CN106126537A - Method and device is recommended in a kind of application - Google Patents

Abstract

The present invention provides a kind of application recommendation method, including: according to the internet records of each user, determine user to be recommended and the similar users of user to be recommended；Determine that similar users used but user to be recommended original application k, calculate similar users degree of adhesion and freshness to applying k；According to similar users to the application degree of adhesion of k, degree of association between freshness and similar users and user to be recommended, calculate the recommendation index of application k；And determine whether to recommend described application k for user to be recommended according to the recommendation index of described application k.Method is recommended in application provided by the present invention, it is possible to avoid by similar users recently not in use by application or use the lowest application of frequency to recommend user to be recommended, thus improve the availability being recommended to apply.

Description

Method and device is recommended in a kind of application

Technical field

The present invention relates to areas of information technology, be specifically related to a kind of application and recommend method and device.

Background technology

Along with developing rapidly of information technology, the various application APP (application of APPLITION cell phone software) on mobile terminal Become in the application shop that service provider provides the user the main channel of various value-added service, Fructus Mali pumilae and Google, various application Total amount is above million, and unanimously keeps the trend quickly increased.In the face of the application of magnanimity, when user needs to spend high Between after application screened and tries out by cost, just can find oneself application interested.

At present, telecom operators and service provider, can recommend for it according to the statistics of the most used application of user The application of same type, also according to the application having the user of identical use interest to be used with it.But this according to whether make Being used as the standard whether recommended to user, there is the biggest inaccuracy in the application recommended, available such as the application recommended Property is the highest.As recommended the application not continuing to use the most quickly of the user after great majority are installed, in this case, it pushes away The application recommended is exactly inaccurate, and the application quality recommended is the highest, or invalid recommendation.

The problem the most fundamentally solving accurately to recommend application for user, it is achieved combine the change of the use interest of user Recommend, for user, the application that availability is high, improve the accuracy that application is recommended, be areas of information technology problem demanding prompt solutions.

Summary of the invention

The technical problem to be solved be for prior art in the presence of drawbacks described above, it is provided that a kind of application Recommend method and device, can not be that it recommends available degree according to the interests change of user in order to solve present in prior art The problem of high application.

For achieving the above object, the present invention provides a kind of application recommendation method, including:

A kind of application recommendation method, described method comprises the steps:

According to the internet records of user each in the range of default duration, determine user to be recommended and similar users, described phase It is to use the user of same application with user to be recommended like user；

Determine that similar users used but user to be recommended original application k, calculate similar users and application k is glued Right and freshness, described degree of adhesion is that similar users uses application k in described preset duration for the first time and makes for the last time By the time difference of application k, described freshness be similar users in described preset duration the last time using application k with Time difference between described preset duration deadline；

According to similar users to the application degree of adhesion of k, relevant between freshness and similar users to user to be recommended Degree, calculates the recommendation index of application k；

Recommendation index according to described application k and default first threshold, it is determined whether recommend described for user to be recommended Application k.

Preferably, described determine that user to be recommended and similar users include:

Described preset duration is divided at least two statistical time range, calculate each user use in each statistical time range respectively should With the flow spent；

Calculate the time weighting coefficient of the corresponding each statistical time range of each user；

Determine and use each user q of same application j, according to described flow and time weighting system with described user to be recommended Number, calculates each user q interest score to each application；

According to user to be recommended use the application interest score of j, user to be recommended use each application average interest score, User q uses the application interest score of j, user q to use the average interest score of each application, calculates user to be recommended and user q Between degree of association；

According to the degree of association between described user to be recommended and described each user q, determine the similar users of user to be recommended.

Preferably, in the time weighting coefficient of the corresponding each statistical time range of each user of described calculating, described time weighting coefficient f_iaT () calculates according to below equation (1):

$f_{ia}\left(t\right)=e^{\[\frac{t_{ia}}{t_{iA}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

ti_aFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

Preferably, described calculating similar users in the application degree of adhesion of k and freshness, degree of adhesion f_vkT the calculating of () is public Formula is:

$f_{\mathrm{vk}}\left(t\right)=e^{t_{\mathrm{vk}1}};---\left(2\right)$

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

Preferably, described calculating similar users in the application degree of adhesion of k and freshness, freshness g_vkT the calculating of () is public Formula is:

$g_{\mathrm{vk}}\left(t\right)={\log }_{\frac{1}{2}}{t}_{\mathrm{vk}2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.

The present invention also provides for a kind of application recommendation apparatus, including:

Similar users module, for according to the internet records of each user in the range of the duration preset, determining user to be recommended And similar users, described similar users is to use the user of same application with user to be recommended；

Degree of adhesion freshness module, is used for determining that similar users used but user to be recommended original application k, meter Calculate similar users to the application degree of adhesion of k and freshness, described degree of adhesion be similar users in described preset duration for the first time Using application k and the last time difference using application k, described freshness is that similar users is last in described preset duration Time difference between time of first use application k and described preset duration deadline；

Recommending module, is used for according to similar users the application degree of adhesion of k, freshness and similar users and use to be recommended Degree of association between family, calculates the recommendation index of application k；And for the recommendation index according to described application k and default first Threshold value, it is determined whether recommend described application k for user to be recommended.

Described similar users module, including:

Flow rate calculation unit, for described preset duration is divided at least two statistical time range, calculates each user respectively The flow that each application is spent is used in statistical time range；

Time weighting unit, for calculating the time weighting coefficient of the corresponding each statistical time range of each user；

Interest score unit, uses each user q of same application j, according to described for determining with described user to be recommended Flow and time weighting coefficient, calculate each user q interest score to each application；

Correlation unit, for using the application interest score of j, user to be recommended to use each application according to user to be recommended Average interest score, user q use application the interest score of j, user q use each application average interest score, calculating is treated Recommend the degree of association between user and user q, and according to the degree of association between described user to be recommended and described each user q, really The similar users of fixed user to be recommended.

Described time weighting unit, specifically for calculating described time weighting coefficient, described time weighting coefficient f_iaT () presses Calculate according to below equation (1):

$f_{ia}\left(t\right)=e^{\[\frac{t_{ia}}{t_{iA}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

t_iaFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

Described degree of adhesion freshness module, specifically for calculating degree of adhesion f_vk(t), described viscous

Right f_vkT the computing formula of () is:

$f_{\mathrm{vk}}\left(t\right)=e^{t_{\mathrm{vk}1}};---\left(2\right)$

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

Described degree of adhesion freshness module, specifically for calculating freshness g_vk(t), described freshness g_vkT the calculating of () is public Formula is:

$g_{vk}\left(t\right)={\log }_{\frac{1}{2}}{t}_{vk2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.

Method and device is recommended in application provided by the present invention, it is possible to recommend, for user to be recommended, the application that availability is high, Concrete implementation mode is, finds and has same application to use the similar users of interest with user to be recommended, uses in similar users Cross and in the not used application of user to be recommended, find similar users degree of adhesion is high and freshness is high application to use to be recommended Family recommend, described degree of adhesion be similar users use institute recommendation apply persistent period length, described freshness be similar use Family uses the last time gap being recommended application to recommend the time nearest.Method, energy are recommended in application provided by the present invention Enough avoid by similar users recently not in use by application, or use the lowest application of frequency to recommend user to be recommended, from And improve the availability being recommended application.

Accompanying drawing explanation

For the technical scheme in the clearer explanation embodiment of the present invention, in embodiment being described below required for make Accompanying drawing do and introduce simply, it should be apparent that, the accompanying drawing in describing below is some embodiments of the present invention, for ability From the point of view of the those of ordinary skill of territory, on the premise of not paying creative work, it is also possible to obtain the attached of other according to these accompanying drawings Figure.

The schematic flow sheet of the application recommendation method first embodiment that Fig. 1 provides for the present invention；

The schematic flow sheet of application recommendation method the second embodiment that Fig. 2 provides for the present invention；

The structural representation of the application recommendation apparatus that Fig. 3 provides for the present invention；

Fig. 4 applies the structural representation of similar users module in recommendation apparatus for what the present invention provided.

Detailed description of the invention

For making those skilled in the art be more fully understood that technical scheme, below in conjunction with the accompanying drawings with embodiment to this Invention is described in further detail.Obviously, described embodiment is a part of embodiment of the present invention rather than whole enforcement Example.Based on the embodiment in the present invention, those of ordinary skill in the art are obtained under not making creative work premise Every other embodiment, broadly falls into the scope of protection of the invention.

The schematic flow sheet of the application recommendation method first embodiment that Fig. 1 provides for the present invention, application as shown in Figure 1 pushes away The method of recommending comprises the steps:

Step S101, according to the internet records of user each in the range of default duration, determines user to be recommended and similar use Family, described similar users is to use the user of same application with user to be recommended.

Concrete, only use the user of same application with user to be recommended, just there is the reference value recommending application, In the used application of user to be recommended, the user that at least an application is same, it is possible to be considered user to be recommended Similar users.

Step S102, determines that similar users used but user to be recommended original application k, calculates similar users pair The application degree of adhesion of k and freshness, described degree of adhesion be similar users use for the first time in described preset duration application k with The time difference of rear first use application k, described freshness is that similar users is last in described preset duration uses application k Time and described preset duration deadline between time difference.

Concrete, due to need to get rid of user the most not in use by or use the application of underfrequency, need to consider Similar users is for the degree of adhesion of application to be recommended and freshness.Wherein, described degree of adhesion f_vkT the computing formula of () is:

$f_{\mathrm{vk}}\left(t\right)=e^{t_{\mathrm{vk}1}};---\left(2\right)$

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

Described freshness g_vkT the computing formula of () is:

$g_{vk}\left(t\right)={\log }_{\frac{1}{2}}{t}_{vk2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.

Calculate user v for the application freshness of k and degree of adhesion after, just can be by user v for applying the use feelings of k Condition makes statistics.

Step S103, according to similar users to the application degree of adhesion of k, freshness and similar users and user to be recommended it Between degree of association, calculate application k recommendation index.

Concrete, the computing formula recommending index that the present invention provides is:

$P_{ik}=\frac{{\Σ}_{v\∈Vi}sim(i,v)(r_{v,k}-\stackrel{\‾}{r_v})g_{ik}\left(t\right)f_{vk}\left(t\right)}{{\Σ}_{v\∈Vi}sim(i,v)};---\left(6\right)$

Wherein, i represents user i to be recommended；

P_ikRepresent that user i is for applying the recommendation index of k；

V_iSet for similar users v of user i；

Represent that user i uses the average interest score of its all application；

Represent that user v uses the average interest score of its all application；

r_v,kRepresent the similar users v interest score to application k；

(i v) represents user i and the degree of association of similar users v to sim.

Result of calculation for making recommendation index is more directly perceived and unified, and the present invention also provides for below equation as recommending index Another one computing formula:

${P^{\′}}_{ik}={\stackrel{\‾}{r}}_i+\frac{{\Σ}_{v\∈Vi}sim(i,v)(r_{v,k}-\stackrel{\‾}{r_v})g_{ik}\left(t\right)f_{vk}\left(t\right)}{{\Σ}_{v\∈Vi}sim(i,v)};---\left(7\right)$

Step S104, according to recommendation index and the default first threshold of described application k, it is determined whether for user to be recommended Recommend described application k.

Concrete, according to default first threshold, qualified application k is recommended user to be recommended.Can also be by After described recommendation index is ranked up, chooses and recommend the application k that index is big to recommend to user to be recommended.

Method is recommended in application provided by the present invention, it is possible to recommend, for user to be recommended, the application that availability is high, concrete Implementation is, searching and user to be recommended have the similar users of same application, use and user to be recommended in similar users In not used application, find the application that similar users degree of adhesion is high and freshness is high and recommend to user to be recommended, described Degree of adhesion be similar users use institute recommendations application persistent period length, described freshness by similar users use recommended answer The last time gap recommend the time nearest.Method is recommended in application provided by the present invention, it is possible to avoid similar use Family recently not in use by application, or use the lowest application of frequency to recommend user to be recommended, thus improve and is recommended to answer Availability.

The schematic flow sheet of application recommendation method the second embodiment that Fig. 2 provides for the present invention, application as shown in Figure 2 pushes away The method of recommending comprises the steps:

Step S201, is divided at least two statistical time range by described preset duration, calculates each user at each statistical time range The flow that each application of interior use is spent.

Concrete, each user of described calculating uses the flow value of various application, all application used including all users, Flow value therein, by the statistics of each flow value applied that each user is used every day in the range of described default duration.

As being a statistical time range by the statistics duration of month according to 5 days, 6 statistical time ranges can be divided into, but in reality In border, owing to each user uses application to belong to the use habit of subjectivity, the method carrying out unifying Time segments division by its statistical time range May make statistical result produce deviation, therefore, the division methods of statistical time range of the present invention, be by each user according to The time being used application divides respectively, prevents from considering different user behavior according to unified standard.

For convenience of the follow-up detailed description to scheme, it is illustrated below, when presetting statistics a length of one month, extracts 2015 Each user in January in year (only 3 users of citing) uses the internet records of each application (only 2 application of citing), calculates each use The flow value of the family various application of use and the statistics of time such as table 1:

User's name	Apply Names	Flow value (M)	Time
				XXXX 1	Excellent extremely	2	2015-1-1
XXXX 1	Excellent extremely	5	2015-1-4
				XXXX 1	Excellent extremely	6	2015-1-6
XXXX 1	Excellent extremely	3	2015-1-13
				XXXX 2	Taobao	2	2015-1-1
XXXX 2	Taobao	6	2015-1-3
				XXXX 2	Taobao	1	2015-1-12
XXXX 2	Excellent extremely	3	2015-1-20
				XXXX 3	Taobao	6	2015-1-4
XXXX 3	Taobao	1	2015-1-6
				XXXX 3	Taobao	3	2015-1-13

Table 1

In Table 1, according to the criteria for classifying that 5 days is a statistical time range, user 1 can be divided into 3 statistical time ranges, uses Family 2 is divided into 4 statistical time ranges, and user 3 is divided into 3 statistical time ranges.

Step S202, calculates the time weighting coefficient of the corresponding each statistical time range of each user.

Concrete, in the time weighting coefficient of the corresponding each statistical time range of each user of described calculating, described time weighting coefficient f_iaT () calculates according to below equation (1):

$f_{ia}\left(t\right)=e^{\[\frac{t_{ia}}{t_{iA}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

t_iaFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

Then according to the calculating of formula (1),

The time weighting coefficient of user 1 is 0.51,0.72,1；

The time weighting coefficient of user 2 is 0.47,0.60,0.78,1；

The time weighting coefficient of user 3 is 0.51,0.72,1.

Step S203, determines and uses each user q of same application j with described user to be recommended, according to described flow and time Between weight coefficient, calculate each user q interest score to each application.

Concrete, it is first determined a user to be recommended is user i, with each use that user i to be recommended uses same application j Family q, could be used for calculating the degree of association between user i to be recommended.Otherwise, complete with the application that user i to be recommended is used Difference, then it is the most uncorrelated with user i to be recommended.

Concrete, the formula of the interest score calculating user i use application j is:

$r_{ij}=\underset{a\∈A}{\Σ}{b}_{ia,j}{f}_{ia}\left(t\right);---\left(4\right)$

Wherein, b_iaWithin a period, the flow value of application j is used for user i.

Illustrate, according to the interest score of formula (4) calculating user 1 to the 3 each application of use:

User 1 uses the excellent cruel interest score to be: (2+5) * 0.51+6*0.72+3*1=10.89；

User 2 uses the interest score of Taobao to be: (2+6) * 0.47+1*0.78=4.54；

User 2 uses the excellent cruel interest score to be: 3*1=3；

User 3 uses the interest score of Taobao to be: 6*0.51+1*0.72+3*1=6.78.

If each user of statistics is M altogether, each user altogether for N number of, then used and respectively should by each application that each user is used Interest score, can represent with following matrix:

$\left(\begin{array}{c}{r}_{11},r_{12},\mathrm{...},r_{1j},\mathrm{...},r_{1N}\\ \mathrm{...}\\ r_{i1},r_{i2},\mathrm{...},r_{ij},\mathrm{...},r_{iN}\\ \mathrm{...}\\ r_{M1},r_{M2},\mathrm{...},r_{Mj},\mathrm{...},r_{MN}\end{array}\right)$

Step S204, uses the application interest score of j, user to be recommended to use the average of each application according to user to be recommended Interest score, user q use the application interest score of j, user q to use the average interest score of each application, calculate use to be recommended Degree of association between family and user q.

Concrete, use Pearson correlation coefficient formula, calculate the degree of association between user i and user q, such as formula (5) Shown in,

$sim(i,q)=\frac{{\Σ}_{j\∈cn}(r_{i,j}-\stackrel{\‾}{r_i})(r_{q,j}-\stackrel{\‾}{r_q})}{\sqrt{{\Σ}_{j\∈cn}{(r_{i,j}-\stackrel{\‾}{r_i})}^2}\sqrt{{\Σ}_{j\∈cn}{(r_{q,j}-\stackrel{\‾}{r_q})}^2}};---\left(5\right)$

Wherein, q is and each user q of user i to be recommended use same application j

Q is the set of user q, q ∈ Q；

Represent that user i uses the average interest score of its all application；

Represent that user q uses the average interest score of its all application；

Cn represents the set of the application j that user i and user q is used in conjunction with.

Step S205, according to the degree of association between described user to be recommended and described each user q, determines user's to be recommended Similar users.

Concrete, after the result of calculation of degree of association is ranked up according to order from big to small, can be according to default Screening rule, selects the highest user of degree of correlation as similar users v of user i to be recommended.

It is understood that similar users v is part or all of in user q, above-mentioned formula (5) is equally applicable to user The calculating of the similarity between i and user v, it may be assumed that

$sim(i,v)=\frac{{\Σ}_{j\∈cn}(r_{i,j}-\stackrel{\‾}{r_i})(r_{v,j}-\stackrel{\‾}{r_v})}{\sqrt{{\Σ}_{j\∈cn}{(r_{i,j}-\stackrel{\‾}{r_i})}^2}\sqrt{{\Σ}_{j\∈cn}{(r_{v,j}-\stackrel{\‾}{r_v})}^2}}$

Method is recommended in application provided by the present invention, just considers user when determining similar users and uses the change of interest Changing, the different periods of user gives different time weighting coefficients, period weight coefficient more early is the least, nearest period power Weight coefficient is maximum, and therefore, when certain similar users is applied not in use by some in the recent period, it is according to time weighting and flow value The use interest calculated can be on the low side, and the described similar users calculated also can drop therewith with the degree of association of user to be recommended Low such that it is able to find out, with user to be recommended, there is the similar users of identical interests change, improve the availability that application is recommended.

The structural representation of the application recommendation apparatus that Fig. 3 provides for the present invention, application recommendation apparatus bag as shown in Figure 3 Include:

Similar users module 31, for according to the internet records of each user in the range of the duration preset, determining use to be recommended Family and similar users, described similar users is to use the user of same application with user to be recommended.

Degree of adhesion freshness module 32, is used for determining that similar users used but user to be recommended original application k, Calculating similar users to the application degree of adhesion of k and freshness, described degree of adhesion is that similar users is in described preset duration first Secondary use application k and the last time difference using application k, described freshness be similar users in described preset duration Time difference between time of rear first use application k and described preset duration deadline；

Specifically for calculating degree of adhesion f_vk(t), described degree of adhesion f_vkT the computing formula of () is:

$f_{\mathrm{vk}}\left(t\right)=e^{t_{\mathrm{vk}1}};---\left(2\right)$

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

Specifically for calculating freshness g_vk(t), described freshness g_vkT the computing formula of () is:

$g_{vk}\left(t\right)={\log }_{\frac{1}{2}}{t}_{vk2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.

Recommending module 33, is used for according to similar users applying the degree of adhesion of k, freshness and similar users with to be recommended Degree of association between user, calculates the recommendation index of application k；And for recommending index and default the according to described application k One threshold value, it is determined whether recommend described application k for user to be recommended.

Application recommendation apparatus provided by the present invention, it is possible to recommend the application that availability is high for user to be recommended, concrete Implementation is, finds and has same application to use the similar users of interest with user to be recommended, used in similar users and treated In the application that recommendation user is not used, find the application that similar users degree of adhesion is high and freshness is high and push away to user to be recommended Recommend, described degree of adhesion be similar users use institute recommendation apply persistent period length, described freshness be that similar users makes Recommend the time nearest by the last time gap being recommended application.Method is recommended in application provided by the present invention, it is possible to keep away Exempt from by similar users recently not in use by application, or use the lowest application of frequency to recommend user to be recommended, thus carry The high availability being recommended application.

The structural representation of similar users module in the application recommendation apparatus that Fig. 4 provides for the present invention, as shown in Figure 4 should Include by similar users module in recommendation apparatus:

Flow rate calculation unit 311, for described preset duration is divided at least two statistical time range, calculates each user and exists The flow that each application is spent is used in each statistical time range.

Time weighting unit 312, for calculating the time weighting coefficient of the corresponding each statistical time range of each user；Specifically for meter Calculate described time weighting coefficient, described time weighting coefficient f_iaT () calculates according to below equation (1):

$f_{ia}\left(t\right)=e^{\[\frac{t_{ia}}{t_{iA}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

t_iaFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

Interest score unit 313, uses each user q of same application j, according to institute for determining with described user to be recommended State flow and time weighting coefficient, calculate each user q interest score to each application.

Correlation unit 314, for using the application interest score of j, user to be recommended to use respectively according to user to be recommended The average interest score of application, user q use the application interest score of j, user q to use the average interest score of each application, meter Calculate the degree of association between user to be recommended and user q, and relevant according between described user to be recommended to described each user q Degree, determines the similar users of user to be recommended.

Similar users module provided by the present invention, just considers user and uses the change of interest when determining similar users Changing, the different periods of user gives different time weighting coefficients, period weight coefficient more early is the least, nearest period power Weight coefficient is maximum, and therefore, when certain similar users is applied not in use by some in the recent period, it is according to time weighting and flow value The use interest calculated can be on the low side, and the described similar users calculated also can drop therewith with the degree of association of user to be recommended Low such that it is able to find out, with user to be recommended, there is the similar users of identical interests change, improve the availability that application is recommended.

In several embodiments provided herein, it should be understood that disclosed method, apparatus and system, permissible Realize by another way.Such as, apparatus embodiments described above is only schematic, drawing of described functional module Point, the division of a kind of logic function, actual can have other dividing mode when realizing, and the most multiple modules can be in conjunction with Or it is desirably integrated into another system, or some features can be ignored, or do not perform.

Last it is noted that above example is only in order to illustrate technical scheme, it is not intended to limit；Although With reference to previous embodiment, the present invention is described in detail, it will be understood by those within the art that: it still may be used So that the technical scheme described in foregoing embodiments to be modified, or wherein portion of techniques feature is carried out equivalent； And these amendment or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and Scope.

Claims (10)

1. an application recommendation method, it is characterised in that described method comprises the steps:

According to the internet records of user each in the range of default duration, determine user to be recommended and similar users, described similar use Family is to use the user of same application with user to be recommended；

Determine that similar users used but user to be recommended original application k, calculate the similar users degree of adhesion to application k And freshness, described degree of adhesion is that similar users k of use application for the first time in described preset duration answers with last use By the time difference of k, described freshness is that similar users is last in described preset duration uses the time of application k with described Time difference between preset duration deadline；

According to similar users to the application degree of adhesion of k, degree of association between freshness and similar users and user to be recommended, meter Calculate the recommendation index of application k；

Recommendation index according to described application k and default first threshold, it is determined whether recommend described application for user to be recommended k。

2. application recommendation method as claimed in claim 1, it is characterised in that described determine user to be recommended and similar users bag Include:

Described preset duration is divided at least two statistical time range, calculates each user in each statistical time range, use each application institute The flow spent；

Calculate the time weighting coefficient of the corresponding each statistical time range of each user；

Determine and use each user q of same application j with described user to be recommended, according to described flow and time weighting coefficient, meter Calculate each user q interest score to each application；

The application interest score of j, user to be recommended is used to use the average interest score of each application, user according to user to be recommended Q uses the application interest score of j, user q to use the average interest score of each application, calculates between user to be recommended and user q Degree of association；

According to the degree of association between described user to be recommended and described each user q, determine the similar users of user to be recommended.

3. application recommendation method as claimed in claim 2, it is characterised in that the corresponding each statistical time range of each user of described calculating In time weighting coefficient, described time weighting coefficient f_iaT () calculates according to below equation (1):

$f_{i_a}\left(t\right)=e^{\[\frac{t_{i_a}}{t_{i_A}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

t_iaFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

4. application recommendation method as claimed in claim 1, it is characterised in that the described calculating similar users bonding to application k In degree and freshness, degree of adhesion f_vkT the computing formula of () is:

f_vk(t)=e^tvk1； (2)

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

5. application recommendation method as claimed in claim 1, it is characterised in that the described calculating similar users bonding to application k In degree and freshness, freshness g_vkT the computing formula of () is:

$g_{vk}\left(t\right)={\log }_{\frac{1}{2}}{t}_{vk2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.

6. an application recommendation apparatus, it is characterised in that including:

Similar users module, for according to the internet records of each user in the range of the duration preset, determining user to be recommended and phase Like user, described similar users is to use the user of same application with user to be recommended；

Degree of adhesion freshness module, is used for determining that similar users used but user to be recommended original application k, calculates phase Like user to the application degree of adhesion of k and freshness, described degree of adhesion is that similar users uses in described preset duration for the first time Application k and the last time difference using application k, described freshness is that similar users is last in described preset duration Time difference between the time and the described preset duration deadline that use application k；

Recommending module, for according to similar users to the application degree of adhesion of k, freshness and similar users and user to be recommended it Between degree of association, calculate application k recommendation index；And for the recommendation index according to described application k and default first threshold, Determine whether to recommend described application k for user to be recommended.

Apply recommendation apparatus the most as claimed in claim 6, it is characterised in that:

Described similar users module, including:

Flow rate calculation unit, for described preset duration is divided at least two statistical time range, calculates each user in each statistics The flow that each application is spent is used in period；

Time weighting unit, for calculating the time weighting coefficient of the corresponding each statistical time range of each user；

Interest score unit, uses each user q of same application j, according to described flow for determining with described user to be recommended With time weighting coefficient, calculate each user q interest score to each application；

Correlation unit, for using the application interest score of j, user to be recommended to use the flat of each application according to user to be recommended All interest score, user q use the application interest score of j, user q to use the average interest score of each application, calculate to be recommended Degree of association between user and user q, and according to the degree of association between described user to be recommended and described each user q, determine and treat Recommend the similar users of user.

Apply recommendation apparatus the most as claimed in claim 7, it is characterised in that:

Described time weighting unit, specifically for calculating described time weighting coefficient, described time weighting coefficient f_ia(t) according to Lower formula (1) calculates:

$f_{i_a}\left(t\right)=e^{\[\frac{t_{i_a}}{t_{i_A}}-1\]};---\left(1\right)$

Wherein, i be user to be recommended be user i；

t_iAQuantity for the time period that user i divides；

t_iaFor the user i a period within described A the time period；

I.e. t_ia∈(1…t_ia…t_iA)。

Apply recommendation apparatus the most as claimed in claim 6, it is characterised in that:

Described degree of adhesion freshness module, specifically for calculating degree of adhesion f_vk(t), described degree of adhesion f_vkT the computing formula of () is:

f_vk(t)=e^tvk1； (2)

Wherein, v is similar users v；

t_vk1The time of application k and the time difference accessing application k for the first time is accessed for the last time for similar users v.

Apply recommendation apparatus the most as claimed in claim 6, it is characterised in that:

Described degree of adhesion freshness module, specifically for calculating freshness g_vk(t), described freshness g_vkT the computing formula of () is:

$g_{vk}\left(t\right)={\log }_{\frac{1}{2}}{t}_{vk2};---\left(3\right)$

Wherein, v is similar users v；

t_vk2Represent the last time difference accessing application k and the minima of current time in each similar users v.