CN101266620A - Method and apparatus for providing target information to user - Google Patents

Abstract

The invention discloses a method of providing target information for users, comprising: gaining historical behavior data of the user for browsing information, wherein, the historical behavior data comprises content of the information and browsing time; processing to classify for the content of the information to gain a probability of the information vesting in setting category; gaining an interesting value of the user in setting time according to the probability and the browsing time; providing the target information for the user at the setting time according to the interesting value. The invention simultaneously discloses a device for providing the target information for the user. Adopting the invention can quantize and multi-granularity describe user interest for the information according to user requirement, dynamically reflect a changing process of the user interest, and simplify user operation.

Description

The method and apparatus of target information is provided to the user

Technical field

The present invention relates to networking technology area, relate in particular to the method and apparatus that target information is provided to the user.

Background technology

Have the information of magnanimity on the internet, abundant in content, form is different.The network user wishes to obtain the information that satisfies demands of individuals from the information resources of vastness.Search engine, META Search Engine and other some research tools can help our information extraction from huge information network.General can provide less search key when the user retrieves, and search engine can return the Search Results of enormous amount.If lack the interest analysis to the user, if promptly do not set up user model accurately, the user will be submerged in the ocean of information.

This shows, set up that user model is extremely important for obtaining of ownership goal information accurately.User modeling has become the core research contents of network data excavation, such as, personalized search, advertising objective client location, information recommendation, market business decision-making and customer relation management or the like.

The main at present method that has two kinds of user modelings, a kind of is static subscriber's modeling, a kind of is dynamic subscriber's modeling.

The personal information that static subscriber's modeling analysis user provides, such as, log-on message and survey.The inventor by analysis, the method of discovery static subscriber modeling can only be done a rough description to user's interest, there are some problems in actual applications, at first, because this user model is static, therefore the user interest of model description is only effective in a certain period, can not reflect the situation of change of following user interest; Secondly, the personal information that the user submits to is a kind of input of subjectivity, can not objectively respond the feature of its interest.

Activity and the behavior of dynamic subscriber's modeling analysis user on the website, the inventor finds that this model is very little to the description granularity of user model by analysis, specific to some points of interest, the description that neither one is summarized.

In addition, also have the method for some other user modeling to need the user that feedback information is provided, can make user's operation become complicated like this, influence the normal behavior of user.

Summary of the invention

The embodiment of the invention provides a kind of and provides the method for target information to the user, and in order to quantizing and to describe the interest of user to information by the many granularities of user's request ground, the change procedure of dynamic reflection user interest is simplified user's operation, and this method comprises:

Obtain the historical behavior data of user's browsing information, described historical behavior data comprise the content of information and the moment of browsing;

Content to described information is classified, and obtains described information and belongs to the probability of setting classification;

According to the described probability and the described moment of browsing, obtain the user and setting interest value constantly;

According to described interest value, constantly provide target information to the user in described setting.

Preferable, described historical behavior data also comprise user ID;

According to user ID, described probability and the described moment of browsing, obtain different user and setting interest value constantly;

According to the user ID of described interest value and reception, constantly provide target information to relative users in described setting.

Preferable, described user ID is user's login name or IP address.

Preferable, the content of described information is classified, obtain described information and belong to the probability of setting classification and comprise:

Content to described information is classified, and obtains the number of the classification of described information ownership;

Number according to the classification of described information ownership obtains described information and belongs to the probability of setting classification.

Preferable, according to the described probability and the described moment of browsing, obtain the user and comprise in the interest value of setting the moment:

By following formula, obtain the interest value of user in the described moment of browsing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}{\mathrm{weight}}_0}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t})$

Wherein, k is a forgetting factor, k _αBe memory fact, weight is described probability, and t is the described moment of browsing;

By following formula, obtain the user and setting interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t})$

Wherein, t is the described moment of browsing, and t+ τ is the described setting moment, time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is described probability.

Preferable, described information is the object in website, webpage or the webpage.

Preferable, according to described interest value, constantly provide target information to comprise to the user in described setting:

Described interest value and threshold value are carried out size relatively;

Described set constantly to the user provide the interest value that is not less than described threshold value the information of corresponding classification.

The embodiment of the invention also provides a kind of and provides the equipment of target information to the user, and in order to quantizing and to describe the interest of user to information by the many granularities of user's request ground, the change procedure of dynamic reflection user interest is simplified user's operation, and this equipment comprises:

Acquisition module is used to obtain the historical behavior data of user's browsing information, and described historical behavior data comprise the content of information and the moment of browsing;

Sort module is used for the content of described information is classified, and obtains described information and belongs to the probability of setting classification;

Processing module was used for according to the described probability and the described moment of browsing, and obtained the user and was setting interest value constantly;

Module is provided, is used for, constantly provide target information to the user in described setting according to described interest value.

Preferable, described historical behavior data also comprise user ID;

Described processing module was further used for according to user ID, described probability and the described moment of browsing, and obtained different user and was setting interest value constantly;

The described module that provides comprises:

Receiving element is used to receive user ID;

First provides the unit, is used for the user ID according to described interest value and reception, constantly provides target information to relative users in described setting.

Preferable, described sort module comprises:

First taxon is used for the content of described information is classified, and obtains the number of the classification of described information ownership;

Second taxon is used for the number according to the classification of described information ownership, obtains described information and belongs to the probability of setting classification.

Preferable, described processing module comprises:

First processing unit is used for by following formula, obtains the interest value of user in the described moment of browsing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}{\mathrm{weight}}_0}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, k is a forgetting factor, k _αBe memory fact, weight is described probability, and t is the described moment of browsing;

Second processing unit is used for by following formula, obtains the user and is setting interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, t is the described moment of browsing, and t+ τ is the described setting moment, time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is described probability.

Preferable, the described module that provides comprises:

Comparing unit is used for described interest value and threshold value are carried out size relatively;

Second provides the unit, be used for described set constantly to the user provide the interest value that is not less than described threshold value the information of corresponding classification.

In the embodiment of the invention, by obtaining the historical behavior data of user's browsing information, described historical behavior data comprise the content of information and the moment of browsing; Content to described information is classified, and obtains described information and belongs to the probability of setting classification; According to the described probability and the described moment of browsing, obtain the user and setting interest value constantly; According to described interest value, constantly provide target information in described setting to the user, not only can quantize and describe the interest of user information by the many granularities of user's request ground, the change procedure of dynamic reflection user interest, interest trend to user future is made a prediction, also need not the user during enforcement feedback information is provided, make the user operate relative simplification.

Description of drawings

Fig. 1 is for providing the process flow diagram of target information to the user in the embodiment of the invention;

The information that Fig. 2 browses for user in the embodiment of the invention belongs to the probability curve diagram of setting classification;

Fig. 3 is for providing the structural representation of the equipment of target information to the user in the embodiment of the invention;

Fig. 4, Fig. 7 are for providing the structural representation of module in the embodiment of the invention;

Fig. 5 is the structural representation of sort module in the embodiment of the invention;

Fig. 6 is the structural representation of processing module in the embodiment of the invention.

Embodiment

Below in conjunction with Figure of description the embodiment of the invention is elaborated.

As shown in Figure 1, in the embodiment of the invention, provide the flow process of target information as follows to the user:

Step 11, obtain the historical behavior data of user's browsing information, these historical behavior data comprise the content of information and the moment of browsing.

Step 12, the content of information is classified, obtain this information and belong to the probability of setting classification.

Step 13, according to the probability that obtains and the moment of browsing, obtain the user and setting interest value constantly.

Step 14, according to this interest value, constantly provide target information setting to the user.

In the flow process shown in Figure 1, the information that the user browsed can be the website, also can be webpage, can also be the object in the webpage, specifically can be according to the user's request setting.Those of ordinary skills understand easily, in different application scenarios, can carry out above-mentioned processing at the needed target information of user, greatly to a website, little commodity in webpage, all applicable embodiment of the invention method, thus realize describing the user to the interested degree of information by the many granularities of user's request ground.

Among the embodiment, when being a plurality of, the historical behavior data in the step 11 can also comprise user ID, in order to the unique identification user the user.This user ID can be login name, IP address of user etc.

Table 1 is that webpage is an example with the information that the user was browsed, and user's historical behavior data are described:

Table 1 user's historical behavior data

User ID	The moment of browsing	The content of information (only listing the chained address of the information content herein)
			userID1	2008-01-01	http://idoican/page1.html
userID2	2008-01-02	http://idoican/2008.mp3
			userID1	2008-01-04	http://idoican/page2.html

If the user is a plurality of, then in step 12, can obtain different user and set interest value constantly according to user ID, described probability and the described moment of browsing.According to requirements of different users, can preset different taxonomic hierarchieses, can be individual layer, also can be multilayer, can also be other sorting technique.No matter adopt what sorting technique, its final purpose all be the content classification of information that the user is browsed in some or certain several classification, and obtain the Probability p that information that the user browses belongs to some classifications.For example, in the table 1, the information content of http://idoican/page1.html belongs to Probability p=0.8 of news category, the information content of http://idoican/2008.mp3 belongs to Probability p=1.0 of music categories, and the information content of http://idoican/page2.html belongs to the Probability p of Sport Class=0.6.

Among the embodiment, can classify to the content of information earlier, obtain the number of the classification of this information ownership, the number of follow-up classification according to this information ownership obtains this information and belongs to the probability of setting classification.For example, certain information had both comprised the content of news category, comprised the content of music categories again, also comprised the content of Sport Class, and then this information can belong to news, music, 3 classifications of physical culture, and probability is p=0.33.And for example, certain information only comprises the content of news category, and then this information belongs to 1 classification of news, and probability is p=1.Also can take the concrete formation of the information content in the enforcement into consideration, for example, the content of certain information 80% belongs to news category, 20% the happy classification of content dominant, can think that then the probability that this information belongs to news category is p=0.8, the probability that belongs to music categories is p=0.2.

Among the embodiment, after the content of the information that the user is browsed is classified according to taxonomic hierarchies, can in each classification, sort to probability, can get sorted historical behavior data as shown in table 2 after data processing is finished according to time sequencing.Wherein, Topic is the some classifications in the taxonomic hierarchies, and weight is that the information that the user browses belongs to the probability of setting classification.

The sorted historical behavior data of table 2

With table 1 be applied to table 2 for example, can obtain table 3:

The instantiation of the sorted historical behavior data of table 3

Among the embodiment, forget rule according to great this memory of Chinese mugwort guest, 1. step 14 can by formula obtain the interest value of user in the moment of browsing when implementing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}{\mathrm{weight}}_0}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t})$ ①

Wherein, k is a forgetting factor, for example gets k=0.1; k _αBe memory fact, for example get k _α=0.9; Weight is a probability, and t is the moment of browsing.As seen, formula 1. describe the user in the moment of browsing the degree to some this interest of classification, be worth greatly more, interest is big more, actual is a kind of quantization means of interest.

Derive by iterative relation, can by formula 2. obtain the user and set interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t})$ ②

Wherein, t is the moment of browsing, and t+ τ is for setting constantly time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is a probability.This shows that at t+ τ constantly, user's interest value is the superposition value of last moment interest value of leaving over and the interest value that this moment increases newly.

Bring the time series of each topic of each user in the table 2 into interest model that 2. formula can obtain each user, in each user interest model, the corresponding interest value Z (t) of each topic just can dope certain time user's in future interests change according to Z (t).

By formula shown in Figure 2 function curve diagram 2. as can be known, the variation that the interest value in the embodiment of the invention can the dynamic reflection user interest, as the generation of new interest, You Xingqu disappearance or enhancing.All can calculate the interest value of user at any time to certain classification information.Fig. 2 only is an example, and as can be seen from the figure, the user strengthens to some extent to the interest of cuisines (cooking) and education (education) these two classifications, and the interest of this classification of family (home) weakens to some extent.

Can calculate the interest value of user to different information according to the user model in the embodiment of the invention, during browsing information, commending contents that can interest value is big is given the user to the user on the website.Specifically when implementing, can set a threshold value, interest value and this threshold value of calculating gained are carried out size relatively; Follow-up set constantly to the user provide the interest value that is not less than this threshold value the information of corresponding classification.

If when the user was a plurality of, the user need submit user ID to, step 14 can constantly provide target information to relative users in setting according to the interest value of aforementioned acquisition and the user ID of reception at embodiment.

Based on same inventive concept, the present invention also provides a kind of and provides the equipment of target information to the user, and its structure comprises as shown in Figure 3: acquisition module 31, sort module 32, processing module 33, provide module 34; Wherein, acquisition module 31 is used to obtain the historical behavior data of user's browsing information, and these historical behavior data comprise the content of information and the moment of browsing; Sort module 32 is used for the content of information is classified, and obtains this information and belongs to the probability of setting classification; Processing module 33 was used for according to this probability and moment of browsing, and obtained the user and was setting interest value constantly; Module 34 is provided, is used for, constantly provide target information to the user in setting according to this interest value.

Among the embodiment, the historical behavior data also comprise user ID; At this moment, processing module 33 can also be used for according to the probability of user ID, aforementioned acquisition and the moment of browsing, and obtained different user and was setting interest value constantly; As shown in Figure 4, provide module 34 to comprise this moment: receiving element 341 is used to receive user ID; First provides unit 342, is used for the user ID according to interest value and reception, constantly provides target information to relative users in setting.

As shown in Figure 5, among the embodiment, sort module 32 can comprise: first taxon 321 is used for the content of information is classified the number of the classification of acquired information ownership; Second taxon 322 is used for the number according to the classification of information ownership, and acquired information belongs to the probability of setting classification.

As shown in Figure 6, among the embodiment, processing module 33 comprises: first processing unit 331, be used for by following formula, and obtain the interest value of user in the moment of browsing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}{\mathrm{weight}}_0}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, k is a forgetting factor, k _αBe memory fact, weight is a probability, and t is the moment of browsing; Second processing unit 332 is used for by following formula, obtains the user and is setting interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, t is the moment of browsing, and t+ τ is for setting constantly time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is described probability.

As shown in Figure 7, among the embodiment, provide module 34 to comprise: comparing unit 343 is used for interest value and threshold value are carried out size relatively; Second provides unit 344, be used for set constantly to the user provide the interest value that is not less than threshold value the information of corresponding classification.

One of ordinary skill in the art will appreciate that all or part of step in the foregoing description method is to instruct relevant hardware to finish by program, this program can be stored in the computer-readable recording medium, and storage medium can comprise: ROM, RAM, disk or CD etc.

In the embodiment of the invention, by obtaining the historical behavior data of user's browsing information, these historical behavior data comprise the content of information and the moment of browsing; Content to information is classified, and obtains this information and belongs to the probability of setting classification; According to the probability that obtains and the moment of browsing, obtain the user and setting interest value constantly; According to interest value, constantly provide target information in setting to the user, not only can quantize and describe the interest of user information by the many granularities of user's request ground, the change procedure of dynamic reflection user interest, interest trend to user future is made a prediction, also need not the user during enforcement feedback information is provided, make the user operate relative simplification.

Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.

Claims (12)

1, a kind of method that provides target information to the user is characterized in that this method comprises:

Obtain the historical behavior data of user's browsing information, described historical behavior data comprise the content of information and the moment of browsing;

Content to described information is classified, and obtains described information and belongs to the probability of setting classification;

According to the described probability and the described moment of browsing, obtain the user and setting interest value constantly;

According to described interest value, constantly provide target information to the user in described setting.

2, the method for claim 1 is characterized in that, described historical behavior data also comprise user ID;

According to user ID, described probability and the described moment of browsing, obtain different user and setting interest value constantly;

According to the user ID of described interest value and reception, constantly provide target information to relative users in described setting.

3, method as claimed in claim 2 is characterized in that, described user ID is user's login name or IP address.

4, the method for claim 1 is characterized in that, the content of described information is classified, and obtains described information and belongs to the probability of setting classification and comprise:

Content to described information is classified, and obtains the number of the classification of described information ownership;

Number according to the classification of described information ownership obtains described information and belongs to the probability of setting classification.

5, the method for claim 1 is characterized in that, according to the described probability and the described moment of browsing, obtains the user and comprises in the interest value of setting the moment:

By following formula, obtain the interest value of user in the described moment of browsing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}\mathrm{weigh}{t}_0}{k_{\mathrm{\α}}-k}\left(e^{-\mathrm{kt}}{-e}^{-k_{\mathrm{\α}}t}\right)$

Wherein, k is a forgetting factor, k _αBe memory fact, weight is described probability, and t is the described moment of browsing;

By following formula, obtain the user and setting interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t})$

Wherein, t is the described moment of browsing, and t+ τ is the described setting moment, time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is described probability.

6, the method for claim 1 is characterized in that, described information is the object in website, webpage or the webpage.

7, as each described method of claim 1 to 6, it is characterized in that,, constantly provide target information to comprise to the user in described setting according to described interest value:

Described interest value and threshold value are carried out size relatively;

Described set constantly to the user provide the interest value that is not less than described threshold value the information of corresponding classification.

8, a kind ofly provide the equipment of target information, it is characterized in that, comprising to the user:

Acquisition module is used to obtain the historical behavior data of user's browsing information, and described historical behavior data comprise the content of information and the moment of browsing;

Sort module is used for the content of described information is classified, and obtains described information and belongs to the probability of setting classification;

Processing module was used for according to the described probability and the described moment of browsing, and obtained the user and was setting interest value constantly;

Module is provided, is used for, constantly provide target information to the user in described setting according to described interest value.

9, equipment as claimed in claim 8 is characterized in that, described historical behavior data also comprise user ID;

Described processing module was further used for according to user ID, described probability and the described moment of browsing, and obtained different user and was setting interest value constantly;

The described module that provides comprises:

Receiving element is used to receive user ID;

First provides the unit, is used for the user ID according to described interest value and reception, constantly provides target information to relative users in described setting.

10, equipment as claimed in claim 8 is characterized in that, described sort module comprises:

First taxon is used for the content of described information is classified, and obtains the number of the classification of described information ownership;

Second taxon is used for the number according to the classification of described information ownership, obtains described information and belongs to the probability of setting classification.

11, equipment as claimed in claim 8 is characterized in that, described processing module comprises:

First processing unit is used for by following formula, obtains the interest value of user in the described moment of browsing:

$Z\left(t\right)=\frac{k_{\mathrm{\α}}{\mathrm{weight}}_0}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, k is a forgetting factor, k _αBe memory fact, weight is described probability, and t is the described moment of browsing;

Second processing unit is used for by following formula, obtains the user and is setting interest value constantly:

$Z^{\left(n\right)}\left(t\right)=Z^{(n-1)}(t+\mathrm{\τ})+\frac{k_{\mathrm{\α}}{\mathrm{weight}}_n}{k_{\mathrm{\α}}-k}(e^{-\mathrm{kt}}-e^{-k_{\mathrm{\α}}t});$

Wherein, t is the described moment of browsing, and t+ τ is the described setting moment, time _n≤ t≤time _N+1, τ=time _n-time _N-1, weight is described probability.

As each described equipment of claim 8 to 11, it is characterized in that 12, the described module that provides comprises:

Comparing unit is used for described interest value and threshold value are carried out size relatively;

Second provides the unit, be used for described set constantly to the user provide the interest value that is not less than described threshold value the information of corresponding classification.

Images