CN103383702A - Method and system for recommending personalized news based on ranking of votes of users - Google Patents

Abstract

The invention is applicable to the field of communication, and provides a method for recommending personalized news based on ranking of votes of users. The method comprises the following steps: collecting vote information of the news in which the Internet users are interested; selecting a candidate news recommendation item according to the vote information; through a collaborative filtering recommendation algorithm, predicting the news in which the target users are interested in the candidate news recommendation item; sending the predicted news in which the target users are interested to the target users. The invention further provides a system for recommending the personalized news based on the ranking of votes of the users. The method and system for recommending the personalized news based on ranking of votes of the users can recommend the personalized news to the users quickly and accurately, so that the user experience is improved.

Description

A kind of based on user's method and system that the Personalize News of rank recommends of voting

Technical field

The present invention relates to the communications field, relate in particular to a kind of based on user's method and system that the Personalize News of rank recommends of voting.

Background technology

Solve one of effective ways of information overload the Internet era that commending system being, but also have the problem of cold start-up simultaneously, namely newly join user in system or to be recommended, owing to lacking corresponding historical record, and cause recommending and normally to launch.Another spinoff that cold start-up brings is exactly: the longer project of retention time in system, the probability that obtains recommending is larger.In the main application of commending system, recommended project has common feature at present, and namely the user large variation can not occur to the interest of recommended project in long segment limit.

Yet, when similar system is applied to ageing stronger news recommendation, may be because of the reason of effect of time for news, causing recommended may be " out-of-date " news that the user has lost interest, and that is to say, in the news commending system, due to the ageing and susceptibility of news, be doomed that it has the shorter cycle that exists, often in the scope of a day or even several hours, exceed this actual effect scope, often can not get affirming of user.

Therefore, need a kind of Personalize News commending system of design badly, to realize quick, the accurately recommendation of internet news.

Summary of the invention

In view of this, it is a kind of based on user's method and system that the Personalize News of rank recommends of voting that the purpose of the embodiment of the present invention is to provide, be intended to solve can't realize in prior art internet news fast, the problem of accurately recommending.

The embodiment of the present invention is achieved in that a kind of based on user's method that the Personalize News of rank recommends of voting, and described method comprises:

Gather the Internet user to the vote information of news interested separately;

Select candidate's the news recommended project according to described vote information;

Pass through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project;

The interested news of target of prediction user is sent to described targeted customer.

Preferably, describedly select the step of candidate's the news recommended project specifically to comprise according to described vote information:

Determine the clicking rate rank order of all kinds of news according to described vote information,

Based on the clicking rate rank order of described all kinds of news, select candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor.

Preferably, described in described candidate's the news recommended project step by the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user specifically comprise:

According to browsing record search other users close with described targeted customer's Interest Similarity;

Sort from big to small and select front n other users to form the similar users group according to similarity;

Predict the interested news of described targeted customer according to the record of browsing of user in described similar users group.

Preferably, the described basis step of browsing record search other users close with described targeted customer's Interest Similarity specifically comprises:

Pass through formula

Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j.

Preferably, the described step of browsing the interested news of the record described targeted customer of prediction according to user in described similar users group specifically comprises:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

On the other hand, it is a kind of based on the user system that the Personalize News of rank recommends of voting that the present invention also provides, and described system comprises:

Information acquisition module be used for to gather the Internet user to the vote information of news interested separately;

The candidate selects module, is used for selecting according to described vote information candidate's the news recommended project;

The prediction recommending module is used for passing through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project;

Sending module, be used for the interested news of target of prediction user is sent to described targeted customer as a result.

Preferably, described candidate selects module specifically to comprise:

Order is determined submodule, is used for determining according to described vote information the clicking rate rank order of all kinds of news,

The candidate recommends submodule, is used for the clicking rate rank order based on described all kinds of news, selects candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, T _PresentRepresent the current time, T _posTime when t represents to give a news briefing, G represents the time weighting adjustment factor.

Preferably, described prediction recommending module specifically comprises:

The record search submodule is used for according to browsing record search other users close with described targeted customer's Interest Similarity;

Similar chooser module is used for sorting from big to small and selecting front n other users to form the similar users group according to similarity;

News predictor module is used for predicting the interested news of described targeted customer according to described similar users group user's the record of browsing.

Preferably, described record search submodule specifically is used for passing through formula

Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j.

Preferably, described news predictor module specifically is used for:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

In embodiments of the present invention, technical scheme provided by the invention, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

Description of drawings

Fig. 1 is based on user's method flow diagram that the Personalize News of rank recommends of voting in an embodiment of the present invention;

Fig. 2 is the concrete methods of realizing process flow diagram of step S12 shown in Figure 1 in an embodiment of the present invention;

Fig. 3 is the concrete methods of realizing process flow diagram of step S13 shown in Figure 1 in an embodiment of the present invention;

Fig. 4 is based on user's system architecture schematic diagram that the Personalize News of rank recommends of voting in an embodiment of the present invention;

Fig. 5 is the concrete structure schematic diagram that candidate shown in Figure 4 in an embodiment of the present invention selects module;

Fig. 6 is the concrete structure schematic diagram of prediction recommending module shown in Figure 4 in an embodiment of the present invention.

Embodiment

In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, is not intended to limit the present invention.

It is a kind of based on user's method that the Personalize News of rank recommends of voting that the specific embodiment of the invention provides, and mainly comprises the steps:

S11, gather the Internet user to the vote information of news interested separately;

S12, select candidate's the news recommended project according to described vote information;

S13, in described candidate's the news recommended project by the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user;

S14, the interested news of target of prediction user is sent to described targeted customer.

Provided by the present invention a kind of based on user's method that the Personalize News of rank recommends of voting, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

Below will a kind ofly be elaborated based on user's method that the Personalize News of rank recommends of voting to provided by the present invention.

See also Fig. 1, in an embodiment of the present invention based on user's method flow diagram that the Personalize News of rank recommends of voting.

In step S11, gather the Internet user to the vote information of news interested separately.

In the present embodiment, each in this commending system on internet news option has two buttons that need the Internet user to select, i.e. " UP " button and " DOWN " button, if the user is interested in this news, can select " UP " button, otherwise, if the user loses interest in to this news, can select " DOWN " button.

In the present embodiment, select " UP " button still to select the mode of " DOWN " button to gather the vote information of Internet user's news interested to it by counting user, simultaneously, step S11 not only can be by the internet the user's of colony voting behavior decide the popular degree of news, can also be used for the corrigendum to user profile, and then make the result of recommendation more accurate.

In step S12, select candidate's the news recommended project according to described vote information.

In the present embodiment, select the step S12 of candidate's the news recommended project specifically to comprise step S121-S122 according to described vote information, as shown in Figure 2.

See also Fig. 2, be the concrete methods of realizing process flow diagram of step S12 shown in Figure 1 in an embodiment of the present invention.

In step S121, determine the clicking rate rank order of all kinds of news according to described vote information.In the present embodiment, gathered Internet user's vote information at step S11 after, determine the clicking rate rank order of all kinds of news by these vote information.

In step S122, based on the clicking rate rank order of described all kinds of news, select candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, if P _upGreater than P _DownSign(P _up-P _Down) numerical value be 1, if P _upLess than P _DownSign(P _up-P _Down) numerical value be-1, if P _upEqual P _DownSign(P _up-P _Down) numerical value be 0, T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor.

In the present embodiment, the molecular moiety explanation of described default preference pattern: user's clicks becomes positive correlation with the rank of news, it is the high rank that can improve news of clicks, also be subject to simultaneously the impact of colony's user preferences, both quantitatively more greatly different, impact on the news rank is larger, wherein passes through sign(P _up-P _Down) user is clicked adjust, when using logarithmic function to make the user less, the increase of clicks is larger on the rank impact, its impact on rank is less more at most and click, the phenomenon that so can not get showing because clicks is less with regard to the news that has prevented from newly putting up occurs, and then has solved the existing cold start-up problem of commending system in prior art.

In the present embodiment, the denominator of described default preference pattern has partly embodied the impact of time on the news rank, wherein, and T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor, the G value can be made as according to actual needs the real number greater than 1, so the denominator of described default preference pattern has guaranteed that partly the news information of new issue can obtain relatively large displaying probability, has equally effectively solved the existing cold start-up problem of commending system in prior art.

In the present embodiment, can in time be paid close attention to for the news that guarantees the new issue in internet, when news briefing, be given simultaneously its P _ViewBetween 100-200; and the inferior number average of the corresponding news that user ballot is liked or do not liked is designated as 0; due to the news that is new issue; can obtain thus a relatively large recommended probability; then; when the clicked number of the news of this new issue greatly after certain numerical value between 1000-2000; when calculating the clicking rate rank of all kinds of news; the value that needs initially to give cuts; with the fairness that guarantees to recommend; so the method notes protecting the news of new issue, " Matthew effect " in can the reduction commending system of relative efficiency.

Please continue to consult Fig. 1, in step S13, pass through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project.

In the present embodiment, the step S13 by the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project specifically comprises step S131-S132, as shown in Figure 3.

See also Fig. 3, be the concrete methods of realizing process flow diagram of step S13 shown in Figure 1 in an embodiment of the present invention.

In step S131, according to browsing record search other users close with described targeted customer's Interest Similarity.

In the present embodiment, specifically comprise according to the step S131 that browses record search other users close with described targeted customer's Interest Similarity:

Pass through formula

Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j, | S _i∩ S _j| represent that user i and user j browsed the quantity of identical news, | S _i∪ S _j| expression user i and user j browsed the quantity of all news.

In step S132, sort from big to small and select front n other users to form the similar users group according to similarity.In the present embodiment, n is the natural number greater than 1.

In step S133, predict the interested news of described targeted customer according to the record of browsing of user in described similar users group.

In the present embodiment, the step S133 that browses the interested news of the record described targeted customer of prediction according to user in described similar users group specifically comprises:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

In the present embodiment, m is the natural number greater than 1.

Please continue to consult Fig. 1, in step S14, the interested news of target of prediction user be sent to described targeted customer.

Provided by the present invention a kind of based on user's method that the Personalize News of rank recommends of voting, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

It is a kind of based on the user system 10 that the Personalize News of rank recommends of voting that the specific embodiment of the invention also provides, and mainly comprises:

Information acquisition module 101 be used for to gather the Internet user to the vote information of news interested separately;

The candidate selects module 102, is used for selecting according to described vote information candidate's the news recommended project;

Prediction recommending module 103 is used for passing through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project;

Sending module 104 as a result, are used for the interested news of target of prediction user is sent to described targeted customer.

Provided by the present invention a kind of based on the user system 10 that the Personalize News of rank recommends of voting, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

Below will a kind ofly be elaborated based on the user system 10 that the Personalize News of rank recommends of voting to provided by the present invention.

See also Fig. 4, be depicted as in an embodiment of the present invention the vote structural representation of the system 10 that the Personalize News of rank recommends based on the user.In the present embodiment, comprise that based on user vote system 10 that the Personalize News of rank recommends information acquisition module 101, candidate select module 102, prediction recommending module 103 and sending module 104 as a result.

Information acquisition module 101 be used for to gather the Internet user to the vote information of news interested separately.

In the present embodiment, each in this commending system on internet news option has two buttons that need the Internet user to select, i.e. " UP " button and " DOWN " button, if the user is interested in this news, can select " UP " button, otherwise, if the user loses interest in to this news, can select " DOWN " button.

In the present embodiment, select " UP " button still to select the mode of " DOWN " button to gather the Internet user to the vote information of news interested separately by counting user, the user's of colony that not only can be by the internet voting behavior decides the popular degree of news, can also be used for the corrigendum to user profile, and then make the result of recommendation more accurate.

The candidate selects module 102, is used for selecting according to described vote information candidate's the news recommended project.

In the present embodiment, the candidate selects module 102 to comprise that specifically the definite submodule 1021 of order and candidate recommend submodule 1022, as shown in Figure 5.

See also Fig. 5, select the concrete structure schematic diagram of module 102 for candidate shown in Figure 4 in an embodiment of the present invention.

Order is determined submodule 1021, is used for determining according to described vote information the clicking rate rank order of all kinds of news.In the present embodiment, gathered Internet user's vote information at information acquisition module 101 after, determine the clicking rate rank order of all kinds of news by these vote information.

The candidate recommends submodule 1022, is used for the clicking rate rank order based on described all kinds of news, selects candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, if P _upGreater than P _DownSign(P _up-P _Down) numerical value be 1, if P _upLess than P _DownSign(P _up-P _Down) numerical value be-1, if P _upEqual P _DownSign(P _up-P _Down) numerical value be 0, T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor.

In the present embodiment, the molecular moiety explanation of described default preference pattern: user's clicks becomes positive correlation with the rank of news, it is the high rank that can improve news of clicks, also be subject to simultaneously the impact of colony's user preferences, both quantitatively more greatly different, impact on the news rank is larger, wherein passes through sign(P _up-P _Down) user is clicked adjust, when using logarithmic function to make the user less, the increase of clicks is larger on the rank impact, its impact on rank is less more at most and click, the phenomenon that so can not get showing because clicks is less with regard to the news that has prevented from newly putting up occurs, and then has solved the existing cold start-up problem of commending system in prior art.

In the present embodiment, the denominator of described default preference pattern has partly embodied the impact of time on the news rank, wherein, and T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor, the G value can be made as according to actual needs the real number greater than 1, so the denominator of described default preference pattern has guaranteed that partly the news information of new issue can obtain relatively large displaying probability, has equally effectively solved the existing cold start-up problem of commending system in prior art.

In the present embodiment, can in time be paid close attention to for the news that guarantees the new issue in internet, when news briefing, be given simultaneously its P _ViewBetween 100-200; and the inferior number average of the corresponding news that user ballot is liked or do not liked is designated as 0; due to the news that is new issue; can obtain thus a relatively large recommended probability; then; when the clicked number of the news of this new issue greatly after certain numerical value between 1000-2000; when calculating the clicking rate rank of all kinds of news; the value that needs initially to give cuts; with the fairness that guarantees to recommend; so the method notes protecting the news of new issue, " Matthew effect " in can the reduction commending system of relative efficiency.

Please continue to consult Fig. 4, prediction recommending module 103 is used for passing through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project.

In the present embodiment, prediction recommending module 103 specifically comprises record search submodule 1031, similar chooser module 1032 and news predictor module 1033, as shown in Figure 6.

See also Fig. 6, be the concrete structure schematic diagram of prediction recommending module 103 shown in Figure 4 in an embodiment of the present invention.

Record search submodule 1031 is used for according to browsing record search other users close with described targeted customer's Interest Similarity.

In the present embodiment, record search submodule 1031 is concrete for passing through formula Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j, | S _i∩ S _j| represent that user i and user j browsed the quantity of identical news, | S _i∪ S _j| expression user i and user j browsed the quantity of all news.

Similar chooser module 1032 is used for sorting from big to small and selecting front n other users to form the similar users group according to similarity.In the present embodiment, n is the natural number greater than 1.

News predictor module 1033 is used for predicting the interested news of described targeted customer according to described similar users group user's the record of browsing.

In the present embodiment, news predictor module 1033 specifically is used for:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

In the present embodiment, m is the natural number greater than 1.

Please continue to consult Fig. 4, sending module 104 as a result, are used for the interested news of target of prediction user is sent to described targeted customer.

Provided by the present invention a kind of based on the user system 10 that the Personalize News of rank recommends of voting, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

In embodiments of the present invention, technical scheme provided by the invention, the factors such as ageing, the popular degree of news of the hobby by considering the Internet user, news, and the behavior of browsing by the user of colony is in conjunction with corresponding proposed algorithm, quick, the accurately recommendation of realization to Internet user's Personalize News, and then raising user's experience.

It should be noted that in above-described embodiment, included unit is just divided according to function logic, but is not limited to above-mentioned division, as long as can realize corresponding function; In addition, the concrete title of each functional unit also just for the ease of mutual differentiation, is not limited to protection scope of the present invention.

In addition, one of ordinary skill in the art will appreciate that all or part of step that realizes in the various embodiments described above method is to come the relevant hardware of instruction to complete by program, corresponding program can be stored in a computer read/write memory medium, described storage medium is as ROM/RAM, disk or CD etc.

The above is only preferred embodiment of the present invention, not in order to limiting the present invention, all any modifications of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., within all should being included in protection scope of the present invention.

Claims (10)

1. one kind based on user's method that the Personalize News of rank recommends of voting, and it is characterized in that, described method comprises:

Gather the Internet user to the vote information of news interested separately;

Select candidate's the news recommended project according to described vote information;

Pass through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project;

The interested news of target of prediction user is sent to described targeted customer.

2. as claimed in claim 1ly it is characterized in that based on user's method that the Personalize News of rank recommends of voting, describedly select the step of candidate's the news recommended project specifically to comprise according to described vote information:

Determine the clicking rate rank order of all kinds of news according to described vote information,

Based on the clicking rate rank order of described all kinds of news, select candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor.

3. as claimed in claim 1ly it is characterized in that based on user's method that the Personalize News of rank recommends of voting, described in described candidate's the news recommended project step by the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user specifically comprise:

According to browsing record search other users close with described targeted customer's Interest Similarity;

Sort from big to small and select front n other users to form the similar users group according to similarity;

Predict the interested news of described targeted customer according to the record of browsing of user in described similar users group.

4. as claimed in claim 3ly it is characterized in that based on user's method that the Personalize News of rank recommends of voting, the step that described basis is browsed record search other users close with described targeted customer's Interest Similarity specifically comprises:

Pass through formula

Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j.

5. as claimed in claim 3ly it is characterized in that based on user's method that the Personalize News of rank recommends of voting, the described step of browsing the interested news of the record described targeted customer of prediction according to user in described similar users group specifically comprises:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

6. one kind based on the user system that the Personalize News of rank recommends of voting, and it is characterized in that, described system comprises:

Information acquisition module be used for to gather the Internet user to the vote information of news interested separately;

The candidate selects module, is used for selecting according to described vote information candidate's the news recommended project;

The prediction recommending module is used for passing through the interested news of Collaborative Filtering Recommendation Algorithm target of prediction user in described candidate's the news recommended project;

Sending module, be used for the interested news of target of prediction user is sent to described targeted customer as a result.

7. as claimed in claim 6ly it is characterized in that based on the user system that the Personalize News of rank recommends of voting, described candidate selects module specifically to comprise:

Order is determined submodule, is used for determining according to described vote information the clicking rate rank order of all kinds of news,

The candidate recommends submodule, is used for the clicking rate rank order based on described all kinds of news, selects candidate's the news recommended project by default preference pattern;

Wherein, described default preference pattern is specific as follows:

$\frac{{\log }_{\mathrm{base}}{P}_{\mathrm{view}}+\mathrm{sign}(P_{\mathrm{up}}-P_{\mathrm{down}}){\log }_{\mathrm{base}}|P_{\mathrm{up}}-P_{\mathrm{down}}|}{{\left(\frac{T_{\text{present}}-T_{\mathrm{post}}}{1000*3600}\right)}^G};$

Wherein, the natural number of the setting value of base between 10-20, P _ViewThe total degree of expression user's all kinds of news of click, P _upThe number of times of corresponding news, P are liked in expression user ballot _DownThe number of times of corresponding news, sign(P are not liked in expression user ballot _up-P _Down) the expression sign bit, T _PresentRepresent the current time, T _PostTime during the expression news briefing, G represents the time weighting adjustment factor.

8. as claimed in claim 6ly it is characterized in that based on the user system that the Personalize News of rank recommends of voting, described prediction recommending module specifically comprises:

The record search submodule is used for according to browsing record search other users close with described targeted customer's Interest Similarity;

Similar chooser module is used for sorting from big to small and selecting front n other users to form the similar users group according to similarity;

News predictor module is used for predicting the interested news of described targeted customer according to described similar users group user's the record of browsing.

9. as claimed in claim 8ly it is characterized in that based on the user system that the Personalize News of rank recommends of voting, described record search submodule specifically is used for passing through formula

Search other users close with described targeted customer's Interest Similarity;

Wherein, S _iThe set of the news that expression user i had browsed, S _jThe set of the news that expression user j had browsed, sim _i,jThe Interest Similarity of expression user i and user j.

10. as claimed in claim 8ly it is characterized in that based on the user system that the Personalize News of rank recommends of voting, described news predictor module specifically is used for:

By the formula ∑ _{U ∈ M}sim _u/ | M| calculates described targeted customer to the possible score value of each news;

According to the height of the described targeted customer who calculates to the possible score value of each news, before selecting, m bar news is as predicting the outcome.

Images