CN109933731A - A kind of friend recommendation method, apparatus, equipment and storage medium - Google Patents
A kind of friend recommendation method, apparatus, equipment and storage medium Download PDFInfo
- Publication number
- CN109933731A CN109933731A CN201910204096.6A CN201910204096A CN109933731A CN 109933731 A CN109933731 A CN 109933731A CN 201910204096 A CN201910204096 A CN 201910204096A CN 109933731 A CN109933731 A CN 109933731A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- label
- friend
- target
- 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 discloses a kind of friend recommendation method, apparatus, equipment and storage mediums, comprising: updates rule according to preset data, obtains the user data of at least one user of update；According to the user data of update and the label model tree of training generation in advance, user tag information corresponding with each user is determined；Wherein, label model tree is distributed frame model；According to user tag information and default recommendation rules, the friend information recommended to each user is determined.The embodiment of the present invention can be improved the accuracy and timeliness of recommended friend information, while reduce the requirement to hardware is executed.
Description
Technical field
The present embodiments relate to data recommendation technology more particularly to a kind of friend recommendation method, apparatus, equipment and storage
Medium.
Background technique
With the continuous development of information technology, user oriented application type is increasing, and is in the application user's recommendation
Other users have become and promote to interact between user, and then realize the important means of application function.
In the prior art, collaborative filtering is one of the important algorithm for realizing information recommendation, based on the algorithm and
Friend recommendation method after improving the algorithm is widely used, but the implementation procedure of these methods is due to lacking data
Dynamic processing and study, tend not in time be the reasonable information of user's recommendation, and it is more to work as user data, the calculating of algorithm
When larger, also there is higher requirement to the hardware for executing this method.
Summary of the invention
The embodiment of the present invention provides a kind of friend recommendation method, apparatus, equipment and storage medium, with improve recommended it is good
The accuracy and timeliness of friendly information, while reducing the requirement to hardware is executed.
In a first aspect, the embodiment of the invention provides a kind of friend recommendation methods, comprising:
Rule is updated according to preset data, obtains the user data of at least one user of update；
According to the user data of update and the label model tree of training generation in advance, determining and each user
Corresponding user tag information；Wherein, the label model tree is distributed frame model；
According to the user tag information and default recommendation rules, the good friend's letter recommended to each user is determined
Breath.
Second aspect, the embodiment of the invention also provides a kind of friend recommendation devices, comprising:
User data obtains module, for updating rule according to preset data, obtains the use of at least one user of update
User data；
Label information determining module, the label model for the user data and training generation in advance according to update
Tree, determining user tag information corresponding with each user；Wherein, the label model tree is distributed frame model；
Friend information determining module, for determining to each according to the user tag information and default recommendation rules
The friend information that the user recommends.
The third aspect, the embodiment of the invention also provides a kind of equipment, comprising:
One or more processors；
Storage device, for storing one or more programs；
When one or more of programs are executed by one or more of processors, so that one or more of processing
Device realizes the friend recommendation method that any embodiment of that present invention provides.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer
Program realizes the friend recommendation method that any embodiment of that present invention provides when the program is executed by processor.
The present invention obtains the user data of continuous renewal, and according to the use of update by updating rule according to preset data
User data and the distributed tags model tree that training generates in advance, determine user tag information corresponding with each user,
In conjunction with user tag information and default recommendation rules, the friend information recommended to each user is determined, i.e. the present invention realizes pair
The dynamic of user data is reported and is handled, and the pressure of data load and operation is alleviated by establishing distributed frame model.
It solves the dynamic processing and learning process for lacking user data in the prior art, and has to the hardware for executing the prior art
Higher the problem of requiring, realizes the accuracy and timeliness for improving recommended friend information, while reducing hard to executing
The effect of the requirement of part.
Detailed description of the invention
Fig. 1 is a kind of flow chart for friend recommendation method that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow chart of friend recommendation method provided by Embodiment 2 of the present invention；
Fig. 3 is a kind of flow chart for friend recommendation method that the embodiment of the present invention three provides；
Fig. 4 is a kind of structural schematic diagram for friend recommendation device that the embodiment of the present invention four provides；
Fig. 5 is a kind of structural schematic diagram for equipment that the embodiment of the present invention five provides.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 be the embodiment of the present invention one provide a kind of friend recommendation method flow chart, the present embodiment be applicable to
The case where user's commending friends information in platform, typically, the platform is user oriented application program, and this method can be with
It is executed by friend recommendation device, which can be by hardware and/or software sharing, and can integrate in various general purpose computers
In equipment, typically, the general purpose computing device is Platform Server.The present embodiment specifically comprises the following steps:
Step 110 updates rule according to preset data, obtains the user data of at least one user of update.
Wherein, it is to carry out the rule of dynamic update to user data based on actual conditions that preset data, which updates rule, described
Actual conditions are arranged by backstage according to the specific object of user data and corresponding application program, and the actual conditions can be base
In the condition of time, for example, being updated every 1 hour to user data, it is also possible to the condition based on data volume, for example,
User data is every to increase by 10,000 trigger datas updates, and certainly, the actual conditions are not limited to above two.In the present embodiment
In, user data includes at least the data in this platform, when this platform gets the authorization of other third-party platforms, number of users
According to that can also include third-party platform data, to this and without limitation.
In the present embodiment, rule is updated according to preset data, triggers the continually changing user of Platform Server dynamic acquisition
Data provide timely and effectively data source for subsequent determining user tag information, avoid existing friend recommendation method
In, lack the dynamical linkage between user data processing and user data acquisition, causes data source that cannot reflect user's current state
The problem of.
Step 120, the label model tree generated according to the user data and preparatory training of update, determining and each user
Corresponding user tag information；Wherein, label model tree is distributed frame model.
Wherein, label model tree is that training generates in advance based on user data, the multistage model with distributed frame
System, for label model tree for determining user tag information corresponding with each user, user tag information is according to number of users
According to determining user property value, for example, certain user often pushes shopping information, then " shopping enthusiasts " can be added to the user
Label.
In the present embodiment, firstly, label model tree can have multistage mould from the point of view of the lateral setting of label model tree
Type, junior's model are the micronization processes submodel of upper level model (for example, upper level model determines the label of user are as follows: movement hobby
Person, the confirmable user tag of junior's model are as follows: swimming fan, running fan or rock-climber etc.), although often
The amount of user data of grade model treatment is identical, but since data are calculated as successively deep form, reduce every grade of model and exist
Calculation amount when processes user data；Secondly, from label model tree it is longitudinally disposed from the point of view of, label model tree can have multiple
Independent labeling, i.e. there may be multiple independent categories of model (for example, upper level model can be from fortune in every grade of model
Three aspects of dynamic, diet and culture are that user determines that label, corresponding junior's model can be from the specific of movement, diet and culture
Form refines the label of user), in this way, for relatively independent model at the same level, the user of each independent model processing
Data are reduced；Finally, no matter the distributed frame of label model tree, ensure that from lateral or longitudinal direction, be all easily achieved to mould
The extended operation of type tree increases new labeling, it is seen then that the label model tree of distributed frame in the present embodiment, it can be with
Solve bottleneck problem when the centralized model of tradition determines user tag, improve reliability, the availability of label model tree with
And scalability.
Step 130, according to user tag information and default recommendation rules, determine that the good friend recommended to each user believes
Breath.
Wherein, default recommendation rules, for handling the user tag information of acquisition, are obtained by being arranged from the background
The rule of the final friend information recommended to each user, for example, default recommendation rules can be with are as follows: will have same subscriber mark
The user of label information is recommended as mutually good friend.
The technical solution of the present embodiment, it is regular by being updated according to preset data, the user data of continuous renewal is obtained, and
According to the user data of update and the distributed tags model tree of training generation in advance, user corresponding with each user is determined
Label information is combining user tag information and default recommendation rules, is determining the friend information recommended to each user, i.e. this reality
The technical solution for applying example, which is realized, to be reported and handles to the dynamic of user data, and is alleviated by establishing distributed frame model
The pressure of data load and operation.Solve the dynamic processing and learning process and right for lacking user data in the prior art
The hardware for executing the prior art has higher the problem of requiring, and realizes the accuracy for improving recommended friend information and in time
Property, while reducing the effect to the requirement for executing hardware.
Embodiment two
Fig. 2 is a kind of flow chart of friend recommendation method provided by Embodiment 2 of the present invention, the present embodiment can with it is above-mentioned
Each optinal plan combines in one or more embodiment, provides and is updating rule according to preset data, obtains update
Specific implementation step before user data.Below with reference to Fig. 2 to a kind of friend recommendation method provided by Embodiment 2 of the present invention
It is illustrated, comprising the following steps:
Step 210, usage history user data, are trained master pattern tree, obtain label model tree；
Wherein, label model tree includes: parent label model and at least one level subclass label model；By parent label model
Determining user tag information includes the user tag information that corresponding subclass label model determines.
In the present embodiment, it before determining user tag by master pattern tree, needs to instruct using historical use data
Practice label model tree, the method for training label model tree are as follows: the master pattern tree with distributed frame is established, to historical user
Data are pre-processed, the user tag information that mark historical use data is embodied, by above-mentioned pretreated historical user
Training data of the data as master pattern tree determines final label model tree by continuous correction model parameter, wherein
The learning method of training label model tree is including but not limited to support vector machines (Support Vector Machine, SVM), volume
Product neural network (Convolutional Neural Networks, CNN) and deep neural network (Deep Neural
Networks, DNN).
In label model tree, including parent label model and at least one level subclass label model, the parent label mould
Type is level-one label model, for carrying out preliminary classification, acquiring way or feature based on user data to user data
Information obtains corresponding level-one label information, and the subclass label model is the label model of second level or more, for being based on father
Class label model further refines user tag, the corresponding second level for generating user and the above label information.
Step 220 updates rule according to preset data, obtains the user data of at least one user of update.
Step 230, the label model tree generated according to the user data and preparatory training of update, determining and each user
Corresponding user tag information；Wherein, label model tree is distributed frame model.
Step 240, according to user tag information and default recommendation rules, determine that the good friend recommended to each user believes
Breath.
This implementation does not explain in detail place, please be detailed in previous embodiment, details are not described herein.
The technical solution of the present embodiment, by obtain update user data before, usage history user data, to mark
Quasi- model tree is trained, and obtains the models at different levels in label model tree, so that the user data constantly updated in later use
When labelling to user, user data can be classified operation, and treatment process deeply, avoids disposable load mass data layer by layer
Caused by operation response not in time, substantially increase operation efficiency, reduce the requirement to arithmetic hardware.
Optionally, user tag information includes: user tag and weighted value corresponding with user tag.
Wherein, user tag is the specific tag expression formula obtained, for example, cuisines intelligent, body-building intelligent, travel enthusiasts
With literature and art youth etc..Weighted value corresponding with user tag is determined according at least one setting standard, with each user's
The corresponding weighted value of each user tag, the setting standard can be obtained according to data rule correspondence standard (for example,
In all customer data of one user, how much weight is determined according to the corresponding data volume of different user label, data volume is more,
Weight is bigger), the standard that is also possible to artificially to be arranged (for example, according to the function Promotion Strategy in current platform, will with wait promote
The corresponding weight proportion of user tag of function association be turned up, to meet platform requirement), can also be data rule standard with
The combining form of artificial setting standard, determines the setting standard in terms of user and platform two.
This optional technical solution, provides the particular content of user tag information, i.e., user tag and marks with user
Corresponding weighted value is signed, increases data foundation for subsequent calculating commending friends information, finally calculated recommend can be made
Friendly information is more accurate.
Embodiment three
Fig. 3 be the embodiment of the present invention three provide a kind of friend recommendation method flow chart, the present embodiment can with it is above-mentioned
Each optinal plan combines in one or more embodiment, provides according to user tag information and default recommendation rules,
Determine the specific implementation step for the friend information recommended to each user.The embodiment of the present invention three is provided below with reference to Fig. 3
A kind of friend recommendation method is illustrated, further comprising the steps of:
Step 310, usage history user data, are trained master pattern tree, obtain label model tree；
Wherein, label model tree includes: parent label model and at least one level subclass label model；By parent label model
Determining user tag information includes the user tag information that corresponding subclass label model determines.
Step 320 updates rule according to preset data, obtains the user data of at least one user of update.
Step 330, the label model tree generated according to the user data and preparatory training of update, determining and each user
Corresponding user tag and weighted value corresponding with user tag；Wherein, label model tree is distributed frame model.
Step 340, using the user of current good friend to be recommended as target user.
In the present embodiment, the user data of at least one user of update is often got, it will be based on the user of update
Data and label model tree update the user tag information of at least one corresponding user, and in user tag information update
After, it regard preset user (can be all users, be also possible to specified user) as good friend user to be recommended, platform needs
It will be to each good friend user's commending friends to be recommended, wherein currently will determine that the user of commending friends makees by operation
For target user.
Step 350, the user tag for obtaining target user and weighted value corresponding with user tag.
Step 360 sorts the corresponding weighted value of the user tag of target user, and the weighted value for obtaining preset ratio is corresponding
User tag, the target labels as target user.
Wherein, target labels are eventually for the label for determining commending friends, that is to say, that each target user may gather around
There are multiple user tags, platform needs, using the user tag of wherein preset ratio as target labels, to do so by screening
Purpose is；Target labels are the labels that can embody a concentrated reflection of user behavior and platform requirement, are recommended in this, as final determine
The label of good friend can effectively improve the accuracy of recommended friend information.
In the present embodiment, the corresponding weighted value of the user tag of target user is sorted, since weighted value can reflect
The importance of corresponding user tag is tied so ranking results also can really reflect the behavioral characteristic of target user from sequence
In fruit, the corresponding user tag of weighted value of preset ratio is obtained, as the target labels of target user, for example, will be arranged in
Target labels of preceding 40% user tag as target user.
Step 370, in user tag information, obtain corresponding user tag and at least one target mark of target user
Identical user is signed, as the good friend recommended to target user.
In the present embodiment, when at least one target labels of target user are identical as the user tag of some user,
There are certain similarities for the user behavior both thought, so, which can be used as the commending friends of target.
This implementation does not explain in detail place, please be detailed in previous embodiment, details are not described herein.
The technical solution of the present embodiment is provided according to user tag information and default recommendation rules, is determined to each
The specific implementation step for the friend information that user recommends determines the final label of target user using the ranking results of weighted value,
That is target labels enable the target labels obtained with the behavioral characteristic of high probability reflection user, ensure that based on target
The commending friends and target user's matching degree with higher that label obtains.
Optionally, the corresponding weighted value of the user tag of target user is sorted, comprising:
Calculate the corresponding weighted value of user tag of target user and the fitting weighted value of time attenuation function；
By the corresponding fitting weighted value sequence of the user tag of target user.
Wherein, time attenuation function refers to the proportional function of the decaying of functional value and time, and typically, the time declines
Subtraction function can be decaying exponential function, for example, Chinese mugwort this great forgetting curve of guest.
Use in this optional technical solution, using time attenuation function as time decay factor, with target user
Label corresponding weighted value in family is fitted calculating, obtains fitting weighted value, which can be embodied based on the time uses
The temperature of family label, i.e. fitting weighted value is bigger, and care label temperature is higher, and the time that corresponding user data generates is closer
Current time.
This optional technical solution, by calculating the fitting weighted value of weighted value and time attenuation function, so that obtain
User tag information has time variation, influence of the time factor to data is considered when using user data, so that combining
The user tag information of fitting weighted value can sufficiently reflect the behavior of user from time angle, help finally to obtain accurate
Commending friends information.
Optionally, obtain corresponding user tag user identical at least one target labels of target user it
Afterwards, further includes:
Calculate the similarity of the user data of target user and the user data of the user of acquisition；
Judge whether similarity is greater than default similarity, is used if so, being used as user corresponding with similarity to target
The good friend that family is recommended.
Wherein, similarity is to evaluate the measurement of close degree between two groups of user data, and user data is closer, similarity
It is bigger, typically, when calculating the similarity of data, the related coefficient of data can be calculated.Default similarity is by setting from the background
Set, judge calculated similarity foundation whether up to standard.
In this optional technical solution, when the user data phase of the other users of the user data and acquisition of target user
Higher like spending, when reaching default similarity, then the other users that will acquire are as the good friend recommended to target user.
This optional technical solution, increases and further determines that the processing step of target user's commending friends, that is, calculate
Similarity between user data thinks that user corresponding with the similarity is that can push away when the similarity meets preset condition Shi Caineng
The good friend recommended realizes the further screening to user data, ensure that the commending friends finally obtained are and target user exists
Some aspects have the user of high consistency, improve the user quality of recommended good friend.
Example IV
Fig. 4 is a kind of structural schematic diagram for friend recommendation device that the embodiment of the present invention four provides, which includes: user
Data acquisition module 410, label information determining module 420 and good friend's information determination module 430, in which:
User data obtains module 410, for updating rule according to preset data, obtains at least one user's of update
User data；
Label information determining module 420, the label model for user data and training generation in advance according to update
Tree determines user tag information corresponding with each user；Wherein, label model tree is distributed frame model；
Friend information determining module 430, for determining to each use according to user tag information and default recommendation rules
The friend information that family is recommended.
The technical solution of the present embodiment, it is regular by being updated according to preset data, the user data of continuous renewal is obtained, and
According to the user data of update and the distributed tags model tree of training generation in advance, user corresponding with each user is determined
Label information is combining user tag information and default recommendation rules, is determining the friend information recommended to each user, i.e. this reality
The technical solution for applying example, which is realized, to be reported and handles to the dynamic of user data, and is alleviated by establishing distributed frame model
The pressure of data load and operation.Solve the dynamic processing and learning process and right for lacking user data in the prior art
The hardware for executing the prior art has higher the problem of requiring, and realizes the accuracy for improving recommended friend information and in time
Property, while reducing the effect to the requirement for executing hardware.
Optionally, before user data obtains module 410, further includes:
Model tree determining module is used for usage history user data, is trained to master pattern tree, obtains label model
Tree；
Wherein, label model tree includes: parent label model and at least one level subclass label model；By parent label model
Determining user tag information includes the user tag information that corresponding subclass label model determines.
Optionally, user tag information includes: user tag and weighted value corresponding with user tag.
Optionally, friend information determining module 430, comprising:
Target user's determination unit, for using the user of current good friend to be recommended as target user；
Label information acquiring unit, for obtaining the user tag and weight corresponding with user tag of target user
Value；
Target labels determination unit obtains default ratio for the corresponding weighted value of the user tag of target user to sort
The corresponding user tag of weighted value of example, the target labels as target user；
Label same subscriber determination unit is used in user tag information, obtaining corresponding user tag and target
The identical user of at least one target labels at family, as the good friend recommended to target user.
Optionally, target labels determination unit is specifically used for:
Calculate the corresponding weighted value of user tag of target user and the fitting weighted value of time attenuation function；
By the corresponding fitting weighted value sequence of the user tag of target user.
Optionally, after label same subscriber determination unit, further includes:
Similarity calculated, for calculate the user data of target user and the user of acquisition user data it is similar
Degree；
Commending friends determination unit, for judging whether similarity is greater than default similarity, if so, will be with similarity pair
The user answered is as the good friend recommended to target user.
A kind of friend recommendation device provided by the embodiment of the present invention can be performed provided by any embodiment of the invention one
Kind friend recommendation method, has the corresponding functional module of execution method and beneficial effect.
Embodiment five
Fig. 5 is a kind of structural schematic diagram for equipment that the embodiment of the present invention five provides, as shown in figure 5, the equipment includes place
Manage device 50 and memory 51；The quantity of processor 50 can be one or more in equipment, be with a processor 50 in Fig. 5
Example；Processor 50 in equipment can be connected with memory 51 by bus or other modes, to be connected as by bus in Fig. 5
Example.
Memory 51 is used as a kind of computer readable storage medium, can be used for storing software program, journey can be performed in computer
Sequence and module, if the corresponding program instruction/module of one of embodiment of the present invention friend recommendation method is (for example, good friend pushes away
The user data recommended in device obtains module 410, label information determining module 420 and good friend's information determination module 430).Processing
Software program, instruction and the module that device 50 is stored in memory 51 by operation, are answered thereby executing the various functions of equipment
With and data processing, that is, realize above-mentioned friend recommendation method.
Memory 51 can mainly include storing program area and storage data area, wherein storing program area can store operation system
Application program needed for system, at least one function；Storage data area, which can be stored, uses created data etc. according to terminal.This
Outside, memory 51 may include high-speed random access memory, can also include nonvolatile memory, for example, at least a magnetic
Disk storage device, flush memory device or other non-volatile solid state memory parts.In some instances, memory 51 can be further
Including the memory remotely located relative to processor 50, these remote memories can pass through network connection to equipment.It is above-mentioned
The example of network includes but is not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
Embodiment six
The embodiment of the present invention six also provides a kind of storage medium comprising computer executable instructions, and the computer can be held
Row instruction is used to execute a kind of friend recommendation method when being executed by computer processor, this method comprises:
Rule is updated according to preset data, obtains the user data of at least one user of update；
According to the user data of update and the label model tree of training generation in advance, use corresponding with each user is determined
Family label information；Wherein, label model tree is distributed frame model；
According to user tag information and default recommendation rules, the friend information recommended to each user is determined.
It certainly, include the storage medium of computer executable instructions provided by the embodiment of the present invention, computer can be held
The method operation that row instruction is not limited to the described above, can also be performed friend recommendation method provided by any embodiment of the invention
In relevant operation.
By the description above with respect to embodiment, it is apparent to those skilled in the art that, the present invention
It can be realized by software and required common hardware, naturally it is also possible to which by hardware realization, but in many cases, the former is more
Good embodiment.Based on this understanding, technical solution of the present invention substantially in other words contributes to the prior art
Part can be embodied in the form of software products, which can store in computer readable storage medium
In, floppy disk, read-only memory (Read-Only Memory, ROM), random access memory (Random such as computer
Access Memory, RAM), flash memory (FLASH), hard disk or CD etc., including some instructions are with so that a computer is set
Standby (can be personal computer, server or the network equipment etc.) executes method described in each embodiment of the present invention.
It is worth noting that, included each unit and module are only in a kind of embodiment of above-mentioned friend recommendation device
It is to be divided according to the functional logic, but be not limited to the above division, as long as corresponding functions can be realized；Separately
Outside, the specific name of each functional unit is also only for convenience of distinguishing each other, the protection scope being not intended to restrict the invention.
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (10)
1. a kind of friend recommendation method characterized by comprising
Rule is updated according to preset data, obtains the user data of at least one user of update；
It is corresponding with each user according to the user data of update and the label model tree of training generation in advance, determination
User tag information；Wherein, the label model tree is distributed frame model；
According to the user tag information and default recommendation rules, the friend information recommended to each user is determined.
2. the method according to claim 1, wherein obtaining the use of update updating rule according to preset data
Before user data, further includes:
Usage history user data is trained master pattern tree, obtains the label model tree；
Wherein, the label model tree includes: parent label model and at least one level subclass label model；By the parent label
The user tag information that model determines includes the user tag information that corresponding subclass label model determines.
3. method according to claim 1 or 2, which is characterized in that the user tag information include: user tag and
Weighted value corresponding with the user tag.
4. according to the method described in claim 3, it is characterized in that, according to the user tag information and default recommendation rule
Then, the friend information recommended to each user is determined, comprising:
Using the user of current good friend to be recommended as target user；
Obtain the target user user tag and weighted value corresponding with the user tag；
By the corresponding weighted value sequence of the user tag of the target user, the corresponding use of the weighted value of preset ratio is obtained
Family label, the target labels as the target user；
In the user tag information, at least one the target labels phase of corresponding user tag with the target user is obtained
Same user, as the good friend recommended to the target user.
5. according to the method described in claim 4, it is characterized in that, by the corresponding weighted value of the user tag of the target user
Sequence, comprising:
Calculate the corresponding weighted value of user tag of the target user and the fitting weighted value of time attenuation function；
By the corresponding fitting weighted value sequence of the user tag of the target user.
6. according to the method described in claim 4, it is characterized in that, obtaining corresponding user tag and the target user extremely
After few identical user of a target labels, further includes:
Calculate the similarity of the user data of the target user and the user data of the user of acquisition；
Judge whether the similarity is greater than default similarity, if so, being used as user corresponding with the similarity to institute
State the good friend of target user's recommendation.
7. a kind of friend recommendation device characterized by comprising
User data obtains module, for updating rule according to preset data, obtains the number of users of at least one user of update
According to；
Label information determining module, for according to the user data of update and training generates in advance label model tree,
Determining user tag information corresponding with each user；Wherein, the label model tree is distributed frame model；
Friend information determining module, for determining to each described according to the user tag information and default recommendation rules
The friend information that user recommends.
8. device according to claim 7, which is characterized in that before the user data obtains module, further includes:
Model tree determining module is used for usage history user data, is trained to master pattern tree, obtains the label model
Tree；
Wherein, the label model tree includes: parent label model and at least one level subclass label model；By the parent label
The user tag information that model determines includes the user tag information that corresponding subclass label model determines.
9. a kind of equipment, which is characterized in that the equipment includes:
One or more processors；
Storage device, for storing one or more programs；
When one or more of programs are executed by one or more of processors, so that one or more of processors are real
Now such as friend recommendation method as claimed in any one of claims 1 to 6.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
Such as friend recommendation method as claimed in any one of claims 1 to 6 is realized when execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910204096.6A CN109933731A (en) | 2019-03-18 | 2019-03-18 | A kind of friend recommendation method, apparatus, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910204096.6A CN109933731A (en) | 2019-03-18 | 2019-03-18 | A kind of friend recommendation method, apparatus, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109933731A true CN109933731A (en) | 2019-06-25 |
Family
ID=66987455
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910204096.6A Pending CN109933731A (en) | 2019-03-18 | 2019-03-18 | A kind of friend recommendation method, apparatus, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109933731A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112016003A (en) * | 2020-08-19 | 2020-12-01 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
| CN113343013A (en) * | 2019-12-24 | 2021-09-03 | 北京旷视科技有限公司 | Target object determination method and device and electronic equipment |
| CN116361566A (en) * | 2023-03-29 | 2023-06-30 | 竞速信息技术(廊坊)有限公司 | User relationship recommendation method and device based on big data |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103106199A (en) * | 2011-11-09 | 2013-05-15 | 中国移动通信集团四川有限公司 | Text retrieval method and test retrieval device |
| CN103116588A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技（深圳）有限公司 | Method and system for personalized recommendation |
| CN104239399A (en) * | 2014-07-14 | 2014-12-24 | 上海交通大学 | Method for recommending potential friends in social network |
| CN105045907A (en) * | 2015-08-10 | 2015-11-11 | 北京工业大学 | Method for constructing visual attention-label-user interest tree for personalized social image recommendation |
| CN106484764A (en) * | 2016-08-30 | 2017-03-08 | 江苏名通信息科技有限公司 | User's similarity calculating method based on crowd portrayal technology |
| CN106611008A (en) * | 2015-10-26 | 2017-05-03 | 中国移动通信集团公司 | Method and device for managing internet content labels |
| CN106919695A (en) * | 2017-03-07 | 2017-07-04 | 维沃移动通信有限公司 | A kind of information recommendation method and terminal |
| CN107451216A (en) * | 2017-07-17 | 2017-12-08 | 广州特道信息科技有限公司 | The granularity attribute recognition methods of label and device |
| CN109344255A (en) * | 2018-09-26 | 2019-02-15 | 平安科技（深圳）有限公司 | The fill method and terminal device of label |
-
2019
- 2019-03-18 CN CN201910204096.6A patent/CN109933731A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103106199A (en) * | 2011-11-09 | 2013-05-15 | 中国移动通信集团四川有限公司 | Text retrieval method and test retrieval device |
| CN103116588A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技（深圳）有限公司 | Method and system for personalized recommendation |
| CN104239399A (en) * | 2014-07-14 | 2014-12-24 | 上海交通大学 | Method for recommending potential friends in social network |
| CN105045907A (en) * | 2015-08-10 | 2015-11-11 | 北京工业大学 | Method for constructing visual attention-label-user interest tree for personalized social image recommendation |
| CN106611008A (en) * | 2015-10-26 | 2017-05-03 | 中国移动通信集团公司 | Method and device for managing internet content labels |
| CN106484764A (en) * | 2016-08-30 | 2017-03-08 | 江苏名通信息科技有限公司 | User's similarity calculating method based on crowd portrayal technology |
| CN106919695A (en) * | 2017-03-07 | 2017-07-04 | 维沃移动通信有限公司 | A kind of information recommendation method and terminal |
| CN107451216A (en) * | 2017-07-17 | 2017-12-08 | 广州特道信息科技有限公司 | The granularity attribute recognition methods of label and device |
| CN109344255A (en) * | 2018-09-26 | 2019-02-15 | 平安科技（深圳）有限公司 | The fill method and terminal device of label |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113343013A (en) * | 2019-12-24 | 2021-09-03 | 北京旷视科技有限公司 | Target object determination method and device and electronic equipment |
| CN112016003A (en) * | 2020-08-19 | 2020-12-01 | 重庆邮电大学 | Social crowd user tag mining and similar user recommending method based on CNN |
| CN116361566A (en) * | 2023-03-29 | 2023-06-30 | 竞速信息技术(廊坊)有限公司 | User relationship recommendation method and device based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109933731A (en) | A kind of friend recommendation method, apparatus, equipment and storage medium | |
| CN109360097A (en) | Prediction of Stock Index method, apparatus, equipment and storage medium based on deep learning | |
| CN109815980A (en) | Prediction technique, device, electronic equipment and the readable storage medium storing program for executing of user type | |
| CN108229986A (en) | Feature construction method, information distribution method and device in Information prediction | |
| US10621137B2 (en) | Architecture for predicting network access probability of data files accessible over a computer network | |
| US20220230203A1 (en) | Generating and utilizing machine-learning models to create target audiences with customized auto-tunable reach and accuracy | |
| CN115471283B (en) | Advertisement batch delivery method, device, equipment and storage medium | |
| CN107833088A (en) | Content providing, device and smart machine | |
| CN109308475A (en) | A kind of character recognition method and device | |
| CN108446351A (en) | The hotel's screening technique and system based on user preference of OTA platforms | |
| CN104221019A (en) | Method and apparatus for enhancing context intelligence in random index based system | |
| CN110135889A (en) | Method, server and the storage medium of intelligent recommendation book list | |
| CN114186084A (en) | Online multi-mode Hash retrieval method, system, storage medium and equipment | |
| CN114764471A (en) | Recommendation method, recommendation device and storage medium | |
| CN115329683A (en) | Aviation luggage online loading planning method, device, equipment and medium | |
| CN115982463A (en) | Resource recommendation method, device, equipment and storage medium | |
| CN105260458A (en) | Video recommendation method for display apparatus and display apparatus | |
| CN110489131B (en) | Gray level user selection method and device | |
| CN113869377A (en) | Training method and device and electronic equipment | |
| CN117312979A (en) | Object classification method, classification model training method and electronic equipment | |
| CN111489196A (en) | Prediction method and device based on deep learning network, electronic equipment and medium | |
| CN110852833A (en) | Taxi booking order processing method and device | |
| CN114092057A (en) | Project model construction method and device, terminal equipment and storage medium | |
| CN112818241B (en) | Content promotion method and device, computer equipment and storage medium | |
| Jabeen | The use of AI in marketing: Its impact and future |
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: 20190625 |
|
| RJ01 | Rejection of invention patent application after publication |