CN109886300A - A kind of user's clustering method, device and equipment - Google Patents
A kind of user's clustering method, device and equipment Download PDFInfo
- Publication number
- CN109886300A CN109886300A CN201910043942.0A CN201910043942A CN109886300A CN 109886300 A CN109886300 A CN 109886300A CN 201910043942 A CN201910043942 A CN 201910043942A CN 109886300 A CN109886300 A CN 109886300A
- Authority
- CN
- China
- Prior art keywords
- user
- clustered
- cluster
- feature
- class cluster
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention provides a kind of user's clustering method, device and equipment, this method comprises: the history pushed information that each user to be clustered clicked is determined, respectively as the characteristic information of each user to be clustered;According to the characteristic information of each user to be clustered, the similarity between every two user to be clustered is calculated;According to the characteristic information of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.User is clustered using each scheme provided in an embodiment of the present invention, can be improved the accuracy of cluster result.
Description
Technical field
The present invention relates to field of computer technology, more particularly to a kind of user's clustering method, device and equipment.
Background technique
With popularizing for mobile terminal, operator often passes through the mobile terminal actively used to user and sends information
Mode, the push of Lai Shixian information.For example, in such a way that the mobile terminal that active is used to user sends advertisement, it is wide to realize
Accuse push.
In process of information push, in order to increase the probability that information is clicked, usually according to the hobby of user come
The possible interested information of user is pushed, and in order to accelerate the efficiency of information push, usually simultaneously to similar interests
The same class user of hobby sends identical information.The prior art is usually the basic letter such as gender, age, occupation for collecting user
Breath, and judge whether the user is a kind of user that there are similar interests to like using the essential information being collected into.
However, inventor has found in the implementation of the present invention, at least there are the following problems for the prior art:
User may provide the essential information of some inaccuracy, cause to collect basic to protect oneself privacy
Information inaccuracy, so that determining that the hobby of user is not accurate enough according to essential information, then determining according to hobby
Same class user in there may be the actual interest of certain user hobby be from the actual interest of other users hobby be different
's.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of user's clustering method, device and equipment, to solve the prior art
The problem of.Specific technical solution is as follows:
The one side that the present invention is implemented provides a kind of user's clustering method, which comprises
For each user to be clustered, the history pushed information that the user to be clustered clicked is determined, and according to determining
History pushed information obtain the feature of the user to be clustered;
According to the feature of user to be clustered, the similarity between every two user to be clustered is calculated;
According to the feature of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.
Optionally, the feature of the basis is calculated similarity and each user to be clustered, treat cluster user into
After the step of row cluster, further includes:
Each existing subscriber's class cluster is extended using following manner:
The user class cluster that cluster obtains is compared with the feature of user in existing subscriber's class cluster, determines what cluster obtained
The number of user in user class cluster with similar features;
From the user class cluster that cluster obtains, the first preset quantity user class cluster is chosen, wherein selected user class
The number of user in cluster with similar features is all larger than the number of the user in the user class cluster that do not choose with similar features;
User in the user class cluster of selection is added in existing subscriber's class cluster.
Optionally, the step of history pushed information according to determined by obtains the feature of the user to be clustered described it
Afterwards, this method further include:
From the feature of user to be clustered obtained, the second preset quantity feature is randomly selected as cluster centre;
The feature of similarity and each user to be clustered that the basis is calculated, treats what cluster user was clustered
Step, comprising:
For each user to be clustered, according to the similarity being calculated, the user to be clustered and any described poly- is determined
Similarity between class center, and judge whether the user to be clustered belongs to the cluster centre and correspond to according to identified similarity
User class cluster;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, which is added to
In the corresponding user class cluster of the cluster centre;
For each user class cluster, according to the feature of the user to be clustered in the user class cluster, the user class cluster is calculated
Average characteristics, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by the user class
The cluster centre of cluster is updated to the average characteristics being calculated;And return it is described be directed to each user to be clustered, according to calculating
The similarity arrived determines the similarity between the user to be clustered and any cluster centre, and according to identified similar
Degree judges whether the user to be clustered belongs to the corresponding user class cluster of the cluster centre;If the user to be clustered belongs in the cluster
The user to be clustered is then added to the step in the corresponding user class cluster of the cluster centre by the corresponding user class cluster of the heart, until
When the average characteristics of each user class cluster are identical as the cluster centre of the user class cluster, the user class cluster clustered at this time is made
For cluster result.
Optionally, the feature according to user to be clustered calculates separately the similarity between every two user to be clustered
The step of, comprising:
The phase between every two user to be clustered is calculated using cosine similarity algorithm according to the feature of user to be clustered
Like degree;Alternatively,
Using following formula calculate separately the similarity factor between every two user to be clustered, and according to being calculated
Similarity factor determines the similarity between described two users to be clustered:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate use to be clustered
The feature vector of family i, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate user i to be clustered feature vector and to
The intersection of the feature vector of cluster user j, | Ui|Uj| indicate the feature vector of user i to be clustered and the feature of user j to be clustered
The union of vector.
The another aspect that the present invention is implemented, additionally provides a kind of user's clustering apparatus, and described device includes:
Determining module, for determining the history pushed information that the user to be clustered clicked for each user to be clustered,
And the feature of the user to be clustered is obtained according to identified history pushed information;
Computing module calculates the similarity between every two user to be clustered for the feature according to user to be clustered;
Cluster module treats cluster user for the feature according to the similarity and each user to be clustered being calculated
It is clustered.
Optionally, described device further include:
Expansion module, for being extended to each existing subscriber's class cluster;
Wherein, the expansion module includes:
The feature of Comparative sub-module, the user class cluster for obtaining cluster and user in existing subscriber's class cluster compare
Compared with determining has the number of the user of similar features in the user class cluster for clustering and obtaining;
Submodule is chosen, for choosing the first preset quantity user class cluster from the user class cluster that cluster obtains,
In, the number of the user with similar features is all larger than in the user class cluster that do not choose with similar in selected user class cluster
The number of the user of feature;
Submodule is extended, is added in existing subscriber's class cluster for the user in the user class cluster by selection.
Optionally, described device further include:
Module is chosen, for randomly selecting the second preset quantity feature from the feature of user to be clustered obtained
As cluster centre;
Correspondingly, the cluster module, comprising:
Submodule is added, for determining the use to be clustered according to the similarity being calculated for each user to be clustered
Similarity between family and any cluster centre, and judge whether the user to be clustered belongs to according to identified similarity
The corresponding user class cluster of the cluster centre;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, this is waited for
Cluster user is added in the corresponding user class cluster of the cluster centre;
Submodule is updated, for the feature according to the user to be clustered in the user class cluster, calculates the flat of the user class cluster
Equal feature, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by the user class cluster
Cluster centre is updated to the average characteristics being calculated, and triggers the addition submodule, until each user class cluster is averaged
When feature is identical as the cluster centre of the user class cluster, using the user class cluster clustered at this time as cluster result.
Optionally, the computing module, is specifically used for,
The phase between every two user to be clustered is calculated using cosine similarity algorithm according to the feature of user to be clustered
Like degree;Alternatively,
Using following formula calculate separately the similarity factor between every two user to be clustered, and according to being calculated
Similarity factor determines the similarity between described two users to be clustered:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate use to be clustered
The feature vector of family i, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate user i to be clustered feature vector and to
The intersection of the feature vector of cluster user j, | Ui|Uj| indicate the feature vector of user i to be clustered and the feature of user j to be clustered
The union of vector.
The another aspect that the present invention is implemented additionally provides a kind of electronic equipment, including processor, communication interface, memory
And communication bus, wherein processor, communication interface, memory complete mutual communication by communication bus;
Memory, for storing computer program;
Processor when for executing the program stored on memory, realizes any of the above-described user's clustering method.
At the another aspect that the present invention is implemented, a kind of computer readable storage medium is additionally provided, it is described computer-readable
Instruction is stored in storage medium, when run on a computer, so that computer executes any of the above-described user and gathers
Class method.
At the another aspect that the present invention is implemented, the embodiment of the invention also provides a kind of, and the computer program comprising instruction is produced
Product, when run on a computer, so that computer executes any of the above-described user's clustering method.
When carrying out user's cluster using scheme provided in an embodiment of the present invention, it can determine that each user to be clustered clicked
History pushed information, respectively as the characteristic information of each user to be clustered;According to the characteristic information of each user to be clustered,
Calculate separately the similarity between every two user to be clustered;According to the spy of the similarity and each user to be clustered that are calculated
Reference breath, treats cluster user and is clustered.Since the history pushed information that user clicked is by the true of user's generation
Click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked can generally also compare
It is more similar, thus user is clustered using scheme provided in an embodiment of the present invention, it can be improved the accuracy of cluster result,
And the user for including in each the user class cluster obtained on this basis for cluster carries out information push, can accelerate
While information pushing efficiency, so that meeting the hobby of user to the information that user pushes.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.
Fig. 1 is a kind of flow diagram of user's clustering method provided in an embodiment of the present invention;
Fig. 2 is a kind of structural schematic diagram of user's clustering apparatus provided in an embodiment of the present invention;
Fig. 3 is the structural schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention is described.
Fig. 1 is a kind of flow diagram of user's clustering method provided in an embodiment of the present invention, this method comprises:
S100 determines the history pushed information that the user to be clustered clicked, and according to institute for each user to be clustered
Determining history pushed information obtains the feature of the user to be clustered.
From time dimension, above-mentioned history pushed information refers to the information pushed.Come from content dimension
It says, above-mentioned history pushed information can be with are as follows: advertising information, short video information and news information etc..
The server for sending pushed information to user can recorde the history pushed information that each user clicked, except this it
Outside, the number etc. that each user clicks each history pushed information can also be recorded, for ease of description, the above- mentioned information recorded
It is properly termed as historical record.It is above-mentioned when the executing subject of the embodiment of the present invention is to send the server of pushed information to user
Server can directly determine the history pushed information that user to be clustered clicked according to the information recorded in historical record;And work as
When the executing subject of the embodiment of the present invention is the other equipment except the server for sending pushed information to user, other equipment are then needed
The history pushed information that user to be clustered clicked is determined from above-mentioned server acquisition historical record.
Each user has different characteristics, and can be described from a variety of different angles when describing a user.
Inventor has found that the pushed information that a user clicked is often related to its hobby during the experiment, it is, tool
The pushed information that the user for having same interest to like clicked is similar, in consideration of it, in the embodiment of the present invention, using user institute point
The pushed information hit characterizes the feature of user.
Based on above content, in a kind of implementation of the invention, the push that user to be clustered clicked can use
The attributes such as the type of information, the keyword for including characterize the feature of user to be clustered.For example, what user to be clustered clicked
The type of pushed information are as follows: shopping, the keyword for including have: nest's milk powder, correspondingly, the feature of the user to be clustered can wrap
It includes: shopping, nest's milk powder.
In another implementation of the invention, the click condition of each history pushed information can also be clicked using user
To characterize the feature of user.Specifically, can indicate the feature of user to be clustered in a manner of vector, user to be clustered is to institute
There is the click condition of each of history pushed information information to constitute the corresponding feature vector of feature of the user to be clustered
In each element, that is, an element representation user to be clustered in feature vector pushes the click of information to a history
Situation.
Specifically, whether above-mentioned click condition can click the feelings of the history pushed information for expression user to be clustered
Condition, for example, the feature vector U of user i to be clusterediAn element UijWhen=1, indicate that user i clicked history pushed information
J, and element UijWhen=0, then it represents that user i did not clicked on history pushed information j.
In addition, above-mentioned click condition may be to indicate user to be clustered to the feelings of the history pushed information number of clicks
Condition, for example, the feature vector U of user i to be clusterediAn element UijWhen=3, indicate that user i clicked history pushed information j
Number be 3 times, the feature vector U of user i to be clusterediAn element UijWhen=0, indicate that user i clicked history push
The number of information j is 0 time.
For convenient for handling in subsequent process the feature of each user to be clustered, the feature of each user to be clustered can
To be described based on identical history pushed information, that is, the feature of its feature is indicated for each user to be clustered
The number for the element for including in vector is identical, therefore, it is possible to using matrix U indicate the feature of each user to be clustered to
Amount.
Wherein, in a kind of situation, every a line of matrix U can correspond to a user to be clustered, and each column corresponding one is gone through
The case where history pushed information, each user to be clustered of element representation in matrix clicks each history pushed information.Such case
Under, every a line in matrix U can be used as the feature vector of a user to be clustered, indicate that a user to be clustered goes through to all
The click condition of history pushed information.
In another case, each column of matrix U can correspond to a user to be clustered, the corresponding history of every a line is pushed away
It delivers letters breath, the case where each user to be clustered of the element representation in matrix clicks each history pushed information.In this case, square
Each column in battle array U can be used as the feature vector of a user to be clustered, indicate that a user to be clustered pushes away all history
It delivers letters the click condition of breath.
S110 calculates the similarity between every two user to be clustered according to the feature of user to be clustered.
In the first implementation, every two can be calculated using cosine similarity algorithm according to the feature of user to be clustered
Similarity between a user to be clustered.
In second of implementation, two can be calculated to poly- using following formula according to the feature of user to be clustered
Similarity between class user:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate use to be clustered
The feature vector of family i, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate user i to be clustered feature vector and to
The intersection of the feature vector of cluster user j, | Ui|Uj| indicate the feature vector of user i to be clustered or the feature of user j to be clustered
The union of vector.Specifically, as the feature vector U of user i to be clusterediFor the historical information of its opposite push of user i to be clustered
Click condition, the feature vector U of user j to be clusteredjFor the click condition of the historical information of its opposite push of user i to be clustered
When, | Ui&Uj| for the number of user i to be clustered and user the j to be clustered identical history pushed information clicked;|Ui|Uj| for
The number for the history pushed information that the number for the history pushed information that cluster user i was clicked or user j to be clustered were clicked.
In the third implementation, in the case where feature is indicated in the form of feature vector, the distance between feature vector
It can be used to indicate that similarity between two feature vectors, i.e. the distance between two feature vectors are smaller, show two features
Similarity between vector is higher.
Specifically, can Euclidean distance between the feature vector by calculating two users to be clustered, using calculating
To Euclidean distance indicate the similarity between two users to be clustered, that is, the Euclidean distance being calculated is smaller, two
Similarity between a user to be clustered is higher.
The similarity factor between two users to be clustered can also be calculated first with formula one, and formula two is recycled to calculate this
The distance between the feature of two users to be clustered:
D (i, j)=1-s (i, j) formula two
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate use to be clustered
The feature vector of family i, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate user i to be clustered feature vector and to
The intersection of the feature vector of cluster user j, | Ui|Uj| indicate the feature vector of user i to be clustered or the feature of user j to be clustered
The union of vector;D (i, j) indicates the distance between user i and user j to be clustered to be clustered.Specifically, as user i to be clustered
Feature vector UiFor the click condition of the historical information of its opposite push of user i to be clustered, the feature vector of user j to be clustered
UjFor the historical information of its opposite push of user i to be clustered click condition when, | Ui&Uj| it is user i to be clustered and use to be clustered
The number for the identical history pushed information that family j was clicked;|Ui|Uj| for of the user i to be clustered history pushed information clicked
The number for the history pushed information that user j several or to be clustered was clicked.
When indicating the similarity between every two user to be clustered with calculated distance, the spy of user to be clustered
The distance between sign is smaller, then it represents that the similarity between the feature of user to be clustered is higher.
S120 treats cluster user and is clustered according to the feature of the similarity and each user to be clustered that are calculated.
Before being clustered, after being clustered according to the class of subscriber quantity setting intentionally got during concrete application
The quantity of obtained class cluster, i.e. the second preset quantity.For example, it is desirable to obtain 3 class of subscribers, then can be set before cluster
Obtained class number of clusters amount is 3 after fixed cluster.Since each class cluster can have a cluster centre in cluster process,
When being clustered, the number of the user class cluster that can according to need determines the number of cluster centre, that is, one poly-
One user class cluster for needing to obtain of class center representative.
Correspondingly, the numerical value of the second preset quantity is bigger, then it represents that and the user class cluster to be obtained after cluster is more, that is,
It is thinner to treat the classification results that cluster user is classified.
In a kind of implementation, it can use hierarchical clustering algorithm and treat cluster user and clustered, it is default to obtain second
Quantity user class cluster, then using the center of obtained user class cluster as the cluster centre of the user class cluster.
In another implementation, the feature that the user to be clustered is obtained according to identified history pushed information it
Afterwards, can also include:
Step A randomly selects the second preset quantity feature as in cluster from the feature of each user to be clustered
The heart.
Correspondingly, above-mentioned S120 may include steps of:
Step B, for each user to be clustered, according to the similarity being calculated, determine the user to be clustered with it is any
Similarity between the cluster centre, and judge whether the user to be clustered belongs in the cluster according to identified similarity
The corresponding user class cluster of the heart;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, by the user to be clustered
It is added in the corresponding user class cluster of the cluster centre.
Randomly selected from the feature of each user to be clustered due to cluster centre, and user to be clustered be characterized in
User clicks what the case where pushed information indicated, and user clicked some pushed information and illustrates user to this push
Information is interested, so, the feature of user to be clustered and the similarity of cluster centre are higher, then show user to be clustered and cluster
Hobby similarity degree between the corresponding user in center is higher, and a cluster centre represents a user class cluster,
It so has found and has also determined that user belonging to user to be clustered with the high cluster centre of the characteristic similarity of user to be clustered
Class cluster.
Step C calculates the use according to the feature of the user to be clustered in the user class cluster for each user class cluster
The average characteristics of family class cluster, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by this
The cluster centre of user class cluster is updated to the average characteristics being calculated;And return step A, until each user class cluster is averaged
When feature is identical as the cluster centre of the user class cluster, using the user class cluster clustered at this time as cluster result.
The average value of the feature for the user to be clustered for including in average characteristics i.e. user class cluster, each user class cluster are equal
Correspondence possesses an average characteristics.
In the case where user to be clustered is characterized in the number determination for the history pushed information clicked according to user, than
It such as, include three users in a user class cluster, the feature of user 1 is 2,3,4;The feature of user 2 is 5,6,5;The spy of user 3
Sign is 5,9,6;When calculating average characteristics, first element of average characteristics is (2+5+5)/3=4;Second element is (3+6+
9)/3=6;Third element is (4+5+6)/3=5.
And in the case where user to be clustered is characterized in that whether clicking history pushed information according to user determines, than
It such as, include three users in a user class cluster, the feature of user 1 is 1,0,1;The feature of user 2 is 1,0,1;The spy of user 3
Sign is 0,0,1;When calculating average characteristics, first element of average characteristics is (1+1+0)/3=0.66;Second element is (0
+ 0+0)/3=0;Third element is (1+1+1)/3=1.
Since above-mentioned average characteristics illustrate the average value of the feature for the user to be clustered for including in user class cluster, so working as
Under the cluster centre of selection and the different situation of the average characteristics being calculated, then show that selected cluster centre is not
The actual center of user class cluster, cluster result at this time may have error, it is then desired to by the cluster of the user class cluster
Center is updated to the average characteristics being calculated, and since each class cluster can have a cluster centre, cluster centre becomes
Change correspondingly user class cluster also just to change, thus, after cluster centre updates, need to re-start cluster, it is poly- to improve
The accuracy of class result.
When the cluster centre of selection is identical as the average characteristics being calculated, show each user for including in user class cluster
It is centered around around cluster centre and is evenly distributed, belong to same type of user;When the cluster centre of selection be calculated
Average characteristics it is not identical when, it may be possible to be not belonging to of a sort user due to existing in user class cluster with other users and generate
Error, it is then desired to need to re-start cluster, after cluster centre updates to improve the accuracy of cluster result.
The quantity of user is ever-increasing in practical application, and in order to accelerate to cluster speed, it can be only to a period of time
The user inside newly increased clusters, then will the user for including in the obtained user class cluster of cluster be added to this time it
In the preceding each existing subscriber's class cluster clustered to user, specifically, in a kind of implementation of the embodiment of the present invention,
It treats after cluster user clustered, each existing subscriber's class cluster can also be expanded using the user class cluster that cluster obtains
Exhibition, specifically includes following steps D-F:
The user class cluster that cluster obtains is compared by step D with the feature of user in existing subscriber's class cluster, determines cluster
The number of user in obtained user class cluster with similar features.
Existing subscriber's class cluster is clustered to obtain namely before this clusters the user newly increased to user
User class cluster.
User with similar features user namely similar or identical to the click condition of history pushed information;
In a kind of situation, the quantity of identical history pushed information is greater than pre- in the history pushed information that two users clicked
When fixed number value, it may be considered that the two users have similar features, for example, predetermined value is 2, user 1, clicked history and pushes away
Deliver letters breath H, I, G, K;User 2, click history pushed information H, I, G, L;The identical history that user 1 and user 2 clicked pushes away
Information is H, I, G, and quantity is 3 and greater than 2, then user 1 and user 2 is determined as the user with similar features;
In another case, the history pushed information that two users clicked is identical, it may be considered that the two are used
Family has similar features, for example, user 1, clicked history pushed information H, I, G;User 2, clicked history pushed information H,
I,G;It is H, I, G that the history that user 1 and user 2 clicked, which pushes away information, then user 1 and user 2 is determined as having similar spy
The user of sign.
Step E chooses the first preset quantity user class cluster from the user class cluster that cluster obtains.
Wherein, the number of the user in selected user class cluster with similar features is all larger than the user class cluster that do not choose
In with similar features user number.
In a kind of implementation, the user class cluster that cluster is obtained, according to the number of the user in cluster with similar features
Descending sort is carried out, preceding first preset quantity user class cluster is chosen;Wherein, the number of the user in cluster with similar features is
It is determined in step D.
In another implementation, optional cluster in the set of the user class cluster obtained from cluster, determining has in the cluster
The number of the user of similar features determines whether the number is to have the number of the user of similar features maximum in all clusters;
If so, determining that the user class cluster is the user class cluster chosen, and the use is deleted from the set for the user class cluster that cluster obtains
Family class cluster, the step of returning again to the optional cluster from the set of user class cluster that cluster obtains, until the user class cluster of selection reaches
To the first preset quantity.
First preset quantity, which can according to need, determines the number of users that existing subscriber's class cluster is extended, due to cluster
The number of users for including in obtained each user class cluster is fixed, then, when needing to be extended to existing subscriber's class cluster
Number of users it is more, correspondingly, then it is also more to need to cluster obtained user class cluster, that is, the first preset quantity is got over
Greatly.
For example, the number comprising the user with same interest hobby in an existing user class cluster is 200, need
When the user's number for including in the user class cluster being expanded to 1000, that is, needing to extend 800 users and existing subscriber
The user for including in user class cluster L, M, N, O, P that there is class cluster the cluster of user's number of same characteristic features from high to low to obtain
Number is respectively as follows: 300,300,200,200,300;So, then it needs using the user for including in user class cluster L, M, N to this
There is user class cluster to be extended.And in a kind of situation, when the user's number for including in the user class cluster is expanded to 1100 by needs
It is a, that is, when needing to extend 900 users, due to the user for including in user class cluster L, M, N lazy weight with by this
There is the number of users for including in user class cluster to extend 900, it is possible to which optional 100 users are added in user class cluster O
In the user class cluster, realizes and the user's number for including in the user class cluster is expanded to 1100.
User in the user class cluster of selection is added in existing subscriber's class cluster by step F.
The case where history pushed information that two users clicked, is closer, then shows that the hobby of the two users is got over
It is close, then when the user's number with similar features for including in the user class cluster and existing subscriber's class cluster that cluster obtains is got over
It is more, that is, in two user class clusters comprising the identical user's number of hobby it is more when, then show that two user class clusters are
It include that a possibility that user is same class user is bigger, therefore, it is possible to utilize the use with existing subscriber's class cluster with same characteristic features
The user for including in user class cluster more than the number of family is extended existing subscriber's class cluster.
In each scheme provided in an embodiment of the present invention, since the history pushed information that user clicked is to be produced by user
Raw true click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked is logical
Often also can be more similar, thus the scheme provided through the embodiment of the present invention clusters user, can be improved cluster result
Accuracy, and carry out information push for the user for including in obtained each the user class cluster of cluster on this basis,
It can be while accelerating pushing efficiency, so that meeting the hobby of user to the information that user pushes.
It referring to fig. 2, is a kind of structural schematic diagram of user's clustering apparatus provided in an embodiment of the present invention, which includes:
Determining module 200 determines the history push letter that the user to be clustered clicked for being directed to each user to be clustered
It ceases, and obtains the feature of the user to be clustered according to identified history pushed information;
Computing module 210 calculates similar between every two user to be clustered for the feature according to user to be clustered
Degree;
Cluster module 220, for the feature according to the similarity and each user to be clustered being calculated, to use to be clustered
Family is clustered.
In a kind of implementation of the embodiment of the present invention, above-mentioned apparatus further include:
Expansion module, for being extended to each existing subscriber's class cluster, wherein the expansion module includes:
The feature of Comparative sub-module, the user class cluster for obtaining cluster and user in existing subscriber's class cluster compare
Compared with determining has the number of the user of similar features in the user class cluster for clustering and obtaining;
Submodule is chosen, for choosing the first preset quantity user class cluster from the user class cluster that cluster obtains,
In, the number of the user with similar features is all larger than in the user class cluster that do not choose with similar in selected user class cluster
The number of the user of feature;
Submodule is extended, is added in existing subscriber's class cluster for the user in the user class cluster by selection.
In a kind of implementation of the embodiment of the present invention, described device further include:
Module is chosen, for randomly selecting the second preset quantity feature from the feature of user to be clustered obtained
As cluster centre;
Correspondingly, the cluster module, comprising:
Submodule is added, for determining the use to be clustered according to the similarity being calculated for each user to be clustered
Similarity between family and any cluster centre, and judge whether the user to be clustered belongs to according to identified similarity
The corresponding user class cluster of the cluster centre;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, this is waited for
Cluster user is added in the corresponding user class cluster of the cluster centre;
Submodule is updated, for the feature according to the user to be clustered in the user class cluster, calculates the flat of the user class cluster
Equal feature, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by the user class cluster
Cluster centre is updated to the average characteristics being calculated, and triggers the addition submodule, until each user class cluster is averaged
When feature is identical as the cluster centre of the user class cluster, using the user class cluster clustered at this time as cluster result.
In a kind of implementation of the embodiment of the present invention, the computing module is specifically used for,
The phase between every two user to be clustered is calculated using cosine similarity algorithm according to the feature of user to be clustered
Like degree;Alternatively,
Using following formula calculate separately the similarity factor between every two user to be clustered, and according to being calculated
Similarity factor determines the similarity between described two users to be clustered:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate use to be clustered
The feature vector of family i, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate user i to be clustered feature vector and to
The intersection of the feature vector of cluster user j, | Ui|Uj| indicate the feature vector of user i to be clustered and the feature of user j to be clustered
The union of vector.
In each scheme provided in an embodiment of the present invention, since the history pushed information that user clicked is to be produced by user
Raw true click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked is logical
Often also can be more similar, thus the scheme provided through the embodiment of the present invention clusters user, can be improved cluster result
Accuracy, and carry out information push for the user for including in obtained each the user class cluster of cluster on this basis,
It can be while accelerating pushing efficiency, so that meeting the hobby of user to the information that user pushes.
The embodiment of the invention also provides a kind of electronic equipment, as shown in figure 3, include processor 001, communication interface 002,
Memory 003 and communication bus 004, wherein processor 001, communication interface 002, memory 003 are complete by communication bus 004
At mutual communication,
Memory 003, for storing computer program;
Processor 001 when for executing the program stored on memory 003, realizes use provided in an embodiment of the present invention
Family clustering method.
Specifically, above-mentioned user's clustering method includes:
For each user to be clustered, the history pushed information that the user to be clustered clicked is determined, and according to determining
History pushed information obtain the feature of the user to be clustered;
According to the feature of user to be clustered, the similarity between every two user to be clustered is calculated;
According to the feature of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.
It should be noted that above-mentioned processor 001, which executes the program stored on memory 003, realizes user's clustering method
Other embodiments, with preceding method embodiment part provide embodiment it is identical, which is not described herein again.
In each scheme provided in an embodiment of the present invention, since the history pushed information that user clicked is to be produced by user
Raw true click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked is logical
Often also can be more similar, thus the scheme provided through the embodiment of the present invention clusters user, can be improved cluster result
Accuracy, and carry out information push for the user for including in obtained each the user class cluster of cluster on this basis,
It can be while accelerating pushing efficiency, so that meeting the hobby of user to the information that user pushes.
The communication bus that above-mentioned electronic equipment is mentioned can be Peripheral Component Interconnect standard (Peripheral Component
Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.;It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
In another embodiment provided by the invention, a kind of computer readable storage medium is additionally provided, which can
It reads to be stored with computer program in storage medium, realizes that the embodiment of the present invention provides when the computer program is executed by processor
User's clustering method.
Specifically, above-mentioned user's clustering method includes:
For each user to be clustered, the history pushed information that the user to be clustered clicked is determined, and according to determining
History pushed information obtain the feature of the user to be clustered;
According to the feature of user to be clustered, the similarity between every two user to be clustered is calculated;
According to the feature of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.
It should be noted that the other embodiments of user's clustering method are realized by above-mentioned computer readable storage medium,
Identical as the embodiment that preceding method embodiment part provides, which is not described herein again.
In each scheme provided in an embodiment of the present invention, since the history pushed information that user clicked is to be produced by user
Raw true click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked is logical
Often also can be more similar, thus the scheme provided through the embodiment of the present invention clusters user, can be improved cluster result
Accuracy, and carry out information push for the user for including in obtained each the user class cluster of cluster on this basis,
It can be while accelerating pushing efficiency, so that meeting the hobby of user to the information that user pushes.
In another embodiment provided by the invention, a kind of computer program product comprising instruction is additionally provided, when it
When running on computers, so that computer executes user's clustering method provided by the above embodiment.
Specifically, above-mentioned user's clustering method includes:
For each user to be clustered, the history pushed information that the user to be clustered clicked is determined, and according to determining
History pushed information obtain the feature of the user to be clustered;
According to the feature of user to be clustered, the similarity between every two user to be clustered is calculated;
According to the feature of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.
It should be noted that the other embodiments of user's clustering method are realized by above-mentioned computer program product, and it is preceding
The embodiment for stating the offer of embodiment of the method part is identical, and which is not described herein again.
In each scheme provided in an embodiment of the present invention, since the history pushed information that user clicked is to be produced by user
Raw true click data, thus reliability is higher, and the history pushed information that the similar user of hobby clicked is logical
Often also can be more similar, thus the scheme provided through the embodiment of the present invention clusters user, can be improved cluster result
Accuracy, and carry out information push for the user for including in obtained each the user class cluster of cluster on this basis,
It can be while accelerating pushing efficiency, so that meeting the hobby of user to the information that user pushes.
In the above-described embodiments, can come wholly or partly by software, hardware, firmware or any combination thereof real
It is existing.When implemented in software, it can entirely or partly realize in the form of a computer program product.The computer program
Product includes one or more computer instructions.When loading on computers and executing the computer program instructions, all or
It partly generates according to process or function described in the embodiment of the present invention.The computer can be general purpose computer, dedicated meter
Calculation machine, computer network or other programmable devices.The computer instruction can store in computer readable storage medium
In, or from a computer readable storage medium to the transmission of another computer readable storage medium, for example, the computer
Instruction can pass through wired (such as coaxial cable, optical fiber, number from a web-site, computer, server or data center
User's line (DSL)) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer, server or
Data center is transmitted.The computer readable storage medium can be any usable medium that computer can access or
It is comprising data storage devices such as one or more usable mediums integrated server, data centers.The usable medium can be with
It is magnetic medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium (such as solid state hard disk
Solid State Disk (SSD)) etc..
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device,
For electronic equipment, computer readable storage medium and computer program product embodiments, since it is substantially similar to method
Embodiment, so being described relatively simple, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (10)
1. a kind of user's clustering method, which is characterized in that the described method includes:
For each user to be clustered, the history pushed information that the user to be clustered clicked is determined, and go through according to identified
History pushed information obtains the feature of the user to be clustered;
According to the feature of user to be clustered, the similarity between every two user to be clustered is calculated;
According to the feature of the similarity and each user to be clustered that are calculated, treats cluster user and clustered.
2. the method as described in claim 1, which is characterized in that the similarity and each use to be clustered that the basis is calculated
The feature at family, after treating the step of cluster user is clustered, further includes:
Each existing subscriber's class cluster is extended using following manner:
The user class cluster that cluster obtains is compared with the feature of user in existing subscriber's class cluster, determines the user that cluster obtains
The number of user in class cluster with similar features;
From the user class cluster that cluster obtains, the first preset quantity user class cluster is chosen, wherein in selected user class cluster
The number of user with similar features is all larger than the number of the user in the user class cluster that do not choose with similar features;
User in the user class cluster of selection is added in existing subscriber's class cluster.
3. method according to claim 1 or 2, which is characterized in that obtained in the history pushed information according to determined by
After the step of feature of the user to be clustered, this method further include:
From the feature of user to be clustered obtained, the second preset quantity feature is randomly selected as cluster centre;
The feature of similarity and each user to be clustered that the basis is calculated, treats the step that cluster user is clustered
Suddenly, comprising:
For each user to be clustered, according to the similarity being calculated, determine in the user to be clustered and any cluster
Similarity between the heart, and judge whether the user to be clustered belongs to the corresponding use of the cluster centre according to identified similarity
Family class cluster;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, which is added to this and is gathered
In the corresponding user class cluster in class center;
The flat of the user class cluster is calculated according to the feature of the user to be clustered in the user class cluster for each user class cluster
Equal feature, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by the user class cluster
Cluster centre is updated to the average characteristics being calculated;And each user to be clustered is directed to described in returning, according to what is be calculated
Similarity determines the similarity between the user to be clustered and any cluster centre, and is sentenced according to identified similarity
Whether the user to be clustered of breaking belongs to the corresponding user class cluster of the cluster centre;If the user to be clustered belongs to the cluster centre pair
The user to be clustered is then added to the step in the corresponding user class cluster of the cluster centre by the user class cluster answered, until each
When the average characteristics of user class cluster are identical as the cluster centre of the user class cluster, using the user class cluster clustered at this time as poly-
Class result.
4. method according to claim 1 or 2, which is characterized in that the feature according to user to be clustered calculates every two
The step of similarity between user to be clustered, comprising:
The similarity between every two user to be clustered is calculated using cosine similarity algorithm according to the feature of user to be clustered;
Alternatively,
The similarity factor between every two user to be clustered is calculated using following formula, and according to the similar system being calculated
Number, determines the similarity between described two users to be clustered:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate user i's to be clustered
Feature vector, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate the feature vector of user i to be clustered and to be clustered
The intersection of the feature vector of user j, | Ui|Uj| indicate the feature vector of user i to be clustered and the feature vector of user j to be clustered
Union.
5. a kind of user's clustering apparatus, which is characterized in that described device includes:
Determining module determines the history pushed information that the user to be clustered clicked, and root for being directed to each user to be clustered
The feature of the user to be clustered is obtained according to identified history pushed information;
Computing module calculates the similarity between every two user to be clustered for the feature according to user to be clustered;
Cluster module treats cluster user progress for the feature according to the similarity and each user to be clustered being calculated
Cluster.
6. device as claimed in claim 5, which is characterized in that described device further include:
Expansion module, for being extended to each existing subscriber's class cluster;
Wherein, the expansion module includes:
Comparative sub-module, the user class cluster for obtaining cluster are compared with the feature of user in existing subscriber's class cluster, really
The number of the user in obtained user class cluster with similar features is clustered calmly;
Submodule is chosen, for choosing the first preset quantity user class cluster, wherein institute from the user class cluster that cluster obtains
The number of the user with similar features, which is all larger than in the user class cluster that do not choose, in the user class cluster of selection has similar features
User number;
Submodule is extended, is added in existing subscriber's class cluster for the user in the user class cluster by selection.
7. such as device described in claim 5 or 6, which is characterized in that described device further include:
Module is chosen, for from the feature of user to be clustered obtained, randomly selecting the second preset quantity feature conduct
Cluster centre;
Correspondingly, the cluster module, comprising:
Add submodule, for be directed to each user to be clustered, according to the similarity being calculated, determine the user to be clustered with
Similarity between any cluster centre, and judge whether the user to be clustered belongs to this and gather according to identified similarity
The corresponding user class cluster in class center;If the user to be clustered belongs to the corresponding user class cluster of the cluster centre, this is to be clustered
User is added in the corresponding user class cluster of the cluster centre;
It updates submodule and calculates the average spy of the user class cluster for the feature according to the user to be clustered in the user class cluster
Sign, in the case where the average characteristics being calculated are different from the cluster centre of the user class cluster, by the cluster of the user class cluster
Center is updated to the average characteristics being calculated, and triggers the addition submodule, until the average characteristics of each user class cluster
When identical as the cluster centre of the user class cluster, using the user class cluster clustered at this time as cluster result.
8. such as device described in claim 5 or 6, which is characterized in that the computing module is specifically used for,
The similarity between every two user to be clustered is calculated using cosine similarity algorithm according to the feature of user to be clustered;
Alternatively,
The similarity factor between every two user to be clustered is calculated separately using following formula, and similar according to what is be calculated
Coefficient determines the similarity between described two users to be clustered:
Wherein, s (i, j) indicates the similarity factor between user i and user j to be clustered to be clustered, UiIndicate user i's to be clustered
Feature vector, UjIndicate the feature vector of user j to be clustered, | Ui&Uj| indicate the feature vector of user i to be clustered and to be clustered
The intersection of the feature vector of user j, | Ui|Uj| indicate the feature vector of user i to be clustered and the feature vector of user j to be clustered
Union.
9. a kind of electronic equipment, which is characterized in that including processor, communication interface, memory and communication bus, wherein processing
Device, communication interface, memory complete mutual communication by communication bus;
Memory, for storing computer program;
Processor when for executing the program stored on memory, realizes any method and step of claim 1-4.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer in the computer readable storage medium
Program realizes claim 1-4 any method and step when the computer program is executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910043942.0A CN109886300A (en) | 2019-01-17 | 2019-01-17 | A kind of user's clustering method, device and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910043942.0A CN109886300A (en) | 2019-01-17 | 2019-01-17 | A kind of user's clustering method, device and equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109886300A true CN109886300A (en) | 2019-06-14 |
Family
ID=66926156
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910043942.0A Pending CN109886300A (en) | 2019-01-17 | 2019-01-17 | A kind of user's clustering method, device and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109886300A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111444933A (en) * | 2019-11-26 | 2020-07-24 | 北京邮电大学 | Object classification method and device |
CN111784412A (en) * | 2020-07-15 | 2020-10-16 | Oppo广东移动通信有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN114880580A (en) * | 2022-06-15 | 2022-08-09 | 北京百度网讯科技有限公司 | Information recommendation method and device, electronic equipment and medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104063801A (en) * | 2014-06-23 | 2014-09-24 | 广州优蜜信息科技有限公司 | Mobile advertisement recommendation method based on cluster |
CN106228188A (en) * | 2016-07-22 | 2016-12-14 | 北京市商汤科技开发有限公司 | Clustering method, device and electronic equipment |
CN108108419A (en) * | 2017-12-15 | 2018-06-01 | 百度在线网络技术(北京)有限公司 | A kind of information recommendation method, device, equipment and medium |
CN108647293A (en) * | 2018-05-07 | 2018-10-12 | 广州虎牙信息科技有限公司 | Video recommendation method, device, storage medium and server |
CN108898432A (en) * | 2018-06-25 | 2018-11-27 | 武汉斗鱼网络科技有限公司 | Advertisement putting effect evaluation method and device and electronic equipment |
-
2019
- 2019-01-17 CN CN201910043942.0A patent/CN109886300A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104063801A (en) * | 2014-06-23 | 2014-09-24 | 广州优蜜信息科技有限公司 | Mobile advertisement recommendation method based on cluster |
CN106228188A (en) * | 2016-07-22 | 2016-12-14 | 北京市商汤科技开发有限公司 | Clustering method, device and electronic equipment |
CN108108419A (en) * | 2017-12-15 | 2018-06-01 | 百度在线网络技术(北京)有限公司 | A kind of information recommendation method, device, equipment and medium |
CN108647293A (en) * | 2018-05-07 | 2018-10-12 | 广州虎牙信息科技有限公司 | Video recommendation method, device, storage medium and server |
CN108898432A (en) * | 2018-06-25 | 2018-11-27 | 武汉斗鱼网络科技有限公司 | Advertisement putting effect evaluation method and device and electronic equipment |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111444933A (en) * | 2019-11-26 | 2020-07-24 | 北京邮电大学 | Object classification method and device |
CN111444933B (en) * | 2019-11-26 | 2023-10-10 | 北京邮电大学 | Object classification method and device |
CN111784412A (en) * | 2020-07-15 | 2020-10-16 | Oppo广东移动通信有限公司 | Information pushing method and device, electronic equipment and storage medium |
CN114880580A (en) * | 2022-06-15 | 2022-08-09 | 北京百度网讯科技有限公司 | Information recommendation method and device, electronic equipment and medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10277480B2 (en) | Method, apparatus, and system for determining a location corresponding to an IP address | |
US9405746B2 (en) | User behavior models based on source domain | |
US10547618B2 (en) | Method and apparatus for setting access privilege, server and storage medium | |
CN107341716A (en) | A kind of method, apparatus and electronic equipment of the identification of malice order | |
US20140280548A1 (en) | Method and system for discovery of user unknown interests | |
EP2438539A1 (en) | Co-selected image classification | |
CN108021708B (en) | Content recommendation method and device and computer readable storage medium | |
CN109886300A (en) | A kind of user's clustering method, device and equipment | |
US20120295633A1 (en) | Using user's social connection and information in web searching | |
CN108366012B (en) | Social relationship establishing method and device and electronic equipment | |
CN109241403A (en) | Item recommendation method, device, machinery equipment and computer readable storage medium | |
CN108345601A (en) | Search result ordering method and device | |
CN110928739B (en) | Process monitoring method and device and computing equipment | |
US10762122B2 (en) | Method and device for assessing quality of multimedia resource | |
US20150302088A1 (en) | Method and System for Providing Personalized Content | |
CN105574030A (en) | Information search method and device | |
US8750629B2 (en) | Method for searching and ranking images clustered based upon similar content | |
CN108154024A (en) | A kind of data retrieval method, device and electronic equipment | |
WO2017095413A1 (en) | Incremental automatic update of ranked neighbor lists based on k-th nearest neighbors | |
US10733244B2 (en) | Data retrieval system | |
CN109885729B (en) | Method, device and system for displaying data | |
US10516684B1 (en) | Recommending and prioritizing computer log anomalies | |
CN114997327A (en) | Target object classification method and device, storage medium and electronic equipment | |
CN113868373A (en) | Word cloud generation method and device, electronic equipment and storage medium | |
CN112015924A (en) | Streaming media caching method and device and computer 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 |
Application publication date: 20190614 |
|
RJ01 | Rejection of invention patent application after publication |