CN109145146A - A kind of data object recommended method, device and electronic equipment - Google Patents
A kind of data object recommended method, device and electronic equipment Download PDFInfo
- Publication number
- CN109145146A CN109145146A CN201811043564.8A CN201811043564A CN109145146A CN 109145146 A CN109145146 A CN 109145146A CN 201811043564 A CN201811043564 A CN 201811043564A CN 109145146 A CN109145146 A CN 109145146A
- Authority
- CN
- China
- Prior art keywords
- target
- label
- information
- client
- group
- 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 data object recommended method, device and electronic equipments.The described method includes: determining target property information when client is to user's recommending data object；Wherein, the target property information includes target time information, and the target time information is determined based on present system time；Based on the target property information, target group's label corresponding to the user of the currently used client is determined；Search the target data objects to match with target group's label；The identification information of the target data objects is exported in the client.It using the embodiment of the present invention, can be realized in the case where more natural person's shared devices or client account, improve and recommend targetedly purpose.
Description
Technical field
The present invention relates to data object fields, more particularly to a kind of data object recommended method, device and electronic equipment.
Background technique
In order to improve user experience, user, in the client can be to user when using client access data object
Recommend certain data objects.For example: when user watches video using the client of video class, in the client of the video class
It is middle to recommend certain videos to user；Alternatively, when user listens to music using the client of music class, in the client of the music class
It is middle to recommend certain music to user.
In the prior art, data object recommended method includes: the network identity of the equipment based on user or the visitor of user
Family end account obtains the history access information of user, the determining and matched data object of history access information, and will determine
Data object out recommends user.
But in actual scene, often there are the feelings that equipment or client account are used in conjunction in multiple natural persons
Condition, for example, one family shares an equipment or a client account.So, in this case, using above-mentioned data pair
When as recommended method, since the history access information utilized when recommending may relate to multiple natural persons, will lead to difference
The recommendation specific aim of user group is not strong.For example: when for multiple natural persons using same client account viewing video,
It is children for active user, the video for being suitble to adult viewing may be recommended to children；Or for active user be it is adult,
The video of suitable children's viewing may be recommended to adult.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of data object recommended method, device and electronic equipment, to realize
In the case where more natural person's shared devices or client account, improves and recommend targetedly purpose.Specific technical solution is such as
Under:
In a first aspect, the embodiment of the invention provides a kind of data object recommended methods, which comprises
When client is to user's recommending data object, target property information is determined；Wherein, the target property information packet
Target time information is included, the target time information is determined based on present system time；
Based on the target property information, determine that target group corresponding to the user of the currently used client mark
Label；
Search the target data objects to match with target group's label；
The identification information of the target data objects is exported in the client.
Optionally, the target property information further includes target location information；Wherein, the target location information are as follows: institute
State location information when the currently used client of user.
Optionally, described based on the target property information, corresponding to the user for determining the currently used client
Target group's label, comprising:
The target property information is input to neural network model trained in advance, obtains the currently used client
User corresponding to target group's label；
Wherein, the neural network model is according to sample attribute information and the corresponding sample cluster of the sample attribute information
Body label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history of the client
Access information is obtained.
Optionally, the training process of the neural network model, comprising:
From the history access information about the client, multiple sample attribute information and each sample attribute information are determined
Corresponding accessed data object；
For each sample attribute information, group's mark of the corresponding accessed data object of the sample attribute information is determined
Label, and based on identified group's label, determine sample populations label corresponding to the sample attribute information；
Using the multiple sample attribute information and the corresponding sample populations label of each sample attribute information, training is pre-
The initial neural network model first constructed, obtains the neural network model.
Optionally, the target data objects that the lookup matches with target group's label, comprising:
Determine multiple data objects for being provided with group's label；
In the multiple data object for being provided with group's label, the number of targets with target group's label is searched
According to object.
Optionally, the method also includes:
After the identification information for exporting the target data objects in the client every time, detection is in the first pre- timing
Whether object event occurs in long；Wherein, the object event are as follows: user accesses the first object, alternatively, user accesses first pair
As and to first object access duration be more than scheduled duration threshold value；First object is with institute's output identification information
Target data objects other than object；
When detecting that the number that the object event occurs in the second scheduled duration reaches pre-determined number, re -training institute
State neural network model；
Second scheduled duration is greater than first scheduled duration.
Second aspect, the embodiment of the invention provides a kind of data object recommendation apparatus, described device includes:
First determining module, for determining target property information when client is to user's recommending data object；Wherein,
The target property information includes target time information, and the target time information is determined based on present system time；
Second determining module determines the user institute of the currently used client for being based on the target property information
Corresponding target group's label；
Searching module, for searching the target data objects to match with target group's label；
Output module, for exporting the identification information of the target data objects in the client.
Optionally, the target property information further includes target location information；Wherein, the target location information are as follows: institute
State location information when the currently used client of user.
Optionally, second determining module, is specifically used for:
The target property information is input to neural network model trained in advance, obtains the currently used client
User corresponding to target group's label；
Wherein, the neural network model is according to sample attribute information and the corresponding sample cluster of the sample attribute information
Body label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history of the client
Access information is obtained.
Optionally, the training process of the neural network model, comprising:
From the history access information about the client, multiple sample attribute information and each sample attribute information are determined
Corresponding accessed data object；
For each sample attribute information, group's mark of the corresponding accessed data object of the sample attribute information is determined
Label, and based on identified group's label, determine sample populations label corresponding to the sample attribute information；
Using the multiple sample attribute information and the corresponding sample populations label of each sample attribute information, training is pre-
The initial neural network model first constructed, obtains the neural network model.
Optionally, the searching module, is specifically used for:
Determine multiple data objects for being provided with group's label；
In the multiple data object for being provided with group's label, the number of targets with target group's label is searched
According to object.
Optionally, described device further include:
Detection module, after the identification information for exporting the target data objects in the client every time, inspection
It surveys and whether object event occurs in the first scheduled duration；Wherein, the object event are as follows: user accesses the first object, alternatively,
User accesses the first object and is more than scheduled duration threshold value to the access duration of first object；First object be with
Object other than the target data objects of institute's output identification information；
Training module detects that the number that the object event occurs in the second scheduled duration reaches pre-determined number for working as
When, neural network model described in re -training；
Second scheduled duration is greater than first scheduled duration.
The third aspect, the embodiment of the invention provides a kind of electronic equipment, including processor and memory, wherein
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes that the embodiment of the present invention is provided
Data object recommended method the step of.
Scheme provided by the embodiment of the present invention, it is contemplated that use can be embodied using attribute informations such as the times of client
Therefore group's identity at family when client is to user's recommending data object, can determine target property information；Based on described
Target property information determines target group's label of the user of currently used client, and then is determining target group's label
Afterwards, target data objects to be recommended are determined based on target group's label.As it can be seen that can be realized by this programme mostly natural
In the case where people's shared device or client account, improves and recommend targetedly purpose.
Certainly, it implements any of the products of the present invention or method must be not necessarily required to reach all the above excellent simultaneously
Point.
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 data object recommended method provided by the embodiment of the present invention；
Fig. 2 is the flow diagram of another kind data object recommended method provided by the embodiment of the present invention；
Fig. 3 is a kind of structural schematic diagram of data object recommendation apparatus provided by the embodiment of the present invention；
Fig. 4 is the structural schematic diagram of a kind of electronic equipment provided by the 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.
In order to realize in the case where more natural person's shared devices or client account, improves and recommend targetedly mesh
, the embodiment of the invention provides a kind of data object recommended method, device, electronic equipment and storage mediums.
It should be noted that a kind of executing subject of data object recommended method provided by the embodiment of the present invention can be
A kind of data object recommendation apparatus, the device can be run in client or server, certainly, however it is not limited to this.
It is introduced in the following, being provided for the embodiments of the invention a kind of data object recommended method first.
As shown in Figure 1, a kind of data object recommended method provided by the embodiment of the present invention, may include steps of:
S101 determines target property information when client is to user's recommending data object.
Wherein, the data object may include: video, picture, music, books and commodity etc..
In embodiments of the present invention, client may is that the opportunity of user's recommending data object and open in the client
When dynamic；Alternatively, being also possible at the end of the client terminal start-up latter predetermined amount of time；It is pushed away alternatively, can also be and receive
When recommending instruction, this is all reasonable.Therefore, mesh can be determined for client to the specific opportunity of user's recommending data object
Mark attribute information.
Wherein, the target property information includes target time information, when the target time information is based on current system
Between determine.It is specific:
The target time information may include present system time；So, determine that the process of present system time can be with
Are as follows: operating system is transferred using API (Application Programming Interface, application programming interface)
Time etc. determines that the mode of present system time is without being limited thereto certainly.
Alternatively, the target time information may include target time section；So, determine that the process of target time section can be with
Are as follows: first determine present system time, then the period that the determining present system time is subordinate to from multiple predetermined amount of time, so
Afterwards using the determining period as target time section.
It should be noted that in embodiments of the present invention, multiple predetermined amount of time can be divided by one day, wherein appoint
The duration of a predetermined amount of time of anticipating can be for one hour or one minute etc., certainly, the duration of each predetermined amount of time
It can not be identical.Alternatively, multiple predetermined amount of time are also possible to Monday to Sunday etc., this is all reasonable.
S102 is based on the target property information, determines target complex corresponding to the user of the currently used client
Body label.
In embodiments of the present invention, group's label is the label of the classification about user, for example, group's label can be with
It is the label about user's gender, is also possible to the label etc. about age of user.In embodiments of the present invention, each data
Object is provided with corresponding group's label, for example, in the brief introduction of a video, can there is suitable sight in the client of video class
The age for the object seen.
In a kind of optional scheme, group's label may include age bracket.Wherein, the age bracket can be adult
Or it is teenage etc.；Or children, teenager, youth, middle age and old age etc., certainly, the age bracket may be specific
The range of age of numerical value, such as 3~10 years old etc..It is to subsequent, be different age group using age bracket as group's label
User recommend to meet the data object of the age bracket demand, to improve the specific aim recommended.
It should be noted that the content due to the target property information is different, it is based on the target property information, is determined
There are a variety of for the specific implementation of target group's label corresponding to the user of the currently used client.In order to which scheme is clear
Chu and layout are clear, for being based on the target property information, determine mesh corresponding to the user of the currently used client
The process of group's label is marked, it is subsequent to be introduced in conjunction with concrete implementation mode.
S103 searches the target data objects to match with target group's label.
In embodiments of the present invention, the process for searching the target data objects to match with target group's label can be with
Include the following steps a and b:
A determines multiple data objects for being provided with group's label；
It, can be in the client, or in the visitor as previously mentioned, each data object is both provided with group's label
In the corresponding server in family end, multiple data objects for being provided with group's label are determined, here, not in the embodiment of the present invention
The positions of multiple data objects for being provided with group's label be defined.
B searches the target with target group's label in the multiple data object for being provided with group's label
Data object.
It in embodiments of the present invention, can be by multiple data objects for being provided with group's label, one by one and with the mesh
The target data objects of mark group's label are compared, so that it is determined that the target data objects.
S104 exports the identification information of the target data objects in the client.
In embodiments of the present invention, the identification information that the target data objects can be shown in the client, than
The identification information as described in being shown with recommendation list or window form；
Alternatively, the identification information of the target data objects can also be exported in the form of sound etc. in the client
Deng this is all reasonable.
Wherein, the identification information can be the title of data object or access address etc..
Scheme provided by the embodiment of the present invention, it is contemplated that use can be embodied using attribute informations such as the times of client
Therefore group's identity at family when client is to user's recommending data object, can determine target property information；Based on described
Target property information determines target group's label of the user of currently used client, and then is determining target group's label
Afterwards, target data objects to be recommended are determined based on target group's label.As it can be seen that can be realized by this programme mostly natural
In the case where people's shared device or client account, improves and recommend targetedly purpose.
Due in the history access information of the client, may include time and the access of user accesses data object
Group's label of data object etc., it is possible in the history access information, by searching for mode determination currently make
Target group's label corresponding to user with the client.
Hereinafter, the particular content of combining target attribute information, in such a way that two specific implementations are introduced respectively in this kind
Under, it is described to be based on the target property information, determine target group's label corresponding to the user of the currently used client
Specific steps.
Optionally, in the first specific implementation, the target property information is present system time, then, institute
It states based on the target property information, determines target group's label corresponding to the user of the currently used client, it can be with
Include:
Based on the present system time, in the history access information of the client, the currently used visitor is determined
Target group's label corresponding to the user at family end.
It may include the time of user's access and the data object of access in history access information, can also include data pair
The information such as group's label of elephant.It is possible to from history access information corresponding to multiple times, the determining and current system
History access information corresponding to the time identical time of uniting determines access in the determining history access information
Data object group's label, and using determining group's label as mesh corresponding to the user of the currently used client
Mark group's label.
For example, can be looked into the history access information of the previous day if present system time is 8 points of evening on the same day
8 points of history access information of evening on the same day is found, and in 8 points of evening of the history access information of the previous day found, determines access
Data object group's label, and using determining group's label as mesh corresponding to the user of the currently used client
Mark group's label.
It certainly, in embodiments of the present invention, can also be in the history of multiple times identical with the present system time
In the corresponding group's label of access information, determine group's label according to pre-defined rule, and using determining group's label as
Target group's label, wherein the pre-defined rule can be with are as follows: one that frequency of occurrence is most is selected in multiple group's labels
It is a, or: in multiple group's labels, corresponding access duration longest one etc. is selected, here, not to this hair
The pre-defined rule in bright embodiment is defined.
For example, determining that group marks within past one week, in the corresponding group's label of multiple late 8 points of history access informations
Signing most one of frequency of occurrence is target group's label.
Optionally, in second of specific implementation, the target property information is target time section, then, it is described
Based on the target property information, determines target group's label corresponding to the user of the currently used client, can wrap
It includes:
Based on the target time section, in the history access information of the client, the currently used client is determined
Target group's label corresponding to the user at end.
It may include the data object accessed in multiple predetermined amount of time and each predetermined amount of time in history access information,
It can also include the information such as group's label of data object.It is possible to the access of the history corresponding to multiple predetermined amount of time
In information, history access information corresponding to a predetermined amount of time identical with the target time section is determined, determining
In the history access information, group's label of the data object of access is determined, and using determining group's label as currently making
Target group's label corresponding to user with the client.
For example, can be looked into the history access information of the previous day if target time section is 9 points of evening of 8 points of evening-
The history access information in late 8 points -9 points of evening is found, and in the history access information in 8 points -9 points of evening of the evening of the previous day found
In, determine group's label of the data object of access, and using determining group's label as the use of the currently used client
Target group's label corresponding to family.
It should be noted that in 9 points of history access information of evening in 8 points of the evening of the previous day-, if only group's mark
Label, can be using group's label as target group's label；It, can be in multiple group's labels if there is multiple group's labels
It is middle to determine group's label according to pre-defined rule, and using determining group's label as target group's label.About pre-
The content of set pattern then, details are not described herein.
It is of course also possible to which the history access information in multiple predetermined amount of time identical with the target time section is corresponding
In group's label, target group's label is determined, this is all reasonable.For example, interior, late 8 points multiple -9 points of evening past one week
In the corresponding group's label of history access information, group's label etc. is determined according to pre-defined rule.
It is understood that above-mentioned be based on the present system time or the target time section, in the client
In history access information, by searching for mode determine target group corresponding to the user of the currently used client mark
Label, can simply and quickly determine target group's label corresponding to the user of the currently used client.
Optionally, the target property information further includes target location information；Wherein, the target location information are as follows: institute
State location information when the currently used client of user.
Wherein, the target location information can be the geographical coordinate etc. of the equipment of the user, correspondingly, described in determining
The mode of target location information can pass through the modes such as GPS (Global Positioning System, global positioning system).
The target location information may be the label about place such as family or company, it should be noted that at this
In kind mode, the geographical coordinate of the equipment of the user can be determined first, then according to pre-set, user equipment
Corresponding relationship between geographical coordinate and each label determines target location information, that is, specifically about the mark in place
Label.
Due to that in the history access information of the client, can also include the place of user accesses data object, then,
For previously described, in the history access information, by searching for mode determine the use of the currently used client
Target group's label corresponding to family can introduce target location letter on the basis of above two specific implementation respectively
Breath cooperates the determination of corresponding target time information completion target group's label with the target location information.
Specifically, the first specific implementation can be with are as follows:
Based on the present system time and target location information, in the history access information of the client, determine
Target group's label corresponding to the user of the currently used client.
Specifically, a kind of optional realize that process can be with are as follows: from history access information corresponding to multiple times, determine
History access information corresponding to multiple times identical with the present system time, in determining multiple history access informations
In, determine a location information history access information identical with target location information, and from the determining history access information
In, determine group's label of the data object of access, and using determining group's label as the use of the currently used client
Target group's label corresponding to family.
For example, target location information is X if present system time is 8 points of evening.Then can it for the previous period
In the history access information of (such as the previous moon etc.), multiple late 8 points of history access informations are found；And it is finding
In multiple late 8 points of history access informations, determine that location information is a history access information of X, and in the determining history
In access information, group's label of the data object of access is determined, and using determining group's label as the currently used visitor
Target group's label corresponding to the user at family end.
Certainly, in the examples described above, if location information be X history access information have it is multiple, can be in multiple places
Information is to determine group's label according to pre-defined rule, and will determine in group's label corresponding to the history access information of X
Group's label as target group's label.About the content of pre-defined rule, details are not described herein.
Certainly, concrete implementation process is not limited to the above.
Second specific implementation can be with are as follows:
Based on the target time section and target location information, in the history access information of the client, determination is worked as
Target group's label corresponding to the preceding user using the client.
Specifically, a kind of optional realize that process can be with are as follows: the history access information corresponding to multiple predetermined amount of time
In, history access information corresponding to identical with the target time section multiple predetermined amount of time is determined, determining multiple
In history access information, a location information history access information identical with target location information is determined, and should from determining
In history access information, group's label of the data object of access is determined, and using determining group's label as currently used institute
State target group's label corresponding to the user of client.
For example, target location information is X if target time section is 9 points of evening of 8 points of evening-.Then can the last period when
Between (such as the previous moon etc.) history access information in, find the history access information in multiple late 8 points -9 points of evening；And
In multiple late 8 points found -9 points of history access information of evening, determine that location information is a history access information of X, and
In the determining history access information, determine access data object group's label, and using determining group's label as
Target group's label corresponding to the user of the currently used client.
Certainly, in the examples described above, if location information be X history access information have it is multiple, can be in multiple places
Information is to determine group's label according to pre-defined rule, and will determine in group's label corresponding to the history access information of X
Group's label as target group's label.About the content of pre-defined rule, details are not described herein.
Certainly, concrete implementation process is not limited to the above.
In the above scheme, the target location information cooperation target time information that can use active user, determines current
Target group's label corresponding to user using the client, thus in subsequent determination and target group's label phase
The target data objects matched are recommended.Since user is when using client access data object, place to use also has one
Fixed regularity, such as children are usually to use at home, therefore, be can be improved really using the target location information of active user
The accuracy for determining ownership goal group label is conducive to improve the specific aim and accuracy recommended.
Optionally, in embodiments of the present invention, it can use advantage of the neural network model in terms of deep learning, realize
It is described to be based on the target property information, determine the mistake of target group's label corresponding to the user of the currently used client
Journey, the process may include:
The target property information is input to neural network model trained in advance, obtains the currently used client
User corresponding to target group's label；
Wherein, the neural network model is according to sample attribute information and the corresponding sample cluster of the sample attribute information
Body label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history of the client
Access information is obtained.
Wherein, the training process of the neural network model, may comprise steps of:
The first step determines multiple sample attribute information and each sample from the history access information about the client
The corresponding accessed data object of attribute information；
Second step determines the corresponding accessed data object of the sample attribute information for each sample attribute information
Group's label determine sample populations label corresponding to the sample attribute information and based on identified group's label；
Third step utilizes the multiple sample attribute information and the corresponding sample populations mark of each sample attribute information
Label, the initial neural network model that training constructs in advance, obtain the neural network model.
It should be noted that in embodiments of the present invention, if the target property information is present system time, institute
Stating sample attribute information is sample time；If the target property information is target time section, the sample attribute information
For sample time section；If the target property information is present system time and target location information, the sample attribute
Information is sample time and sample site information；If the target property information be target time section and target location information,
Then the sample attribute information is sample time section and sample site information.
Hereinafter, illustrating training for the neural network model so that the target property information is present system time as an example
Journey, the training process are specifically as follows:
1) training set that the multiple sample time and the corresponding sample populations label of each sample time are constituted is inputted into institute
State initial neural network model；Wherein, using the corresponding sample populations label of sample time as the initial neural network model
True value；
2) parameter in (0,1) range in the initial neural network model of random initializtion, the parameter include neuron
Connection weight etc..
3) sample time and the corresponding sample populations label of the sample time are passed through into the initial neural network
The training of model obtains training result.
4) training result and corresponding true value are compared, obtain output result；
5) according to output as a result, calculating the value of the loss function Loss of the initial neural network model；
6) according to the value of the Loss, the parameter of initial neural network model is adjusted, and re-starts 3) -6) step, directly
Value to the Loss has reached certain condition of convergence, that is, the value of the Loss reaches minimum, at this moment, determines initial mind
Parameter through network model completes the training of initial neural network model, obtains the neural network model that training is completed.
The training process of target property information neural network model corresponding when being other content, with above-mentioned training process
Similar, details are not described herein.
In embodiments of the present invention, corresponding to the user that the currently used client is determined using neural network model
Target group's label can use advantage of the neural network model in terms of classification, saves the cost of data processing, improves determination
The accuracy of target group's label.
Optionally, on the basis of using the implementation of neural network model, in embodiments of the present invention, the method
Can also include step c and d:
C, after the identification information for exporting the target data objects in the client every time, detection is predetermined first
Whether object event occurs in duration；Wherein, the object event are as follows: user accesses the first object, alternatively, user's access first
Object and to the access duration of first object be more than scheduled duration threshold value；First object is with institute's output identification letter
Object other than the target data objects of breath；
D, when detecting that the number that the object event occurs in the second scheduled duration reaches pre-determined number, re -training
The neural network model.
Second scheduled duration is greater than first scheduled duration.Wherein, first scheduled duration and second makes a reservation for
Duration may be set according to actual conditions, and the present invention is without limitation.For example: first scheduled duration can be 5 minutes, phase
It answers, which can be 15 minutes；Alternatively, first scheduled duration can be 10 minutes, correspondingly, second is pre-
Timing length can be 20 minutes, etc..
Since in actual scene, possible multiple natural persons are used in conjunction with time or the place of equipment or client account
Information is changed, then, the corresponding user of present system time may be distinct from history access information and work as with described
Corresponding user of preceding system time identical time, or when may be distinct from history access information with the current system
Between user or the corresponding user of current location information corresponding to the corresponding target time section identical period, may be simultaneously
Different from user corresponding to location information identical with the current location information in history access information；So current use
Family may and not like the data object of recommendation, thus therefore generation object event in embodiments of the present invention, can pass through
Above-mentioned steps are detected and are judged to object event, when detecting the number that the object event occurs in the second scheduled duration
When reaching pre-determined number, neural network model described in re -training is determined, to adapt to the user of variation, utilize the mind of re -training
Through network model in the subsequent satisfaction for improving the specific aim and user recommended.
Hereinafter, being video for data object, illustrate that a kind of data object provided herein pushes away with specific embodiment
Recommend the implementation process of method.Referring to fig. 2, Fig. 2 is the stream of another kind data object recommended method provided by the embodiment of the present invention
Journey schematic diagram；Video recommendation method in the present embodiment may comprise steps of:
S201 determines present system time and the currently used visitor of user when client recommends video to user
Target location information when the end of family；
Wherein, S201 is the data object when being video, a kind of specific implementation of S101.When about current system
Between and the specific method of determination of target location information repeat no more.
For example, identified present system time can be evening 8:30, and identified target location information can be
Label " family ".
S202 determines the target time section that the present system time is subordinate in multiple predetermined amount of time；
In the present embodiment, multiple predetermined amount of time can be divided into advance by one day time as unit of one hour.
In this step, it is evening 8:30 for present system time, can determines evening 8 in multiple predetermined amount of time:
30 target time sections being subordinate to are 8 points to 9 points of evening of evening.
S203, the neural network model that the target time section and the target location information input are trained in advance, obtains
The age bracket exported to the neural network model；
Wherein, the neural network model is trained according to sample time section, sample site information and sample age bracket
It arriving, the sample time section, the sample site information and the sample age bracket are obtained based on historical viewing information,
And the sample time section, the sample site information and the sample age bracket are one-to-one.In the present embodiment,
Each video is provided with corresponding age bracket, and the age bracket can be children or adult.
Training process about the neural network model repeats no more.
It in this step, is 8 points to late 9 points of evening and target location information " family " input instruction in advance by target time section
Experienced neural network model obtains the age bracket of the neural network model output；Such as the neural network model output
Age bracket is children.
S204, using the age bracket exported as target age section corresponding to the user of the currently used client；
In this step, using children as target age section corresponding to the user of the currently used client.
Wherein, it is video that S202-S204, which is the data object, when group's label is age bracket, one kind of S102
Specific implementation.
S205 determines multiple videos for being provided with age bracket；
Multiple videos for being provided with age bracket can be determined in the client.
S206 searches the target video with the target age section in the multiple video for being provided with age bracket.
In the multiple video for being provided with age bracket, the video that age bracket is children, and the institute that will be found are searched
Video is stated as target video.
Wherein, it is video that S205-S206, which is the data object, when group's label is age bracket, one kind of S103
Specific implementation.
S207 shows the identification information of the target video in the client.
In the client, can in the form of recommendation list displaying target video identification information, mark letter
Breath can be the title of video or URL (Uniform Resource Locator, uniform resource locator) of video etc..
Wherein, it is video that S207, which is the data object, when group's label is age bracket, a kind of specific reality of S104
Existing mode.
In the present embodiment, it is contemplated that group's body of user can be embodied using the when and where information of client
Part, therefore, when recommending video to user in the client, it can use and believed based on the system time and place recommended when video
Breath is based on target year and then after determining target age section to analyze the target age section of user of currently used client
Age section determines target video to be recommended.As it can be seen that can be realized by this programme in more natural person's shared devices or client
In the case where account, improves and recommend targetedly purpose.And neural network model algorithm is used, data processing can be saved
Cost improves the accuracy of age bracket judgement.Video is recommended according to age of user in the present embodiment, may be implemented to all ages and classes
The user of section recommends the purpose of the favorite video of the age bracket, thus is conducive to improve the accuracy recommended, and can be enhanced
User uses the experience of client.Meanwhile, it is capable to children is avoided to watch the video outside age bracket limitation, it can also be to avoid children family
It is long by equipment to children in use, need the video of moment monitoring children's viewing, and when children watch unsuitable video into
Therefore row video handover operation can save the time of Parents.
Corresponding to above method embodiment, the embodiment of the invention also provides a kind of data object recommendation apparatus, such as Fig. 3 institute
Show, which includes:
First determining module 301, for determining target property information when client is to user's recommending data object；Its
In, the target property information includes target time information, and the target time information is determined based on present system time；
Second determining module 302 determines the user of the currently used client for being based on the target property information
Corresponding target group's label；
Searching module 303, for searching the target data objects to match with target group's label；
Output module 304, for exporting the identification information of the target data objects in the client.
Optionally, in embodiments of the present invention, the target property information further includes target location information；Wherein, described
Target location information are as follows: location information when the currently used client of the user.
Optionally, in embodiments of the present invention, second determining module 302, is specifically used for:
The target property information is input to neural network model trained in advance, obtains the currently used client
User corresponding to target group's label；
Wherein, the neural network model is according to sample attribute information and the corresponding sample cluster of the sample attribute information
Body label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history of the client
Access information is obtained.
Optionally, in embodiments of the present invention, the training process of the neural network model, comprising:
From the history access information about the client, multiple sample attribute information and each sample attribute information are determined
Corresponding accessed data object；
For each sample attribute information, group's mark of the corresponding accessed data object of the sample attribute information is determined
Label, and based on identified group's label, determine sample populations label corresponding to the sample attribute information；
Using the multiple sample attribute information and the corresponding sample populations label of each sample attribute information, training is pre-
The initial neural network model first constructed, obtains the neural network model.
Optionally, in embodiments of the present invention, the searching module 303, is specifically used for:
Determine multiple data objects for being provided with group's label；
In the multiple data object for being provided with group's label, the number of targets with target group's label is searched
According to object.
Optionally, in embodiments of the present invention, described device further include:
Detection module, after the identification information for exporting the target data objects in the client every time, inspection
It surveys and whether object event occurs in the first scheduled duration；Wherein, the object event are as follows: user accesses the first object, alternatively,
User accesses the first object and is more than scheduled duration threshold value to the access duration of first object；First object be with
Object other than the target data objects of institute's output identification information；
Training module detects that the number that the object event occurs in the second scheduled duration reaches pre-determined number for working as
When, neural network model described in re -training；
Second scheduled duration is greater than first scheduled duration.
Scheme provided by the embodiment of the present invention, it is contemplated that use can be embodied using attribute informations such as the times of client
Therefore group's identity at family when client is to user's recommending data object, can determine target property information；Based on described
Target property information determines target group's label of the user of currently used client, and then is determining target group's label
Afterwards, target data objects to be recommended are determined based on target group's label.Wherein, the target property information includes the object time
Information, the target time information are determined based on present system time.As it can be seen that can be realized by this programme mostly natural
In the case where people's shared device or client account, improves and recommend targetedly purpose.
Corresponding to above method embodiment, the embodiment of the invention also provides a kind of electronic equipment, as shown in figure 4, can be with
Including processor 401 and memory 402, wherein
The memory 402, for storing computer program；
The processor 401 when for executing the program stored on the memory 402, realizes the embodiment of the present invention
The step of provided data object recommended method.
Above-mentioned memory may include RAM (Random Access Memory, random access memory), also may include
NVM (Non-Volatile Memory, nonvolatile memory), for example, at least a magnetic disk storage.Optionally, memory
It can also be that at least one is located away from the storage device of above-mentioned processor.
Above-mentioned processor can be general processor, including CPU (Central Processing Unit, central processing
Device), NP (Network Processor, network processing unit) etc.；Can also be DSP (Digital Signal Processor,
Digital signal processor), ASIC (Application Specific Integrated Circuit, specific integrated circuit),
FPGA (Field-Programmable Gate Array, field programmable gate array) or other programmable logic device are divided
Vertical door or transistor logic, discrete hardware components.
It by above-mentioned electronic equipment, can be realized: in view of the attribute informations such as time using client can embody
Therefore group's identity of user when client is to user's recommending data object, can determine target property information；Based on institute
State target property information determine currently used client user target group's label, and then determining target group mark
After label, target data objects to be recommended are determined based on target group's label.As it can be seen that can be realized by this programme mostly certainly
In the case where right people's shared device or client account, improves and recommend targetedly purpose.
In addition, the embodiment of the invention provides one kind corresponding to data object recommended method provided by above-described embodiment
Computer readable storage medium is stored with computer program in the computer readable storage medium, and computer program is by processor
The step of data object recommended method provided by the embodiment of the present invention is realized when execution.
Above-mentioned computer-readable recording medium storage has executes data object provided by the embodiment of the present invention at runtime
The application program of recommended method, therefore can be realized: in view of the attribute informations such as time using client can embody use
Therefore group's identity at family when client is to user's recommending data object, can determine target property information；Based on described
Target property information determines target group's label of the user of currently used client, and then is determining target group's label
Afterwards, target data objects to be recommended are determined based on target group's label.As it can be seen that can be realized by this programme mostly natural
In the case where people's shared device or client account, improves and recommend targetedly purpose.
For electronic equipment and computer readable storage medium embodiment, method content base as involved in it
Originally it is similar to embodiment of the method above-mentioned, so being described relatively simple, referring to the part explanation of embodiment of the method in place of correlation
?.
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 system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.
The foregoing is merely alternative embodiments of the invention, are 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 (13)
1. a kind of data object recommended method, which is characterized in that the described method includes:
When client is to user's recommending data object, target property information is determined；Wherein, the target property information includes mesh
Temporal information is marked, the target time information is determined based on present system time；
Based on the target property information, target group's label corresponding to the user of the currently used client is determined；
Search the target data objects to match with target group's label；
The identification information of the target data objects is exported in the client.
2. the method according to claim 1, wherein the target property information further includes target location information；
Wherein, the target location information are as follows: location information when the currently used client of the user.
3. method according to claim 1 or 2, which is characterized in that it is described to be based on the target property information, it determines current
Target group's label corresponding to user using the client, comprising:
The target property information is input to neural network model trained in advance, obtains the use of the currently used client
Target group's label corresponding to family；
Wherein, the neural network model is according to sample attribute information and the corresponding sample populations mark of the sample attribute information
Label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history access of the client
Information is obtained.
4. according to the method described in claim 3, it is characterized in that, the training process of the neural network model, comprising:
From the history access information about the client, multiple sample attribute information and each sample attribute information difference are determined
Corresponding accessed data object；
For each sample attribute information, group's label of the corresponding accessed data object of the sample attribute information is determined,
And based on identified group's label, sample populations label corresponding to the sample attribute information is determined；
Utilize the multiple sample attribute information and the corresponding sample populations label of each sample attribute information, the preparatory structure of training
The initial neural network model built, obtains the neural network model.
5. method according to claim 1 or 2, which is characterized in that the lookup matches with target group's label
Target data objects, comprising:
Determine multiple data objects for being provided with group's label；
In the multiple data object for being provided with group's label, the target data pair with target group's label is searched
As.
6. according to the method described in claim 3, it is characterized in that, the method also includes:
After the identification information for exporting the target data objects in the client every time, detect in the first scheduled duration
Whether object event is occurred；Wherein, the object event are as follows: user access the first object, alternatively, user access the first object and
Access duration to first object is more than scheduled duration threshold value；First object is the mesh with institute's output identification information
Mark the object other than data object；
When detecting that the number that the object event occurs in the second scheduled duration reaches pre-determined number, mind described in re -training
Through network model；
Second scheduled duration is greater than first scheduled duration.
7. a kind of data object recommendation apparatus, which is characterized in that described device includes:
First determining module, for determining target property information when client is to user's recommending data object；Wherein, described
Target property information includes target time information, and the target time information is determined based on present system time；
Second determining module is used for based on the target property information, corresponding to the user for determining the currently used client
Target group's label；
Searching module, for searching the target data objects to match with target group's label；
Output module, for exporting the identification information of the target data objects in the client.
8. device according to claim 7, which is characterized in that the target property information further includes target location information；
Wherein, the target location information are as follows: location information when the currently used client of the user.
9. device according to claim 7 or 8, which is characterized in that second determining module is specifically used for:
The target property information is input to neural network model trained in advance, obtains the use of the currently used client
Target group's label corresponding to family；
Wherein, the neural network model is according to sample attribute information and the corresponding sample populations mark of the sample attribute information
Label training obtains, and the sample attribute information and the sample populations label are based on the corresponding history access of the client
Information is obtained.
10. device according to claim 9, which is characterized in that the training process of the neural network model, comprising:
From the history access information about the client, multiple sample attribute information and each sample attribute information difference are determined
Corresponding accessed data object；
For each sample attribute information, group's label of the corresponding accessed data object of the sample attribute information is determined,
And based on identified group's label, sample populations label corresponding to the sample attribute information is determined；
Utilize the multiple sample attribute information and the corresponding sample populations label of each sample attribute information, the preparatory structure of training
The initial neural network model built, obtains the neural network model.
11. device according to claim 7 or 8, which is characterized in that the searching module is specifically used for:
Determine multiple data objects for being provided with group's label；
In the multiple data object for being provided with group's label, the target data pair with target group's label is searched
As.
12. device according to claim 9, which is characterized in that described device further include:
Detection module, after the identification information for exporting the target data objects in the client every time, detection exists
Whether object event occurs in first scheduled duration；Wherein, the object event are as follows: user accesses the first object, alternatively, user
It accesses the first object and is more than scheduled duration threshold value to the access duration of first object；First object is with defeated
Object other than the target data objects of identification information out；
Training module, for when detecting that the number that the object event occurs in the second scheduled duration reaches pre-determined number,
Neural network model described in re -training；
Second scheduled duration is greater than first scheduled duration.
13. a kind of electronic equipment, which is characterized in that including processor and memory, wherein
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes any side claim 1-6
Method step.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811043564.8A CN109145146A (en) | 2018-09-07 | 2018-09-07 | A kind of data object recommended method, device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811043564.8A CN109145146A (en) | 2018-09-07 | 2018-09-07 | A kind of data object recommended method, device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109145146A true CN109145146A (en) | 2019-01-04 |
Family
ID=64823855
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811043564.8A Pending CN109145146A (en) | 2018-09-07 | 2018-09-07 | A kind of data object recommended method, device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109145146A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147497A (en) * | 2019-05-15 | 2019-08-20 | 中国搜索信息科技股份有限公司 | A kind of personalization content recommendation method towards younger population |
| CN110728370A (en) * | 2019-09-16 | 2020-01-24 | 北京达佳互联信息技术有限公司 | Training sample generation method and device, server and storage medium |
| CN111475691A (en) * | 2020-03-06 | 2020-07-31 | 拉扎斯网络科技（上海）有限公司 | Method and device for acquiring recommended object data and electronic equipment |
| WO2020257993A1 (en) * | 2019-06-24 | 2020-12-30 | 深圳市欢太科技有限公司 | Content pushing method and apparatus, server, and storage medium |
| CN112561636A (en) * | 2020-12-14 | 2021-03-26 | 深圳市傲天科技股份有限公司 | Recommendation method, recommendation device, terminal equipment and medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105207897A (en) * | 2015-09-25 | 2015-12-30 | 北京奇艺世纪科技有限公司 | Method and device for recommending videos to mobile equipment |
| CN105868254A (en) * | 2015-12-25 | 2016-08-17 | 乐视网信息技术（北京）股份有限公司 | Information recommendation method and apparatus |
| CN107341272A (en) * | 2017-08-25 | 2017-11-10 | 北京奇艺世纪科技有限公司 | A kind of method for pushing, device and electronic equipment |
-
2018
- 2018-09-07 CN CN201811043564.8A patent/CN109145146A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105207897A (en) * | 2015-09-25 | 2015-12-30 | 北京奇艺世纪科技有限公司 | Method and device for recommending videos to mobile equipment |
| CN105868254A (en) * | 2015-12-25 | 2016-08-17 | 乐视网信息技术（北京）股份有限公司 | Information recommendation method and apparatus |
| CN107341272A (en) * | 2017-08-25 | 2017-11-10 | 北京奇艺世纪科技有限公司 | A kind of method for pushing, device and electronic equipment |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147497A (en) * | 2019-05-15 | 2019-08-20 | 中国搜索信息科技股份有限公司 | A kind of personalization content recommendation method towards younger population |
| WO2020257993A1 (en) * | 2019-06-24 | 2020-12-30 | 深圳市欢太科技有限公司 | Content pushing method and apparatus, server, and storage medium |
| CN110728370A (en) * | 2019-09-16 | 2020-01-24 | 北京达佳互联信息技术有限公司 | Training sample generation method and device, server and storage medium |
| CN110728370B (en) * | 2019-09-16 | 2020-08-14 | 北京达佳互联信息技术有限公司 | Training sample generation method and device, server and storage medium |
| CN111475691A (en) * | 2020-03-06 | 2020-07-31 | 拉扎斯网络科技（上海）有限公司 | Method and device for acquiring recommended object data and electronic equipment |
| CN112561636A (en) * | 2020-12-14 | 2021-03-26 | 深圳市傲天科技股份有限公司 | Recommendation method, recommendation device, terminal equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109145146A (en) | A kind of data object recommended method, device and electronic equipment | |
| US11127020B2 (en) | Generating an activity inference model from contextual data | |
| US10341741B2 (en) | Methods, systems, and media for presenting suggestions of media content | |
| CN103617075B (en) | application program recommending method, system and server | |
| ES2261527T3 (en) | SYSTEM, PROCEDURE AND APPLICATION OF SOFTWARE FOR DIRECT ADVERTISING THROUGH A GROUP OF BEHAVIOR MODELS, AND PROGRAMMING PREFERENCES BASED ON BEHAVIOR MODEL GROUPS. | |
| CN105282565A (en) | Video recommendation method and device | |
| CN102855309B (en) | A kind of information recommendation method based on user behavior association analysis and device | |
| US10002194B2 (en) | Event location with social network integration | |
| KR101312927B1 (en) | Advertisement providing system | |
| EP2608502A2 (en) | Context activity tracking for recommending activities through mobile electronic terminals | |
| US20160105520A1 (en) | Method, apparatus, and system for pushing network content | |
| KR101424382B1 (en) | Method for recommending point of interest using user preferences and moving patterns | |
| US11118926B2 (en) | Computer-implemented method for recommending booths-to-visit | |
| CN110020148A (en) | A kind of information recommendation method, device and the device for information recommendation | |
| CN104933134A (en) | User feature analysis method and user feature analysis device | |
| CN111177568B (en) | Object pushing method based on multi-source data, electronic device and storage medium | |
| US20140143804A1 (en) | System and method for providing advertisement service | |
| CN104503975A (en) | Method and device for customizing recommended card | |
| US11521238B2 (en) | Method and system for determining fact of visit of user to point of interest | |
| CN113454669A (en) | Characterizing a place by user visited features | |
| CN114339417A (en) | Video recommendation method, terminal device and readable storage medium | |
| CN108694211B (en) | Application distribution method and device | |
| US20130013625A1 (en) | Estimating apparatus, estimating method, and program | |
| CN112559649B (en) | Method and device for identifying interest point state | |
| CN109753585B (en) | Method and device for determining recommended video, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |