CN105956161A - Information recommendation method and apparatus - Google Patents

Abstract

An embodiment of the invention provides an information recommendation method and relates to the technical field of advertisement systems. The method comprises the steps of making statistics on a weight of each piece of network address information in each search word according to a search behavior history record of each user; making statistics on a click frequency of each piece of network address information in each user identifier according to a browsing behavior history record of each user; according to the click frequency of each piece of the network address information in each user identifier and the weight of each piece of the network address information in each search word, calculating a weight of each search word of each piece of the network address information in each user identifier; and when the user accesses new network address information, obtaining the user identifier and the network address information, generating recommendation information according to the search word with the relatively high weight, and performing display. According to the method, the problem that an advertisement system cannot recommend accurate information for a network address not accessed by the user through a search behavior is solved and the effects of improving the accuracy of information recommendation during direct access of the network address by the user and reducing system resource waste are achieved.

Description

A kind of information recommendation method and device

Technical field

The present invention relates to ad system technical field, particularly relate to a kind of information recommendation method and device.

Background technology

In ad system, system usually can be by drawing the demand of user to the analysis of user's routine behavior Purpose, thus recommend, to user, the information that some users need.

In actual applications, ad system can obtain, according to the search behavior of user, the letter that user needs Breath, wherein, the search behavior of user includes: user inputs in the search function bar that ad system provides Search word and user input search word etc. in other search engines, and ad system is according to above-mentioned user behavior Analysis result generates recommendation information, and shows user on ad system screen.Such as, user wants Obtain the information about " school ", search box or search engine input search word " school ", is According to the above-mentioned search behavior analysis of user, system can show that " school " is the information that user needs, thus can To recommend the relevant information about " school " to user.

, owing to there is not search behavior, extensively in the website information accessed for user not by search behavior Announcement system cannot know the intention of user, thus cannot recommend information accurately for user, so that waste system System resource.

Summary of the invention

In view of the above problems, it is proposed that the present invention is to provide one to overcome the problems referred to above or at least partly Ground solves information recommendation method and the device of the problems referred to above.

According to one aspect of the present invention, it is provided that a kind of information recommendation method, including:

Search behavior historical record according to each user, adds up each website information under each search word Weight；

Navigation patterns historical record according to each user, adds up each network address letter under each ID The number of clicks of breath；

Each under number of clicks according to each website information under each ID and each search word The weight of website information, calculates the power of each search word of each website information under each ID Weight；

When an ID place client newly accesses a website information, described recommendation request obtain and use Family mark and website information, and according to each search word of the described website information under described ID Weight, extracts and sorts at least one forward search word and generate recommendation information based on described search word；

Return described recommendation information to be shown to client.

Preferably, the described search behavior historical record according to each user, add up under each search word The step of the weight of each website information, including:

For the website information under a search word, the number of clicks of described website information is searched divided by described Under rope word, the number of clicks sum of all website information, obtains the first result；

Described first result is carried out smoothing computation, obtains described website information power under described search word Weight.

Preferably, described described first result is carried out smoothing computation, obtain described website information described The step of the weight under search word, including:

The number of clicks of website information all under described search word is taken the logarithm with non-zero normal number sum, To smoothing parameter；

Described first result is multiplied with smoothing parameter, obtains described website information under described search word Weight.

Preferably, described when an ID place client newly accesses a website information, described user The weight of each search word of the described website information under Biao Shi, extracts at least one search that sequence is forward Word step based on described search word generation recommendation information, including:

Receive the recommendation request that client sends；Described recommendation request is in an ID place client Send to Advertisement Server when accessing a website information；

ID and website information is obtained by described recommendation request, and according to the institute under described ID State the weight of each search word of website information, extract at least one forward search word of sequence and based on institute State search word and generate recommendation information.

Preferably, the described step generating recommendation information based on described search word, including:

The advertisement link of corresponding described search word is searched from advertisement base；

Recommendation information is generated based on described advertisement link and search word.

Preferably, return described recommendation information to after the step that client is shown, also include:

Receive the access request in the client clicking operation of a recommendation information triggered by user；

According to described access request, return corresponding advertisement page to client.

Preferably, the step that the described recommendation information of described return is shown to client, including:

Return described recommendation information to browser client, to generate in pop-up in browser client Row is shown.

Preferably, the described search behavior historical record according to each user, add up under each search word The step of the weight of each website information, including:

To the website information of record under search word each in search history behavior record, according to website information institute The theme merger of corresponding content；

Add up under each search word the weight of the website information after merger.

Preferably, the described navigation patterns historical record according to each user, add up under each ID The step of number of clicks of each website information, including:

To the website information in the navigation patterns historical record of each user, in corresponding to website information The theme merger held；

Add up under each ID the number of clicks of the website information after merger.

According to another aspect of the present invention, it is provided that a kind of information recommending apparatus, including:

Website information weight statistical module, for the search behavior historical record according to each user, statistics The weight of each website information under each search word；

Website information number of clicks statistical module, for the navigation patterns historical record according to each user, Add up the number of clicks of each website information under each ID；

Search word weight computation module, for the click according to each website information under each ID time The weight of each website information under number and each search word, calculates each network address under each ID The weight of each search word of information；

Recommendation information generation module, for newly accessing a website information in an ID place client Time, according to the weight of each search word of the described website information under described ID, extract sequence forward At least one search word and based on described search word generate recommendation information；

Recommendation information display module, is used for returning described recommendation information and is shown to client.

Preferably, described website information weight statistical module, including:

First result calculating sub module, for for the website information under a search word, by described network address The number of clicks of information, divided by the number of clicks sum of website information all under described search word, obtains first Result；

First result smooths submodule, for described first result is carried out smoothing computation, obtains described net Location information weight under described search word.

Preferably, described first result smooths submodule, including:

Smoothing parameter obtains submodule, for the number of clicks of website information all under described search word with Non-zero normal number sum is taken the logarithm, and obtains smoothing parameter；

Website information weight the first calculating sub module, for described first result is multiplied with smoothing parameter, Obtain described website information weight under described search word.

Preferably, described recommendation information generation module, including:

Recommendation request receives submodule, for receiving the recommendation request that client sends；Described recommendation request For sending to Advertisement Server when ID place client-access one website information；

Recommendation information first generates submodule, for being obtained ID and network address letter by described recommendation request Breath, and the weight of each search word according to the described website information under described ID, extract sequence At least one forward search word also generates recommendation information based on described search word.

Preferably, described recommendation information first generates submodule, including:

Advertisement link searches submodule, for searching the advertisement link of corresponding described search word from advertisement base；

Recommendation information second generates submodule, for generating recommendation based on described advertisement link and search word Breath.

Preferably, it is recommended that after information display module, also include:

Clicking operation receiver module, for receiving the click to a recommendation information in the client by user The access request that operation is triggered；

Page jump module, for according to described access request, returns corresponding advertisement page to client.

Preferably, described recommendation information display module, including:

Submodule is shown in pop-up, is used for returning described recommendation information to browser client, with at browser Client generates in pop-up and is shown.

Preferably, described website information weight statistical module, including:

Website information the first merger submodule, for remembering under search word each in search history behavior record The website information of record, according to the theme merger of content corresponding to website information；

Website information weight merger submodule, for adding up under each search word the website information after merger Weight.

Preferably, described website information number of clicks statistical module, including:

Website information the second merger submodule, for the net in the navigation patterns historical record of each user Location information, according to the theme merger of content corresponding to website information；

Number of clicks merger submodule, for adding up under each ID the point of the website information after merger Hit number of times.

Information recommendation method according to the present invention and device, can add up from search behavior historical record and search The weight of website information under rope word, the website information under counting user mark from navigation patterns historical record Number of clicks.So for an ID, each website information that it is accessed, can basis The weight of each website information of each aforementioned search word obtained, and this ID is to each website information Access weight, associates with website information, calculates the weight of each search word under this ID.So, When user takes notice of with this ID and to access certain network address in client, ad system then can extract power One or more search word reordering forward, and according to this search word to user's recommendation information, thus Solve the network address that user is accessed not by search behavior, owing to there is not search behavior, advertisement System cannot know the intention of user, thus cannot recommend the problem of information accurately for user, achieves Can the situation of access network address direct to user, search the intention relevant to this network address, such that it is able to recommend Accurate information, reduces the beneficial effect of the waste of system resource.

Described above is only the general introduction of technical solution of the present invention, in order to better understand the skill of the present invention Art means, and can being practiced according to the content of description, and in order to allow, the present invention's is above and other Objects, features and advantages can become apparent, below especially exemplified by the detailed description of the invention of the present invention.

Accompanying drawing explanation

By reading the detailed description of hereafter preferred implementation, various other advantage and benefit for this Field those of ordinary skill will be clear from understanding.Accompanying drawing is only used for illustrating the purpose of preferred implementation, And it is not considered as limitation of the present invention.And in whole accompanying drawing, be denoted by the same reference numerals Identical parts.In the accompanying drawings:

Fig. 1 shows the flow chart of steps of the information recommendation method embodiment one of the present invention；

Fig. 2 shows the flow chart of steps of the information recommendation method embodiment two of the present invention；

Fig. 3 shows the flow chart of steps of the information recommendation method embodiment three of the present invention；

Fig. 4 shows the structured flowchart of the information recommendation system embodiment four of the present invention；

Fig. 5 shows the structured flowchart of the information recommending apparatus embodiment five of the present invention；

Fig. 6 shows the structured flowchart of the information recommending apparatus embodiment six of the present invention.

Detailed description of the invention

It is more fully described the exemplary embodiment of the disclosure below with reference to accompanying drawings.Although accompanying drawing shows The exemplary embodiment of the disclosure, it being understood, however, that may be realized in various forms the disclosure and not Should be limited by embodiments set forth here.On the contrary, it is provided that these embodiments are able to more thoroughly Understand the disclosure, and complete for the scope of the present disclosure can be conveyed to those skilled in the art.

The core concept of the present invention is, adds up website information under search word from search behavior historical record Weight, the number of clicks of website information under counting user mark from navigation patterns historical record.That For an ID, each website information that it is accessed, can according to each aforementioned obtain search The weight of each website information of rope word, and the access weight that this ID is to each website information, with net Location information association, calculates the weight of each search word under this ID.So, user with this user Mark takes notice of that when accessing certain network address in client, ad system then can extract that weight sequencing is forward Individual or multiple search words, and according to this search word to user's recommendation information, thus solve for user The network address accessed not by search behavior, owing to there is not search behavior, ad system cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Embodiment one

With reference to Fig. 1, it is shown that according to the steps flow chart of a kind of information recommendation method embodiment one of the present invention Figure, specifically may include steps of:

Step 110, search behavior historical record according to each user, that adds up under each search word is each The weight of individual website information；

The system architecture of the embodiment of the present invention may include that client, ad system, third-party server, Wherein, client can be browser or other software browsing website information can be provided；Ad system According to user, navigation patterns historical record and the search behavior historical record of website information are calculated each searching The weight of each website information under rope word generating according to the search word that weight under website information is bigger pushes away Recommend information；Third-party server receives the client access request to website information, and does access request Response message is returned after respective handling.

It is appreciated that the system architecture of the embodiment of the present invention can also include search engine server, advertisement System can obtain search behavior historical record from search engine server.

In actual applications, user can be provided by ad system function of search or other various search Engine obtains the information relevant to search word, and therefore, user the most once searches for, Search engine will record this time search behavior of user, when user is carried out for several times after search behavior, and this is searched Index holds up the search behavior historical record that can generate user, so the server of ad system can be searched from this This search behavior historical record of middle acquisition held up in index.

Wherein, in search behavior historical record, every record can include following information: what user inputted searches The website information that rope word and user click under search word.Wherein, website information can be URL (Uniform Resource Locator, URL) form, URL includes: resources-type Type, deposit the host name of resource, resource file name, wherein, resource type can be https agreement or Http etc., host name refers to deposit the domain name system host name of the server of resource or IP address, resource file Name refers to resource file title.

It is appreciated that it practice, for a search word, be to use this search word to carry out all users The historical record clicked on merges, and then adds up the weight of each website information under this search word.Such as, User A by search word W1 click on website information N1, the number of times of N2, N3 be respectively C1, C2, It is logical that C3, user B are respectively C4, user C by the number of times of search word W1 click website information N3 Cross search word W1 click website information N1, the number of times of N2 is respectively C5, C6.Then, by useful for institute The historical record that family uses search word W1 to click on each website information obtains after merging: under search word W1 The number of clicks of website information N1 is that user A clicks on the number of times of website information N1 by search word W1 C1, user C click on the number of times C5 sum of website information N1, search word W1 by search word W1 The number of clicks of lower website information N2 is that user A clicks on the secondary of website information N2 by search word W1 Number C2, user C click on the number of times C6 sum of website information N2, search word by search word W1 Under W1, the number of clicks of website information N3 is that user A clicks on website information N3 by search word W1 Number of times C3, user B by search word W1 click on website information N3 number of times C4 sum.

Preferably, step 110 can include sub-step 111 to 112:

Sub-step 111, for the website information under a search word, by the click of described website information time Number, divided by the number of clicks sum of website information all under described search word, obtains the first result；

Equation below specifically can be used to calculate:

$F(Q_n,I_{n,i})=\frac{C(Q_n,I_{n,i})}{\underset{i=1}{\overset{k_n}{\Σ}}C(Q_n,I_{n,i})}---\left(1\right)$

Wherein, F (Q_n,I_n,i) it is for the n-th search word Q in m search word_nUnder i-th network address letter Breath I_n,iThe first result, certainly, the different I under search word_n,iCan be different, n is little more than or equal to 1 In the integer equal to m, k_nRepresent the n-th search word Q_nUnder website information sum, different search words Under website information sum can be different, for the integer more than or equal to 1, C (Q_n,I_n,i) represent that n-th searches Rope word Q_nUnder i-th website information I_n,iNumber of clicks.

It can be seen that the first result is added up, to n-th according to same user's difference website information Individual search word Q_n, this Q_nUnder i-th website information I_n,iNumber of clicks C (Q_n,I_n,i) the biggest, the first knot Really F (Q_n,I_n,i) the biggest, otherwise, the first result F (Q_n,I_n,i) the biggest, show the n-th search word Q_nUnder I-th website information I_n,iNumber of clicks C (Q_n,I_n,i) the biggest.

Sub-step 112, described first result is carried out smoothing computation, obtain described website information described Weight under search word.

The number of clicks sum of the first result that sub-step 111 obtains website information under two kinds of search words Following situation, such as, the click of website information N1, N2 under search word B1 is there will be when differing greatly Number of times is 1 time, then the first result calculated according to formula (1) is 0.5, under search word B2 Website information N1, the number of clicks of N2 are respectively 10,900 times, then calculate according to formula (1) First result is respectively 0.1 and 0.9, now, the first result of website information N1 under search word B1 0.5 is bigger than the first result 0.1 of website information N1 under search word B2, but actually search word B1 Under the point of website information N1 that is but far smaller than under search word B2 of the number of clicks 1 of website information N1 Hit number of times, thus the first result accessing the lowest network address of search rate can be caused very big, but it is to searching The contribution of rope word is the least, but it is big on the impact of weight, so the first result directly describes network address letter Breath weight under search word is inaccurate, in order to solve the problems referred to above, needs by smoothing computation first Result is adjusted, so that result of calculation is more reasonable.

Website information can be calculated in described search according to calculated first result of sub-step 111 Weight under word.

Preferably, sub-step 112 can include sub-step 112A to 112B:

Sub-step 112A, to the number of clicks of website information all under described search word and non-zero normal number Sum is taken the logarithm, and obtains smoothing parameter；

Equation below specifically can be used to calculate:

$P(Q_n,I_{n,i})={\log }_a\left(\underset{i=1}{\overset{k_n}{\Σ}}C\right(Q_n,I_{n,i})+b)---\left(2\right)$

Wherein, P (Q_n,I_n,i) it is the n-th search word Q_nI-th website information I_n,iSmoothing parameter, become Amount Q_nAnd I_n,iMeaning with reference to described in formula (1), truth of a matter a can take and include 10 and e be more than or equal to Any number of 1, smoothing factor b can take any number more than or equal to 1.

Sub-step 112B, described first result is multiplied with smoothing parameter, obtains described website information and exist Weight under described search word.

Equation below specifically can be used to calculate:

$V(Q_n,I_{n,i})=F(Q_n,I_{n,i})*P(Q_n,I_{n,i})=\frac{C(Q_n,I_{n,i})}{\underset{i=1}{\overset{k_n}{\Σ}}C(Q_n,I_{n,i})}*{\log }_a\left(\underset{i=1}{\overset{k_n}{\Σ}}C\right(Q_n,I_{n,i})+b)---\left(3\right)$

Wherein, V (Q_n,I_i) it is the n-th search word Q_nI-th website information I_n,iWeight, other are each The meaning of variable is with reference to described in formula (1) and formula (2).

Step 120, navigation patterns historical record according to each user, add up under each ID The number of clicks of each website information；

Wherein, ID can uniquely identify the identity of active user, it will be understood that the present invention implements In example, this client is as a example by browser client, and its ID can be the Accounting system of browser In user account, the Accounting system of this browser client is independent of the account system of third-party server System, and dock with ad system.Certainly, ad system can also be with the Accounting system of third-party server Docking, directly using user's user account in the Accounting system of third-party server is ID. It is, of course, also possible to the physical identity address directly using client place user terminal is ID.When So, the concrete form of ID is not limited by the present invention.

In actual applications, user can browse the website information of needs by various modes, including: clear Look at and input the URL of website information on device, from the webpage of current presentation, click on URL thus directly redirect To target website information etc., user browses website information can be there to be other modes, and the present invention is clear to user The mode of website information of looking at does not limits.It is hereby achieved that the browsing histories behavior record of each user, should In browsing histories behavior record, record is that user is not hit by Searching point and navigation patterns to network address.

In embodiments of the present invention, when user browses website information, it is clear can to pass through client records Look at historical behavior, then have client that this browsing histories behavior is sent the server to advertisement.The most also The service to advertisement can be then sent to by third-party server record user this time navigation patterns Device.This browsing histories behavior includes: ID, website information URL etc..

The navigation patterns historical record of the user A that client or server are generated by ID is (Ua, N1), (Ua, N2), (Ua, N3), (Ua, N3), (Ua, N4), (Ua, N2), (Ua, N3), (Ua, N4) }, the navigation patterns historical record of user B be (Ub, N1), (Ub, N3), (Ub, N2), (Ub, N1), (Ub, N4), (Ub, N1), (Ub, N3) }, wherein, then under user A The number of clicks of N1 website information is 1, and the number of clicks of the N2 website information under user A is 2, uses The number of clicks of the N3 website information under the A of family is 3, the click of the N4 website information under user A time Number is 2, and the number of clicks of the N1 website information under user B is 3, the N2 website information under user B Number of clicks be 1, the number of clicks of the N3 website information under user B is 2, the N4 under user B The number of clicks of website information is 1.If being appreciated that when a website information is not accessed by a user, The access times of this website information are 0.

Step 130, according to the number of clicks of each website information under each ID and each search word Under the weight of each website information, calculate each search of each website information under each ID The weight of word；

For each website information under all search words described in step 110, calculate each searching one by one Each website information under rope word, for the weight of different user, specifically can use equation below to calculate:

V(U_t,Q_n,I_n,i)=C (U_t,I_n,i)*V(Q_n,I_n,i) (4)

Wherein, V (U_t,Q_n,I_n,i) it is the t user U_tUnder i-th website information I of the n-th search word_n,i's Weight, V (Q_n,I_n,i) it is the n-th search word Q of calculating of formula (3)_nLower i-th website information I_n,iPower Weight, C (U_t,I_n,i) it is the t user U_tTo the n-th search word Q_nI-th website information I_n,iClick time Number, In ad system each website information number of clicks result from each ID that step 120 obtains Obtain website information mark and website information I_n,iNumber of clicks corresponding to the identical record of network address mark make It is the n-th search word Q_nLower i-th website information I_n,iNumber of clicks.

Step 140, when an ID place client newly accesses a website information, according to described use The weight of each search word of the described website information under the mark of family, extracts at least one search that sequence is forward Word also generates recommendation information based on described search word；

Wherein, client can be arranged on terminal unit website information access function can be provided Software, including: browser or other software of website information access function, terminal unit can be provided Including desktop computer, notebook computer, mobile phone etc., the present invention is to the form of client and terminal unit not Limit.

Certainly, the client of the present invention is not when clicking on a website information by search results pages, to the 3rd Side's server sends the request accessing website information, and this request includes ID.Third-party server This website information and ID are sent to Advertisement Server.Advertisement Server then searches this use of record One or more the forward search word that sorts of this website information under the mark of family, then searches according to these Rope word generates recommendation information and is then back to client, and client can show this recommendation information.

It is corresponding that the Advertisement Server of the present invention obtains above-mentioned ID from the search word that step 130 obtains All search words, and by search word according to weight descending, extract the search word that sequence is forward, will The link corresponding with search word of described search word is as a recommendation information.Wherein, bigger due to weight Search word is more relevant to the navigation patterns of user, thus the recommendation generated by the search word that weight is bigger Breath more meets user's request.

Step 150, return described recommendation information and be shown to client.

Server generating after recommendation information, can return this recommendation information, and then client can be Position is specified to show this recommendation information.

Wherein, it is intended that position can be any region of client screen, but in order to not affect user The application used, preferably can be placed on the window of recommendation information in screen edges and corners.Show recommendation The mode of breath specifically may include that and advertisement chain link corresponding with search word for search word be presented on system screen On curtain, if there is multiple search word, then arrange aobvious according to order from big to small according to the weight of search word Show on screen, at this point it is possible to arrange the maximum of recommendation information number, when the number of recommendation information is big When equal to above-mentioned maximum, the recommendation information number of displaying is maximum.

Preferably, receiving the non-browser client of this recommendation information, then browser client can be given birth to Becoming pop-up, and be illustrated in by above-mentioned recommendation information among described pop-up, wherein, the size of pop-up can root Reasonably arrange according to the size of system screen, under normal circumstances preferably no more than the percent of screen Ten, the position of pop-up can be the lower left corner of screen, the lower right corner or other do not hinder user normally to use system The position etc. of system, when there is multiple recommendation information and needing to show, can suitably reduce the little of recommendation information Window, and wicket is illustrated on screen according to horizontal, longitudinal or other layouts, should in reality In with, when the number of search word is more than 1, the number of corresponding recommendation information is also greater than 1, then according to searching The order that the weight of rope word is descending shows the recommendation information that search word is corresponding.It is appreciated that, it is also possible to Have other modes to show recommendation information, the present invention to show the mode of recommendation information, position, size, Shape, layout do not limit.

When occurring the window of recommendation information on screen, user can select to check or close as required. At client-side,

Step 160, receive by user what clicking operation to a recommendation information in the client was triggered Access request；

Step 170, according to described access request, return corresponding advertisement page to client.

If user needs to check certain recommendation information, then can click on the recommendation shown in pop-up Breath.Access request would generally be sent to Advertisement Server when user clicks on recommendation information.Owing to the present invention is real Execute example and every recommendation information has been hung on backstage link, then when user clicks on recommendation information, can call clear Look at device kernel to Advertisement Server generation access request, therefore this access request be actually a webpage please Ask.After Advertisement Server receives this access request, then return corresponding wide according to the link in this request The announcement page gives this client, and this client then shows this advertisement page.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Embodiment two

With reference to Fig. 2, it is shown that according to a kind of information recommendation method embodiment two of the present invention, concrete steps May include that

Step 210, under search word each in search history behavior record record website information, according to The theme merger of content corresponding to website information；

Under normal circumstances, the different website information that same search word is corresponding are to be mutually related, and network address is believed The content of breath belongs to same subject, in this case, in order to reduce amount of calculation, and can be first to search Behavior historical record carries out merger according to theme, so that the quantity participating in the website information of calculating subtracts Few.

Under one search word in multiple website information of record, the content of certain a few website information is relevant or phase With, then can be a website information group by these network address merger, the click time to described website information group Number is equal to the number of clicks sum of multiple website information in group.Such as, the network address letter of record under a search word Number of clicks corresponding for breath A1, A2, A3, A4, A5 is respectively C1, C2, C3, C4, C5, Wherein, website information A1 be about school S chemistry in terms of Teaching Advantages, A3 is about school S Advantage in terms of scientific research, A4 is about school S advantage in terms of educational alternative, these three network address Information corresponding content belongs to same subject, is all the advantage about school S, then by website information A1, A3 and A4 merger is website information group D, and the number of clicks corresponding to D is C1, C3, C4 sum C, the result obtained after merger is D, A2, A5, and corresponding number of clicks is respectively C, C2, C5.

The method of the theme obtaining website information content may include that Statistics-Based Method, knowledge based Method and based on both combine method, wherein, Statistics-Based Method utilize characteristic item frequency, The position of characteristic item, the contribution information etc. of characteristic item obtain theme, and Knowledge based engineering method uses knowledge Storehouse such as script, machine readable dictionary carry out grammatical and semantic analysis to obtain theme, the side combined based on both Method improves overall performance by two kinds of methods being combined.The present invention theme to obtaining website information content Method does not limits.

Step 220, add up under each search word the weight of the website information after merger.

Calculate according still further to the sub-step 111,112 that embodiment is a kind of on the basis of website information after merger The weight of website information under above-mentioned search word.

Step 230, to the website information in the navigation patterns historical record of each user, believe according to network address The theme merger of the corresponding content of breath；

Due to when calculating the weight of website information under search word to website information according to the master of corresponding content Topic has carried out merger, and in like manner, under calculating ID, the number of clicks of website information is also required to network address Information carries out same merger according to the theme of corresponding content, so that the navigation patterns history note of user In website information in record and search history behavior record, the website information under search word is the most corresponding.

Step 240, add up under each ID the number of clicks of the website information after merger.

After in system, the multiple website information under an ID are according to the theme merger of corresponding content, total point Hit number of times equal to the number of clicks sum of multiple website information in group.

Step 250, according to the number of clicks of each website information under each ID and each search word Under the weight of each website information, calculate each search of each website information under each ID The weight of word；

Step 260, when an ID place client newly accesses a website information, according to described use The weight of each search word of the described website information under the mark of family, extracts at least one search that sequence is forward Word also generates recommendation information based on described search word；

Step 270, return described recommendation information and be shown to client.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Further, the embodiment of the present invention is to the network address letter of record under search word each in search history behavior record Website information in the navigation patterns historical record of breath and each user, according to content corresponding to website information Theme merger, thus decrease and add up under each search word the weight of the website information after merger and each Under ID, the amount of calculation of the number of clicks of website information after merger, finally improves information recommendation Efficiency.

Embodiment three

With reference to Fig. 3, it is shown that according to a kind of information recommendation method embodiment three of the present invention, concrete steps May include that

Step 310, search behavior historical record according to each user, that adds up under each search word is each The weight of individual website information；

Step 320, navigation patterns historical record according to each user, add up under each ID The number of clicks of each website information；

Step 330, according to the number of clicks of each website information under each ID and each search word Under the weight of each website information, calculate each search of each website information under each ID The weight of word；

The recommendation request that step 340, reception client send；Described recommendation request is in an ID Send to Advertisement Server during client-access one website information of place；

When user the most in the client non-search result page click on website information, client Can send access request to third-party server, third-party server can be various application server.With Time, client sends the recommendation request of recommendation information to Advertisement Server.Or third-party server is to extensively Accusing server and send recommendation request, this recommendation request includes the physical address information of client, such as IP Location, facilitates Advertisement Server to connect client.

Step 350, obtained ID and website information by described recommendation request, and according to described user The weight of each search word of the described website information under Biao Shi, extracts at least one search that sequence is forward Word also generates recommendation information based on described search word；

Client-server initiates the recommendation request of recommendation information to Advertisement Server, wherein, it is recommended that information Recommendation request in carry ID, and website information URL etc..Then server can be from Resolve user identity and website information in this recommendation request, then according to the described net under described ID The weight of each search word of location information, extracts at least one forward search word of sequence and searches based on described Rope word generates recommendation information.

If being sent recommendation request by third-party server, the IP address of client wherein can also be included. Preferably, step 350 generates based on described search word the step of recommendation information, sub-step can be included 351 to 352:

Sub-step 351, from advertisement base search corresponding described search word advertisement link；

Sub-step 352, generate recommendation information based on described advertisement link and search word.

Wherein, advertisement base is the data base of the advertisement link storing all search words and its correspondence, permissible For existing various types of data bases, the type of data base is not limited by the present invention.In actual application, First to search word set up index, in retrieving by search word with index comparison one by one, if search Word mates with certain index, then will currently index the corresponding advertisement link advertisement link as search word, institute State link and search word as one group of recommendation information.

Step 360, return described recommendation information and be shown to client.

If client directly transmits recommendation request to Advertisement Server, then client and Advertisement Server Directly establishing link, Advertisement Server can be directly according to returning this recommendation with linking of client Breath.

Recommend to ask if third-party server sends to Advertisement Server, the advertisement clothes of the embodiment of the present invention Business device then can establish the link with client, then according to the IP address of the client in recommendation request This recommendation information is returned client.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Further, the embodiment of the present invention is by combining work by advertisement link corresponding with search word for search word For recommendation information, thus user can understand the theme of recommendation information according to the search word in recommendation information, Click on search word and can jump directly to advertisement page, decrease user and open the operation step of recommendation information Suddenly, bring conveniently for user.

For embodiment of the method, in order to be briefly described, therefore it is all expressed as a series of combination of actions, But those skilled in the art should know, the embodiment of the present invention not limiting by described sequence of movement System, because according to the embodiment of the present invention, some step can use other orders or carry out simultaneously.Its Secondary, those skilled in the art also should know, embodiment described in this description belongs to be preferable to carry out Example, necessary to the involved action not necessarily embodiment of the present invention.

Embodiment four

With reference to Fig. 4, it is shown that according to the structured flowchart of the present invention a kind of information recommending apparatus embodiment four, Specifically can include such as lower module:

Website information weight statistical module 410, for the search behavior historical record according to each user, Add up the weight of each website information under each search word；

Website information number of clicks statistical module 420, remembers for the navigation patterns history according to each user Record, adds up the number of clicks of each website information under each ID；

Search word weight computation module 430, for the point according to each website information under each ID Hit the weight of each website information under number of times and each search word, calculate each under each ID The weight of each search word of website information；

Recommendation information generation module 440, for newly accessing a network address letter in an ID place client During breath, according to the weight of each search word of the described website information under described ID, extract sequence and lean on At least one front search word also generates recommendation information based on described search word；

Recommendation information display module 450, is used for returning described recommendation information and is shown to client.

Preferably, described website information weight statistical module, including:

First result calculating sub module, for for the website information under a search word, by described network address The number of clicks of information, divided by the number of clicks sum of website information all under described search word, obtains first Result；

First result smooths submodule, for described first result is carried out smoothing computation, obtains described net Location information weight under described search word.

Preferably, described first result smooths submodule, including:

Smoothing parameter obtains submodule, for the number of clicks of website information all under described search word with Non-zero normal number sum is taken the logarithm, and obtains smoothing parameter；

Website information weight the first calculating sub module, for described first result is multiplied with smoothing parameter, Obtain described website information weight under described search word.

Preferably, described recommendation information generation module, including:

Recommendation request receives submodule, for receiving the recommendation request that client sends；Described recommendation request For sending to Advertisement Server when ID place client-access one website information；

Recommendation information first generates submodule, for being obtained ID and network address letter by described recommendation request Breath, and the weight of each search word according to the described website information under described ID, extract sequence At least one forward search word also generates recommendation information based on described search word.

Preferably, described recommendation information first generates submodule, including:

Advertisement link searches submodule, for searching the advertisement link of corresponding described search word from advertisement base；

Recommendation information second generates submodule, for generating recommendation based on described advertisement link and search word Breath.

Preferably, it is recommended that after information display module, also include:

Clicking operation receiver module, for receiving the click to a recommendation information in the client by user The access request that operation is triggered；

Page jump module, for according to described access request, returns corresponding advertisement page to client.

Preferably, described recommendation information display module, including:

Submodule is shown in pop-up, is used for returning described recommendation information to browser client, with at browser Client generates in pop-up and is shown.

Preferably, described website information weight statistical module, including:

Website information the first merger submodule, for remembering under search word each in search history behavior record The website information of record, according to the theme merger of content corresponding to website information；

Website information weight merger submodule, for adding up under each search word the website information after merger Weight.

Preferably, described website information number of clicks statistical module, including:

Website information the second merger submodule, for the net in the navigation patterns historical record of each user Location information, according to the theme merger of content corresponding to website information；

Number of clicks merger submodule, for adding up under each ID the point of the website information after merger Hit number of times.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Embodiment five

With reference to Fig. 5, it is shown that according to the structured flowchart of the present invention a kind of information recommending apparatus embodiment four, Specifically can include such as lower module:

Website information weight statistical module 510, for the search behavior historical record according to each user, Add up the weight of each website information under each search word；Specifically include:

Website information the first merger submodule 511, for searching each in search history behavior record The website information of record under rope word, according to the theme merger of content corresponding to website information；

Website information weight merger submodule 512, for adding up under each search word the net after merger The weight of location information.

Website information number of clicks statistical module 520, remembers for the navigation patterns history according to each user Record, adds up the number of clicks of each website information under each ID；Specifically include:

Website information the second merger submodule 521, for remembering the navigation patterns history of each user Website information in record, according to the theme merger of content corresponding to website information；

Number of clicks merger submodule 522, for adding up under each ID the network address after merger The number of clicks of information.

Search word weight computation module 530, for the point according to each website information under each ID Hit the weight of each website information under number of times and each search word, calculate each under each ID The weight of each search word of website information；

Recommendation information generation module 540, for newly accessing a network address letter in an ID place client During breath, according to the weight of each search word of the described website information under described ID, extract sequence and lean on At least one front search word also generates recommendation information based on described search word；

Recommendation information display module 550, is used for returning described recommendation information and is shown to client.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

Further, the embodiment of the present invention is to the network address letter of record under search word each in search history behavior record Website information in the navigation patterns historical record of breath and each user, according to content corresponding to website information Theme merger, thus decrease and add up under each search word the weight of the website information after merger and each Under ID, the amount of calculation of the number of clicks of website information after merger, finally improves information recommendation Efficiency.

Embodiment six

With reference to Fig. 6, it is shown that according to the structured flowchart of the present invention a kind of information recommending apparatus embodiment four, Specifically can include such as lower module:

Website information weight statistical module 610, for the search behavior historical record according to each user, Add up the weight of each website information under each search word；

Website information number of clicks statistical module 620, remembers for the navigation patterns history according to each user Record, adds up the number of clicks of each website information under each ID；

Search word weight computation module 630, for the point according to each website information under each ID Hit the weight of each website information under number of times and each search word, calculate each under each ID The weight of each search word of website information；

Recommendation information generation module 640, for newly accessing a network address letter in an ID place client During breath, according to the weight of each search word of the described website information under described ID, extract sequence and lean on At least one front search word also generates recommendation information based on described search word；Specifically include:

Recommendation request receives submodule 641, for receiving the recommendation request that client sends；Described push away Request of recommending is for sending to Advertisement Server when ID place client-access one website information；

Recommendation information first generates submodule 642, for by described recommendation request obtain ID and Website information, and the weight of each search word according to the described website information under described ID, carry Take at least one forward search word of sequence and generate recommendation information based on described search word.

Recommendation information display module 650, is used for returning described recommendation information and is shown to client.

The embodiment of the present invention can add up the power of website information under search word from search behavior historical record Weight, the number of clicks of the website information under counting user mark from navigation patterns historical record.So pin To an ID, each website information that it is accessed, can be according to each aforementioned search word obtained The weight of each website information, and the access weight that this ID is to each website information, believe with network address Breath association, calculates the weight of each search word under this ID.So, user with this ID Take notice of that when accessing certain network address in client, Advertisement Server then can extract that weight sequencing is forward Or multiple search words, and according to this search word to user's recommendation information, thus solve for user not Being the network address accessed by search behavior, owing to there is not search behavior, Advertisement Server cannot know use The intention at family, thus the problem of information accurately cannot be recommended for user, achieving can be direct to user Access the situation of network address, search the intention relevant to this network address, such that it is able to recommend accurate information, subtract The beneficial effect of the waste of few system resource.

And the embodiment of the present invention is by combining work by advertisement link corresponding with search word for search word For recommendation information, thus user can understand the theme of recommendation information according to the search word in recommendation information, Click on search word and can jump directly to advertisement page, decrease user and open the operation step of recommendation information Suddenly, bring conveniently for user.

Algorithm and display be not solid with any certain computer, virtual system or miscellaneous equipment provided herein Have relevant.Various general-purpose systems can also be used together with based on teaching in this.As described above, Construct the structure required by this kind of system to be apparent from.Additionally, the present invention is also not for any specific Programming language.It is understood that, it is possible to use various programming languages realize the content of invention described herein, And the description done language-specific above is the preferred forms in order to disclose the present invention.

In description mentioned herein, illustrate a large amount of detail.It is to be appreciated, however, that this Inventive embodiment can be put into practice in the case of not having these details.In some instances, not It is shown specifically known method, structure and technology, in order to do not obscure the understanding of this description.

Similarly, it will be appreciated that in order to simplify the disclosure and help understand in each inventive aspect one Or multiple, above in the description of the exemplary embodiment of the present invention, each feature of the present invention is sometimes It is grouped together in single embodiment, figure or descriptions thereof.But, should be by the disclosure Method be construed to reflect an intention that i.e. the present invention for required protection require ratio in each claim The middle more feature of feature be expressly recited.More precisely, as the following claims reflect As, inventive aspect is all features less than single embodiment disclosed above.Therefore, it then follows Claims of detailed description of the invention are thus expressly incorporated in this detailed description of the invention, the most each right Requirement itself is all as the independent embodiment of the present invention.

Those skilled in the art are appreciated that and can carry out the module in the equipment in embodiment certainly Change adaptively and they are arranged in one or more equipment different from this embodiment.Permissible Module in embodiment or unit or assembly are combined into a module or unit or assembly, and in addition may be used To put them into multiple submodule or subelement or sub-component.Except such feature and/or process or Outside at least some in unit excludes each other, can use any combination that (this specification is included companion With claim, summary and accompanying drawing) disclosed in all features and so disclosed any method or All processes of person's equipment or unit are combined.Unless expressly stated otherwise, this specification (includes companion With claim, summary and accompanying drawing) disclosed in each feature can by provide identical, equivalent or phase Replace like the alternative features of purpose.

Although additionally, it will be appreciated by those of skill in the art that embodiments more described herein include it Some feature included in its embodiment rather than further feature, but the group of the feature of different embodiment Close and mean to be within the scope of the present invention and formed different embodiments.Such as, in following power In profit claim, one of arbitrarily can mode making in any combination of embodiment required for protection With.

The all parts embodiment of the present invention can realize with hardware, or to process at one or more The software module run on device realizes, or realizes with combinations thereof.Those skilled in the art should Understand, microprocessor or digital signal processor (DSP) can be used in practice to realize basis The some or all functions of the some or all parts in the information recommendation equipment of the embodiment of the present invention. The present invention is also implemented as part or all the equipment for performing method as described herein Or device program (such as, computer program and computer program).Such realize the present invention Program can store on a computer-readable medium, or can have the shape of one or more signal Formula.Such signal can be downloaded from internet website and obtain, or provides on carrier signal, or Person provides with any other form.

The present invention will be described rather than limits the invention to it should be noted above-described embodiment, And those skilled in the art can design replacement without departing from the scope of the appended claims Embodiment.In the claims, any reference marks that should not will be located between bracket is configured to right The restriction required.Word " comprises " and does not excludes the presence of the element or step not arranged in the claims.Position Word "a" or "an" before element does not excludes the presence of multiple such element.The present invention can With by means of including the hardware of some different elements and realizing by means of properly programmed computer. If in the unit claim listing equipment for drying, several in these devices can be by same Individual hardware branch specifically embodies.Word first, second and third use do not indicate that any order. Can be title by these word explanations.

The invention discloses A1, a kind of information recommendation method, including:

Search behavior historical record according to each user, adds up each website information under each search word Weight；

Navigation patterns historical record according to each user, adds up each network address letter under each ID The number of clicks of breath；

Each under number of clicks according to each website information under each ID and each search word The weight of website information, calculates the power of each search word of each website information under each ID Weight；

When an ID place client newly accesses a website information, described recommendation request obtain and use Family mark and website information, and according to each search word of the described website information under described ID Weight, extracts and sorts at least one forward search word and generate recommendation information based on described search word；

Return described recommendation information to be shown to client.

A2, method as described in A1, the described search behavior historical record according to each user, statistics The step of the weight of each website information under each search word, including:

For the website information under a search word, the number of clicks of described website information is searched divided by described Under rope word, the number of clicks sum of all website information, obtains the first result；

Described first result is carried out smoothing computation, obtains described website information power under described search word Weight.

A3, method as described in A2, described carry out smoothing computation by described first result, obtains described The step of website information weight under described search word, including:

The number of clicks of website information all under described search word is taken the logarithm with non-zero normal number sum, To smoothing parameter；

Described first result is multiplied with smoothing parameter, obtains described website information under described search word Weight.

A4, such as A1-A3 method as described in one of them, described new in an ID place client When accessing a website information, the weight of each search word of the described website information under described ID, Extract and sort at least one forward search word and generate the step of recommendation information based on described search word, bag Include:

Receive the recommendation request that client sends；Described recommendation request is in an ID place client Send to Advertisement Server when accessing a website information；

ID and website information is obtained by described recommendation request, and according to the institute under described ID State the weight of each search word of website information, extract at least one forward search word of sequence and based on institute State search word and generate recommendation information.

A5, method as described in A4, the described step generating recommendation information based on described search word, bag Include:

The advertisement link of corresponding described search word is searched from advertisement base；

Recommendation information is generated based on described advertisement link and search word.

A6, method as described in A1, return step that described recommendation information to client is shown it After, also include:

Receive the access request in the client clicking operation of a recommendation information triggered by user；

According to described access request, return corresponding advertisement page to client.

A7, method as described in A1, the step that the described recommendation information of described return is shown to client Suddenly, including:

Return described recommendation information to browser client, to generate in pop-up in browser client Row is shown.

A8, method as described in 1, the described search behavior historical record according to each user, statistics is every The step of the weight of each website information under individual search word, including:

To the website information of record under search word each in search history behavior record, according to website information institute The theme merger of corresponding content；

Add up under each search word the weight of the website information after merger.

A9, method as described in A8, the described navigation patterns historical record according to each user, statistics The step of the number of clicks of each website information under each ID, including:

To the website information in the navigation patterns historical record of each user, in corresponding to website information The theme merger held；

Add up under each ID the number of clicks of the website information after merger.

The invention also discloses B10, a kind of information recommending apparatus, including:

Website information weight statistical module, for the search behavior historical record according to each user, statistics The weight of each website information under each search word；

Website information number of clicks statistical module, for the navigation patterns historical record according to each user, Add up the number of clicks of each website information under each ID；

Search word weight computation module, for the click according to each website information under each ID time The weight of each website information under number and each search word, calculates each network address under each ID The weight of each search word of information；

Recommendation information generation module, for newly accessing a website information in an ID place client Time, according to the weight of each search word of the described website information under described ID, extract sequence forward At least one search word and based on described search word generate recommendation information；

Recommendation information display module, is used for returning described recommendation information and is shown to client.

B11, device as described in B10, described website information weight statistical module, including:

First result calculating sub module, for for the website information under a search word, by described network address The number of clicks of information, divided by the number of clicks sum of website information all under described search word, obtains first Result；

First result smooths submodule, for described first result is carried out smoothing computation, obtains described net Location information weight under described search word.

B12, device as described in B11, described first result smooths submodule, including:

Smoothing parameter obtains submodule, for the number of clicks of website information all under described search word with Non-zero normal number sum is taken the logarithm, and obtains smoothing parameter；

Website information weight the first calculating sub module, for described first result is multiplied with smoothing parameter, Obtain described website information weight under described search word.

B13, such as 10-12 device as described in one of them, described recommendation information generation module, including:

Recommendation request receives submodule, for receiving the recommendation request that client sends；Described recommendation request For sending to Advertisement Server when ID place client-access one website information；

Recommendation information first generates submodule, for being obtained ID and network address letter by described recommendation request Breath, and the weight of each search word according to the described website information under described ID, extract sequence At least one forward search word also generates recommendation information based on described search word.

B14, device as described in 13, described recommendation information first generates submodule, including:

Advertisement link searches submodule, for searching the advertisement link of corresponding described search word from advertisement base；

Recommendation information second generates submodule, for generating recommendation based on described advertisement link and search word Breath.

B15, device as described in B10, it is recommended that after information display module, also include:

Clicking operation receiver module, for receiving the click to a recommendation information in the client by user The access request that operation is triggered；

Page jump module, for according to described access request, returns corresponding advertisement page to client.

B16, device as described in B10, described recommendation information display module, including:

Submodule is shown in pop-up, is used for returning described recommendation information to browser client, with at browser Client generates in pop-up and is shown.

B17, device as described in B10, described website information weight statistical module, including:

Website information the first merger submodule, for remembering under search word each in search history behavior record The website information of record, according to the theme merger of content corresponding to website information；

Website information weight merger submodule, for adding up under each search word the website information after merger Weight.

B18, device as described in B17, described website information number of clicks statistical module, including:

Website information the second merger submodule, for the net in the navigation patterns historical record of each user Location information, according to the theme merger of content corresponding to website information；

Number of clicks merger submodule, for adding up under each ID the point of the website information after merger Hit number of times.

Claims (10)

1. an information recommendation method, including:

Search behavior historical record according to each user, adds up each website information under each search word Weight；

Navigation patterns historical record according to each user, adds up each network address letter under each ID The number of clicks of breath；

Each under number of clicks according to each website information under each ID and each search word The weight of website information, calculates the power of each search word of each website information under each ID Weight；

When an ID place client newly accesses a website information, described recommendation request obtain and use Family mark and website information, and according to each search word of the described website information under described ID Weight, extracts and sorts at least one forward search word and generate recommendation information based on described search word；

Return described recommendation information to be shown to client.

Method the most according to claim 1, it is characterised in that described searching according to each user Suo Hangwei historical record, adds up the step of the weight of each website information under each search word, including:

For the website information under a search word, the number of clicks of described website information is searched divided by described Under rope word, the number of clicks sum of all website information, obtains the first result；

Described first result is carried out smoothing computation, obtains described website information power under described search word Weight.

Method the most according to claim 2, it is characterised in that described described first result is entered Row smoothing computation, obtains the step of described website information weight under described search word, including:

The number of clicks of website information all under described search word is taken the logarithm with non-zero normal number sum, To smoothing parameter；

Described first result is multiplied with smoothing parameter, obtains described website information under described search word Weight.

4. according to one of them described method of claim 1-3, it is characterised in that described in a use When mark place, family client newly accesses a website information, the described website information under described ID The weight of each search word, extracts at least one forward search word of sequence and generates based on described search word The step of recommendation information, including:

Receive the recommendation request that client sends；Described recommendation request is in an ID place client Send to Advertisement Server when accessing a website information；

ID and website information is obtained by described recommendation request, and according to the institute under described ID State the weight of each search word of website information, extract at least one forward search word of sequence and based on institute State search word and generate recommendation information.

Method the most according to claim 4, it is characterised in that described raw based on described search word Become the step of recommendation information, including:

The advertisement link of corresponding described search word is searched from advertisement base；

Recommendation information is generated based on described advertisement link and search word.

6. an information recommending apparatus, including:

Website information weight statistical module, for the search behavior historical record according to each user, statistics The weight of each website information under each search word；

Website information number of clicks statistical module, for the navigation patterns historical record according to each user, Add up the number of clicks of each website information under each ID；

Search word weight computation module, for the click according to each website information under each ID time The weight of each website information under number and each search word, calculates each network address under each ID The weight of each search word of information；

Recommendation information generation module, for newly accessing a website information in an ID place client Time, according to the weight of each search word of the described website information under described ID, extract sequence forward At least one search word and based on described search word generate recommendation information；

Recommendation information display module, is used for returning described recommendation information and is shown to client.

Device the most according to claim 6, it is characterised in that described website information weight is added up Module, including:

First result calculating sub module, for for the website information under a search word, by described network address The number of clicks of information, divided by the number of clicks sum of website information all under described search word, obtains first Result；

First result smooths submodule, for described first result is carried out smoothing computation, obtains described net Location information weight under described search word.

Device the most according to claim 7, it is characterised in that described first result smooths submodule Block, including:

Smoothing parameter obtains submodule, for the number of clicks of website information all under described search word with Non-zero normal number sum is taken the logarithm, and obtains smoothing parameter；

Website information weight the first calculating sub module, for described first result is multiplied with smoothing parameter, Obtain described website information weight under described search word.

9. according to one of them described device of claim 6-8, it is characterised in that described recommendation Breath generation module, including:

Recommendation request receives submodule, for receiving the recommendation request that client sends；Described recommendation request For sending to Advertisement Server when ID place client-access one website information；

Recommendation information first generates submodule, for being obtained ID and network address letter by described recommendation request Breath, and the weight of each search word according to the described website information under described ID, extract sequence At least one forward search word also generates recommendation information based on described search word.

Device the most according to claim 9, it is characterised in that described recommendation information first generates Submodule, including:

Advertisement link searches submodule, for searching the advertisement link of corresponding described search word from advertisement base；

Recommendation information second generates submodule, for generating recommendation based on described advertisement link and search word Breath.