CN107370614A - Network user active degree appraisal procedure and Forecasting Methodology - Google Patents
Network user active degree appraisal procedure and Forecasting Methodology Download PDFInfo
- Publication number
- CN107370614A CN107370614A CN201610318215.7A CN201610318215A CN107370614A CN 107370614 A CN107370614 A CN 107370614A CN 201610318215 A CN201610318215 A CN 201610318215A CN 107370614 A CN107370614 A CN 107370614A
- Authority
- CN
- China
- Prior art keywords
- value
- time
- activity
- cycle
- time cycle
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
Abstract
The invention discloses the method and network user active degree Forecasting Methodology that a kind of liveness to the network user within the time cycle is assessed.Liveness appraisal procedure includes:Collect activity data of the network user within the time cycle, the activity data is identified and including multiple activity attributes by the unit interval;The activity data is identified by the time cycle, obtains active evaluation data;The value of each activity attributes in the active evaluation data is normalized;And according to the value by normalized each activity attributes, determine the network user active degree value.
Description
Technical field
The present invention relates to internet arena, is carried out more particularly to a kind of liveness to the network user within the time cycle
The method and network user active degree Forecasting Methodology of assessment.
Background technology
In electric business field, typically only any active ues directly or indirectly could create value for website.Any active ues
Definition is varied, and the user for having a key operations or behavior reaches when some is required typically is defined as into any active ues.Each
Website all can define any active ues according to the product feature of itself, and use it for the customer volume that analyzing web site is really grasped.
User activity can help web analytics user data, carry out further user and the action such as keep, promote.Than
Such as, electric business website can take the mode for distributing reward voucher, excitation user's weight after a collection of less active user is identified
Access its website again, and and then the behavior such as facilitate user to be placed an order, buy.And for example, a series of high liveness are being identified
After user, some franchise or special signs of this crowd of user can be assigned, to lift the sense of ownership of this part population.In a word, it is right
Correctly divided in website any active ues it is extremely important, and electric business formulate publicity and marketing strategy important references.
In traditional user activity computational methods, the normally only daily line duration of counting user.For example certain chat is soft
The judgement that part enlivens day for user has following rule:The same day (0:00-23:59) use is 2 hours (and more than 2 hours), when
It is calculated to enliven day, is enlivened number of days for it and is accumulated 1 day;The same day (0:00-23:59) use was at 0.5 hour to 2 hours, for its work
The number of days that jumps accumulates 0.5 day;The same day (0:00-23:59) using below 0.5 hour, number of days is not enlivened for its accumulation.
In electric business field, weigh has a Multiple factors using the liveness of user, including user's day unlatching rate, stay time,
Interactive frequency etc..Ideally, the number that an application starts daily is more, and the daily residence time that starts is longer, is answering
With interior more frequent with other users interaction, show that the liveness of the user is higher.
However, existing user activity computational methods, can only be generally according to the login frequency of user, residence time etc.
Index is counted, and it is artificial boundary is marked off to liveness, each user is corresponded into each different section.It is this
Method is not to be defined with data, but the liveness of user is defined with business.
In addition, conventional method is all based on the statistics of historical data, what is calculated is the current liveness state of user, and
The liveness in user's next cycle is not predicted.Reality electric business application in, it would be desirable to be frequently not current
Data, but gone to speculate the following possible behavior of user according to historical data.
The content of the invention
In order to solve the above-mentioned problems in the prior art, the present invention propose it is a kind of to the network user in the time cycle
The method and network user active degree Forecasting Methodology that interior liveness is assessed.
According to an aspect of the present invention, it is proposed that a kind of liveness to the network user within the time cycle is assessed
Method, including:Activity data of the network user within the time cycle is collected, when the activity data passes through unit
Between identify and including multiple activity attributes;The activity data is identified by the time cycle, obtains active evaluation
Data;The value of each activity attributes in the active evaluation data is normalized;And according to by normalized each
The value of individual activity attributes, determine the network user active degree value.
Preferably, the activity attributes select from the following:Login times, residence time, effectively descend odd number,
Collecting commodities number, searching times, evaluate number, get reward voucher number.
Preferably, the activity data is identified by the time cycle obtaining active evaluation data includes:By described in
The value of the respective activity attribute of the activity data corresponding to each unit interval in time cycle is separately summed, and is commented as activity
Estimate the value of the respective activity attribute of data.
Preferably, the value of the activity attributes X in the active evaluation data is normalized by below equation:
XNormal=AX·X/XMax, wherein, XNormalIt is the activity attributes X normalized value of process, XMaxIt is activity attributes X maximum
Value, AXIt is weight coefficient.
Preferably, determine that the network user active degree value includes according to by the value of normalized each activity attributes:
By will be added by the value of normalized each activity attributes, export the network user enlivens angle value.
Preferably, methods described also includes:Enlivened by what is obtained for multiple network users within multiple time cycles
Angle value, determine the liveness rank of the network user.
Preferably, according to the average value of the liveness Value Data obtained for multiple network users within multiple time cycles
And standard deviation, divide the liveness level range of the network user.
According to another aspect of the present invention, a kind of network user active degree Forecasting Methodology is additionally provided, including:Collection network
Activity data of the user within the cycle very first time, the activity data are identified and including multiple activities by the unit interval
Attribute, the very first time cycle arrest is in the previous day of current one time;The work is identified by the second time cycle
Dynamic data, obtain active evaluation data, the cycle very first time is the multiple of second time cycle;The activity is commented
The value for estimating each activity attributes in data is normalized;According to the value by normalized each activity attributes, net is determined
Angle value is enlivened in each second time cycle of the network user within the cycle very first time;According to each second time
Angle value is enlivened in cycle, liveness of the network user in next second time cycle is predicted.
Preferably, the activity attributes select from the following:Login times, residence time, effectively descend odd number,
Collecting commodities number, searching times, evaluate number, get reward voucher number.
Preferably, obtaining active evaluation data by second time cycle identifying the activity data includes:Will
The activity data corresponding to each unit interval in each second time cycle in the cycle very first time is mutually taken on service jobs
The value of dynamic attribute is separately summed, the value as the respective activity attribute of the active evaluation data of second time cycle.
Preferably, for each second time cycle in the cycle very first time, by below equation to the work
The dynamic value for assessing the activity attributes X in data is normalized:XNormal=AX·X/XMax, wherein, XNormalIt is activity attributes X
The normalized value of process, XMaxIt is activity attributes X maximum, AXIt is weight coefficient.
Preferably, determine the network user in the cycle very first time according to by the value of normalized each activity attributes
Angle value of enlivening in interior each second time cycle includes:By will be added by the value of normalized each activity attributes,
Export in each second time cycle of the network user within the cycle very first time and enliven angle value.
Preferably, angle value is enlivened to the network user in next second time in each second time cycle
Liveness in cycle be predicted including:By enlivening angle value matched curve in each second time cycle;Utilize
The curve fitted, predict liveness of the network user in next second time cycle.
Preferably, angle value is enlivened to the network user in next second time in each second time cycle
Liveness in cycle be predicted including:Angle value is enlivened in each second time cycle, determines the network
The liveness rank of user;Calculate the liveness rank changing value in each second time cycle;When counting described each second
Between there is equal liveness rank changing value with last second time cycle in the cycle very first time in the cycle
All second time cycles ensuing second time cycle liveness rank changing value;By the statistical result
The liveness rank that the maximum liveness rank changing value of middle occurrence probability was defined as in next second time cycle becomes
Change value;The liveness rank of last the second time cycle in the cycle very first time and identified next
Liveness rank changing value in the individual time cycle, liveness level of the calculating network user in next second time cycle
Not.
Preferably, according to the flat of the liveness Value Data obtained for multiple network users within multiple second time cycles
Average and standard deviation, divide the liveness level range of the network user.
Preferably, the liveness rank changing value calculated in each second time cycle includes:By making the network user exist
Liveness rank in one the second time cycle subtracts liveness rank of the network user within previous second time cycle,
Liveness changing value of the derived grid user within second time cycle.
By using network user active degree appraisal procedure proposed by the invention and Forecasting Methodology, more adduction is not only able to
Reason and the liveness for assessing user exactly, but also can be to a certain extent to the network user in following active degree
It is predicted, contributes to the activity decision of electric business.
Brief description of the drawings
Fig. 1 shows commenting liveness of the network user within the time cycle according to an embodiment of the present
The flow chart for the method estimated;
Fig. 2 shows the flow chart of network user active degree Forecasting Methodology according to an embodiment of the present.
Embodiment
The present invention is specifically described below with reference to accompanying drawing.
First, referring to Fig. 1.Fig. 1 show according to an embodiment of the present to the network user within the time cycle
The flow chart of method 100 assessed of liveness.Methods described includes starting from step S110, collects the network user
Activity data within the time cycle, the activity data were identified and belonged to including multiple activities by the unit interval
Property.Next, in the step s 120, the activity data is identified by the time cycle, active evaluation data are obtained.So
Afterwards in step s 130, the value of each activity attributes in the active evaluation data is normalized.Finally, in step
In S140, according to the value by normalized each activity attributes, the network user active degree value is determined.
In order to assess liveness of the network user within a time cycle, it is necessary first to receive in step s 110
Collect activity data of the network user within a time cycle.The time cycle be according to task it needs to be determined that, can
To be for example, 1 year, half a year, a season, one month, one week etc..Activity data refers to specific user and server interaction
Caused data record, it can be collected from the following:System journal, user profile, user activity information etc..
In one embodiment, the activity data, and the activity of each unit interval are identified by the unit interval
Data can include multiple activity attributes.For example, it is day that can make the unit interval, and so described activity data is according to day
Come what is divided, i.e., by the unit interval " my god " identify the activity data.As an example, a certain user is shown in table 1 below
Activity data within the special time cycle (such as three days on the 4th -2 months on the 2nd 2 months).
Date/attribute | Login times | Residence time (second) | Effectively place an order | Collecting commodities | Searching times |
2 days 2 months | 5 | 3000 | 1 | 1 | 2 |
3 days 2 months | 1 | 600 | 0 | 2 | 2 |
4 days 2 months | 4 | 3500 | 2 | 0 | 1 |
Table 1
As it can be seen from table 1 according to the date, the activity data of 2 months -2 months this three days specific users on the 4th on the 2nd is listed.
In addition, the activity data gathered from system is probably numerous and diverse, can if necessary to which it is preferably used
In the form of the activity data of unit interval is expressed as into multiple activity attributes as needed, as shown in table 1.The activity attributes
Selected from least the following:Login times, residence time, effectively descend odd number, collecting commodities number, searching times, comment
Valency number, get reward voucher number.As for needing which activity attributes chosen, can be set as needed, and can be with
Add other activity attributes thereto at any time and/or remove and select attribute.
Next, in the step s 120, the activity data is changed to identify by the time cycle, by described
The activity data that time cycle identifies is also referred to as active evaluation data.
Specifically, if the time cycle is three days, the activity data in table 1 is marked by the time cycle
Can be with as shown in table 2 below during knowledge.
Date/attribute | Login times | Residence time (second) | Effectively place an order | Collecting commodities | Searching times |
4 days -2 months on the 2nd 2 months | 10 | 7100 | 3 | 3 | 5 |
Table 2
It can be seen in fig. 2 that by the time cycle come the activity data that identifies during, by the week time
In phase (that is, this three days 2 days 2 months to 4 days 2 months) each unit interval (that is, 2 days 2 months, 3 days 2 months, it is each in 4 days 2 months
My god) corresponding to the value of respective activity attribute of activity data be separately summed, as mutually taking on service jobs for obtained active evaluation data
The value of dynamic attribute.For example, 4 days -2 months on the 2nd 2 months this three days login times are respectively 5 times, 1 time and 4 times, then at 2 days to 22 months
In 4 this time cycles of the moon, login times are 5+1+4=10 times, and this is i.e. in the time cycle by three days to activity data
It is identified the value of the activity attributes " login times " of rear resulting active evaluation data.
It should be understood that the invention is not restricted to belonged to by the form simply summed to obtain the activity of active evaluation data
The value of property.This can also be determined by other means in other embodiments, such as weighted sum etc..
Next, in step s 130, the value of each activity attributes in the active evaluation data is normalized.
As can be seen from Table 1 and Table 2, the numerical difference of different activity attributes away from may quite it is big (such as " login times " and
Three orders of magnitude of the numerical difference of " residence time "), for accurate evaluation, it is necessary to reasonably be handled data.The present invention
One embodiment in, using normalized method come processing data.Preferably for activity attributes X normalization by with
Lower formula is realized:
XNormal=X/XMax, wherein, XNormalIt is the activity attributes X normalized value of process, XMaxBe activity attributes X most
Big value.By this processing, can largely eliminate by the different institutes of different activity attributes numerical value scales (or magnitude)
The Under Tendency Influence brought.
In addition, in the case of numerical value scale is not considered, it will also be understood that different activity attributes are to active
The percentage contribution of degree should be different.Therefore, it is necessary to which different activity attributes are assigned with different weights, so as to more accurate
Assessed on ground.Such as " login times " and " effectively placing an order " the two activity attributes, it is believed that " under effectively
The influence that influence of the list " to liveness assessment result is greater than " login times " (is checked in fact, user logs in a certain electric business
Commodity, but finally may be placed an order in another electric business, therefore, in this meaning, it is believed that the contribution of logon data is low
In the contribution of lower forms data).
Realized preferably for activity attributes X weighting normalization by below equation:XNormal=AX·X/XMax,
Wherein, XNormalIt is the activity attributes X normalized value of process, XMaxIt is activity attributes X maximum, AXIt is weight coefficient.
The size of weight coefficient can initially be manually specified, and described specify can be arbitrary or can be according to data
Statistics is made.The size of weight coefficient can also be entered according to the assessment or the result of prediction that are carried out with the progress of assessment
Row adaptability amendment, so as to finally give satisfied weight coefficient value.
Finally, in step S140, according to the value by normalized each activity attributes, determine that the network user exists
It is described to enliven angle value.
In one embodiment, by will be added by the value of normalized each activity attributes, exporting the network and using
Enliven angle value in family.Angle value is enlivened to can be used for carrying out quantitative assessment to the liveness of user.
Preferably, after obtaining enlivening angle value, liveness rank can also be further determined that.It is one to enliven angle value
Absolute numerical value, it is that any active ues or inactive users have directly to the user with particular active angle value that it can not allow people sometimes
The judgement of sight.If resulting can be enlivened angle value with multiple different users multiple for a certain special time cycle
The statistics for enlivening angle value in the different time cycle (Cycle Length is identical) is compared, then can be to the liveness of user
Rank has more direct, clearly judgement.
Preferably, can be counted according to the liveness Value Data obtained for multiple network users within multiple time cycles
Its average value and standard deviation are calculated, the liveness level range of the network user is then divided using average value and standard deviation.
For example if the data of 1 year were divided by 3 days for a cycle, each user shares the data in 120 cycles,
Website registered user a total of nearly 80,000,000.Finally, this 120 × 80,000,000 data point is all projected in by we
One plane, it can learn that the data meet normal distribution.
The average value for setting these data is μ, standard deviation sigma, then can define following liveness rank:Section (- ∞, μ -2
It is σ) rank of being sunk into sleep, is expressed as R0;Rank is enlivened in section (μ -2 σ, μ-σ) to be low, is expressed as R1;(μ-σ, μ+σ) is active in section
Rank, it is expressed as R2;Section (μ+σ, μ+2 σ) is that height enlivens rank, is expressed as R3;Section (μ+2 σ ,+∞) is that superelevation enlivens level
Not, it is expressed as R4.
, can be with according to the network use determined in step S140 after liveness rank is determined according to as above method
Angle value is enlivened in family in the time cycle, judges which liveness level the network user belongs in the time cycle
Not.
Next, Fig. 2 shows the stream of network user active degree Forecasting Methodology 200 according to an embodiment of the present
Cheng Tu.Methods described 200 starts from step S210, activity data of the collection network user within the cycle very first time, wherein, institute
Activity data is stated by the unit interval to identify and including multiple activity attributes, the very first time cycle arrest is in current single
The previous day of position time.Next, in step S220, the activity data is identified by the second time cycle, is lived
It is dynamic to assess data, wherein, the cycle very first time is the multiple of second time cycle.Then, it is right in step S230
The value of each activity attributes in the active evaluation data is normalized;Next, in step S240, according to by returning
The value of the one each activity attributes changed, was determined in each second time cycle of the network user within the cycle very first time
Enliven angle value.Finally, in step s 250, angle value is enlivened in each second time cycle, the network user is existed
Liveness in next second time cycle is predicted.
First, in step S210, activity data of the collection network user within the cycle very first time.Step S210 and figure
S110 in 1 is similar, and still, the time cycle can be selected cycle any time in Fig. 1, and in method 200
In step S210, due to needing to be predicted the ensuing liveness of user, it is necessary to additionally require the week very first time
Final the previous day for terminating in the current one time.In one embodiment, with " my god " for the unit time, if the current one time
Be September 21, then the last day in the cycle very first time should be September 20.
It is pointed out that in practical application, the size in the cycle very first time is significantly greater tnan in method 100
" time cycle ".Because method 200 is a kind of Forecasting Methodology, usually, if gathering the " time than being predicted
Activity data in much longer time cycle in cycle " (as described below the second time cycle), then can realize more accurate
True prediction.In one embodiment, it is necessary to predict liveness of the user within three days futures, user was have collected in the past 120
Activity data in it, i.e., the above-mentioned cycle very first time are 120 days.
Next, in step S220, the activity data is identified by the second time cycle, obtains active evaluation number
According to.
Step S220 is similar with the step S120 in method 100, but is used for carrying out activity data in step S220
Mark is not the cycle very first time used in step S210, but second time cycle smaller than it.In a reality
Apply in example, the cycle very first time is the multiple of second time cycle.
In one embodiment, by each unit interval in each second time cycle in the cycle very first time
The value of the respective activity attribute of corresponding activity data is separately summed, the active evaluation data as second time cycle
Respective activity attribute value.
Then, in step S230, the value of each activity attributes in the active evaluation data is normalized.
Wherein, for each second time cycle in the cycle very first time, by below equation to the activity
The value for assessing the activity attributes X in data is normalized:XNormal=AX·X/XMax, wherein, XNormalIt is activity attributes X
By normalized value, XMaxIt is activity attributes X maximum, AXIt is weight coefficient.
Next, in step S240, according to the value by normalized each activity attributes, determine the network user in institute
State in each second time cycle in the cycle very first time and enliven angle value.Unlike the step S140 in method 100,
In step S240, it is thus necessary to determine that multiple to enliven angle value.This is equivalent to the week each time for needing to be directed in the cycle very first time
Phase performs the step S140 in method 100.
In one embodiment, by will be added by the value of normalized each activity attributes, exporting the network and using
Angle value is enlivened in each second time cycle of the family within the cycle very first time.
Finally, in step s 250, angle value is enlivened in each second time cycle, to the network user under
Liveness in one the second time cycle is predicted.
In one embodiment, the step S250 includes:By enlivening angle value in each second time cycle
Matched curve;Using the curve fitted, liveness of the network user in next second time cycle is predicted.
In another embodiment, the step S250 is realized by dividing liveness rank.Specifically, the step
In S250, the liveness rank enlivened angle value, determine the network user first in each second time cycle.
The step can be found in the liveness level calculation method in method 100.
Such as can as in method 100 by the network user each second time cycle data it is flat
Mean μ and standard deviation sigma are calculated.And define following liveness rank:Section (- ∞, μ -2 σ) is rank of being sunk into sleep, and is expressed as
R0;Rank is enlivened in section (μ -2 σ, μ-σ) to be low, is expressed as R1;Section (μ-σ, μ+σ) is expressed as R2 to enliven rank;Section
(μ+σ, μ+2 σ) is that height enlivens rank, is expressed as R3;Section (μ+2 σ ,+∞) is that superelevation enlivens rank, is expressed as R4.
Next, the liveness rank changing value in each second time cycle can be calculated.Specifically, by making network
Liveness rank of the user within second time cycle subtracts the network user's enlivening within previous second time cycle
Spend rank, liveness changing value of the derived grid user within second time cycle.Such as if in certain second round
Liveness rank is R3, and the liveness rank in its previous second round is R2, then second time cycle liveness level
Other changing value is 1.If the liveness rank in certain second round is R3, the liveness rank in its previous second round is
R4, then the liveness rank changing value of second time cycle is -1, by that analogy.
Then, count in each second time cycle with last second time in the cycle very first time
Cycle has ensuing second time cycle of all second time cycles of equal liveness rank changing value
Liveness rank changing value.In this step, all liveness rank changing values and last the second time cycle are found out
The second equal time cycle of liveness rank changing value, and count the liveness rank of their next second time cycle
Changing value.
Then, it is described next for the maximum liveness rank changing value of occurrence probability in the statistical result is defined as
Liveness rank changing value in individual second time cycle.For example, if in statistical result, " -1 " occurs 3 times,
" 1 " occurs 2 times, and " 0 " occurs 1 time, other results does not occur, then can predict next upcoming second time
The liveness rank changing value in cycle is " -1 ".
Finally, the liveness rank of last the second time cycle in the cycle very first time and institute are true
Liveness rank changing value in fixed next time cycle, work of the calculating network user in next second time cycle
Jerk rank.If the liveness rank of last second time cycle is R1, and determines and next will arrive
The liveness rank changing value for the second time cycle come is " -1 ", then can be predicted described next i.e. by the second time of reason
The liveness rank in cycle is R0.
Method 200 described in Fig. 2 includes many similar or identical operations or content with the method 100 described in Fig. 1.
For simplicity, when method 200 is described, these contents are not repeated.Those skilled in the art should understand that
, these describe and limited the step that may be equally applicable in Fig. 2, and vice versa.
Although combined the preferred embodiments of the present invention show the present invention above, those skilled in the art will
It will be appreciated that without departing from the spirit and scope of the present invention, various modifications can be carried out to the present invention, replaces and changes
Become.Therefore, the present invention should not be limited by above-described embodiment, and should be limited by appended claims and its equivalent.
Claims (16)
1. a kind of method that liveness to the network user within the time cycle is assessed, including:
Activity data of the network user within the time cycle is collected, the activity data is identified by the unit interval
And including multiple activity attributes;
The activity data is identified by the time cycle, obtains active evaluation data;
The value of each activity attributes in the active evaluation data is normalized;And
According to the value by normalized each activity attributes, the network user active degree value is determined.
2. according to the method for claim 1, wherein, the activity attributes select from the following:Login times,
Residence time, odd number, collecting commodities number are effectively descended, searching times, number is evaluated, gets reward voucher number.
3. according to the method for claim 1, wherein, the activity data is identified by the time cycle and obtains activity
Assessment packet includes:
The value of the respective activity attribute of activity data corresponding to each unit interval in the time cycle is separately summed,
Value as the respective activity attribute of active evaluation data.
4. the method according to claim 11, wherein, by below equation to the activity attributes in the active evaluation data
X value is normalized:
XNormal=AX·X/XMax,
Wherein, XNormalIt is the activity attributes X normalized value of process, XMaxIt is activity attributes X maximum, AXIt is weight coefficient.
5. according to the method for claim 1, wherein, the net is determined according to by the value of normalized each activity attributes
Network user activity value includes:
By will be added by the value of normalized each activity attributes, export the network user enlivens angle value.
6. the method according to claim 11, in addition to:By being obtained for multiple network users within multiple time cycles
The liveness rank enlivened angle value, determine the network user.
7. the method according to claim 11, wherein, according to what is obtained for multiple network users within multiple time cycles
The average value and standard deviation of liveness Value Data, divide the liveness level range of the network user.
8. a kind of network user active degree Forecasting Methodology, including:
Activity data of the collection network user within the cycle very first time, the activity data by the unit interval identifying and
Including multiple activity attributes, the very first time cycle arrest is in the previous day of current one time;
The activity data is identified by the second time cycle, obtains active evaluation data, the cycle very first time is institute
State the multiple of the second time cycle;
The value of each activity attributes in the active evaluation data is normalized;
According to the value by normalized each activity attributes, each the of the network user within the cycle very first time is determined
Angle value is enlivened in two time cycles;
Angle value is enlivened in each second time cycle, to work of the network user in next second time cycle
Jerk is predicted.
9. according to the method for claim 8, wherein, the activity attributes select from the following:Login times,
Residence time, odd number, collecting commodities number are effectively descended, searching times, number is evaluated, gets reward voucher number.
10. according to the method for claim 8, wherein, obtained by second time cycle to identify the activity data
Include to active evaluation data:
By the activity data corresponding to each unit interval in each second time cycle in the cycle very first time
The value of respective activity attribute is separately summed, the respective activity attribute as the active evaluation data of second time cycle
Value.
11. according to the method for claim 8, wherein, for each second time cycle in the cycle very first time,
The value of the activity attributes X in the active evaluation data is normalized by below equation:
XNormal=AX·X/XMax,
Wherein, XNormalIt is the activity attributes X normalized value of process, XMaxIt is activity attributes X maximum, AXIt is weight coefficient.
12. according to the method for claim 8, wherein, network is determined according to by the value of normalized each activity attributes
Angle value of enlivening in each second time cycle of the user within the cycle very first time includes:
By will be added by the value of normalized each activity attributes, the network user is exported in the cycle very first time
Angle value is enlivened in interior each second time cycle.
13. according to the method for claim 8, wherein, angle value is enlivened to net in each second time cycle
Liveness of the network user in next second time cycle be predicted including:
By enlivening angle value matched curve in each second time cycle;
Using the curve fitted, liveness of the network user in next second time cycle is predicted.
14. according to the method for claim 8, wherein, angle value is enlivened to net in each second time cycle
Liveness of the network user in next second time cycle be predicted including:
The liveness rank enlivened angle value, determine the network user in each second time cycle;
Calculate the liveness rank changing value in each second time cycle;
Count in each second time cycle has with last second time cycle in the cycle very first time
The liveness level of ensuing second time cycle of all second time cycles of equal liveness rank changing value
Other changing value;
The maximum liveness rank changing value of occurrence probability in the statistical result is defined as next week second time
Interim liveness rank changing value;
The liveness rank of last the second time cycle in the cycle very first time and identified next
Liveness rank changing value in the individual time cycle, liveness level of the calculating network user in next second time cycle
Not.
15. according to the method for claim 14, wherein, according to for multiple network users within multiple second time cycles
The average value and standard deviation of obtained liveness Value Data, divide the liveness level range of the network user.
16. according to the method for claim 14, wherein, calculate the liveness rank changing value in each second time cycle
Including:
When subtracting the network user previous second by making liveness rank of the network user within second time cycle
Between liveness rank in the cycle, liveness changing value of the derived grid user within second time cycle.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610318215.7A CN107370614A (en) | 2016-05-13 | 2016-05-13 | Network user active degree appraisal procedure and Forecasting Methodology |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610318215.7A CN107370614A (en) | 2016-05-13 | 2016-05-13 | Network user active degree appraisal procedure and Forecasting Methodology |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107370614A true CN107370614A (en) | 2017-11-21 |
Family
ID=60304696
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610318215.7A Pending CN107370614A (en) | 2016-05-13 | 2016-05-13 | Network user active degree appraisal procedure and Forecasting Methodology |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107370614A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108519996A (en) * | 2018-03-06 | 2018-09-11 | 江苏掌源智能科技有限公司 | A kind of client's liveness method for digging |
CN109543940A (en) * | 2018-10-12 | 2019-03-29 | 中国平安人寿保险股份有限公司 | Active evaluation method, apparatus, electronic equipment and storage medium |
CN109727070A (en) * | 2018-12-28 | 2019-05-07 | 中国银行股份有限公司 | A kind of determination method and device of potential any active ues |
CN109934626A (en) * | 2019-03-05 | 2019-06-25 | 重庆锐云科技有限公司 | A kind of Real Estate Marketing management method and device based on public platform bean vermicelli liveness |
CN110147493A (en) * | 2019-04-15 | 2019-08-20 | 中国平安人寿保险股份有限公司 | Enliven determination method, apparatus, computer equipment and the storage medium of the factor |
CN110740193A (en) * | 2019-10-30 | 2020-01-31 | 江苏满运软件科技有限公司 | platform activity prediction method and device, storage medium and electronic equipment |
CN110971673A (en) * | 2019-11-15 | 2020-04-07 | 广东浪潮大数据研究有限公司 | Computer device and method for acquiring user activity of deep learning platform |
CN111401433A (en) * | 2020-03-12 | 2020-07-10 | 腾讯科技(深圳)有限公司 | User information acquisition method and device, electronic equipment and storage medium |
CN111461793A (en) * | 2020-04-27 | 2020-07-28 | 吉林省桥王智能科技有限公司 | Integral chain consensus method based on activity probability selection |
CN112291625A (en) * | 2020-10-16 | 2021-01-29 | 腾讯科技(北京)有限公司 | Information quality processing method, information quality processing device, electronic equipment and storage medium |
CN112612826A (en) * | 2020-12-21 | 2021-04-06 | 北京达佳互联信息技术有限公司 | Data processing method and device |
CN112734156A (en) * | 2020-09-29 | 2021-04-30 | 红盾大数据(北京)有限公司 | Enterprise activity evaluation method, device, equipment and storage medium |
CN112818244A (en) * | 2021-02-24 | 2021-05-18 | 深圳市网联安瑞网络科技有限公司 | Method, system and terminal for judging activity of channel, group and group user |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101960427A (en) * | 2007-12-26 | 2011-01-26 | 斯曼泰克公司 | The balance consistance hash of distributed resource management |
US20110169603A1 (en) * | 2008-02-05 | 2011-07-14 | International Business Machines Corporation | Distinguishing between user physical exertion biometric feedback and user emotional interest in a media stream |
CN102394798A (en) * | 2011-11-16 | 2012-03-28 | 北京交通大学 | Multi-feature based prediction method of propagation behavior of microblog information and system thereof |
CN102609786A (en) * | 2012-01-18 | 2012-07-25 | 广州市动景计算机科技有限公司 | Method and device for forecasting whether user is off network |
CN102982466A (en) * | 2012-07-17 | 2013-03-20 | 华东师范大学 | Graded forecasting method based on user liveness |
CN104035987A (en) * | 2014-05-30 | 2014-09-10 | 南京邮电大学 | Method for ranking microblog network user influence |
CN104506894A (en) * | 2014-12-22 | 2015-04-08 | 合一网络技术(北京)有限公司 | Method and device for evaluating multi-media resources |
CN104504124A (en) * | 2014-12-31 | 2015-04-08 | 合一网络技术(北京)有限公司 | Method for presenting entity popularity through video searching and playing behaviors |
CN105005909A (en) * | 2015-06-17 | 2015-10-28 | 深圳市腾讯计算机系统有限公司 | Method and device for predicting lost users |
-
2016
- 2016-05-13 CN CN201610318215.7A patent/CN107370614A/en active Pending
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101960427A (en) * | 2007-12-26 | 2011-01-26 | 斯曼泰克公司 | The balance consistance hash of distributed resource management |
US20110169603A1 (en) * | 2008-02-05 | 2011-07-14 | International Business Machines Corporation | Distinguishing between user physical exertion biometric feedback and user emotional interest in a media stream |
CN102394798A (en) * | 2011-11-16 | 2012-03-28 | 北京交通大学 | Multi-feature based prediction method of propagation behavior of microblog information and system thereof |
CN102609786A (en) * | 2012-01-18 | 2012-07-25 | 广州市动景计算机科技有限公司 | Method and device for forecasting whether user is off network |
CN102982466A (en) * | 2012-07-17 | 2013-03-20 | 华东师范大学 | Graded forecasting method based on user liveness |
CN104035987A (en) * | 2014-05-30 | 2014-09-10 | 南京邮电大学 | Method for ranking microblog network user influence |
CN104506894A (en) * | 2014-12-22 | 2015-04-08 | 合一网络技术(北京)有限公司 | Method and device for evaluating multi-media resources |
CN104504124A (en) * | 2014-12-31 | 2015-04-08 | 合一网络技术(北京)有限公司 | Method for presenting entity popularity through video searching and playing behaviors |
CN105005909A (en) * | 2015-06-17 | 2015-10-28 | 深圳市腾讯计算机系统有限公司 | Method and device for predicting lost users |
Cited By (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108519996A (en) * | 2018-03-06 | 2018-09-11 | 江苏掌源智能科技有限公司 | A kind of client's liveness method for digging |
CN109543940A (en) * | 2018-10-12 | 2019-03-29 | 中国平安人寿保险股份有限公司 | Active evaluation method, apparatus, electronic equipment and storage medium |
CN109543940B (en) * | 2018-10-12 | 2024-04-09 | 中国平安人寿保险股份有限公司 | Activity evaluation method, activity evaluation device, electronic equipment and storage medium |
CN109727070B (en) * | 2018-12-28 | 2021-11-19 | 中国银行股份有限公司 | Method and device for determining potential active users |
CN109727070A (en) * | 2018-12-28 | 2019-05-07 | 中国银行股份有限公司 | A kind of determination method and device of potential any active ues |
CN109934626A (en) * | 2019-03-05 | 2019-06-25 | 重庆锐云科技有限公司 | A kind of Real Estate Marketing management method and device based on public platform bean vermicelli liveness |
CN110147493A (en) * | 2019-04-15 | 2019-08-20 | 中国平安人寿保险股份有限公司 | Enliven determination method, apparatus, computer equipment and the storage medium of the factor |
CN110147493B (en) * | 2019-04-15 | 2023-07-21 | 中国平安人寿保险股份有限公司 | Method, device, computer equipment and storage medium for determining active factors |
CN110740193A (en) * | 2019-10-30 | 2020-01-31 | 江苏满运软件科技有限公司 | platform activity prediction method and device, storage medium and electronic equipment |
CN110971673B (en) * | 2019-11-15 | 2022-08-12 | 广东浪潮大数据研究有限公司 | Computer device and method for acquiring user activity of deep learning platform |
CN110971673A (en) * | 2019-11-15 | 2020-04-07 | 广东浪潮大数据研究有限公司 | Computer device and method for acquiring user activity of deep learning platform |
CN111401433A (en) * | 2020-03-12 | 2020-07-10 | 腾讯科技(深圳)有限公司 | User information acquisition method and device, electronic equipment and storage medium |
CN111401433B (en) * | 2020-03-12 | 2023-10-17 | 腾讯科技(深圳)有限公司 | User information acquisition method and device, electronic equipment and storage medium |
CN111461793A (en) * | 2020-04-27 | 2020-07-28 | 吉林省桥王智能科技有限公司 | Integral chain consensus method based on activity probability selection |
CN111461793B (en) * | 2020-04-27 | 2023-10-10 | 吉林省桥王智能科技有限公司 | Integration chain consensus method based on liveness probability selection |
CN112734156A (en) * | 2020-09-29 | 2021-04-30 | 红盾大数据(北京)有限公司 | Enterprise activity evaluation method, device, equipment and storage medium |
CN112291625A (en) * | 2020-10-16 | 2021-01-29 | 腾讯科技(北京)有限公司 | Information quality processing method, information quality processing device, electronic equipment and storage medium |
CN112291625B (en) * | 2020-10-16 | 2024-03-01 | 腾讯科技(北京)有限公司 | Information quality processing method, information quality processing device, electronic equipment and storage medium |
CN112612826A (en) * | 2020-12-21 | 2021-04-06 | 北京达佳互联信息技术有限公司 | Data processing method and device |
CN112612826B (en) * | 2020-12-21 | 2024-02-06 | 北京达佳互联信息技术有限公司 | Data processing method and device |
CN112818244A (en) * | 2021-02-24 | 2021-05-18 | 深圳市网联安瑞网络科技有限公司 | Method, system and terminal for judging activity of channel, group and group user |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107370614A (en) | Network user active degree appraisal procedure and Forecasting Methodology | |
Yamana et al. | Superensemble forecasts of dengue outbreaks | |
Clifford | Charitable organisations, the great recession and the age of austerity: Longitudinal evidence for England and Wales | |
Kim et al. | Role of tourism price in attracting international tourists: The case of Japanese inbound tourism from South Korea | |
US9224094B2 (en) | Systems, methods, and articles of manufacture to measure online audiences | |
Chen et al. | Seasonal ARIMA forecasting of inbound air travel arrivals to Taiwan | |
Chen et al. | An empirical study of demand forecasting of non-volatile memory for smart production of semiconductor manufacturing | |
Lera-López et al. | Sports spectatorship in Spain: Attendance and consumption | |
CN102156932A (en) | Prediction method and device for secondary purchase intention of customers | |
CN107093120A (en) | Service strategy method for pushing and device | |
Mac Nally et al. | Avian biodiversity monitoring in Australian rangelands | |
CN108038622A (en) | A kind of intelligent perception system recommendation user method | |
De Ornellas et al. | The impact of data realities on conservation planning | |
Pope et al. | A multiple-attribute decision model for retail store location | |
CN114880581A (en) | User data processing method, storage medium and electronic device | |
US20130262191A1 (en) | Systems and methods for measuring and scoring engagement in organizations | |
Morris | Which mutualists are most essential? Buffering of plant reproduction against the extinction of pollinators | |
CN111259922A (en) | Order data processing method and device based on customer order-returning early warning | |
Adeola et al. | Tourism and economic wellbeing in Africa | |
Riera et al. | A choice modelling case study on climate change involving two-way interactions | |
KR101346868B1 (en) | Method for estimating election based on population statistics of selected voter by hour | |
Silva et al. | Quantitative analysis of the effectiveness of public health measures on covid-19 transmission | |
Agostini et al. | Local distributional effects of government cash transfers in Chile | |
Harris et al. | Ageing workforces, ill‐health and multi‐state labour market transitions | |
JP7024663B2 (en) | Evaluation updater, method, and program |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171121 |
|
RJ01 | Rejection of invention patent application after publication |