CN109729376A - A kind of processing method of life cycle, device, equipment and storage medium - Google Patents
A kind of processing method of life cycle, device, equipment and storage medium Download PDFInfo
- Publication number
- CN109729376A CN109729376A CN201910000641.XA CN201910000641A CN109729376A CN 109729376 A CN109729376 A CN 109729376A CN 201910000641 A CN201910000641 A CN 201910000641A CN 109729376 A CN109729376 A CN 109729376A
- Authority
- CN
- China
- Prior art keywords
- life cycle
- live streaming
- user
- spectators
- label
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The invention discloses a kind of processing method of life cycle, device, equipment and storage mediums.This method comprises: determining live streaming life cycle, the live streaming life cycle is for characterizing the one of spectators' user activity section of preset time period；It obtains the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming；The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle is obtained.By the way that this method solve cannot find the problem of product declines user's viscosity in time, the user behavior by the short time is realized, judges the stage of life cycle locating for user, finds the user of phase at a low ebb in time, in the case where user does not completely disengage product, user is recalled.
Description
Technical field
The present invention relates to machine learning techniques more particularly to a kind of processing method of life cycle, device, equipment and storages
Medium.
Background technique
With the development of Internet technology and intelligent mobile terminal equipment, work, life of the various internet products to people
Living to bring many convenient and amusement, in recent years, all kinds of live streaming platforms for net cast emerge one after another, and net cast is to people
Bring more real-time social experience.The development that platform is broadcast live relies on a large amount of main broadcaster users and spectators user.Therefore live streaming platform
Need to pay close attention to the state of main broadcaster user and spectators user, to make adjustment in time to platform.Live streaming platform can generally lead to
Counting user is crossed to the hobby of live streaming, and then determines the business processing mode to user, as pushed its preference type to user
It is broadcast live (or main broadcaster).
In realizing process of the present invention, at least there are the following problems in the prior art for inventor's discovery:
It is not distinguish the activity of the user due to carrying out push to the hobby of live streaming for user, cannot sends out in time
Current family more cannot targetedly be determined to user in the variation of live streaming platform liveness for the variation of user activity
Business processing mode.
Summary of the invention
The present invention provides processing method, device, equipment and the storage medium of a kind of life cycle, must not be sent out in time with solution
The problem of existing product declines user's viscosity.
In a first aspect, the embodiment of the invention provides a kind of processing methods of life cycle, comprising:
Determine live streaming life cycle, the live streaming life cycle be for characterize the one of spectators' user activity section it is default when
Between section；
It obtains the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming；
The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are characterized
The life cycle label of liveness of the spectators user in the live streaming life cycle.
Second aspect, the embodiment of the invention also provides a kind of processing methods of life cycle, comprising:
Determine live streaming life cycle, the live streaming life cycle be for characterize the one of spectators' user activity section it is default when
Between section；
The live streaming watching behavior data that acquisition spectators user generates when watching live streaming in the live streaming life cycle；
Determine life cycle label, spectators user described in the life cycle tag characterization is in the live streaming life cycle
Liveness；
Life cycle model is broadcast live according to the live streaming watching behavior data and the life cycle label training.
The third aspect, the embodiment of the invention also provides a kind of processing units of life cycle, comprising:
Period determination module is broadcast live, for determining live streaming life cycle, the live streaming life cycle is for characterizing spectators
One section of preset time period of user activity；
Behavioral data obtains module, generates for obtaining when spectators user's viewing is broadcast live in the live streaming life cycle
Watching behavior data are broadcast live；
Period label obtains module, for the live streaming watching behavior data to be input to preset live streaming life cycle mould
It is handled in type, obtains the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle.
Fourth aspect, the embodiment of the invention also provides a kind of processing units of life cycle, comprising:
Life cycle determining module, for determining live streaming life cycle, the live streaming life cycle is for characterizing spectators
One section of preset time period of user activity；
Behavioral data acquisition module is generated for acquiring when spectators user watches live streaming in the live streaming life cycle
Watching behavior data are broadcast live；
Period label determining module, for determining life cycle label, spectators described in the life cycle tag characterization are used
Liveness of the family in the live streaming life cycle；
Periodic model training module, for straight according to the live streaming watching behavior data and the life cycle label training
Broadcast life cycle model.
5th aspect, the embodiment of the invention also provides a kind of electronic equipment, comprising:
One or more processors；
Memory, 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 a kind of processing method of life cycle as described in any embodiment.
6th aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, are stored thereon with computer
Program realizes a kind of processing method of life cycle as described in any embodiment when the program is executed by processor.
The present invention watches the live streaming generated when live streaming viewing row by obtaining the spectators user in the live streaming life cycle
For data；The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are characterized
The life cycle label of liveness of the spectators user in the live streaming life cycle.Solve hobby according to user into
The not strong problem of row live streaming push bring specific aim, realizes the live streaming watching behavior data by analyzing spectators user, from
And obtain the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention one provides；
Fig. 2 is a kind of flow chart of the processing method of life cycle provided by Embodiment 2 of the present invention；
Fig. 3 is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention three provides；
Fig. 4 A is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention four provides；
Fig. 4 B is the relation schematic diagram of a kind of duration range and liveness that the embodiment of the present invention four provides；
Fig. 5 is a kind of structure chart of the processing unit for life cycle that the embodiment of the present invention five provides；
Fig. 6 is a kind of structure chart of the processing unit for life cycle that the embodiment of the present invention six provides；
Fig. 7 is the structural schematic diagram for a kind of electronic equipment that the embodiment of the present invention seven 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 is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention one provides.The present embodiment can
It is applicable to analyze the live streaming watching behavior data that spectators user generates, and then determines spectators' the activity of the user
Scene.But it is understood that the present embodiment is readily applicable in other application scenarios, it can be according to user in certain time period
Interior behavioral data determines the liveness of user during this period of time.This method can by a kind of processing unit of life cycle Lai
It executes, which can be realized by the mode of software and/or software, be generally disposed in electronic equipment, be typically arranged in
In processor, such as in the processor of the server in live streaming platform.With reference to Fig. 1, this method is specifically included:
S101, live streaming life cycle is determined.
Life cycle includes the starting point of life cycle and the terminal of life cycle.It is for characterizing spectators that life cycle, which is broadcast live,
One section of preset time period of user activity.For a user, can be made with being continuously not logged on live streaming platform for a period of time
For the mark of end of life.If user is to Add User and (log in for the first time after register account number), is logged in and be broadcast live with user
Platform is the starting point of life cycle；If user is that reflux user (is not to be logged in after register account number for the first time, is continuous still
A period of time logs on live streaming platform after being not logged on live streaming platform), then live streaming platform is logged on as Life Cycle using user
The starting point of phase.Live streaming life cycle be from life cycle determine a period of time, by comparing mass viewer audiences user this when
Between section various actions data, can predict the future behaviour of spectators user.
Live streaming platform generally comprises two class users: spectators user and main broadcaster user.In live streaming platform, on the one hand, registration
There are multiple main broadcaster users, is configured with corresponding user data (such as user account, user's head portrait), which can log in directly
It broadcasts platform, open up direct broadcasting room, and be broadcast live in the direct broadcasting room, for example, media and movable live streaming, game live streaming, show field are straight
It broadcasts, social live streaming, etc.；On the other hand, spectators user can request of loading live streaming platform the direct broadcasting room, viewing live streaming.
Wherein, which can be natural person, or legal person (such as enterprises and institutions) can also be virtual
User's (such as camera for some landscape scenic spot to be broadcast live), the present embodiment is without restriction to this.
And spectators user can be user's (can be indicated with user account) of registration, or nonregistered (NR) user
(can be indicated with device identification), the present embodiment is without restriction to this.
In addition, direct broadcasting room can refer to execute the context of live streaming operation, room address (such as URL is generally comprised
The information such as (Uniform Resource Locator, uniform resource locator), room title, room number, room brief introduction.
The time that spectators user spends in live streaming platform, it is shown that liveness of the spectators user in platform.Certainly, with difference
Time span (such as one week, one month) investigated, spectators' the activity of the user may difference.Spectators are investigated to use
The time span of the liveness at family is called live streaming life cycle.
S102, the live streaming watching behavior number generated when spectators user watches and being broadcast live in the live streaming life cycle is obtained
According to.
Spectators user can generate a variety of live streaming watching behavior data when watching live streaming, such as subscribe to or unsubscribe master
It broadcasts, send out barrage, object etc. of giving gifts, these watching behavior data are the important indicators of evaluation live streaming watching behavior.Certainly, viewing is straight
Broadcast the viewing duration itself generated, and the important indicator of live streaming watching behavior.
Wherein, it may include a variety of sources that the behavioral data of spectators user is a variety of.For example spectators user passes through mobile terminal
Access live streaming platform or spectators user are accessed by computer end is broadcast live platform.In view of the standard of spectators' subscriber lifecycle statistics
True property, thus only using spectators user by mobile terminal accessing be broadcast live data that platform generates as spectators' behavioral data come
Source.Spectators user accesses the data that live streaming platform generates by computer end and is not considered.
The viewing duration for obtaining the spectators user of mobile terminal record, as in live streaming life cycle spectators user's viewing it is straight
The live streaming watching behavior data that sowing time generates.
S103, it the live streaming watching behavior data is input in preset live streaming life cycle model handles, obtain
The life cycle label of liveness of the spectators user in the live streaming life cycle must be characterized.
The viewing duration that will acquire subscribes to or unsubscribes main broadcaster, sends out barrage, the live streaming watching behavior data such as object of giving gifts
The data value is input in trained live streaming life cycle model by the data value being quantified as under identical dimensional
Reason, live streaming life cycle model can generate liveness of the spectators user in current live life cycle according to the data value of input
Life cycle label.Such as, when data value embodiment spectators' the activity of the user of input is high, trained live streaming life cycle
It is handled in model, the high life cycle label of characterization spectators' liveness should be exported.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are obtained
The life cycle label of liveness of the spectators user in the live streaming life cycle must be characterized.It solves according to user's
Hobby is broadcast live the not strong problem of push bring specific aim, realizes the live streaming watching behavior number by analyzing spectators user
According to obtain the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle.
Embodiment two
Fig. 2 is a kind of flow chart of the processing method of life cycle provided by Embodiment 2 of the present invention.The present embodiment be
The refinement carried out on the basis of embodiment one essentially describes acquisition spectators user in the live streaming life cycle and watches live streaming
When the specific method of live streaming watching behavior data that generates.With reference to Fig. 2, this method is specifically included:
S201, live streaming life cycle is determined.
S202, statistics the spectators user in the live streaming life cycle watch the live streaming watching behavior number generated when live streaming
According to the watching behavior data include viewing duration.
General spectators user is larger with the viewing duration diversity ratio of nonworkdays on weekdays, therefore, in order in general feelings
(non-festivals or holidays) fluctuations for covering viewing duration as far as possible under condition are usually the live streaming of unit measurement spectators user with " week " more
Life cycle.
It is that unit counts the live streaming watching behavior data generated when spectators user watches and being broadcast live with " week ".
Spectators user can generate a variety of live streaming watching behavior data when watching live streaming, such as subscribe to or unsubscribe master
It broadcasts, send out barrage, object of giving gifts etc., it can be added according to certain weight, obtain final live streaming watching behavior data.
But subscribe to or unsubscribe main broadcaster, send out barrage, the behaviors such as object of giving gifts are accustomed to dependent on the viewing of spectators user
And subjective desire, the judgement for being included in life cycle can introduce uncertain factor, therefore will can also only watch duration conduct
Watching behavior data are broadcast live.
Viewing duration in S203, multiple live streaming life cycles that add up, obtains viewing total duration.
With current " week " for starting point, multiple viewing durations that spectators user in the periods is broadcast live, total as viewing before adding up
Duration.
S204, judge whether the viewing total duration is more than preset duration threshold value.If so, executing S205.
Preset duration threshold value is the threshold value whether being lost for distinguishing spectators user.
If viewing total duration is not above preset duration threshold value, then it is assumed that the live streaming watching behavior data of the spectators user
In vain.Live streaming watching behavior data invalid refers to that the live streaming watching behavior data do not need to be input to preset live streaming life cycle
It is handled in model, it is understood that be invalid result by the live streaming watching behavior data input model.Usually live streaming
The spectators user of watching behavior data invalid can be used as loss user and handle.
Generally preset duration threshold value can be set as surrounding, even spectators user does not see in surrounding before
It has seen being broadcast live, it is believed that be the spectators user that the spectators user is live streaming platform loss, the spectators user of loss can be no longer
Judged by the way that life cycle is broadcast live.
For be lost user, can row live broadcast service processing, such as: pass through provide gift bag behavior attract spectators user return
Return.Business processing no longer can also be broadcast live to user.
S205, determine that the live streaming watching behavior data are effective.
If watching total duration is more than preset duration threshold value, then it is assumed that the live streaming watching behavior data of the spectators user have
Effect.Live streaming watching behavior data effectively refer to that the live streaming watching behavior data need to be input to preset live streaming life cycle model
In handled.
S206, it the live streaming watching behavior data is input in preset live streaming life cycle model handles, obtain
The life cycle label of liveness of the spectators user in the live streaming life cycle must be characterized.
Effective live streaming behavior viewing data are input in trained live streaming life cycle model and are handled,
Life cycle model, which is broadcast live, to generate liveness of the spectators user in current live life cycle according to the data value of input
Life cycle label.
In one example, life cycle label includes following at least one: trough period, safe period, peak period.Low ebb
The liveness of phase is lower than the liveness in safe period, and the liveness in safe period is lower than the liveness of peak period.
S207, business processing is broadcast live to the spectators user according to the life cycle label.
The mode for formulating each life cycle label corresponding live broadcast service processing in advance, when the life of a spectators user
After life period label is determined, handled according to the mode that pre-set live broadcast service is handled.Such as: to trough period label
Spectators user can execute live streaming and recall behavior, such as attract spectators user to return by providing gift bag behavior.To safe period
The spectators user of label can push its interested direct broadcasting room, main broadcaster or activity to it.
By being directed to different life label, carry out striving the live broadcast service processing to property.Reduce no purpose
Live broadcast service handles (especially message push) and decreases the resources such as the bandwidth of server end to reduce client's end pressure
Consumption.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are obtained
The life cycle label of liveness of the spectators user in the live streaming life cycle must be characterized.It solves according to user's
Hobby is broadcast live the not strong problem of push bring specific aim, realizes the live streaming watching behavior number by analyzing spectators user
According to obtain the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle, Jin Ergen
Different live broadcast service processing is carried out according to different life cycle labels.To which platform liveness is being broadcast live in discovery user in time
Variation more cannot targetedly determine the business processing mode to user for the variation of user activity.
Embodiment three
Fig. 3 is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention three provides.The skill of the present embodiment
Art scheme is optionally suitable for training live streaming life cycle model, the live streaming watching behavior number for generating to spectators user
According to being analyzed, and then determine spectators' the activity of the user.This method can be executed by a kind of processing unit of life cycle,
The device can be realized by the mode of software and/or software, be generally disposed in electronic equipment.With reference to Fig. 1, this method is specifically wrapped
It includes:
S301, live streaming life cycle is determined.
Live streaming life cycle at this time is for training live streaming life cycle model.Therefore, live streaming life cycle at this time is answered
When identical as the live streaming life cycle duration in S101 and/or S201.
Meanwhile in order to enable period for covering of life cycle is comprehensive, spectators' user data in 1 year can be carried out
Acquisition, avoids data fluctuations caused by winter and summer vacation.
The live streaming watching behavior number that S302, acquisition spectators user generate when watching live streaming in the live streaming life cycle
According to.
Live streaming watching behavior data at this time are for training live streaming life cycle model.Therefore, row is watched in live streaming at this time
It should be identical as the live streaming watching behavior data type in S102 and/or S202 for data.Especially, obtain what mobile terminal generated
The viewing duration of spectators user, as the live streaming watching behavior number generated when spectators user watches and being broadcast live in live streaming life cycle
According to.
S303, life cycle label is determined, spectators user described in the life cycle tag characterization is in the live streaming life
Liveness in period.
The viewing duration that will acquire subscribes to or unsubscribes main broadcaster, sends out barrage, the live streaming watching behavior data such as object of giving gifts
The data value being quantified as under identical dimensional.At this point, determining that the mode of live streaming life cycle label can be set as: when data value is super
When crossing a certain threshold value, it is considered as spectators' the activity of the user height, live streaming life cycle label is peak period；When data value is lower than
When a certain threshold value, it is low to be considered as spectators' the activity of the user, and live streaming life cycle label is trough period.Certainly, life is determined
The mode of period label can also be set are as follows: by the data value of the data value of this live streaming life cycle and upper live streaming life cycle
It is compared, if the amplitude of variation of data value is in a certain threshold range, then it is assumed that spectators' the activity of the user is general, live streaming
Life cycle label is safe period；If data value increasing degree is more than a certain threshold range, it is considered as enlivening for the spectators user
Degree is high, and live streaming life cycle label is peak period；If it is more than a certain threshold range that data value, which reduces amplitude, it is considered as spectators use
The liveness at family is low, and live streaming life cycle label is trough period.
Spectators' the activity of the user individually can be characterized with the viewing duration of spectators user.According to acquisition mobile terminal record
Spectators user viewing duration, determine liveness of the spectators user in live streaming life cycle, and then determine spectators user
Life cycle label is broadcast live.At this point, determining that the mode of live streaming life cycle label can be set as: when the viewing of spectators user
When length is more than a certain threshold value, it is considered as spectators' the activity of the user height, live streaming life cycle label is peak period；When spectators use
When the viewing duration at family is lower than a certain threshold value, it is low to be considered as spectators' the activity of the user, and live streaming life cycle label is low ebb
Phase.Certainly, determine that the mode of life cycle label can also be set are as follows: by the viewing duration of this live streaming life cycle and it is upper always
The viewing duration for broadcasting life cycle is compared, if the amplitude of variation of viewing duration is in a certain threshold range, then it is assumed that the sight
Many the activity of the user are general, and live streaming life cycle label is safe period；If watching duration increasing degree is more than a certain threshold value model
It encloses, is considered as spectators' the activity of the user height, live streaming life cycle label is peak period；If viewing duration reduces amplitude
A certain threshold range, it is low to be considered as spectators' the activity of the user, and live streaming life cycle label is trough period.
S304, life cycle model is broadcast live according to the live streaming watching behavior data and the life cycle label training.
The corresponding live streaming life cycle label of the live streaming watching behavior data in life cycle will be broadcast live and be associated with preservation
For a sample data, the corresponding sample data of multiple spectators users in live streaming life cycle, formation base training sample are obtained
Collection, carries out multiple stochastical sampling, sampled result is as training sample set from grounding sample set.From training sample concentrate with
Machine extracts 75% and is used as training set, is left 25% and is used as test set.
By training set training live streaming life cycle model, when live streaming life cycle model meets the inspection result of test set
Afterwards, it is believed that training live streaming life cycle model is completed.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input to machine learning model, evaluation of life cycle is formed with training
Model.Solve the problems, such as that the hobby according to user is broadcast live that push bring specific aim not strong, realizes by the model
By analyzing the live streaming watching behavior data of spectators user, to obtain the characterization spectators user in the live streaming life cycle
In liveness life cycle label.
Example IV
Fig. 4 A is a kind of flow chart of the processing method for life cycle that the embodiment of the present invention four provides.The present embodiment be
The refinement carried out on the basis of embodiment three essentially describes the specific method of determining life cycle label.With reference to Fig. 4 A, the party
Method specifically includes:
S401, live streaming life cycle is determined.
The live streaming watching behavior number that S402, acquisition spectators user generate when watching live streaming in the live streaming life cycle
According to.
Live streaming watching behavior data of multiple spectators users in live streaming life cycle are acquired, multiple spectators are especially acquired
Viewing duration of the user in live streaming life cycle.
S403, multiple duration ranges are divided to the viewing duration, each duration range has the life of characterization liveness
Period label.
Viewing duration can be expressed as continuous numerical value in a section, it can be understood as the length of viewing time and active
Degree is positively correlated.
Fig. 4 B is the relation schematic diagram of a kind of duration range and liveness that the embodiment of the present invention four provides.With reference to Fig. 4 B,
The corresponding relationship for indicating duration range and liveness can be constructed accordingly:
Firstly, the average viewing duration based on viewing all spectators users of duration calculation in the live streaming life cycle,
It is denoted as μ；Secondly, the viewing mark based on the averagely viewing all spectators users in duration calculation family in the live streaming life cycle
Quasi- difference σ；Finally, duration range is arranged in conjunction with average viewing duration and the standard deviation.
It generally, is the inflection point of normpdf at μ ± σ, it is contemplated that actual conditions, as it is desirable that examining earlier
Feel that user has the sign into trough period, therefore divided according to ± 0.8 σ of μ, μ -0.8 σ is denoted as the first duration range, μ -0.8
+ 0.8 σ of σ~μ is denoted as the second duration range, and+0.8 σ of μ is denoted as third duration range.
Average viewing duration of a certain spectators user in live streaming life cycle is denoted as x.X < μ -0.8 σ is denoted as low ebb
Phase；+ 0.8 σ of μ -0.8 σ≤x≤μ is denoted as safe period；+ 0.8 σ of x > μ is denoted as peak period.
S404, duration range belonging to the viewing duration is determined.
Judge a certain spectators user live streaming life cycle in viewing duration described in correspond to the first duration range, second
Duration range or third duration range.
S405, the corresponding life cycle of the duration range is marked in the live streaming life cycle to the spectators user
Label.
According to the corresponding life cycle label of duration range, the duration model is marked in live streaming life cycle to spectators user
Enclose corresponding life cycle label.
S406, life cycle model is broadcast live according to the live streaming watching behavior data and the life cycle label training.
Live streaming life cycle mould can be generally constructed by support vector machines (Support Vector Machine, SVM)
Type.Certainly, live streaming life cycle model, which can also be through the deep learning in neural network, completes, such as convolutional neural networks
(Convolutional Neural Networks, CNN) or shot and long term memory network (Long Short-Term Memory,
LSTM)。
In a specific example, live streaming life cycle model, the library in python can be constructed by random forest
Sklearn has function RandomForestClassifier that can train Random Forest model, which is exactly straight
Broadcast life cycle model.
Optionally, multiple training sample sets are inputted into multiple weak learners respectively, to be trained to multiple weak learners；
According to the different weights to multiple weak learner settings, summation is weighted to multiple weak learners and obtains strong learner；It will be strong
Learner is as live streaming life cycle model.
Optionally, multiple training sample sets are inputted into multiple weak learners respectively, to be trained to multiple weak learners；
All weak learners that training obtains are integrated by aggregation policy, obtain strong learner as live streaming life cycle mould
Type.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input to machine learning model, evaluation of life cycle is formed with training
Model.Solve the problems, such as that the hobby according to user is broadcast live that push bring specific aim not strong, realizes by the model
By analyzing the live streaming watching behavior data of spectators user, to obtain the characterization spectators user in the live streaming life cycle
In liveness life cycle label.
Embodiment five
Fig. 5 is a kind of structure chart of the processing unit for life cycle that the embodiment of the present invention five provides.It include: the live streaming period
Determining module 51, behavioral data obtain module 52 and period label obtains module 53.Wherein:
Period determination module 51 is broadcast live, for determining that live streaming life cycle, the live streaming life cycle are seen for characterizing
One section of preset time period of many user activities；
Behavioral data obtains module 52, generates for obtaining when spectators user watches live streaming in the live streaming life cycle
Live streaming watching behavior data；
Period label obtains module 53, for the live streaming watching behavior data to be input to preset live streaming life cycle
It is handled in model, obtains the life cycle mark for characterizing liveness of the spectators user in the live streaming life cycle
Label.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are obtained
The life cycle label of liveness of the spectators user in the live streaming life cycle must be characterized.It solves according to user's
Hobby is broadcast live the not strong problem of push bring specific aim, realizes the live streaming watching behavior number by analyzing spectators user
According to obtain the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle.
On the basis of the above embodiments, life cycle label includes following at least one:
Trough period, safe period, peak period；
Wherein, the liveness of the trough period is lower than the liveness in the safe period, and the liveness in the safe period is lower than
The liveness of the peak period.
On the basis of the above embodiments, further include Service Processing Module, be used for:
Business processing is broadcast live to the spectators user according to the life cycle label.
On the basis of the above embodiments, Service Processing Module is also used to:
Live streaming is executed to the spectators user and recalls behavior.
On the basis of the above embodiments, behavioral data obtains module and is also used to:
It counts the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming, it is described
Watching behavior data include viewing duration；
Viewing duration in multiple live streaming life cycles that add up, obtains viewing total duration；
Judge whether the viewing total duration is more than preset duration threshold value；
If so, determining that the live streaming watching behavior data are effective；
If not, it is determined that the live streaming watching behavior data invalid.
A kind of processing unit of life cycle provided in this embodiment can be used for executing above-described embodiment one or embodiment two
The processing method of the life cycle of offer has corresponding function and beneficial effect.
Embodiment six
Fig. 6 is a kind of structure chart of the processing unit for life cycle that the embodiment of the present invention six provides.It include: life cycle
Determining module 61, behavioral data acquisition module 62, period label determining module 63 and periodic model training module 64.Wherein:
Life cycle determining module 61, for determining that live streaming life cycle, the live streaming life cycle are seen for characterizing
One section of preset time period of many user activities；
Behavioral data acquisition module 62, generation when watching live streaming for acquiring spectators user in the live streaming life cycle
Live streaming watching behavior data；
Period label determining module 63, for determining life cycle label, spectators described in the life cycle tag characterization
Liveness of the user in the live streaming life cycle；
Periodic model training module 64, for according to the live streaming watching behavior data and the life cycle label training
Life cycle model is broadcast live.
The embodiment of the present invention watches the live streaming generated when live streaming by obtaining the spectators user in the live streaming life cycle
Watching behavior data；The live streaming watching behavior data are input to machine learning model, evaluation of life cycle is formed with training
Model.Solve the problems, such as that the hobby according to user is broadcast live that push bring specific aim not strong, realizes by the model
By analyzing the live streaming watching behavior data of spectators user, to obtain the characterization spectators user in the live streaming life cycle
In liveness life cycle label.
On the basis of the above embodiments, period label determining module is also used to:
Multiple duration ranges are divided to the viewing duration, each duration range has the life cycle mark of characterization liveness
Label；
Determine duration range belonging to the viewing duration；
The corresponding life cycle label of the duration range is marked in the live streaming life cycle to the spectators user.
On the basis of the above embodiments, multiple duration ranges are divided to the viewing duration, comprising:
Average viewing duration based on viewing all spectators users of duration calculation in the live streaming life cycle；
Viewing standard based on the averagely viewing all spectators users in duration calculation family in the live streaming life cycle
Difference；
In conjunction with the averagely viewing duration and the standard deviation, duration range is set.
On the basis of the above embodiments, life cycle label includes following at least one:
Trough period, safe period, peak period；
Wherein, the liveness of the trough period is lower than the liveness in the safe period, and the liveness in the safe period is lower than
The liveness of the peak period.
A kind of processing unit of life cycle provided in this embodiment can be used for executing above-described embodiment three or example IV
The processing method of the life cycle of offer has corresponding function and beneficial effect.
Embodiment seven
Fig. 7 is the structural schematic diagram for a kind of electronic equipment that the embodiment of the present invention seven provides.As shown in fig. 7, the electronics is set
Standby includes processor 70, memory 71, communication module 72, input unit 73 and output device 74；Processor 70 in electronic equipment
Quantity can be one or more, in Fig. 7 by taking a processor 70 as an example；Processor 70, memory 71 in electronic equipment,
Communication module 72, input unit 73 and output device 74 can be connected by bus or other modes, to be connected by bus in Fig. 7
It is connected in example.
Memory 71 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 module of the processing method of one of the present embodiment life cycle is (for example, a kind of life cycle
Live streaming period determination module 51, behavioral data in processing unit obtain module 52 and period label obtains module 53, alternatively, one
Life cycle determining module 61, behavioral data acquisition module 62, period label in the processing unit of kind life cycle determine mould
Block 63 and periodic model training module 64).Processor 70 by operation be stored in memory 71 software program, instruction and
Module realizes a kind of place of above-mentioned life cycle thereby executing the various function application and data processing of electronic equipment
Reason method.
Memory 71 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 according to electronic equipment
Deng.In addition, memory 71 may include high-speed random access memory, it can also include nonvolatile memory, for example, at least
One disk memory, flush memory device or other non-volatile solid state memory parts.In some instances, memory 71 can
It further comprise the memory remotely located relative to processor 70, these remote memories can pass through network connection to electronics
Equipment.The example of above-mentioned network includes but is not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
Communication module 72 for establishing connection with display screen, and realizes the data interaction with display screen.Input unit 73 can
Number for receiving input or character information, and generate key related with the user setting of electronic equipment and function control
Signal input.
The processing for the life cycle that any embodiment of the present invention provides can be performed in a kind of electronic equipment provided in this embodiment
Method, specific corresponding function and beneficial effect.
Embodiment eight
The embodiment of the present invention eight also provides a kind of storage medium comprising computer executable instructions, and the computer can be held
Row is instructed when being executed by computer processor for executing a kind of processing method of life cycle, this method comprises:
Determine live streaming life cycle, the live streaming life cycle be for characterize the one of spectators' user activity section it is default when
Between section；
It obtains the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming；
The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, are characterized
The life cycle label of liveness of the spectators user in the live streaming life cycle.
Or；
Determine live streaming life cycle, the live streaming life cycle be for characterize the one of spectators' user activity section it is default when
Between section；
The live streaming watching behavior data that acquisition spectators user generates when watching live streaming in the live streaming life cycle；
Determine life cycle label, spectators user described in the life cycle tag characterization is in the live streaming life cycle
Liveness；
Life cycle model is broadcast live according to the live streaming watching behavior data and the life cycle label training.
Certainly, a kind of storage medium comprising computer executable instructions, computer provided by the embodiment of the present invention
The method operation that executable instruction is not limited to the described above, can also be performed life cycle provided by any embodiment of the present invention
Processing method 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 used so that a calculatings electromechanics
Sub- equipment (can be personal computer, server or network electronic devices etc.) executes described in each embodiment of the present invention
Method.
It is worth noting that, in the embodiment of the processing unit of above-mentioned life cycle, included each unit and module
It is only divided according to the functional logic, but is not limited to the above division, as long as corresponding functions can be realized；
In addition, 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 (13)
1. a kind of processing method of life cycle characterized by comprising
Determine live streaming life cycle, the live streaming life cycle is for characterizing the one of spectators' user activity section of preset time
Section；
It obtains the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming；
The live streaming watching behavior data are input in preset live streaming life cycle model and are handled, characterized described in
The life cycle label of liveness of the spectators user in the live streaming life cycle.
2. the method according to claim 1, wherein the life cycle label includes following at least one:
Trough period, safe period, peak period；
Wherein, the liveness of the trough period is lower than the liveness in the safe period, and the liveness in the safe period is lower than described
The liveness of peak period.
3. the method according to claim 1, wherein further include:
Business processing is broadcast live to the spectators user according to the life cycle label.
4. according to the method described in claim 3, it is characterized in that, described use the spectators according to the life cycle label
Family is broadcast live business processing, comprising:
Live streaming is executed to the spectators user and recalls behavior.
5. the method according to claim 1, wherein acquisition spectators user in the live streaming life cycle
The live streaming watching behavior data generated when viewing live streaming, comprising:
It counts the spectators user in the live streaming life cycle and watches the live streaming watching behavior data generated when live streaming, the viewing
Behavioral data includes viewing duration；
Viewing duration in multiple live streaming life cycles that add up, obtains viewing total duration；
Judge whether the viewing total duration is more than preset duration threshold value；
If so, determining that the live streaming watching behavior data are effective；
If not, it is determined that the live streaming watching behavior data invalid.
6. a kind of processing method of life cycle characterized by comprising
Determine live streaming life cycle, the live streaming life cycle is for characterizing the one of spectators' user activity section of preset time
Section；
The live streaming watching behavior data that acquisition spectators user generates when watching live streaming in the live streaming life cycle；
Determine life cycle label, work of the spectators user described in the life cycle tag characterization in the live streaming life cycle
Jerk；
Life cycle model is broadcast live according to the live streaming watching behavior data and the life cycle label training.
7. according to the method described in claim 6, it is characterized in that, the live streaming watching behavior data include viewing duration, institute
Stating determining life cycle label includes:
Multiple duration ranges are divided to the viewing duration, each duration range has the life cycle label of characterization liveness；
Determine duration range belonging to the viewing duration；
The corresponding life cycle label of the duration range is marked in the live streaming life cycle to the spectators user.
8. the method according to the description of claim 7 is characterized in that described divide multiple duration ranges to the viewing duration,
Include:
Average viewing duration based on viewing all spectators users of duration calculation in the live streaming life cycle；
Viewing standard deviation based on the averagely viewing all spectators users in duration calculation family in the live streaming life cycle；
In conjunction with the averagely viewing duration and the standard deviation, duration range is set.
9. according to the method any in claim 6,7 or 8, which is characterized in that the life cycle label includes as follows
At least one:
Trough period, safe period, peak period；
Wherein, the liveness of the trough period is lower than the liveness in the safe period, and the liveness in the safe period is lower than described
The liveness of peak period.
10. a kind of processing unit of life cycle characterized by comprising
Period determination module is broadcast live, for determining live streaming life cycle, the live streaming life cycle is for characterizing spectators user
One section of preset time period of liveness；
Behavioral data obtains module, watches the live streaming generated when live streaming for obtaining the spectators user in the live streaming life cycle
Watching behavior data；
Period label obtains module, for the live streaming watching behavior data to be input in preset live streaming life cycle model
It is handled, obtains the life cycle label for characterizing liveness of the spectators user in the live streaming life cycle.
11. a kind of processing unit of life cycle characterized by comprising
Life cycle determining module, for determining live streaming life cycle, the live streaming life cycle is for characterizing spectators user
One section of preset time period of liveness；
Behavioral data acquisition module, for acquiring the live streaming generated when spectators user watches live streaming in the live streaming life cycle
Watching behavior data；
Period label determining module, for determining life cycle label, spectators user described in the life cycle tag characterization exists
Liveness in the live streaming life cycle；
Periodic model training module, for being given birth to according to the live streaming watching behavior data and life cycle label training live streaming
Order periodic model.
12. a kind of electronic equipment characterized by comprising
One or more processors；
Memory, 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
A kind of now processing method of life cycle as described in claim 1-9 is any.
13. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the program is by processor
A kind of processing method of life cycle as described in claim 1-9 is any is realized when execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910000641.XA CN109729376B (en) | 2019-01-02 | 2019-01-02 | Life cycle processing method, life cycle processing device, life cycle processing equipment and life cycle processing storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910000641.XA CN109729376B (en) | 2019-01-02 | 2019-01-02 | Life cycle processing method, life cycle processing device, life cycle processing equipment and life cycle processing storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109729376A true CN109729376A (en) | 2019-05-07 |
| CN109729376B CN109729376B (en) | 2021-12-14 |
Family
ID=66298736
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910000641.XA Active CN109729376B (en) | 2019-01-02 | 2019-01-02 | Life cycle processing method, life cycle processing device, life cycle processing equipment and life cycle processing storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109729376B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569910A (en) * | 2019-09-10 | 2019-12-13 | 广州虎牙科技有限公司 | method, device and equipment for processing live broadcast cycle and storage medium |
| CN113259697A (en) * | 2021-05-12 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Live broadcast state notification method, related device, equipment and storage medium |
| CN113422978A (en) * | 2021-07-14 | 2021-09-21 | 北京达佳互联信息技术有限公司 | Training method and device of dormancy early warning model and dormancy early warning method and device |
| CN113743991A (en) * | 2021-09-03 | 2021-12-03 | 上海幻电信息科技有限公司 | Life cycle value prediction method and device |
| CN114727121A (en) * | 2021-01-06 | 2022-07-08 | 厦门蝉羽网络科技有限公司 | Method, medium, system and equipment for acquiring average dwell time of audiences in live broadcast room |
| CN114742569A (en) * | 2021-01-08 | 2022-07-12 | 广州视源电子科技股份有限公司 | User life stage prediction method and device, computer equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160232131A1 (en) * | 2015-02-11 | 2016-08-11 | Google Inc. | Methods, systems, and media for producing sensory outputs correlated with relevant information |
| CN106228403A (en) * | 2016-07-20 | 2016-12-14 | 武汉斗鱼网络科技有限公司 | A kind of user based on step analysis algorithm is worth methods of marking and system |
| CN107784390A (en) * | 2017-10-19 | 2018-03-09 | 北京京东尚科信息技术有限公司 | Recognition methods, device, electronic equipment and the storage medium of subscriber lifecycle |
| CN108108912A (en) * | 2018-01-10 | 2018-06-01 | 百度在线网络技术（北京）有限公司 | Method of discrimination, device, server and the storage medium of interactive low quality user |
| CN108270842A (en) * | 2017-06-09 | 2018-07-10 | 广州市动景计算机科技有限公司 | Push method, system and the server of equity task |
| CN108647293A (en) * | 2018-05-07 | 2018-10-12 | 广州虎牙信息科技有限公司 | Video recommendation method, device, storage medium and server |
-
2019
- 2019-01-02 CN CN201910000641.XA patent/CN109729376B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160232131A1 (en) * | 2015-02-11 | 2016-08-11 | Google Inc. | Methods, systems, and media for producing sensory outputs correlated with relevant information |
| CN106228403A (en) * | 2016-07-20 | 2016-12-14 | 武汉斗鱼网络科技有限公司 | A kind of user based on step analysis algorithm is worth methods of marking and system |
| CN108270842A (en) * | 2017-06-09 | 2018-07-10 | 广州市动景计算机科技有限公司 | Push method, system and the server of equity task |
| CN107784390A (en) * | 2017-10-19 | 2018-03-09 | 北京京东尚科信息技术有限公司 | Recognition methods, device, electronic equipment and the storage medium of subscriber lifecycle |
| CN108108912A (en) * | 2018-01-10 | 2018-06-01 | 百度在线网络技术（北京）有限公司 | Method of discrimination, device, server and the storage medium of interactive low quality user |
| CN108647293A (en) * | 2018-05-07 | 2018-10-12 | 广州虎牙信息科技有限公司 | Video recommendation method, device, storage medium and server |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110569910A (en) * | 2019-09-10 | 2019-12-13 | 广州虎牙科技有限公司 | method, device and equipment for processing live broadcast cycle and storage medium |
| CN114727121A (en) * | 2021-01-06 | 2022-07-08 | 厦门蝉羽网络科技有限公司 | Method, medium, system and equipment for acquiring average dwell time of audiences in live broadcast room |
| CN114727121B (en) * | 2021-01-06 | 2024-04-23 | 厦门蝉羽网络科技有限公司 | Method, medium, system and equipment for acquiring average stay time of audience in live broadcasting room |
| CN114742569A (en) * | 2021-01-08 | 2022-07-12 | 广州视源电子科技股份有限公司 | User life stage prediction method and device, computer equipment and storage medium |
| CN113259697A (en) * | 2021-05-12 | 2021-08-13 | 腾讯科技（深圳）有限公司 | Live broadcast state notification method, related device, equipment and storage medium |
| CN113259697B (en) * | 2021-05-12 | 2022-04-08 | 腾讯科技（深圳）有限公司 | Live broadcast state notification method, related device, equipment and storage medium |
| CN113422978A (en) * | 2021-07-14 | 2021-09-21 | 北京达佳互联信息技术有限公司 | Training method and device of dormancy early warning model and dormancy early warning method and device |
| CN113422978B (en) * | 2021-07-14 | 2022-08-26 | 北京达佳互联信息技术有限公司 | Training method and device of dormancy early warning model and dormancy early warning method and device |
| CN113743991A (en) * | 2021-09-03 | 2021-12-03 | 上海幻电信息科技有限公司 | Life cycle value prediction method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109729376B (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109729376A (en) | A kind of processing method of life cycle, device, equipment and storage medium | |
| Yang et al. | Revealing pathways from payments for ecosystem services to socioeconomic outcomes | |
| Freeman et al. | Social media campaigns that make a difference: what can public health learn from the corporate sector and other social change marketers | |
| CN109803176B (en) | Auditing monitoring method and device, electronic equipment and storage medium | |
| CN110418153B (en) | Watermark adding method, device, equipment and storage medium | |
| CN109634605B (en) | Simulation data generation method and system for web front end | |
| Yu et al. | Analyzing netizens’ view and reply behaviors on the forum | |
| CN109905738B (en) | Video advertisement abnormal display monitoring method and device, storage medium and electronic equipment | |
| Yuan et al. | The duality of structure in China's national television market: A network analysis of audience behavior | |
| CN110490644A (en) | A kind of online questionnaire method of investigation and study, device and storage medium | |
| CN108521405A (en) | A kind of risk management and control method, device and storage medium | |
| CN109034867B (en) | Click traffic detection method and device and storage medium | |
| CN105872006A (en) | Appointment reminding system and appointment reminding method | |
| CN110363427A (en) | Model quality evaluation method and apparatus | |
| CN109829379A (en) | Information processing method, device, server and storage medium | |
| CN110166789A (en) | Monitor method, computer equipment and the readable storage medium storing program for executing of net cast sensitive information | |
| CN108320168A (en) | A kind of data analysing method and device | |
| CN111581521A (en) | Group member recommendation method, device, server, storage medium and system | |
| CN110237536A (en) | Personalized game service providing method and device, electronic equipment and storage medium | |
| CN106204164A (en) | Method of testing that web advertisement presents and device | |
| CN112019875B (en) | Learning behavior monitoring method and device for online live broadcast and live broadcast platform | |
| CN109660871A (en) | A kind of barrage Role Information determines method, device and equipment | |
| CN110233840A (en) | A kind of method for processing business, device, equipment and storage medium | |
| CN109005423A (en) | A kind of video broadcasting method and device | |
| CN113407831A (en) | Course recommendation method and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |