CN109614556A - Access path prediction, information-pushing method and device - Google Patents
Access path prediction, information-pushing method and device Download PDFInfo
- Publication number
- CN109614556A CN109614556A CN201811284144.9A CN201811284144A CN109614556A CN 109614556 A CN109614556 A CN 109614556A CN 201811284144 A CN201811284144 A CN 201811284144A CN 109614556 A CN109614556 A CN 109614556A
- Authority
- CN
- China
- Prior art keywords
- interest
- user
- point
- vertex type
- access
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a kind of access path prediction technique, information-pushing method and devices, wherein the access path prediction technique includes: the current location and user characteristics parameter for obtaining user in target area；According to the current location, the user characteristics parameter and history access number it is predicted that the user accesses the highest point of interest of probability and/or interest vertex type；Using the highest point of interest of the access probability and/or interest vertex type as the starting point of interest and/or interest vertex type of prediction access path.
Description
Technical field
The present invention relates to mobile internet technical fields, and in particular to a kind of access path prediction technique, information push side
Method and device.
Background technique
The very fast development of current development of Mobile Internet technology, the ownership of the smart machines such as smart phone constantly increase, and promote
It is serviced at location based service (LBS, Location-based Services) and life tight association, leading " at one's side "
Using more and more.For example, businessman can send user for personalized advertising information by the location information of analysis user
On mobile phone, so that the precision of advertisement dispensing is improved, and user can then receive the information of discount, preferential of neighbouring businessman quickly
Certificate etc. greatly have stimulated the potential desire for consumer goods of user.
However, more and more attention has been paid to serious to hinder for LBS bring privacy concern while the value of LBS gets the nod
The development of LBS is hindered.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of access path prediction technique, information-pushing method and device, with
Solve the problems, such as that LBS service is unable to effective protection privacy of user in the prior art.
According in a first aspect, the embodiment of the invention provides a kind of access path prediction techniques, comprising: obtain user in mesh
Mark the current location in region and user characteristics parameter；It is accessed according to the current location, the user characteristics parameter and history
Data predict that the user accesses the highest point of interest of probability and/or interest vertex type；By the highest interest of the access probability
The starting point of interest and/or interest vertex type of point and/or interest vertex type as prediction access path.
Optionally, the history accesses the user characteristics parameter and the target area that data include the multiple users acquired
Interior history access path, the history access path include at least one point of interest and/or interest vertex type, described according to institute
State user described in current location, the user characteristics parameter and history access data acquisition access the highest point of interest of probability and/
Or interest vertex type, comprising: corresponding user group, the user group root are matched for the user according to the user characteristics parameter
It is obtained according to the user characteristics parameter of the multiple user；Obtain point of interest in the current location preset distance and/or
Interest vertex type；User in the matched user group of institute is chosen in acquired point of interest and/or interest vertex type accesses probability
Highest point of interest and/or interest vertex type, the access probability of point of interest and/or interest vertex type is according to institute in each user group
The history access path for stating multiple users obtains.
Optionally, the step of obtaining user group according to the user characteristics parameter of the multiple user includes: to calculate each use
Similarity between the user characteristics parameter at family；The multiple user is divided into multiple user groups according to the similarity.
Optionally, the step of obtaining user group according to the user characteristics parameter of the multiple user includes: using non-supervisory
Formula mode of learning clusters to form multiple user groups the user characteristics parameter of each user.
Optionally, described that corresponding user group is matched for the user according to the user characteristics parameter, comprising: to count respectively
Calculate the similarity of the user characteristics parameter of user in the user characteristics parameter and each user group of the user；It chooses similar
Spending the highest user group matching is the corresponding user group of the user.
Optionally, the method also includes: by the matched user group of institute with the starting point of interest and/or point of interest class
The next point of interest and/or interest of the highest point of interest of the type degree of association and/or interest vertex type as the prediction access path
Vertex type；And/or by matched user group it is highest emerging with the point of interest and/or point of interest type association degree of current predictive
The next point of interest and/or interest vertex type of interest point and/or interest vertex type as the prediction access path.
Optionally, when there is the identical point of interest of access probability and/or interest vertex type or when there are degree of association phases
With point of interest and/or interest vertex type when, according to current time select the user relatively may access point of interest and/or emerging
Interesting vertex type.
Optionally, the history accesses the user characteristics parameter and the target area that data include the multiple users acquired
Interior history access path, the history access path include starting point and at least one point of interest and/or interest vertex type, institute
It states and the highest point of interest of user's access probability and/or interest is obtained according to the current location and the user characteristics parameter
Vertex type, comprising: it is described to obtain that the current location and the user characteristics parameter are input to first nerves network model
User accesses the highest point of interest of probability and/or interest vertex type, and the first nerves network model is by the history access number
It is obtained according to training.
Optionally, the method also includes: the starting point of interest and/or interest vertex type and the user characteristics are joined
Number is input to nervus opticus network model to obtain the subsequent point of interest and/or interest vertex type of the prediction access path, institute
Nervus opticus network model is stated to be obtained by history access data training.
Optionally, the prediction access path includes that n point of interest and/or interest vertex type, n meet the following conditions: t1+
t2+…+tn≤min{Tb-Ta,Tc}；Wherein t1, t2 ... tn be respectively in the matched user group of institute user it is emerging at the n
The mean residence time of interest point and/or interest vertex type, Ta indicate current time, and Tb indicates when closing of shop in target area
Between, Tc indicate average total time-consuming of the user in a history access path, min { Tb-Ta, Tc } table in matched user group
Show the smaller value in Tb-Ta and Tc, n is natural number.
Optionally, the method also includes: to the history access data screen.
Optionally, described that history access data are screened, comprising: to be obtained in the history access data respectively
Each user accesses the residence time of each point of interest and/or each interest vertex type；The residence time is removed less than the predetermined time
Point of interest and/or interest vertex type.
Optionally, described that history access data are screened, comprising: it is each emerging to count each user's access respectively
The access probability of interest point and/or each interest vertex type；Removal access probability is lower than the point of interest and/or point of interest of predetermined threshold
Type.
According to second aspect, the embodiment of the invention provides a kind of information-pushing methods, comprising: is receiving working as user
When front position, the method according to any one of above-mentioned first aspect obtains the prediction access path of the user；According to institute
It states prediction access path and pushes corresponding information to the user.
According to the third aspect, the embodiment of the invention provides a kind of access path prediction meanss, comprising: acquiring unit is used
In current location and user characteristics parameter of the acquisition user in target area；Predicting unit, for according to the current location,
User described in the user characteristics parameter and history access data acquisition accesses the highest point of interest of probability and/or point of interest class
Type；Path unit, for using the highest point of interest of the access probability and/or interest vertex type as of prediction access path
Beginning point of interest and/or interest vertex type.
According to fourth aspect, the embodiment of the invention provides a kind of information push-delivery apparatus, comprising: according to the third aspect
Access path prediction meanss, at the current location for receiving user, obtaining the prediction access path of the user；It pushes away
Unit is sent, for pushing corresponding information to the user according to the prediction access path.
According to the 5th aspect, the embodiment of the invention provides a kind of servers, comprising: memory and processor, it is described to deposit
Connection is communicated with each other between reservoir and the processor, computer instruction is stored in the memory, and the processor passes through
The computer instruction is executed, thereby executing method described in any one of first aspect or second aspect.
It is described computer-readable the embodiment of the invention provides a kind of computer readable storage medium according to the 5th aspect
Storage medium is stored with computer instruction, and the computer instruction is for making the computer execute first aspect or second aspect
Any one of described in method.
Detailed description of the invention
The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings, and attached drawing is schematically without that should manage
Solution is carries out any restrictions to the present invention, in the accompanying drawings:
Fig. 1 shows the flow chart of access path prediction technique according to an embodiment of the present invention；
Fig. 2 shows show the flow chart of access path prediction technique according to another embodiment of the present invention；
Fig. 3 shows the similarity according to an embodiment of the present invention according between user characteristics parameter and is grouped to user
Schematic diagram；
Fig. 4 shows the flow chart of access path prediction technique according to another embodiment of the present invention；
Fig. 5 shows the flow chart of information-pushing method according to an embodiment of the present invention；
Fig. 6 shows the schematic diagram of access path prediction meanss according to an embodiment of the present invention；
Fig. 7 shows the schematic diagram of information push-delivery apparatus according to an embodiment of the present invention；
Fig. 8 shows the schematic diagram of server according to an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those skilled in the art are not having
Every other embodiment obtained under the premise of creative work is made, shall fall within the protection scope of the present invention.
Fig. 1 shows access path prediction technique according to an embodiment of the present invention, and this method can enter target in user
When region (such as some commercial circle), data are accessed to predict the possible access path of user according to the history of multiple users, and
The real-time position information for obtaining user always is not needed, access path may include at least one point of interest and/or point of interest class
Type, each point of interest (POI, Point of Interest) can for example indicate a furniture body shop, a furniture body film
Institute, a furniture body restaurant etc., each point of interest all have a type, and an interest vertex type is the shop with the type
Set, such as interest vertex type are cosmetics, then the interest vertex type includes the set in cosmetics shop in commercial circle.For certain
Scene may not need the specific point of interest of concern, and only need to provide the interested interest vertex type of user, by the quotient under the type
Point of interest in circle all shows user, weighs to the certain unrestricted choice of user, can also equally promote the desire for consumer goods of user.
As shown in Figure 1, this method may include steps of:
S11. current location and user characteristics parameter of the user in target area are obtained.
After user enters some target area (such as some commercial circle), if there are no specific idea next step for he
Which point of interest of the access, user can open the corresponding APP in intelligent mobile terminal, and thus APP can pass through intelligent sliding
The locating module of dynamic terminal obtains the current location of user, and sends current location information to service by intelligent mobile terminal
Device, since user is the registration user of the APP, server can also know the user characteristics parameter that user fills in registration.
User characteristics parameter is and the related parameter of the user itself, such as age of user, gender, income, hobby etc..
However the present invention is not limited thereto, if the intelligent mobile terminal for not having positioning function that user is held,
User can also on some service terminal by swiping the card, the modes such as fingerprint recognition, password confirming authenticate, server passes through
The location information of the service terminal can know the current location of user, and can also know the use that user fills in registration
Family characteristic parameter.
S12. according to current location, user characteristics parameter and history access number it is predicted that user access probability is highest emerging
Interest point and/or interest vertex type.
A large amount of history access data are stored in server, record there are the user characteristics of multiple users in history access data
History access path in parameter and the target area, wherein history access path be multiple users for previously having acquired for example certain
At least one point of interest and/or interest vertex type accessed in the target area of commercial circle etc..By taking point of interest as an example, if user is first
Certain movie theatre A in commercial circle is had accessed, certain bookstore B is then had accessed, finally has accessed certain restaurant C, such movie theatre A, bookstore B and meal
Shop C just constitutes the access path of the user, what access path can be ordered into, be also possible to it is unordered, if user
Tend to access these three points of interest according to the sequence of movie theatre A, bookstore B and restaurant C, without in another order, then should
What access path was ordered into, if user does not incline for the sequencing of access movie theatre A, bookstore B and restaurant C significantly
To then the access path is unordered.Similar user usually has similar consumption propensity, and the dimension of user characteristics parameter is got over
More, matching similar users are more accurate, are analyzed by accessing data to a large amount of history, it is most possible can to predict the user
The point of interest and/or interest vertex type of access.
S13. the highest point of interest of probability and/or interest vertex type will be accessed as the starting point of interest of prediction access path
And/or interest vertex type.
In general, user will not only access a point of interest or a kind of interest vertex type after entering some commercial circle,
The user is also possible to access other points of interest or interest vertex type, therefore the highest point of interest of access probability and/or point of interest
Type can be used as the starting point of interest and/or interest vertex type of prediction access path.
By above method step, server only needs to know the initial position of user, so that it may which predicting user may
The point of interest and/or interest vertex type that can be accessed, the actual position without obtaining user always are ensureing privacy of user
Also there is certain commercial value simultaneously.
Fig. 2 shows access path prediction technique according to another embodiment of the present invention, this method may include walking as follows
It is rapid:
S21. obtaining current location and user characteristics parameter, particular content of the user in target area may refer to step
The description of S11.
It S22. is that the user matches corresponding user group according to user characteristics parameter, which accesses data according to history
In the user characteristics parameter of multiple users obtain.
In the present embodiment, the user characteristics that can be accessed multiple users in data according to history using a variety of methods are joined
Number obtains user group.In some optional embodiments of the present embodiment, the user of multiple users in data is accessed according to history
Characteristic parameter obtains the step of user group and may include
S22a the similarity between the user characteristics parameter of each user) is calculated；
S22b multiple users) are divided by multiple user groups according to similarity.
Measure user characteristics parameter between similarity an index can be the space between user characteristics parameter away from
From, for including the two-dimentional user characteristics parameter at age and income, can by history access data in multiple users age
It indicates at one with income to take in the coordinate system for abscissa, age as ordinate, as shown in figure 3, being indicated with circle
Each user's sample, computer can calculate the space length between two o'clock, so as to space length within a predetermined range
Point be divided into one group, indicated in Fig. 3 with dotted line frame, such as take in low young man, the young man that earns a large income, earn a large income
The elderly etc. can then cast out the sample that certain space lengths between any point are more than preset range, in Fig. 3
Point not within the scope of dotted line frame.Only illustrate the present embodiment by taking the user characteristics parameter of two dimensions as an example in Fig. 3, in reality
More accurate user group division is realized in the application of border, it is understood that there may be the user characteristics of dozens or even hundreds of dimension are joined
Number.Thus user can be grouped according to the similarity between user characteristics parameter.
In other optional embodiments of the present embodiment, above-mentioned steps S22 for example passes through the non-supervisory formula of computer
The mode of habit realizes that computer can automatically cluster user's sample, to form multiple user groups.Non-supervisory formula
Habit can be analyzed by the data to great amount of samples, be classified to untagged sample.Non-supervisory formula study can be based on
PDF estimation is realized, such as maximum likelihood method or the Bayes estimation technique, can also be based on similarity between sample come real
It is existing, such as K-Means algorithm etc..
After having obtained multiple user groups, such as the user characteristics parameter and each use of the user can be calculated separately
The similarity of the user characteristics parameter of user in the group of family, choosing the highest user group matching of similarity is the corresponding user of the user
Group.
S23. point of interest and/or interest vertex type in the current location preset distance are obtained.
Server is at the current location for getting user, such as can dissipate outward along each road of current location,
According to the speed of the normal walking measuring and calculating range that for example 5 minutes walkings can reach, the point of interest within the scope of this is got,
And then corresponding interest vertex type can be obtained according to point of interest.As another optional embodiment, server can also be with
The current location of user is the center of circle, does circle as radius using such as 500 meters of length, obtains the point of interest within the scope of this.
S24. user in the matched user group of institute is chosen in acquired point of interest and/or interest vertex type accesses probability
Highest point of interest and/or interest vertex type, the access probability of point of interest and/or interest vertex type is according to more in each user group
The history access path of a user obtains.
Or for including the two-dimentional user characteristics parameter at age and income, such as the user is matched to receipts by server
Enter the user group of low young man, it is believed that similar consumption propensity can be had by belonging to the low young man of income, be simultaneously
Distance is taken into account, can be chosen in point of interest in the range of user can quickly reach from current location with being used in user group
Family accesses the highest point of interest of probability.Since selected point of interest is that user accesses the highest point of interest of probability in same user group
And in the range of can quickly being reached from current location in user, then it can predict that user is likely to access the point of interest.
Illustrate that the history access path according to multiple users obtains point of interest in each user group below with an example
Access probability.It as shown in table 1 below, include three users in some user group, user A, user B and user C, in table respectively
The amount of access for listing each user's Access Interest point 1-4, it is possible thereby to count the amount of access of the point of interest 1 in the user group
It is 24 times, the amount of access of point of interest 2 is 39 times, and the amount of access of point of interest 3 is 22 times, and the amount of access of point of interest 4 is 15 times, relatively
For 100 amount of access in total, the access probability of point of interest 1 is 24%, and the access probability of point of interest 2 is 39%, interest
The access probability of point 3 is 22%, and the access probability of point of interest 4 is 15%.It counts, can be known in the user group in this way
User is interested in which point of interest.
| Point of interest 1 | Point of interest 2 | Point of interest 3 | Point of interest 4 | |
| User A | 15 | 12 | 3 | |
| User B | 15 | 10 | 15 | |
| User C | 9 | 12 | 9 |
Example of 1 user of table to each point of interest access times
The access probability of interest vertex type in each user group is obtained for the history access path according to multiple users, it can
To extract the interest vertex type of each point of interest, each interest vertex type that user in each user group is accessed is counted respectively
Access probability, and then can know that the user in the user group is interested in which type of point of interest.
S25. using the highest point of interest of access probability and/or interest vertex type as the starting point of prediction access path,
Referring specifically to the description of step S13.
By above method step, server is analyzed by accessing data to a large amount of history, is joined according to user characteristics
History is accessed the user in data and is divided into several user groups by number, and the user is matched in corresponding user group, to take
Business device only can need to know the initial position of user, according to user in the matched user group of user institute for different points of interest
And/or the access probability of interest vertex type, the point of interest and/or interest vertex type that prediction user may access, without
The actual position for obtaining user always, also has certain commercial value while ensureing privacy of user.
As some optional embodiments of the present embodiment, as shown in Fig. 2, above-mentioned access path prediction technique can also wrap
It includes:
S26. using in the matched user group of institute with the highest point of interest of the starting point degree of association and/or interest vertex type as
Predict the next point of interest and/or next interest vertex type of access path.
It is that can be defined with the degree of association each there are certain degree of association in access path, between each point of interest
A point of interest appears in the probability of same access path simultaneously.For example, accessing data by analysis of history it can be found that some is used
The user of family group can buy beverage after a meal, can watch movie after having bought beverage, then the restaurant, the liquor store and the film
The degree of association between these three points of interest of institute is relatively high.What the degree of association between each point of interest can be ordered into, it is also possible to
Unordered.There are a kind of possibility, the user of the user group is mostly to follow first to eat, and then buys beverage, subsequently watches movie
Sequence, eat afterwards without first watching movie, or eat after first buying beverage, it is in this case, interviewed according to point of interest
The sequence asked to calculate the degree of association between each point of interest be highly useful, namely the degree of association from restaurant to liquor store,
The degree of association from liquor store to cinema is very high, and otherwise the degree of association from liquor store to restaurant, from cinema to liquor store
The degree of association is then lower.And for other scenes, the degree of association from cosmetics shop 1 to cosmetics shop 2 with from cosmetics shop 2 to change
Significant difference is simultaneously not present in the degree of association in cosmetic shop 1, then can be arranged to the degree of association under this scene unordered, passes through this
Sample setting can be more advantageous to subsequent accurate prediction.
Similarly, it does not need to pay close attention to specific point of interest in some cases, and prefers to concern interest vertex type, herein
In the case of, the degree of association between each interest vertex type can be calculated.Similarly, the degree of association between each interest vertex type can
With what is be ordered into, it is also possible to unordered.
For the scene that the degree of association between each point of interest and/or each interest vertex type is ordered into, closed according to orderly
Connection degree chooses the highest point of interest of the starting point degree of association and/or interest vertex type；For each point of interest and/or each interest
The degree of association between vertex type is unordered scene, chooses the highest point of interest of the starting point degree of association according to the unordered degree of association
And/or interest vertex type.
S27. by matched user group it is highest emerging with the point of interest and/or point of interest type association degree of current predictive
The next point of interest and/or next interest vertex type of interest point and/or point of interest type prediction as prediction access path.
Next point of interest and/or next emerging can be constantly predicted according to current interest point and/or current interest vertex type
Interesting vertex type, to form the prediction access path including n point of interest and/or interest vertex type, wherein n is nature
Number.
In some rush hour section, server needs in time predict the access path of multiple users, in order to save
The computing resource of server and since the energy of user is limited, limited time, predicts that its access path is also impossible to include
More points of interest and/or interest vertex type, that is to say, that n is limited.In certain optional embodiments, n can be arranged
At a fixed value, such as 5.In certain optional embodiments, n can be a value relevant to user characteristics parameter, should
The matched user group of user institute is different, and the value of n is also different, such as the user group of young woman, and n value can be greater than
The user group of elderly men, n can be set equal to or slightly larger than the user in matched user group the history of user visit
The number of point of interest and/or interest vertex type that diameter of asking the way averagely includes.It, can be to n in other optional embodiments
Carry out more accurate determine.In general user for example when watching movie, eating, get a haircut, is corresponding to the residence time of point of interest
It can be greater than in the residence time in clothes shop, determination n value in order to be more accurate, server is often predicting a point of interest
And/or when interest vertex type, at the same obtain user's being averaged in the point of interest and/or interest vertex type in matched user group
Residence time, then judge whether user exceeds beating for commercial circle shop after the point of interest and/or point of interest type of access
The melt time and smaller value of the user in the average total time-consuming in a history access path in matched user group, if
It has exceeded, stopping continues to predict.In this case, n value meets the following conditions:
t1+t2+…+tn≤min{Tb-Ta,Tc}；
Wherein, t1, t2 ... tn be respectively in the matched user group of institute user in n point of interest and/or interest vertex type
Mean residence time, Ta indicates current time, and Tb indicates that the closing time in shop in target area, Tc indicate the matched use of institute
Average total time-consuming of the user in a history access path in the group of family, min { Tb-Ta, Tc } indicate smaller in Tb-Ta and Tc
Value.
In some cases, it is understood that there may be the identical point of interest of two or more access probabilities and/or interest vertex type, and
It is closely located with user current location.It in this case, as an alternative embodiment, can these access are several
The identical point of interest of rate and/or interest vertex type are all predicted as a point of interest and/or interest vertex type in access path.Make
For another optional embodiment, server can be selected according to current time the user relatively may access point of interest and/
Or interest vertex type.For example, there is the identical point of interest of access probability is cinema and restaurant, and current time is 12:00, clothes
The access point of the user of business device prediction before this does not include restaurant, should may determine that user at this time should be to restaurant more
It is interested, so as to predict that the next point of interest of user should be restaurant according to current time.As another optional real
Mode is applied, server is calculating between each point of interest and/or each interest vertex type that user is accessed in each user group
The degree of association when current time can be taken into account, that is, calculate each point of interest that user in each user group is accessed
And/or the degree of association in different time between each interest vertex type, time interval can flexibly be set according to the actual situation
Set, such as can be set to half an hour or a hour time interval, can also only be arranged morning, noon, at night these three
The degree of association between each point of interest and/or each interest vertex type that user is accessed in each user group under different periods.
In addition, user on weekdays, weekend, festivals or holidays consumption propensity may also be different, also can be set under not same date
The degree of association between each point of interest and/or each interest vertex type that user is accessed in each user group, so as to so that
It must predict that access path is more accurate.
By above method step, server is according to each point of interest and/or interest vertex type in the matched user group of institute
Between the degree of association, the subsequent point of interest that may be accessed of prediction user or interest vertex type do not need to obtain user's always
Actual position also has certain commercial value while ensureing privacy of user.
Fig. 4 shows access path prediction technique according to another embodiment of the present invention, and this method may include walking as follows
It is rapid:
S31. current location and user characteristics parameter of the user in target area are obtained, referring specifically to retouching for step S11
It states.
S32. it is several to obtain the user access current location and user characteristics parameter to be input to first nerves network model
The highest point of interest of rate and/or interest vertex type, the first nerves network model are obtained by history access data training, and history is visited
Ask that the user characteristics parameter and the history access path in the target area that data include multiple users of acquisition, history access road
Diameter includes starting point and at least one point of interest and/or interest vertex type.
In the case where accessing data with a large amount of history, first nerves network can be trained by the way of machine learning
Model predicts that the user accesses the highest point of interest of probability and/or interest vertex type.The first nerves network model for example may be used
It is input with the user characteristics parameter and starting point of multiple users to be feedforward neural network, most to start the point of interest of access
And/or the access probability of interest vertex type is output, accesses data using a large amount of history to instruct to the feedforward neural network
Practice to obtain the first nerves network model.Server, can at the current location and user characteristics parameter for receiving user
To be entered into trained first nerves network model, first nerves network can export the interest that the user may access
The access probability of point and/or interest vertex type, so as to obtain the highest point of interest of access probability and/or interest vertex type.
The data volume that history accesses data is bigger, and the first nerves network model trained measures in advance can be more accurate.
S33. the highest point of interest of probability and/or interest vertex type will be accessed as the starting point of interest of prediction access path
And/or interest vertex type, referring specifically to the description of step S13.
By above method step, in the case where accessing data with a large amount of history, server by utilizing trained the
One neural network model, it is only necessary to know the initial position of user, so that it may predict point of interest that user may access and/
Or interest vertex type, the actual position without obtaining user always also have certain while ensureing privacy of user
Commercial value.
As some optional embodiments of the present embodiment, as shown in figure 4, above-mentioned access path prediction technique can also wrap
It includes:
S34. will starting point of interest and/or interest vertex type and user characteristics parameter be input to nervus opticus network model with
The subsequent point of interest and/or interest vertex type of the prediction access path are obtained, the nervus opticus network model is by history access number
It is obtained according to training.
In the case where accessing data with a large amount of history, nervus opticus network can be trained by the way of machine learning
Model predicts the subsequent point of interest and/or interest vertex type of the prediction access path.The nervus opticus network model for example may be used
To be Recognition with Recurrent Neural Network (RNN, Recurrent Neural Network) or shot and long term memory network (LSTM, Long
Short-Term Memory), which is suitable for the event in predicted time sequence.With the user of multiple users
Characteristic parameter and starting point of interest and/or the input of interest vertex type, with subsequent point of interest for the prediction access path and/or emerging
Interesting vertex type output accesses data using a large amount of history to be trained to obtain the nervus opticus network to the neural network
Model.Server can be entered into instruction when receiving starting point of interest and/or interest vertex type and user characteristics parameter
The nervus opticus network model perfected, nervus opticus network can export the point of interest of the subsequent possible access of the user and/or emerging
Interesting vertex type sequence, so as to obtain prediction access path.Similarly, the data volume of history access data is bigger, trains
Nervus opticus network model out measures in advance can be more accurate.
Similarly, predict that the number n of point of interest included by access path and/or interest vertex type can be according to step
Correlation technique in S27 is arranged, so as to save the computing resource of server.
By above method step, in the case where accessing data with a large amount of history, server by utilizing trained the
Two neural network models, the subsequent point of interest that may be accessed of prediction user or interest vertex type, do not need to obtain user always
Actual position, also there is certain commercial value while ensureing privacy of user.
In embodiment described above, server accesses data according to history, it is only necessary to know the initial position of user,
The point of interest and/or interest vertex type that user may access can be predicted.However history access data in there may be
Many noise datas can impact the accuracy of prediction if server also contemplates these noise datas in prediction.
For this purpose, server, which also needs to access data to history, to be screened in some optional embodiments of the present embodiment.
As in a kind of optional embodiment, above-mentioned the step of screening to history access data, may include:
A) each user in history access data is obtained respectively accesses stopping for each point of interest and/or each interest vertex type
Stay the time；
B) the removal residence time is less than the point of interest and/or interest vertex type of predetermined time.
If only each point of interest that simple acquisition user is accessed, it is likely that a large amount of interference data are brought,
It is unfavorable for subsequent access path prediction.The residence time of each point of interest can be accessed according to user to screen point of interest, when
User access some point of interest and/or interest vertex type residence time it is shorter when, it is likely to because user is for the interest
Point and/or interest vertex type are lost interest in, then the point of interest and/or interest vertex type should be accessed road from the history of the user
Screened out in diameter, for example, user access certain cosmetics shop time be no more than 3 minutes, then can largely judge user to this
Cosmetics shop is lost interest in, so as to delete in its history access path from user.
As another optional embodiment, above-mentioned the step of screening to history access data, may include:
C) each user accesses each point of interest in statistics target area respectively and/or the access of each interest vertex type is several
Rate；
D) point of interest and/or interest vertex type that access probability is lower than predetermined threshold are screened out.
Similarly, if collecting history of some user in some commercial circle accesses data, such as it may find that the use
Family has accessed 1,9 Access Interest points 2 of point of interest for 10 times, and only 1 Access Interest point 3, then it can be concluded that the user for
The access probability of point of interest 1 is 50%, and the access probability of point of interest 2 is 45%, and the access probability of point of interest 3 is 5%, due to this
User is lower than some setting value, such as 10% for the access probability of point of interest 3, then may determine that the user for point of interest
3 and lose interest in, so as to be deleted in its history access path from user, for interest vertex type, and it is same
Reason.The point of interest and/or interest vertex type for being lower than predetermined threshold by screening out access probability, can remove to a certain extent
Data are interfered, subsequent access path prediction is conducive to.
Fig. 5 shows information-pushing method according to an embodiment of the present invention, and this method may include steps of:
S41. at the current location for receiving user, the prediction access path of user is obtained.
Such as can be using the access path prediction technique in embodiment shown in Fig. 1-4, according only to obtaining, user is current
Locating point of interest and user characteristics parameter predict at least one point of interest and/or interest vertex type that user may access,
Its particular content may refer to the specific descriptions in embodiment shown in Fig. 1-4.
S42. corresponding information is pushed to user according to the prediction access path.
Server can push information related with the point of interest after predicting the point of interest that user may access
To user, such as store address, trendy merchandise news, information of discount, discount coupon etc., due to the point of interest, originally user can
The point of interest that can be accessed, user can more inspire the consumption of user after receiving these information.The mode of pushed information has
It is a variety of, such as can be used as the notice of APP, can directly be shown in APP in the form of short message or by the information of point of interest
Homepage, the APP homepage that different user is opened is different, for different user personalized customization.
In some optional embodiments of the present embodiment, each point of interest included by access path can will be predicted
Information is successively pushed to user, such as the information of point of interest is directly being shown in the push mode of APP homepage, in homepage
Successively show the information of each point of interest of prediction access path.However, under in such a way that notice or short message push, if
The information of each point of interest is successively continuously pushed to user, user may once will receive many information, and user often only can
Notice the information of the information namely the last one point of interest that push recently, and the point of interest potential range user current location
Farther out, so that user may neglect the information of all push, it is poor to will lead to information dispensing effect in this way.As a kind of optional reality
Mode is applied, first interest point information of prediction is only pushed to user by server, and according to the matched user group of user institute
The information of second point of interest is pushed to use again after postponing a period of time in the mean residence time of the point of interest by interior user
Family is further continued for pushing subsequent interest point information according to this, so that the information pushed recently received by user is all for he
Next the information for the point of interest that will be accessed enhances the effect of information dispensing.
For prediction access path include be each interest vertex type the case where, can as a kind of optional embodiment
The interest point information of the interest vertex type in target area is all pushed to user, such as predict user's current interest
To watch movie, then can APP homepage show commercial circle in all cinemas information, such as address information, fare information etc., by
User selects the cinema to be gone, such as the user of concern price that can select cheap cinema according to the actual situation,
Wish that the user to watch movie as early as possible can select the cinema nearest from oneself, merely desires to see that the user of certain particular film can select to play
The most suitable cinema of play, compared with the mode for only pushing specific point of interest, it is abundant that push belongs to a kind of multiple points of interest
The selection of user equally can also play good information and launch effect.As a kind of optional embodiment, when in target area
When the point of interest of the interest vertex type is more, user is likely difficult to make a policy, such as predict user's current interest is
It has dinner, however usually there will be many restaurants in commercial circle, even the type of point of interest more refines, such as refine to western-style food,
The point of interest of possible the type is still very much.At this time server can according to the user the consumption of user in matched user group
Tendency, screens out a part of point of interest, for example, the available user of server the consumption valence of user in matched user group
Position, so as to screen out the point of interest not within the scope of the consumption price, so as to facilitate user to make a policy.
Correspondingly, as shown in fig. 6, may include: the embodiment of the invention also provides a kind of access path prediction meanss
Acquiring unit 51, for obtaining current location and user characteristics parameter of the user in target area；
Predicting unit 52, for accessing data acquisition user access according to current location, user characteristics parameter and history
The highest point of interest of probability and/or interest vertex type；
Path unit 53, for being accessed using the highest point of interest of the access probability and/or interest vertex type as prediction
The starting point of interest and/or interest vertex type in path.
The detail of above-mentioned acquiring unit 51, predicting unit 52 and path unit 53 can correspond to refering to fig. 1 to Fig. 4
Shown in embodiment corresponding associated description and effect understood that details are not described herein again.
Correspondingly, as shown in fig. 7, may include: the embodiment of the invention also provides a kind of information push-delivery apparatus
Access path prediction meanss 61, the prediction at the current location for receiving user, obtaining the user access
Path, the access path prediction meanss 61 for example can be the access path prediction meanss in embodiment shown in fig. 6；
Push unit 62, for pushing corresponding information to the user according to prediction access path.
Above-mentioned access path prediction meanss 61 and the detail of push unit 62 can be corresponded to refering to implementation shown in fig. 5
Corresponding associated description and effect are understood that details are not described herein again in example.
As shown in figure 8, the server may include processor 71 and deposit the embodiment of the invention also provides a kind of server
Reservoir 72, wherein processor 71 can be connected with memory 72 by bus or other modes, to be connected by bus in Fig. 8
For.
Processor 71 can be central processing unit (Central Processing Unit, CPU).Processor 71 can be with
For other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), field programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
The combination of the chips such as discrete hardware components or above-mentioned all kinds of chips.
Memory 72 is used as a kind of non-transient computer readable storage medium, can be used for storing non-transient software program, non-
Transient computer executable program and module, the program as corresponding to the method for the starting application program in the embodiment of the present invention
Instruction.The non-transient software instruction that processor 71 is stored in memory 72 by operation, thereby executing the various function of processor
It can apply and data processing, i.e., the access path prediction technique or information-pushing method in realization above method embodiment.
Memory 72 may include high-speed random access memory, can also include non-transient memory, for example, at least one
A disk memory, flush memory device or other non-transient solid-state memories.In some embodiments, memory 72 is optional
Including the memory remotely located relative to processor 71, these remote memories can pass through network connection to processor 71.
The example of above-mentioned network includes but is not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
The detail of above-mentioned server can correspond to corresponding associated description in embodiment referring to FIG. 1 to 5
Understood with effect, details are not described herein again.
It is that can lead to it will be understood by those skilled in the art that realizing all or part of the process in above-described embodiment method
Computer program is crossed to instruct relevant hardware and complete, the program can be stored in a computer-readable storage medium
In, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, the storage medium can for magnetic disk,
CD, read-only memory (Read-Only Memory, ROM), random access memory (Random Access
Memory, RAM), flash memory (Flash Memory), hard disk (Hard Disk Drive, abbreviation: HDD) or solid state hard disk
(Solid-State Drive, SSD) etc.；The storage medium can also include the combination of the memory of mentioned kind.
Although being described in conjunction with the accompanying the embodiment of the present invention, those skilled in the art can not depart from the present invention
Spirit and scope in the case where various modifications and variations can be made, such modifications and variations are each fallen within by appended claims institute
Within the scope of restriction.
Claims (18)
1. a kind of access path prediction technique characterized by comprising
Obtain current location and user characteristics parameter of the user in target area；
According to the current location, the user characteristics parameter and history access number it is predicted that user access probability is highest
Point of interest and/or interest vertex type；
Using the highest point of interest of the access probability and/or interest vertex type as the starting point of interest of prediction access path with/
Or interest vertex type.
2. the method according to claim 1, wherein history access data include multiple users of acquisition
History access path in user characteristics parameter and the target area, the history access path include at least one point of interest
And/or interest vertex type, it is described to be used according to the current location, the user characteristics parameter and history access data acquisition
Family accesses the highest point of interest of probability and/or interest vertex type, comprising:
It is that the user matches corresponding user group according to the user characteristics parameter, the user group is according to the multiple user
User characteristics parameter obtain；
Obtain point of interest and/or interest vertex type in the current location preset distance；
It is highest emerging that user's access probability in the matched user group of institute is chosen in acquired point of interest and/or interest vertex type
Interest is put and/or interest vertex type, and the access probability of point of interest and/or interest vertex type is according to the multiple use in each user group
The history access path at family obtains.
3. according to the method described in claim 2, it is characterized in that, the user characteristics parameter according to the multiple user is used
The step of family group includes:
Calculate the similarity between the user characteristics parameter of each user；
The multiple user is divided into multiple user groups according to the similarity.
4. according to the method described in claim 2, it is characterized in that, the user characteristics parameter according to the multiple user is used
The step of family group includes:
It is clustered using user characteristics parameter of the non-supervisory formula mode of learning to each user to form multiple user groups.
5. according to the method described in claim 2, it is characterized in that, it is described according to the user characteristics parameter be the user
With corresponding user group, comprising:
Calculate separately the user characteristics parameter of the user in each user group the user characteristics parameter of user it is similar
Degree；
Choosing the highest user group matching of similarity is the corresponding user group of the user.
6. according to the method described in claim 2, it is characterized by further comprising:
By in the matched user group of institute with the starting point of interest and/or the highest point of interest of point of interest type association degree and/or
Next point of interest and/or interest vertex type of the interest vertex type as the prediction access path；And/or
By in matched user group with the point of interest and/or the highest point of interest of point of interest type association degree of current predictive and/
Or next point of interest and/or interest vertex type of the interest vertex type as the prediction access path.
7. the method according to any one of claim 2-6, which is characterized in that access the identical point of interest of probability when existing
And/or when interest vertex type or when there are the identical point of interest of the degree of association and/or interest vertex type,
The point of interest and/or interest vertex type for selecting the user that may relatively access according to current time.
8. the method according to claim 1, wherein history access data include multiple users of acquisition
History access path in user characteristics parameter and the target area, the history access path include starting point and at least one
A point of interest and/or interest vertex type, it is described that user's visit is obtained according to the current location and the user characteristics parameter
Ask the highest point of interest of probability and/or interest vertex type, comprising:
The current location and the user characteristics parameter are input to first nerves network model to obtain user's access
The highest point of interest of probability and/or interest vertex type, the first nerves network model are trained by history access data
It arrives.
9. according to the method described in claim 8, it is characterized by further comprising:
By the starting point of interest and/or interest vertex type and the user characteristics parameter be input to nervus opticus network model with
The subsequent point of interest and/or interest vertex type of the prediction access path are obtained, the nervus opticus network model is gone through by described
History access data training obtains.
10. method according to claim 6 or 9, which is characterized in that the prediction access path include n point of interest and/
Or interest vertex type, n meet the following conditions
t1+t2+…+tn≤min{Tb-Ta,Tc}
Wherein t1, t2 ... tn be respectively in the matched user group of institute user in the n point of interest and/or interest vertex type
Mean residence time, Ta indicates current time, and Tb indicates that the closing time in shop in target area, Tc indicate the matched use of institute
Average total time-consuming of the user in a history access path in the group of family, min { Tb-Ta, Tc } indicate smaller in Tb-Ta and Tc
Value, n is natural number.
11. method according to claim 1 to 10, which is characterized in that further include:
History access data are screened.
12. according to the method for claim 11, which is characterized in that it is described that history access data are screened, it wraps
It includes:
The stop that each user in the history access data accesses each point of interest and/or each interest vertex type is obtained respectively
Time；
Remove point of interest and/or interest vertex type of the residence time less than the predetermined time.
13. according to the method for claim 11, which is characterized in that it is described that history access data are screened, it wraps
It includes:
The access probability that each user accesses each point of interest and/or each interest vertex type is counted respectively；
Removal access probability is lower than the point of interest and/or interest vertex type of predetermined threshold.
14. a kind of information-pushing method characterized by comprising
At the current location for receiving user, method according to claim 1 to 13 obtains the user's
Predict access path；
Corresponding information is pushed to the user according to the prediction access path.
15. a kind of access path prediction meanss characterized by comprising
Acquiring unit, for obtaining current location and user characteristics parameter of the user in target area；
Predicting unit, for the user according to the current location, the user characteristics parameter and history access data acquisition
Access the highest point of interest of probability and/or interest vertex type；
Path unit, for using the highest point of interest of the access probability and/or interest vertex type as prediction access path
Originate point of interest and/or interest vertex type.
16. a kind of information push-delivery apparatus characterized by comprising
Access path prediction meanss according to claim 15, for obtaining institute at the current location for receiving user
State the prediction access path of user；
Push unit, for pushing corresponding information to the user according to the prediction access path.
17. a kind of server characterized by comprising memory and processor, between the memory and the processor mutually
Connection is communicated, computer instruction is stored in the memory, the processor, which passes through, executes the computer instruction, thus
Perform claim requires method described in any one of 1-14.
18. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer to refer to
It enables, the computer instruction is used to that the computer perform claim to be made to require method described in any one of 1-14.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811284144.9A CN109614556B (en) | 2018-10-31 | 2018-10-31 | Access path prediction and information push method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811284144.9A CN109614556B (en) | 2018-10-31 | 2018-10-31 | Access path prediction and information push method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109614556A true CN109614556A (en) | 2019-04-12 |
| CN109614556B CN109614556B (en) | 2020-03-13 |
Family
ID=66002097
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811284144.9A Active CN109614556B (en) | 2018-10-31 | 2018-10-31 | Access path prediction and information push method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109614556B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110377825A (en) * | 2019-07-15 | 2019-10-25 | 腾讯科技（深圳）有限公司 | Content providing, device, equipment and storage medium |
| CN110782284A (en) * | 2019-10-24 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Information pushing method and device and readable storage medium |
| CN110781415A (en) * | 2019-09-16 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Interest point recommendation method, device, equipment and medium |
| CN112231594A (en) * | 2019-06-27 | 2021-01-15 | 北京百度网讯科技有限公司 | Information processing method and device |
| CN112667726A (en) * | 2021-01-04 | 2021-04-16 | 拉卡拉支付股份有限公司 | Data extraction method, data extraction device, electronic equipment, storage medium and program product |
| CN112883725A (en) * | 2020-12-29 | 2021-06-01 | 上海讯飞瑞元信息技术有限公司 | File generation method and device, electronic equipment and storage medium |
| CN112902964A (en) * | 2021-01-22 | 2021-06-04 | 南京邮电大学 | Path recommendation method and device based on interest points |
| CN112989207A (en) * | 2021-04-27 | 2021-06-18 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060129312A1 (en) * | 2004-12-15 | 2006-06-15 | Institute For Information Industry | Guide route generation methods and systems |
| CN107092629A (en) * | 2017-01-18 | 2017-08-25 | 北京小度信息科技有限公司 | Recommend method and device |
| CN107491474A (en) * | 2017-06-26 | 2017-12-19 | 北京三快在线科技有限公司 | Information recommendation method and device |
| CN107798557A (en) * | 2017-09-30 | 2018-03-13 | 平安科技（深圳）有限公司 | Electronic installation, the service location based on LBS data recommend method and storage medium |
-
2018
- 2018-10-31 CN CN201811284144.9A patent/CN109614556B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060129312A1 (en) * | 2004-12-15 | 2006-06-15 | Institute For Information Industry | Guide route generation methods and systems |
| CN107092629A (en) * | 2017-01-18 | 2017-08-25 | 北京小度信息科技有限公司 | Recommend method and device |
| CN107491474A (en) * | 2017-06-26 | 2017-12-19 | 北京三快在线科技有限公司 | Information recommendation method and device |
| CN107798557A (en) * | 2017-09-30 | 2018-03-13 | 平安科技（深圳）有限公司 | Electronic installation, the service location based on LBS data recommend method and storage medium |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112231594A (en) * | 2019-06-27 | 2021-01-15 | 北京百度网讯科技有限公司 | Information processing method and device |
| CN112231594B (en) * | 2019-06-27 | 2024-04-12 | 北京百度网讯科技有限公司 | Information processing method and device |
| CN110377825A (en) * | 2019-07-15 | 2019-10-25 | 腾讯科技（深圳）有限公司 | Content providing, device, equipment and storage medium |
| CN110781415A (en) * | 2019-09-16 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Interest point recommendation method, device, equipment and medium |
| CN110782284A (en) * | 2019-10-24 | 2020-02-11 | 腾讯科技（深圳）有限公司 | Information pushing method and device and readable storage medium |
| CN112883725A (en) * | 2020-12-29 | 2021-06-01 | 上海讯飞瑞元信息技术有限公司 | File generation method and device, electronic equipment and storage medium |
| CN112667726A (en) * | 2021-01-04 | 2021-04-16 | 拉卡拉支付股份有限公司 | Data extraction method, data extraction device, electronic equipment, storage medium and program product |
| CN112902964A (en) * | 2021-01-22 | 2021-06-04 | 南京邮电大学 | Path recommendation method and device based on interest points |
| CN112902964B (en) * | 2021-01-22 | 2024-02-23 | 南京邮电大学 | Path recommendation method and device based on interest points |
| CN112989207A (en) * | 2021-04-27 | 2021-06-18 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109614556B (en) | 2020-03-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109614556A (en) | Access path prediction, information-pushing method and device | |
| US9183497B2 (en) | Performance-efficient system for predicting user activities based on time-related features | |
| CN106570722B (en) | Intelligent recommendation system and intelligent recommendation method | |
| WO2019174395A1 (en) | Method and apparatus for information recommendation, and device | |
| CN102591911B (en) | The real time individual of position related entities is recommended | |
| CN108416616A (en) | The sort method and device of complaints and denunciation classification | |
| CN109741146A (en) | Products Show method, apparatus, equipment and storage medium based on user behavior | |
| CN106504099A (en) | A kind of system for building user's portrait | |
| CN104166700A (en) | Search term recommendation method and device | |
| US20200134627A1 (en) | System, Method, and Computer Program Product to Determine Cuisine Type Classifiers | |
| JP6937776B2 (en) | Methods and devices for identifying geographic location attribute information | |
| CN108596695B (en) | Entity pushing method and system | |
| US10592917B2 (en) | Method and systems for determining programmatically expected performances | |
| CN109213802B (en) | User portrait construction method and device, terminal and computer readable storage medium | |
| CN110019699A (en) | Pass through the classification of grammer slot between domain | |
| CN110210882A (en) | Promote position matching process and device, promotion message methods of exhibiting and device | |
| US10178498B2 (en) | Method and device for signal processing | |
| CN110880127B (en) | Consumption level prediction method and device, electronic equipment and storage medium | |
| CN112913210B (en) | Method for determining information content and related product | |
| JP6767952B2 (en) | Estimator, estimation method and estimation program | |
| CN110942338A (en) | Marketing enabling strategy recommendation method and device and electronic equipment | |
| CN109903086A (en) | A kind of similar crowd's extended method, device and electronic equipment | |
| CN111144874A (en) | Payment mode recommendation method, device and equipment | |
| CN111080339A (en) | Method and device for generating category preference data based on scene | |
| Sun et al. | Collaborative nowcasting for contextual recommendation |
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 |