, wherein the method also includes:
The registration of user's sample is calculated according to the login liveness of user's sample, behavior liveness and social liveness
Liveness；
Wherein, the login liveness logs in frequency for characterizing, and the behavior liveness generates the history row for characterizing
For the frequency of event, the social activity liveness is used to characterize the frequency for generating history Social behaviors.
7. according to the method described in claim 6, it is characterized in that, the registration liveness calculates according to the following formula:
In formula, RA is the registration liveness, and rd is registration time length, and la is to log in liveness, and ba is behavior liveness, and sa is society
Turn over a finished item jerk, and N is project online time.
8. the method according to claim 1, wherein the goal behavior sequence of user's sample includes the first row
For sequence and the second behavior sequence；
Wherein, first behavior sequence is the highest behavior sequence of similarity with the behavior sequence of user's sample；
Second behavior sequence is the behavior sequence minimum with the similarity of the behavior sequence of user's sample.
9. a kind of user's conversion ratio prediction technique, which is characterized in that the described method includes:
User's conversion ratio of active user is predicted according to user's conversion ratio prediction model trained in advance；
User's conversion ratio prediction model is trained to method described in 8 any one according to claim 1.
10. according to the method described in claim 9, it is characterized in that, the method also includes:
According to user's conversion ratio to active user's PUSH message, the object letter of target object is carried in the message
Breath, the target object is the maximum object of user's conversion ratio for making active user.
11. a kind of training device of user's conversion ratio prediction model, which is characterized in that described device includes:
Training module, for being trained according to the goal behavior sequence of multiple user's samples to user's conversion ratio prediction model；
Wherein, the goal behavior sequence of any user sample is obtained all in accordance with such as under type:
First computing module, for after obtaining user's sample to the behavior sequence of item objects, according to the user
The registration liveness of sample calculate separately user's sample behavior sequence and each other users sample behavior sequence it
Between similarity；
First choice module, for selecting the user from the behavior sequence of each other users sample according to the similarity
The goal behavior sequence of sample.
12. a kind of user's conversion ratio prediction meanss, which is characterized in that described device includes:
Prediction module, it is pre- for being carried out according to user conversion ratio of the user's conversion ratio prediction model trained in advance to active user
It surveys；
Active user's conversion ratio prediction model is trained to method described in 8 any one according to claim 1.
13. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that described program is processed
The step of claims 1 to 10 any one the method is realized when device executes.
14. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes claims 1 to 10 any one the method when executing described program
The step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810775857.9A CN109063900A (en) | 2018-07-16 | 2018-07-16 | The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810775857.9A CN109063900A (en) | 2018-07-16 | 2018-07-16 | The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109063900A true CN109063900A (en) | 2018-12-21 |
Family
ID=64816774
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810775857.9A Pending CN109063900A (en) | 2018-07-16 | 2018-07-16 | The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109063900A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110740193A (en) * | 2019-10-30 | 2020-01-31 | 江苏满运软件科技有限公司 | platform activity prediction method and device, storage medium and electronic equipment |
| CN111724184A (en) * | 2019-03-20 | 2020-09-29 | 北京嘀嘀无限科技发展有限公司 | Transformation probability prediction method and device |
| CN111767982A (en) * | 2020-05-20 | 2020-10-13 | 北京大米科技有限公司 | Training method and device for user conversion prediction model, storage medium and electronic equipment |
| CN112116393A (en) * | 2020-09-23 | 2020-12-22 | 贝壳技术有限公司 | Method, device and equipment for realizing event user maintenance |
| CN113395200A (en) * | 2020-03-12 | 2021-09-14 | 阿里巴巴集团控股有限公司 | Message pushing method and system, client, storage medium and processor |
| CN114399355A (en) * | 2021-12-22 | 2022-04-26 | 中国电信股份有限公司 | Information pushing method and device based on user conversion rate and electronic equipment |
| CN116883048A (en) * | 2023-07-12 | 2023-10-13 | 广州朝辉智能科技有限公司 | Customer data processing method and device based on artificial intelligence and computer equipment |
| CN117893256A (en) * | 2024-03-14 | 2024-04-16 | 浙江卡赢信息科技有限公司 | Big data-based app user intelligent management system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100114663A1 (en) * | 2008-11-03 | 2010-05-06 | Oracle International Corporation | Hybrid prediction model for a sales prospector |
| CN105446802A (en) * | 2014-08-13 | 2016-03-30 | 阿里巴巴集团控股有限公司 | Operation execution method and device based on conversion rate |
| CN105631538A (en) * | 2015-12-23 | 2016-06-01 | 北京奇虎科技有限公司 | User activity prediction method and device, and application method and system thereof |
| CN107392645A (en) * | 2017-06-20 | 2017-11-24 | 小草数语（北京）科技有限公司 | Usage mining method, apparatus and its equipment |
| CN107767152A (en) * | 2016-08-16 | 2018-03-06 | 平安科技（深圳）有限公司 | Product purchase intention analysis method and server |
| CN107944643A (en) * | 2017-12-22 | 2018-04-20 | 上海斐讯数据通信技术有限公司 | A kind of purchase conversion ratio Forecasting Methodology and system based on Hidden Markov Model |
-
2018
- 2018-07-16 CN CN201810775857.9A patent/CN109063900A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100114663A1 (en) * | 2008-11-03 | 2010-05-06 | Oracle International Corporation | Hybrid prediction model for a sales prospector |
| CN105446802A (en) * | 2014-08-13 | 2016-03-30 | 阿里巴巴集团控股有限公司 | Operation execution method and device based on conversion rate |
| CN105631538A (en) * | 2015-12-23 | 2016-06-01 | 北京奇虎科技有限公司 | User activity prediction method and device, and application method and system thereof |
| CN107767152A (en) * | 2016-08-16 | 2018-03-06 | 平安科技（深圳）有限公司 | Product purchase intention analysis method and server |
| CN107392645A (en) * | 2017-06-20 | 2017-11-24 | 小草数语（北京）科技有限公司 | Usage mining method, apparatus and its equipment |
| CN107944643A (en) * | 2017-12-22 | 2018-04-20 | 上海斐讯数据通信技术有限公司 | A kind of purchase conversion ratio Forecasting Methodology and system based on Hidden Markov Model |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111724184A (en) * | 2019-03-20 | 2020-09-29 | 北京嘀嘀无限科技发展有限公司 | Transformation probability prediction method and device |
| CN110740193A (en) * | 2019-10-30 | 2020-01-31 | 江苏满运软件科技有限公司 | platform activity prediction method and device, storage medium and electronic equipment |
| CN113395200B (en) * | 2020-03-12 | 2023-07-21 | 阿里巴巴集团控股有限公司 | Message pushing method and system, client, storage medium and processor |
| CN113395200A (en) * | 2020-03-12 | 2021-09-14 | 阿里巴巴集团控股有限公司 | Message pushing method and system, client, storage medium and processor |
| CN111767982A (en) * | 2020-05-20 | 2020-10-13 | 北京大米科技有限公司 | Training method and device for user conversion prediction model, storage medium and electronic equipment |
| CN112116393A (en) * | 2020-09-23 | 2020-12-22 | 贝壳技术有限公司 | Method, device and equipment for realizing event user maintenance |
| CN112116393B (en) * | 2020-09-23 | 2022-03-15 | 贝壳找房(北京)科技有限公司 | Method, device and equipment for realizing event user maintenance |
| CN114399355A (en) * | 2021-12-22 | 2022-04-26 | 中国电信股份有限公司 | Information pushing method and device based on user conversion rate and electronic equipment |
| CN114399355B (en) * | 2021-12-22 | 2023-10-10 | 中国电信股份有限公司 | Information pushing method and device based on user conversion rate and electronic equipment |
| CN116883048A (en) * | 2023-07-12 | 2023-10-13 | 广州朝辉智能科技有限公司 | Customer data processing method and device based on artificial intelligence and computer equipment |
| CN116883048B (en) * | 2023-07-12 | 2024-03-15 | 卓盛科技(广州)有限公司 | Customer data processing method and device based on artificial intelligence and computer equipment |
| CN117893256A (en) * | 2024-03-14 | 2024-04-16 | 浙江卡赢信息科技有限公司 | Big data-based app user intelligent management system |
| CN117893256B (en) * | 2024-03-14 | 2024-05-31 | 浙江卡赢信息科技有限公司 | Big data-based app user intelligent management system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109063900A (en) | The training of user's conversion ratio prediction model and user's conversion ratio prediction technique and device | |
| CN104090912B (en) | Information-pushing method and device | |
| CN108256907A (en) | A kind of construction method and computing device of customer grouping model | |
| Araújo et al. | Ensemble forecasting of species distributions | |
| CN109657138A (en) | A kind of video recommendation method, device, electronic equipment and storage medium | |
| CN109067990A (en) | A kind of application service execution method and device | |
| CN108109011A (en) | A kind of anti-cheat method of advertisement and computing device | |
| CN107729532A (en) | A kind of resume matching process and computing device | |
| CN107704485A (en) | A kind of position recommends method and computing device | |
| CN110351318A (en) | Using the method, terminal and computer storage medium of recommendation | |
| CN109636479A (en) | A kind of advertisement recommended method, device, electronic equipment and storage medium | |
| CN112380449B (en) | Information recommendation method, model training method and related device | |
| CN110910201B (en) | Information recommendation control method and device, computer equipment and storage medium | |
| CN110246037A (en) | Transaction feature prediction technique, device, server and readable storage medium storing program for executing | |
| CN107093091A (en) | A kind of data processing method and device | |
| CN106445954A (en) | Business object display method and apparatus | |
| Cheng et al. | Randomly weighted sums of dependent subexponential random variables with applications to risk theory | |
| Jessen et al. | Prediction of outstanding payments in a Poisson cluster model | |
| CN109117442A (en) | A kind of application recommended method and device | |
| CN112288554A (en) | Commodity recommendation method and device, storage medium and electronic device | |
| CN110166498A (en) | Class of subscriber determines method and device, computer equipment and storage medium | |
| CN108667877B (en) | Method and device for determining recommendation information, computer equipment and storage medium | |
| Nasrabadi et al. | A stepwise benchmarking approach to DEA with interval scale data | |
| Stoklosa et al. | Heterogeneous capture–recapture models with covariates: A partial likelihood approach for closed populations | |
| CN111340522A (en) | Resource recommendation method, device, server and storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181221 |