CN103902597B

CN103902597B - The method and apparatus for determining relevance of searches classification corresponding to target keyword

Info

Publication number: CN103902597B
Application number: CN201210581476.XA
Authority: CN
Inventors: 孙宇; 谭广明; 韩彦俊
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2012-12-27
Filing date: 2012-12-27
Publication date: 2019-03-08
Anticipated expiration: 2032-12-27
Also published as: CN103902597A

Abstract

The object of the present invention is to provide a kind of methods and apparatus for determining relevance of searches classification corresponding to target keyword.Specifically, according to the searching order routing information of target keyword, target critical term clustering belonging to the target keyword is determined from one or more keyword clusterings；According to the target critical term clustering, relevance of searches classification corresponding to the target keyword is determined, to be used for subsequent processing.Compared with prior art, the present invention, which passes through, determines target critical term clustering belonging to target keyword, and then relevance of searches classification corresponding to the target keyword, to be used for subsequent processing, relevance of searches classification corresponding to keyword is effectively determined to realize, and the automatic test to batch keyword data, only Optimizing Search engine search sequence does not provide reference, and improves the testing efficiency to search engine relevance.

Description

The method and apparatus for determining relevance of searches classification corresponding to target keyword

Technical field

The present invention relates to Internet technical fields more particularly to a kind of for determining search phase corresponding to target keyword The technology of closing property classification.

Background technique

Currently, with the development of internet technology and Internet application to user learn, work with life infiltration, people More and more by network acquisition information, keyword is such as inputted by search engine, search engine is by taking certain search Sortord, which determines, returns to the search result that user matches with keyword, however the search result and use of search engine return The matching degree of the search sequence of family input largely affects the accuracy that user obtains information.Correspondingly, if can mention The matching degree of the search sequence of search result and user's input that high search engine returns, will greatly improve user and obtain letter The efficiency of breath.Therefore, it is necessary to the correlations to search engine to carry out effective assessment test, such as according to keyword and search result Matching degree classify to keyword, determine relevance of searches classification corresponding to keyword, effectively determine keyword Corresponding relevance of searches classification, and improve the testing efficiency of search engine relevance.

Summary of the invention

The object of the present invention is to provide a kind of for determining relevance of searches class method for distinguishing corresponding to target keyword With equipment.

According to an aspect of the invention, there is provided a kind of for determining relevance of searches class corresponding to target keyword Method for distinguishing, wherein method includes the following steps:

A determines the mesh according to the searching order routing information of target keyword from one or more keyword clusterings Mark target critical term clustering belonging to keyword；

B determines relevance of searches classification corresponding to the target keyword according to the target critical term clustering, with In subsequent processing.

According to another aspect of the present invention, it additionally provides a kind of for determining the correlation of search corresponding to target keyword The classification of property classification determines equipment, wherein the category determines that equipment includes:

Determining device is clustered, for the searching order routing information according to target keyword, from one or more keywords Target critical term clustering belonging to the target keyword is determined in cluster；

Classification determining device, for determining and being searched corresponding to the target keyword according to the target critical term clustering Rope correlation classification, to be used for subsequent processing.

According to a further aspect of the invention, a kind of computer equipment is additionally provided, wherein the computer equipment includes such as The aforementioned classification for determining relevance of searches classification corresponding to target keyword according to a further aspect of the present invention determines Equipment.

According to a further aspect of the invention, it additionally provides a kind of for determining the correlation of search corresponding to target keyword Property classification search engine, wherein the search engine include as it is aforementioned according to a further aspect of the present invention for determining target The classification of relevance of searches classification corresponding to keyword determines equipment.

According to a further aspect of the invention, it additionally provides a kind of for determining the correlation of search corresponding to target keyword The search engine plug-in unit of property classification, wherein the search engine plug-in unit includes such as aforementioned being used for according to a further aspect of the present invention Determine that the classification of relevance of searches classification corresponding to target keyword determines equipment.

Compared with prior art, the present invention, which passes through, determines target critical term clustering belonging to target keyword, and then described Relevance of searches classification corresponding to target keyword effectively determines keyword institute to realize to be used for subsequent processing Corresponding relevance of searches classification, and to the automatic test of batch keyword data, only Optimizing Search engine search is not arranged Sequence provides reference, and improves the testing efficiency to search engine relevance.Moreover, the present invention may further determine that target keyword Corresponding preferred searching order routing information, to adjust the searching order routing information of the target keyword, thus into one Step realizes the sequence of Optimizing Search engine search, improves user information and obtains efficiency.Further, the present invention may further determine that excellent Change keyword set, the Optimizing Search sequence routing information of one or more of keywords to be optimized is determined, for adjusting The searching order routing information of one or more of keywords to be optimized is searched to further realize Optimizing Search engine Rope sequence improves user information and obtains efficiency.

Detailed description of the invention

By reading a detailed description of non-restrictive embodiments in the light of the attached drawings below, of the invention other Feature, objects and advantages will become more apparent upon:

Fig. 1 show one aspect according to the present invention for determining relevance of searches classification corresponding to target keyword Equipment schematic diagram；

Fig. 2 shows in accordance with a preferred embodiment of the present invention for determining relevance of searches corresponding to target keyword The equipment schematic diagram of classification；

Fig. 3 show according to a further aspect of the present invention for determining relevance of searches classification corresponding to target keyword Method flow diagram；

Fig. 4 show in accordance with a preferred embodiment of the present invention for determining relevance of searches corresponding to target keyword The method flow diagram of classification.

The same or similar appended drawing reference represents the same or similar component in attached drawing.

Specific embodiment

Present invention is further described in detail with reference to the accompanying drawing.

Fig. 1 show one aspect according to the present invention for determining relevance of searches classification corresponding to target keyword Classification determines equipment 1, wherein classification determines that equipment 1 includes cluster determining device 11 and classification determining device 12.Specifically, gather Class determining device 11 according to the searching order routing information of target keyword, determined from one or more keyword clusterings described in Target critical term clustering belonging to target keyword；Classification determining device 12 according to the target critical term clustering, determine described in Relevance of searches classification corresponding to target keyword, to be used for subsequent processing.Here, the meaning of described search correlation refers to The matching degree of keyword and search result.Here, classification determine equipment 1 include but is not limited to the network equipment, user equipment or The network equipment and user equipment are integrated constituted equipment by network.Wherein, the network equipment includes but is not limited to net The cloud that network host, single network server, multiple network server collection or multiple servers are constituted.Here, cloud is by based on cloud The a large amount of hosts or network server for calculating (Cloud Computing) are constituted, wherein and cloud computing is one kind of distributed computing, A super virtual computer consisting of a loosely coupled set of computers.The user equipment its include but is not limited to appoint What one kind can carry out the electronic product of human-computer interaction with user by keyboard, remote controler, touch tablet or voice-operated device, such as count Calculation machine, smart phone, PDA or IPTV etc..The network includes but is not limited to internet, wide area network, Metropolitan Area Network (MAN), local area network, VPN Network, wireless self-organization network (Ad Hoc network) etc..Those skilled in the art will be understood that above-mentioned classification determines that equipment 1 is only Citing, other network equipments or user equipment existing or be likely to occur from now on are such as applicable to the present invention, should also be included in Within the scope of the present invention, and it is incorporated herein by reference.

Specifically, cluster determining device 11 carries out clustering processing to multiple sample keywords first according to pre-defined rule, with Obtain one or more keyword clusterings；Further according to the searching order routing information of target keyword, from one or more keys Target critical term clustering belonging to the target keyword is determined in term clustering.It is searched here, the meaning of described search sequence refers to Index, which is held up, analyzes the understanding of user's input keyword and demand, with certain algorithm, in the predetermined web data extracted It is picked out in library and inputs the webpage that keyword matches with user, and provide it to user comprising but be not limited to such as theme The selection of matching degree result is sorted, good result proposes power sequence, cheating click is suppressed, general rise of prices of the stocks and other securities selected ci poem takes sequence, topic/abstract assembles row Sequence etc., wherein it may include many sub- sequences that the result, which proposes power sequence, and such as: web sites authority proposes power, official website proposes power, page Face richness proposes power, click proposes power etc..Here, described search engine includes but is not limited to that the Google of Google company such as is searched for Engine, baidu search engine of Baidu company etc., and as Google ToolBar of Google company, Baidu company hundred Degree searches the search engines plug-in unit such as MSN ToolBar of despot, Microsoft.Here, described search sequence routing information is for showing The sort algorithm code path information that search engine is passed through in determining candidate search sort result information process can be used and be searched Rope sequence ID of trace route path (Strategy Identifier, SID) and branch's mark (Branch Identifier, BID) carry out table Show, wherein branch's mark is subordinated to searching order ID of trace route path, can be transported when carrying out fine-grained mark to searching order With.Those skilled in the art will be understood that above-mentioned searching order, search engine and searching order routing information are only for example, other Searching order or search engine or searching order routing information existing or be likely to occur from now on are such as applicable to the present invention, It should be included within the scope of the present invention, and be incorporated herein by reference.

Specifically, cluster determining device 11 carries out clustering processing to multiple sample keywords, such as first according to pre-defined rule Using unsupervised learnings methods such as k-means, ISODATA, chain methods, to obtain one or more of keyword clusterings. Wherein, the pre-defined rule includes but is not limited to following at least any one:

It is crucial to the multiple sample according to the corresponding searching order routing information of the multiple sample keyword Word carries out clustering processing, to obtain one or more of keyword clusterings；

Information is recorded according to the historical search of the corresponding search user of the multiple sample keyword, to described more A sample keyword carries out clustering processing, to obtain one or more of keyword clusterings；

Meet the statistics in the content of pages information of predetermined quality degree threshold value according to the multiple each leisure of sample keyword Information carries out clustering processing to the multiple sample keyword, to obtain one or more of keyword clusterings.

For example, when the pre-defined rule includes according to the corresponding searching order path letter of the multiple sample keyword Breath carries out clustering processing to the multiple sample keyword, when obtaining one or more of keyword clusterings, it is assumed that multiple Sample keyword such as sample keyword I to VI, corresponding searching order routing information are as shown in table 1 below, wherein S_* Indicate the searching order ID of trace route path or branch's mark in the searching order path that keyword search request processing is passed through:

Sample keyword	Searching order routing information
		I	S_A→S_C→S_D→S_B→S_E→S_G
II	S_A→S_E→S_G→S_C→S_D→S_B
		III	S_A→S_F→S_C→S_D→S_E→S_G
IV	S_A→S_C→S_D→S_F→S_E→S_G
		V	S_A→S_E→S_G→S_C→S_B→S_D
VI	S_M→S_N→S_C→S_B→S_G→S_D

Table 1

Determining device 11 is then clustered according to the corresponding searching order routing information of sample keyword I to VI, using k- The unsupervised learnings methods such as means, ISODATA, chain method carry out clustering processing to sample keyword I to VI, obtain one Or multiple keyword clusterings are such as: sample keyword I, III and VI are such as classified as one kind by 1. the first keyword clustering of cluster1； 2. the second keyword clustering cluster2, is such as classified as one kind for keyword II and V；3. third keyword clustering cluster3, Sample keyword VI is such as classified as one kind；For another example, when the pre-defined rule includes according to the multiple each leisure of sample keyword Meet the statistical information in the content of pages information of predetermined quality degree threshold value, the multiple sample keyword is carried out at cluster Reason, when obtaining one or more of keyword clusterings, then clusters determining device 11 according to each leisure of sample keyword I to VI Meet the statistical information in the content of pages information of predetermined quality degree threshold value, as each comfortable satisfaction of sample keyword I to VI is predetermined The content of pages information of matter metric threshold such as belongs to high confidence and appoints in site page such as http://www.sina.com.cn/ Existing frequency information, using unsupervised learnings methods such as k-means, ISODATA, chain methods, to sample keyword I to VI into Row clustering processing obtains one or more keyword clusterings such as: 1. the first keyword clustering of cluster1, such as by sample key Word I, II and III are classified as one kind；2. sample keyword IV and VI are such as classified as one kind by the second keyword clustering cluster2； 3. third keyword clustering cluster3, is such as classified as one kind for sample keyword V.

Those skilled in the art will be understood that the above-mentioned mode for carrying out clustering processing to the multiple sample keyword is only Citing, other modes for carrying out clustering processing to the multiple sample keyword that are existing or being likely to occur from now on are for example applicable It in the present invention, should also be included within the scope of protection of the present invention, and be incorporated herein by reference.

Then, cluster determining device 11 is closed further according to the searching order routing information of target keyword from one or more Target critical term clustering belonging to the target keyword is determined in keyword cluster.Specifically, cluster determining device 11 is logical first It crosses search engine, browser, provide the application programming interfaces (API) of the third party devices such as target keyword equipment, obtain Target keyword, alternatively, obtaining the target critical that user is inputted by user equipment by dynamic web page techniques such as ASP, JSP Word；Then, cluster determining device 11 is gathered further according to the searching order routing information of target keyword from one or more keywords Target critical term clustering belonging to the target keyword is determined in class.

For example, it is assumed that test man A assessment search engine relevance test process in, in test platform keyword input field Target keyword goal-query is inputted, then clusters determining device 11 by dynamic web page techniques such as ASP, JSP, can get The target keyword goal-query that test man A is inputted by user equipment.

Those skilled in the art will be understood that the mode of above-mentioned acquisition target keyword is only for example, other are existing or modern The mode for the acquisition target keyword being likely to occur afterwards is such as applicable to the present invention, should also be included in the scope of the present invention with It is interior, and be incorporated herein by reference.

Finally, searching order routing information of the cluster determining device 11 further according to target keyword, is closed from one or more Target critical term clustering belonging to the target keyword is determined in keyword cluster.Here, described in cluster determining device 11 is determining The method of target critical term clustering includes but is not limited to following at least any one:

1) by the class searching order path of the searching order routing information of the target keyword and the keyword clustering Information is compared, with target critical term clustering belonging to the determination target keyword.For example, it is assumed that cluster determining device 11 After carrying out clustering processing to sample keyword I to VI as shown in Table 1, obtained each keyword clustering and the characterization key The class searching order routing information of term clustering is as shown in table 2 below:

Keyword clustering	Class searching order routing information
		cluster1	S_A→S_C→S_D→S_E→S_G→S_F
cluster2	S_A→S_E→S_G→S_C→S_D→S_B
		cluster3	S_M→S_N→S_C→S_B→S_G→S_D

Table 2

Assuming that cluster determining device 11 obtain target keyword goal-query searching order routing information be S_A → Sequence in S_C → S_D → S_B → S_E → S_G, with the class searching order routing information of the first keyword clustering cluster1 And searching order ID of trace route path it is all the same routing information it is most, then cluster determining device 11 determine target keyword goal- Target critical term clustering belonging to query is the first keyword clustering cluster1.

2) reference for including in the searching order routing information of the target keyword and the keyword clustering is crucial The searching order routing information of word is compared, with target critical term clustering belonging to the determination target keyword.For example, false If cluster determining device 11 obtain target keyword goal-query searching order routing information be S_A → S_C → S_D → S_B → S_E → S_G, it is identical as the searching order routing information of keyword I in the first keyword clustering cluster1, then gather Class determining device 11 determines that target critical term clustering belonging to target keyword goal-query is the first keyword clustering cluster1。

Those skilled in the art will be understood that the side of target critical term clustering belonging to the above-mentioned determination target keyword Formula is only for example, target critical term clustering belonging to other described target keywords of determination that is existing or being likely to occur from now on Mode is such as applicable to the present invention, should also be included within the scope of protection of the present invention, and is incorporated herein by reference.

Classification determining device 12 can count each right with reference to keyword included by the target critical term clustering first The relevance of searches description information answered determines relevance of searches classification corresponding to target critical term clustering；Then, further according to institute Relevance of searches classification corresponding to target critical term clustering is stated, determines relevance of searches class corresponding to the target keyword Not, to be used for subsequent processing.Here, described search correlation classification includes but is not limited to such as high correlation classification, lower phase Closing property classification, uncorrelated classification, cheating keyword categories etc..Here, the subsequent processing includes but is not limited to such as: 1) to target Keyword carries out Screening Treatment, such as whether as test data etc.；2) the searching order information of optimization aim keyword.This field Technical staff will be understood that above-mentioned relevance of searches classification and subsequent processing mode are only for example, other are existing or from now on may The relevance of searches classification or subsequent processing mode of appearance are such as applicable to the present invention, should also be included in the scope of the present invention with It is interior, and be incorporated herein by reference.

For example, it is assumed that cluster determining device 11 determines that target critical term clustering belonging to target keyword goal-query is First keyword clustering cluster1, and the institute of sample keyword I, II and III included by keyword clustering cluster1 are right The relevance of searches description information answered is respectively that correlation is high, correlation is high, correlation is low, since the high correlation of correlation is retouched State information account for correlation description information total quantity ratio meet be greater than threshold value such as 0.65, then classification determining device 12 determination target Relevance of searches classification corresponding to keyword goal-query is the high classification of correlation.For another example, it is assumed that cluster determining device 11 Determine that target critical term clustering belonging to target keyword goal-query is the first keyword clustering cluster2, and keyword The corresponding relevance of searches description information for clustering sample keyword IV and VI included by cluster2 is respectively correlation It is low, correlation is low, due to the low correlation description information of correlation account for correlation description information total quantity ratio satisfaction be greater than Threshold value such as 0.65, then classification determining device 12 determines that relevance of searches classification corresponding to target keyword goal-query is phase The low classification of closing property.

Those skilled in the art will be understood that relevance of searches classification corresponding to the above-mentioned determination target keyword Mode is only for example, relevance of searches class corresponding to other described target keywords of determination that is existing or being likely to occur from now on It is such as applicable to the present invention otherwise, should also be included within the scope of protection of the present invention, and is contained in by reference herein This.

Classification determines constantly to work between each device of equipment 1.Specifically, cluster determining device 11 continues According to the searching order routing information of target keyword, the target keyword institute is determined from one or more keyword clusterings The target critical term clustering of category；Classification determining device 12 continues to determine the target critical according to the target critical term clustering Relevance of searches classification corresponding to word, to be used for subsequent processing.Here, it should be understood by those skilled in the art that " lasting " refers to Classification determines the determination that each device of equipment 1 constantly carries out target critical term clustering respectively and relevance of searches classification really It is fixed, until classification determines that equipment 1 stops the determination of target critical term clustering in a long time.

Preferably, the keyword clustering includes the class searching order routing information for characterizing the keyword clustering, Cluster determining device 11 includes that comparing unit (not shown), similarity determining unit (not shown) and cluster determination unit (are not shown Out).The preferred embodiment is described below with reference to Fig. 1: comparing unit is by the searching order path of the target keyword Information is compared with class searching order routing information corresponding to one or more of keyword clusterings, to determine State the searching order routing information of target keyword and the smallest edit distance of the class searching order routing information；Similarity is true Order member determines that the searching order routing information of the target keyword and class search are arranged according to the smallest edit distance The sequence similarity of paths of sequence routing information；Determination unit is clustered according to the sequence similarity of paths, determines that the target is closed Keyword cluster.

Specifically, comparing unit carries out clustering processing to multiple sample keywords, such as uses k- first according to pre-defined rule The unsupervised learnings methods such as means, ISODATA, chain method, to obtain described in one or more of keyword clusterings determinations One or more keyword clusterings.Here, mode and cluster that comparing unit obtains one or more of keyword clusterings are really Determine device 11 obtain one or more of keyword clusterings mode it is same or similar, for simplicity, therefore no longer superfluous herein State, and include by reference and this.

Then, comparing unit is by the searching order routing information of the target keyword and one or more of keywords The corresponding class searching order routing information of cluster is compared, with the searching order path of the determination target keyword The smallest edit distance of information and the class searching order routing information.For example, it is assumed that the target keyword that comparing unit obtains The searching order routing information of goal-query is S_A → S_C → S_D → S_B → S_E → S_G, and described in comparing unit determination One or more keyword clusterings are as shown in Table 2 above, then comparing unit is by the searching order of target keyword goal-query Routing information S_A → S_C → S_D → S_B → S_E → S_G is serialized to obtain character string goal-string= " ACDBEG " equally carries out class searching order routing information corresponding to keyword clustering cluster1 to cluster3 Serializing obtains corresponding character string such as cluster1-string=" ACDEGF ", cluster2-string=" AEGSDB ", Cluster3-string=" MNCBGD ", then, comparing unit pass through the smallest edit distances such as Dynamic Programming, matrix method Algorithm calculates separately character string goal-string corresponding to the searching order routing information of target keyword goal-query The corresponding class searching order routing information institute of=" ACDBEG " and keyword clustering cluster1 to cluster3 is right The character string answered such as cluster1-string=" ACDEGF ", cluster2-string=" AEGSDB ", cluster3- The smallest edit distance of string=" MNCBGD ", such as obtains target keyword goal-query and keyword clustering cluster1 To cluster3 pair smallest edit distance be respectively as follows: 2,6 and 6.

Those skilled in the art will be understood that the mode of the above-mentioned determination smallest edit distance is only for example, other are existing Or the mode of the determination smallest edit distance that is likely to occur from now on be such as applicable to the present invention, should also be included in the present invention Within protection scope, and it is incorporated herein by reference.

Similarity determining unit determines the searching order path letter of the target keyword according to the smallest edit distance The sequence similarity of paths of breath and the class searching order routing information.For example, connect example, similarity determining unit is according to comparing The smallest edit distance that unit determines, the sequence similarity of paths is determined by following formula (1):

Wherein, d is smallest edit distance, then similarity determining unit determines target keyword according to above-mentioned formula (1) The searching order routing information of goal-query is searched with the respective class of keyword clustering cluster1 to cluster3 respectively The sequence similarity of paths of rope sequence routing information is respectively as follows: 1/3,1/7 and 1/7.

Preferably, the smallest edit distance that similarity determining unit is determined according to comparing unit, passes through following formula (2) the sequence similarity of paths is determined:

Wherein, α is normalization coefficient,For the average string length of character string corresponding to class searching order routing information, D is smallest edit distance, wherein normalization coefficient α can be calculated by following formula (3):

Wherein, x indicates the statistical length of character string during the test corresponding to class searching order routing information, if α= 0.5, and character string corresponding to the corresponding class searching order routing information of keyword clustering cluster1 to cluster3 Average string length is 6, then similarity determining unit can determine target keyword goal-query's according to above-mentioned formula (2) Searching order routing information is believed with the respective class searching order path keyword clustering cluster1 to cluster3 respectively The sequence similarity of paths of breath is respectively as follows: 1,3/7 and 3/7.

Those skilled in the art will be understood that the mode of the above-mentioned determination sequence similarity of paths is only for example, other are existing The mode of the determination sequence similarity of paths that is having or being likely to occur from now on is such as applicable to the present invention, should also be included in this Within invention protection scope, and it is incorporated herein by reference.

Determination unit is clustered according to the sequence similarity of paths, determines the target critical term clustering, as described in determining The target critical term clustering belonging to target keyword is that the sequence similarity of paths meets predetermined threshold as corresponding to 0.8 Keyword clustering.For example, connecting example, similarity determining unit determines the searching order path of target keyword goal-query The information sequence path with the respective class searching order routing information of keyword clustering cluster1 to cluster3 respectively Similarity is respectively as follows: 1,3/7 and 3/7, then clusters determination unit and determine the target belonging to target keyword goal-query Keyword clustering is cluster1.

Preferably, cluster determining device 11 can also obtain one or more keywords to be measured to be processed first, using as The target keyword；Then, poly- from one or more keywords according to the searching order routing information of the target keyword Target critical term clustering belonging to the target keyword is determined in class；Classification determining device 12 can also be first according to the target Keyword clustering determines relevance of searches classification corresponding to the target keyword；Then, according to described search correlation class Not, Screening Treatment is carried out to the target keyword.

Specifically, cluster determining device 11 can also be set by such as search engine, browser, offer keyword to be measured first The application programming interfaces (API) of standby equal third party devices obtain one or more keywords to be measured to be processed, using as described Target keyword；Then, according to the searching order routing information of the target keyword, from one or more keyword clusterings Determine target critical term clustering belonging to the target keyword.Here, cluster determining device 11 determines the target keyword The mode of affiliated target critical term clustering and aforementioned cluster determining device 11 determine target critical belonging to the target keyword The mode of term clustering is same or similar, and for simplicity, therefore details are not described herein, and include by reference and this.

Then, classification determining device 12 can also determine the target keyword first according to the target critical term clustering Corresponding relevance of searches classification.Here, classification determining device 12 determines that search corresponding to the target keyword is related Property class the side of relevance of searches classification corresponding to the target keyword is determined with aforementioned categories determining device 12 otherwise Formula is same or similar, and for simplicity, therefore details are not described herein, and include by reference and this.

Then, classification determining device 12 carries out at screening the target keyword according to described search correlation classification Reason.For example, it is assumed that the target keyword that cluster determining device 11 obtains include such as query1, query2, query3 and Query4, and classification determining device 12 determines that the target keyword query1, query2, query3 and query4 are corresponding described Relevance of searches classification be respectively correlation is high, in correlation, correlation is low, correlation is high, then 12 basis of classification determining device The corresponding described search correlation classification of target keyword query1, query2, query3 and query4, sieves it Choosing processing, such as the keyword query3 for belonging to the low classification of correlation is screened from keyword set to be measured, with to its into Row later period searching order Advance data quality.

Preferably, classification determines that equipment 1 further includes that set determining device (not shown) and path optimizing determining device (are not shown Out).Specifically, set determining device determines keyword set to be optimized corresponding to the keyword clustering；Path optimizing is true Determine device common search row according to corresponding to the one or more keyword to be optimized that the keyword set to be optimized includes Sequence routing information determines the Optimizing Search sequence routing information of one or more of keywords to be optimized, for adjusting institute State the searching order routing information of one or more keywords to be optimized.

Specifically, the relevance of searches description letter for all keywords that set determining device can include according to keyword clustering Breath, such as relevance of searches is high, relevance of searches is low, finds out the low classification of relevance of searches, using as the keyword clustering Corresponding keyword set to be optimized.Here, the keyword set to be optimized, which corresponds to, belongs to the low classification of relevance of searches Keyword.For example, it is assumed that the keyword for belonging to the low classification of relevance of searches in keyword clustering cluster1 is that sample is crucial The keyword for belonging to the low classification of relevance of searches in word III, keyword clustering cluster2 is sample keyword IV and VI, crucial The keyword for not belonging to the low classification of relevance of searches in term clustering cluster3, then gathering determining device can be by keyword clustering All keywords for belonging to the low classification of relevance of searches that cluster1 includes into cluster3 are as the key to be optimized Set of words such as includes sample keyword III, IV and VI.

Preferably, set determining device can also be by the actual search knot of all keywords included by the keyword clustering Fruit relevant information is compared with system index information, and the keyword set to be optimized is determined from all keywords. Here, the actual search results relevant information includes but is not limited to as returned to search result quantity, obtaining click volume, return station The authority of point, the quality degree of content of pages for returning to website etc..Here, the system index information includes as returned to search knot Fruit quantity, the authority for returning to website etc..For example, set determining device can will cluster the determining keyword of determining device 11 Cluster the letter of all keywords, that is, sample keyword I to VI actual search results correlation included by cluster1 to cluster3 Breath is compared with system index information, the keyword set to be optimized is determined from all keywords, such as by sample The actual search results relevant information of keyword I to VI is unsatisfactory for the keyword of system index information, as the pass to be optimized Keyword set.

Those skilled in the art will be understood that the mode of the above-mentioned determination keyword set to be optimized is only for example, other The mode of the determination keyword set to be optimized that is existing or being likely to occur from now on is such as applicable to the present invention, also should include Within the scope of the present invention, and it is incorporated herein by reference.

Then, the one or more pass to be optimized that path optimizing determining device includes according to the keyword set to be optimized The sequence routing information of common search corresponding to keyword determines the Optimizing Search sequence of one or more of keywords to be optimized Routing information, with the searching order routing information for adjusting one or more of keywords to be optimized.For example, example is connected, Gathering the keyword set to be optimized that determining device determines includes sample keyword III, IV and VI, then path optimizing determines Device can the common search according to corresponding to sample keyword III, IV and VI sort routing information such as S_C → S_D → S-G, make For the Optimizing Search sort routing information, with the searching order path for adjusting one or more of keywords to be optimized Information such as deletes the Optimizing Search sequence path letter for including in the searching order routing information of sample keyword III, IV and VI Breath, alternatively, by the Optimizing Search for including in the searching order routing information of sample keyword III, IV and VI sequence path letter Breath is by other searching order routing information common search as corresponding to the keyword of the high classification of relevance of searches sequence path letter Breath replaces.

Preferably, the above-mentioned classification for being used to determine relevance of searches classification corresponding to target keyword can be determined equipment 1, it is combined with existing search engine, constitutes a kind of new search engine, existing search engine includes but is not limited to such as Google search engine, baidu search engine of Baidu company of Google company etc..

Preferably, the above-mentioned classification for being used to determine relevance of searches classification corresponding to target keyword can be determined equipment 1, it is combined with existing search engine plug-in unit, constitutes a kind of new search engine plug-in unit, existing includes but is not limited to such as Google ToolBar of Google company, the Baidu of Baidu company search the search engines such as MSN ToolBar of despot, Microsoft Plug-in unit.

Fig. 2 shows in accordance with a preferred embodiment of the present invention for determining relevance of searches corresponding to target keyword The equipment schematic diagram of classification, wherein classification determines that equipment 1 includes cluster determining device 11 ', classification determining device 12 ', preferably road Diameter determining device 13 ', adjustment device 14 ' and offer device 15 '.Specifically, cluster determining device 11 ' is according to target keyword Searching order routing information determines that target keyword belonging to the target keyword is poly- from one or more keyword clusterings Class；Classification determining device 12 ' determines relevance of searches corresponding to the target keyword according to the target critical term clustering Classification, to be used for subsequent processing；Preferred path determining device 13 ' determines preferred searching order corresponding to the target keyword Routing information；Device 14 ' is adjusted according to the preferred searching order routing information, adjusts the searching order of the target keyword Routing information；If search sequence matches with the target keyword, device 15 ' is provided according to the target critical adjusted The searching order routing information of word, search result corresponding to the search sequence is supplied to corresponding to the search sequence User.Here, cluster determining device 11 ', classification determining device 12 ' are same or similar with corresponding intrument shown in Fig. 1 respectively, so Place repeats no more, and is incorporated herein by reference.

And specifically, it is preferable to which path determining device 13 ' determines preferred searching order path corresponding to the target keyword Information.Here, preferred path determining device 13 ' determine the preferred searching order routing information mode include but is not limited to Under any one of at least:

1) by the sequence of each common search with reference to corresponding to keyword included by target critical term clustering path Information, as the preferred searching order routing information.For example, it is assumed that cluster determining device 11 ' determines target keyword goal- Target critical term clustering belonging to query is keyword clustering cluster1, then preferred path determining device 13 ' can incite somebody to action Each sequence of the common search with reference to corresponding to keyword, that is, sample keyword I, III and VI path letter included by cluster1 Breath such as S_A → S_C → S_D → S_E → S_G, as the preferred searching order routing information.

2) to each searching order routing information with reference to corresponding to keyword included by the target critical term clustering Processing for statistical analysis, with the determination preferred searching order routing information, as counted the search row for showing that keyword is passed by The frequency is met the path of predetermined threshold, or preferably searched using high frequency searching order path as described by sequence path frequency information Rope sequence routing information.For example, connecting example, preferred path determining device 13 ' can close each reference included by cluster1 Searching order routing information processing for statistical analysis corresponding to keyword, that is, sample keyword I, III and VI, the frequency is met Predetermined threshold such as frequency of occurrence meets 2 searching order path such as S_A → S_C → S_D → S_E → S_G → S_F, as described It is preferred that searching order routing information.

Those skilled in the art will be understood that the mode of the above-mentioned determination preferred searching order routing information is only for example, The mode of other determination preferred searching order routing informations that are existing or being likely to occur from now on is such as applicable to the present invention, It should also be included within the scope of protection of the present invention, and be incorporated herein by reference.

Device 14 ' is adjusted according to the preferred searching order routing information, adjusts the searching order of the target keyword Routing information.For example, it is assumed that the preferred searching order routing information that preferred path determining device 13 ' determines is S_A → S_C → S_D → S_E → S_G, then adjust device 14 ' according to the preferred searching order routing information, by such as adjustment algorithm or Machine learning model such as SVM model adjusts the searching order routing information of target keyword goal-query, such as by target critical The searching order routing information of word goal-query is adjusted to preferentially to execute the preferred searching order routing information S_A → S_C →S_D→S_E→S_G。

If search sequence matches with the target keyword, device 15 ' is provided according to the target critical adjusted The searching order routing information of word, search result corresponding to the search sequence is supplied to corresponding to the search sequence User.Specifically, device 15 ' is provided and obtains search sequence first；Then, judge the search sequence and the target keyword Whether match, if matching, provides device 15 ' according to the searching order routing information of the target keyword adjusted, incite somebody to action Search result corresponding to the search sequence is supplied to user corresponding to the search sequence.Matched contain here, described Including search sequence and the target keyword, completely the same, search sequence is contained in the target keyword to justice.

Specifically, device 15 ' is provided first by dynamic web page techniques such as ASP, JSP, or is provided by search engine Application programming interfaces (API), obtain the search sequence that inputs by user equipment of user.For example, if search user B passes through it PC equipment inputs keyword " fresh flower " in search engine search column, by "enter" key", provides device 15 ' and passes through such as ASP, JSP Or the dynamic web page techniques such as PHP, the keyword " fresh flower " of search user B input can be got.Those skilled in the art should manage The mode for solving above-mentioned acquisition search sequence is only for example, other modes of acquisition search sequence that are existing or being likely to occur from now on It is such as applicable to the present invention, should also be included within the scope of protection of the present invention, and is incorporated herein by reference.

Then, the target keyword that device 15 ' is obtained according to cluster determining device 11 ' is provided, is compared by text Mode, judge whether the search sequence matches with the target keyword.

If matching, device 15 ' is provided according to the searching order routing information of the target keyword adjusted, it will be described Search result corresponding to search sequence is supplied to user corresponding to the search sequence.For example, connecting example, it is assumed that cluster is true The target keyword such as " fresh flower ", " the fresh flower express delivery " for determining the acquisition of device 11 ', then provide device 15 ' and judge search sequence " fresh flower " matches with target keyword such as " fresh flower ", " fresh flower express delivery ", then, provides device 15 ' for search engine according to tune Search result corresponding to the searching order routing information of the target keyword after whole such as " fresh flower Baidu discussion bar ", " fresh flower Picture materials day is off line ", as search result corresponding to search sequence " fresh flower ", and it is dynamic by ASP, JSP or PHP etc. The communication mode of state web technologies or other agreements, such as http or https communication protocol, are supplied to the search sequence Corresponding user, that is, user B browses such as the user equipment of the user for user.

Fig. 3 show according to a further aspect of the present invention for determining relevance of searches classification corresponding to target keyword Method flow diagram.

Specifically, in step sl, classification determines equipment 1 according to the searching order routing information of target keyword, from one Target critical term clustering belonging to the target keyword is determined in a or multiple keyword clusterings；In step s 2, classification is true Locking equipment 1 determines relevance of searches classification corresponding to the target keyword, according to the target critical term clustering to be used for Subsequent processing.Here, the meaning of described search correlation refers to the matching degree of keyword and search result.Here, classification is true Locking equipment 1 includes but is not limited to that the network equipment, user equipment or the network equipment and user equipment are integrated by network and are constituted Equipment.Wherein, the network equipment includes but is not limited to network host, single network server, multiple network server collection Or the cloud that multiple servers are constituted.Here, cloud by based on cloud computing (Cloud Computing) a large amount of hosts or network service Device is constituted, wherein cloud computing is one kind of distributed computing, a super void consisting of a loosely coupled set of computers Quasi- computer.The user equipment its include but is not limited to any one can with user by keyboard, remote controler, touch tablet or The electronic product, such as computer, smart phone, PDA or IPTV etc. of voice-operated device progress human-computer interaction.The network include but It is not limited to internet, wide area network, Metropolitan Area Network (MAN), local area network, VPN network, wireless self-organization network (Ad Hoc network) etc..This field Technical staff will be understood that above-mentioned classification determines that equipment 1 is only for example, other network equipments that are existing or being likely to occur from now on Or user equipment is such as applicable to the present invention, should also be included within the scope of protection of the present invention, and includes by reference herein In this.

Specifically, in step sl, classification determines that equipment 1 according to pre-defined rule, carries out multiple sample keywords first Clustering processing, to obtain one or more keyword clusterings；Further according to the searching order routing information of target keyword, from one Or target critical term clustering belonging to the target keyword is determined in multiple keyword clusterings.Here, described search sequence Meaning refers to that search engine inputs the understanding of keyword to user and demand is analyzed, and with certain algorithm, extracts predetermined Web database in pick out and input the webpage that matches of keyword with user, and do not provide it to user comprising but not Be limited to as theme matching degree result choose sequence, good result mention power sequence, cheating click suppress, general rise of prices of the stocks and other securities selected ci poem takes sequence, topic/ Abstract assembling sequence etc., wherein it may include many sub- sequences that the result, which proposes power sequence, and such as: web sites authority proposes power, official Net proposes power, page richness proposes power, clicks the power of proposing etc..Here, described search engine includes but is not limited to such as Google company Google search engine, baidu search engine of Baidu company etc., and as Google company Google ToolBar, hundred The search engines plug-in unit such as MSN ToolBar of despot, Microsoft are searched by the Baidu of degree company.Here, described search sequence path letter Cease the sort algorithm code path letter passed through in determining candidate search sort result information process for showing search engine Breath can identify (Branch with searching order ID of trace route path (Strategy Identifier, SID) and branch Identifier, BID) indicate, wherein branch's mark is subordinated to searching order ID of trace route path, searching order is carried out it is thin It can be employed when the mark of granularity.Those skilled in the art will be understood that above-mentioned searching order, search engine and searching order road Diameter information is only for example, other searching orders or search engine or searching order routing information existing or be likely to occur from now on It is such as applicable to the present invention, should also be included within the scope of protection of the present invention, and is incorporated herein by reference.

Specifically, in step sl, classification determines that equipment 1 according to pre-defined rule, carries out multiple sample keywords first Clustering processing, such as k-means, ISODATA, chain method unsupervised learning method are used, it is one or more of to obtain Keyword clustering.Wherein, the pre-defined rule includes but is not limited to following at least any one:

For example, when the pre-defined rule includes according to the corresponding searching order path letter of the multiple sample keyword Breath carries out clustering processing to the multiple sample keyword, when obtaining one or more of keyword clusterings, it is assumed that multiple Sample keyword such as sample keyword I to VI, corresponding searching order routing information are as shown in table 3 below, wherein S_* Indicate the searching order ID of trace route path or branch's mark in the searching order path that keyword search request processing is passed through:

Table 3

Then in step sl, classification determines that equipment 1 is believed according to the corresponding searching order path sample keyword I to VI Breath, using unsupervised learnings methods such as k-means, ISODATA, chain methods, carries out at cluster sample keyword I to VI Reason, obtains one or more keyword clusterings such as: 1. the first keyword clustering of cluster1, such as by sample keyword I, III and VI is classified as one kind；2. the second keyword clustering cluster2, is such as classified as one kind for keyword II and V；3. third keyword is poly- Sample keyword VI is such as classified as one kind by class cluster3；For another example, when the pre-defined rule includes being closed according to the multiple sample Each leisure of keyword meets the statistical information in the content of pages information of predetermined quality degree threshold value, to the multiple sample keyword into Row clustering processing, when obtaining one or more of keyword clusterings, then in step sl, classification determines equipment 1 according to sample Each leisure of this keyword I to VI meets the statistical information in the content of pages information of predetermined quality degree threshold value, such as sample keyword I The content of pages information for meeting predetermined quality degree threshold value to each leisure of VI such as belongs to high confidence and appoints site page such as http: // The frequency information occurred in www.sina.com.cn/, using unsupervised learnings sides such as k-means, ISODATA, chain methods Method carries out clustering processing to sample keyword I to VI, obtains one or more keyword clusterings such as: 1. the first keyword clustering of Sample keyword I, II and III are such as classified as one kind by cluster1；2. the second keyword clustering cluster2, such as closes sample Keyword IV and VI are classified as one kind；3. third keyword clustering cluster3, is such as classified as one kind for sample keyword V.

Then, in step sl, classification determines equipment 1 further according to the searching order routing information of target keyword, from one Target critical term clustering belonging to the target keyword is determined in a or multiple keyword clusterings.Specifically, in step sl, Classification determines equipment 1 first by the application of the third party devices such as search engine, browser, offer target keyword equipment Routine interface (API) obtains target keyword, alternatively, obtaining user by dynamic web page techniques such as ASP, JSP and passing through user The target keyword of equipment input；Then, cluster determining device 11 further according to target keyword searching order routing information, from Target critical term clustering belonging to the target keyword is determined in one or more keyword clusterings.

For example, it is assumed that test man A assessment search engine relevance test process in, in test platform keyword input field Input target keyword goal-query, then in step sl, classification determines that equipment 1 passes through the dynamic web page techniques such as ASP, JSP, The target keyword goal-query that test man A is inputted by user equipment can be got.

Finally, in step sl, classification determines equipment 1 further according to the searching order routing information of target keyword, from one Target critical term clustering belonging to the target keyword is determined in a or multiple keyword clusterings.Here, in step sl, class Not Que Ding the method that determines the target critical term clustering of equipment 1 include but is not limited to it is following any one of at least:

1) by the class searching order path of the searching order routing information of the target keyword and the keyword clustering Information is compared, with target critical term clustering belonging to the determination target keyword.For example, it is assumed that in step sl, class Not Que Ding after equipment 1 carries out clustering processing to sample keyword I to VI as shown in table 3, obtained each keyword clustering and The class searching order routing information for characterizing the keyword clustering is as shown in table 4 below:

Table 4

Assuming that in step sl, classification determines the searching order path for the target keyword goal-query that equipment 1 obtains Information is S_A → S_C → S_D → S_B → S_E → S_G, the class searching order path with the first keyword clustering cluster1 In information sequence and searching order ID of trace route path it is all the same routing information it is most, then in step sl, classification determines that equipment 1 is true The target critical term clustering belonging to keyword goal-query that sets the goal is the first keyword clustering cluster1.

2) reference for including in the searching order routing information of the target keyword and the keyword clustering is crucial The searching order routing information of word is compared, with target critical term clustering belonging to the determination target keyword.For example, false If in step sl, classification determine equipment 1 obtain target keyword goal-query searching order routing information be S_A → The searching order routing information of keyword I in S_C → S_D → S_B → S_E → S_G, with the first keyword clustering cluster1 Identical, then in step sl, classification determines that equipment 1 determines that target critical term clustering belonging to target keyword goal-query is First keyword clustering cluster1.

In step s 2, classification determines that equipment 1 can count each reference included by the target critical term clustering first Relevance of searches description information corresponding to keyword determines relevance of searches classification corresponding to target critical term clustering；So Afterwards, it further according to relevance of searches classification corresponding to the target critical term clustering, determines corresponding to the target keyword Relevance of searches classification, to be used for subsequent processing.Here, described search correlation classification includes but is not limited to such as high correlation Classification, compared with low correlation classification, uncorrelated classification, cheating keyword categories etc..Here, the subsequent processing includes but is not limited to Such as: 1) Screening Treatment being carried out to target keyword, such as whether as test data etc.；2) searching order of optimization aim keyword Information.Those skilled in the art will be understood that above-mentioned relevance of searches classification and subsequent processing mode are only for example, other are existing Or the relevance of searches classification or subsequent processing mode that are likely to occur from now on be such as applicable to the present invention, should also be included in this hair Within bright protection scope, and it is incorporated herein by reference.

For example, it is assumed that in step sl, classification determines that equipment 1 determines that target belonging to target keyword goal-query is closed Keyword cluster be the first keyword clustering cluster1, and sample keyword I, II included by keyword clustering cluster1 and The corresponding relevance of searches description information of III is respectively that correlation is high, correlation is high, correlation is low, due to correlation height Correlation description information account for correlation description information total quantity ratio meet be greater than threshold value such as 0.65, then in step s 2, Classification determines that equipment 1 determines that relevance of searches classification corresponding to target keyword goal-query is the high classification of correlation.Again Such as, it is assumed that in step sl, classification determines that equipment 1 determines that target critical term clustering belonging to target keyword goal-query is First keyword clustering cluster2, and sample keyword IV's and VI included by keyword clustering cluster2 is corresponding Relevance of searches description information is respectively that correlation is low, correlation is low, since the low correlation description information of correlation accounts for correlation Property description information total quantity ratio meet be greater than threshold value such as 0.65, then in step s 2, classification determine equipment 1 determine target pass Relevance of searches classification corresponding to keyword goal-query is the low classification of correlation.

Classification determines constantly to work between each step of equipment 1.Specifically, in step sl, classification is true Locking equipment 1 is persistently according to the searching order routing information of target keyword, from one or more keyword clusterings described in determination Target critical term clustering belonging to target keyword；In step s 2, classification determines that equipment 1 continues according to the target keyword Cluster, determines relevance of searches classification corresponding to the target keyword, to be used for subsequent processing.Here, art technology Personnel should understand that " to continue " to refer to that classification determines that each step of equipment 1 constantly carries out target critical term clustering really respectively Fixed and relevance of searches classification determination, until classification determines that equipment 1 stops target critical term clustering really in a long time It is fixed.

Preferably, the keyword clustering includes the class searching order routing information for characterizing the keyword clustering, Step S1 includes step S11 (not shown), step S12 (not shown) and step S13 (not shown).It is excellent to this below with reference to Fig. 3 Select embodiment to be described: in step s 11, classification determines searching order routing information of the equipment 1 by the target keyword It is compared with class searching order routing information corresponding to one or more of keyword clusterings, with the determination mesh Mark the searching order routing information of keyword and the smallest edit distance of the class searching order routing information；In step s 12, Classification determines equipment 1 according to the smallest edit distance, determine the searching order routing information of the target keyword with it is described The sequence similarity of paths of class searching order routing information；In step s 13, classification determines equipment 1 according to the sequence path Similarity determines the target critical term clustering.

Specifically, in step s 11, classification determines that equipment 1 according to pre-defined rule, carries out multiple sample keywords first Clustering processing, such as k-means, ISODATA, chain method unsupervised learning method are used, it is one or more of to obtain Keyword clustering determines one or more of keyword clusterings.Here, in step s 11, it is described that classification determines that equipment 1 obtains For the mode of one or more keyword clusterings in step sl, classification determines that equipment 1 obtains one or more of keywords The mode of cluster is same or similar, and for simplicity, therefore details are not described herein, and include by reference and this.

Then, in step s 11, classification determines equipment 1 by the searching order routing information of the target keyword and institute It states the class searching order routing information corresponding to one or more keyword clusterings to be compared, be closed with the determination target The smallest edit distance of the searching order routing information of keyword and the class searching order routing information.For example, it is assumed that in step In S11, classification determines that the searching order routing information for the target keyword goal-query that equipment 1 obtains is S_A → S_C → S_ D → S_B → S_E → S_G, and in step s 11, classification determines that equipment 1 determines that one or more of keyword clusterings are as above It states shown in table 2, then in step s 11, classification determines searching order routing information of the equipment 1 by target keyword goal-query S_A → S_C → S_D → S_B → S_E → S_G is serialized to obtain character string goal-string=" ACDBEG ", equally will The class searching order routing information, which is serialized, corresponding to keyword clustering cluster1 to cluster3 is corresponded to Character string such as cluster1-string=" ACDEGF ", cluster2-string=" AEGSDB ", cluster3-string= " MNCBGD ", then, in step s 11, classification determines that equipment 1 passes through the smallest edit distances such as Dynamic Programming, matrix method Algorithm calculates separately character string goal-string corresponding to the searching order routing information of target keyword goal-query The corresponding class searching order routing information institute of=" ACDBEG " and keyword clustering cluster1 to cluster3 is right The character string answered such as cluster1-string=" ACDEGF ", cluster2-string=" AEGSDB ", cluster3- The smallest edit distance of string=" MNCBGD ", such as obtains target keyword goal-query and keyword clustering cluster1 To cluster3 pair smallest edit distance be respectively as follows: 2,6 and 6.

In step s 12, classification determines that equipment 1 according to the smallest edit distance, determines searching for the target keyword The sequence similarity of paths of rope sequence routing information and the class searching order routing information.For example, example is connected, in step S12 In, classification determines the smallest edit distance that equipment 1 is determined according to comparing unit, is determined by following formula (4) described Sort similarity of paths:

Wherein, d is smallest edit distance, then similarity determining unit determines target keyword according to above-mentioned formula (4) The searching order routing information of goal-query is searched with the respective class of keyword clustering cluster1 to cluster3 respectively The sequence similarity of paths of rope sequence routing information is respectively as follows: 1/3,1/7 and 1/7.

Preferably, in step s 12, classification determines the smallest edit distance that equipment 1 is determined according to comparing unit, leads to Following formula (5) is crossed to determine the sequence similarity of paths:

Wherein, α is normalization coefficient,For the average string length of character string corresponding to class searching order routing information, D is smallest edit distance, wherein normalization coefficient α can be calculated by following formula (6):

Wherein, x indicates the statistical length of character string during the test corresponding to class searching order routing information, if α= 0.5, and character string corresponding to the corresponding class searching order routing information of keyword clustering cluster1 to cluster3 Average string length is 6, then in step s 12, classification determines that equipment 1 can determine target keyword according to above-mentioned formula (5) The searching order routing information of goal-query is searched with the respective class of keyword clustering cluster1 to cluster3 respectively The sequence similarity of paths of rope sequence routing information is respectively as follows: 1,3/7 and 3/7.

In step s 13, classification determines that equipment 1 according to the sequence similarity of paths, determines that the target keyword is poly- Class, determined, the target critical term clustering belonging to the target keyword meets predetermined threshold for the sequence similarity of paths It is worth the keyword clustering as corresponding to 0.8.For example, connecting example, in step s 12, classification determines that equipment 1 determines target keyword The searching order routing information of goal-query is searched with the respective class of keyword clustering cluster1 to cluster3 respectively The sequence similarity of paths of rope sequence routing information is respectively as follows: 1,3/7 and 3/7, then in step s 13, classification determines that equipment 1 is true The target critical term clustering belonging to keyword goal-query that sets the goal is cluster1.

Preferably, in step sl, classification determines equipment 1 can also obtain one or more key to be measured to be processed first Word, using as the target keyword；Then, according to the searching order routing information of the target keyword, from one or more Target critical term clustering belonging to the target keyword is determined in a keyword clustering；In step s 2, classification determines equipment 1 Also relevance of searches classification corresponding to the target keyword can be determined first according to the target critical term clustering；Then, According to described search correlation classification, Screening Treatment is carried out to the target keyword.

Specifically, in step sl, classification determine equipment 1 can also first by such as search engine, browser, offer to The application programming interfaces (API) for surveying the third party devices such as keyword equipment obtain one or more keyword to be measured to be processed, Using as the target keyword；Then, it according to the searching order routing information of the target keyword, is closed from one or more Target critical term clustering belonging to the target keyword is determined in keyword cluster.Here, in step sl, classification determines equipment 1 determine the mode of target critical term clustering belonging to the target keyword and it is aforementioned in step sl, classification determines that equipment 1 is fixed The mode of target critical term clustering belonging to the target keyword is same or similar, and for simplicity, therefore details are not described herein, And include by reference and this.

Then, in step s 2, classification determine equipment 1 can also first according to the target critical term clustering, determine described in Relevance of searches classification corresponding to target keyword.Here, in step s 2, classification determines that equipment 1 determines that the target is closed Relevance of searches class corresponding to keyword otherwise with it is aforementioned in step s 2, classification determines that equipment 1 determines that the target is closed Relevance of searches class corresponding to keyword is same or similar otherwise, and for simplicity, therefore details are not described herein, and with reference Mode include and this.

Then, in step s 2, classification determines equipment 1 according to described search correlation classification, to the target keyword Carry out Screening Treatment.For example, it is assumed that classification determines that the target keyword that equipment 1 obtains includes such as in step S 1 Query1, query2, query3 and query4, and in step s 2, classification determines that equipment 1 determines the target keyword The corresponding described search correlation classification of query1, query2, query3 and query4 be respectively correlation is high, in correlation, Correlation is low, correlation is high, then in step s 2, classification determine equipment 1 according to target keyword query1, query2, The corresponding described search correlation classification of query3 and query4, carries out Screening Treatment to it, and it is low such as to belong to correlation The keyword query3 of classification is screened from keyword set to be measured, to carry out later period searching order Advance data quality to it.

Preferably, classification determines that equipment 1 further includes step S6 (not shown) and step S7 (not shown).Specifically, in step In rapid S6, classification determines that equipment 1 determines keyword set to be optimized corresponding to the keyword clustering；In the step s 7, class It Que Ding equipment 1 be public according to corresponding to the one or more keywords to be optimized that the keyword set to be optimized includes searches Rope sequence routing information determines the Optimizing Search sequence routing information of one or more of keywords to be optimized, for adjusting The searching order routing information of whole one or more of keywords to be optimized.

Specifically, in step s 6, classification determines the search for all keywords that equipment 1 can include according to keyword clustering Correlation description information, such as relevance of searches is high, relevance of searches is low, finds out the low classification of relevance of searches, using as institute State keyword set to be optimized corresponding to keyword clustering.Here, the keyword set to be optimized, which corresponds to, belongs to search The keyword of the low classification of correlation.For example, it is assumed that belonging to the key of the low classification of relevance of searches in keyword clustering cluster1 Word is sample keyword III, and the keyword that the low classification of relevance of searches is belonged in keyword clustering cluster2 is that sample is crucial The keyword of the low classification of relevance of searches, then in step s 6, class are not belonged in word IV and VI, keyword clustering cluster3 Not Que Ding equipment 1 can include into cluster3 by keyword clustering cluster1 all belong to the low classification of relevance of searches Keyword such as includes sample keyword III, IV and VI as the keyword set to be optimized.

Preferably, in step s 6, classification determines that equipment 1 can also be by all keywords included by the keyword clustering Actual search results relevant information be compared with system index information, from all keywords determination it is described to be optimized Keyword set.Here, the actual search results relevant information includes but is not limited to as returned to search result quantity, obtaining point The amount of hitting, the authority for returning to website, the quality degree of content of pages for returning to website etc..Here, the system index information includes The authority for such as returning to search result quantity, returning to website.For example, in step s 6, classification determines that equipment 1 can be by it in step All keywords, that is, sample keyword I included by the keyword clustering cluster1 to cluster3 determined in rapid S1 is extremely The actual search results relevant information of VI is compared with system index information, is determined from all keywords described to excellent Change keyword set, the actual search results relevant information of sample keyword I to VI is such as unsatisfactory for the pass of system index information Keyword, as the keyword set to be optimized.

Then, in the step s 7, classification determines the one or more that equipment 1 includes according to the keyword set to be optimized The sequence routing information of common search corresponding to keyword to be optimized, determines the optimization of one or more of keywords to be optimized Searching order routing information, with the searching order routing information for adjusting one or more of keywords to be optimized.For example, Example is connected, in step s 6, the keyword set to be optimized that classification determines that equipment 1 determines includes sample keyword III, IV And VI, then in the step s 7, classification determines that equipment 1 can the sequence of the common search according to corresponding to sample keyword III, IV and VI Routing information such as S_C → S_D → S-G, as Optimizing Search sequence routing information, with one or more of for adjusting The searching order routing information of keyword to be optimized, as delete sample keyword III, IV and VI searching order routing information in Including the Optimizing Search sort routing information, alternatively, by the searching order routing information of sample keyword III, IV and VI Including the Optimizing Search sequence routing information by the keyword of the high classification of other searching order routing informations such as relevance of searches Corresponding common search sequence routing information replaces.

Wherein, classification determines that equipment 1 includes step S1 ', step S2 ', step S3 ', step S4 ' and step S5 '.Specifically Ground, in step S1 ', classification determines equipment 1 according to the searching order routing information of target keyword, from one or more keys Target critical term clustering belonging to the target keyword is determined in term clustering；In step S2 ', classification determines 1 basis of equipment The target critical term clustering determines relevance of searches classification corresponding to the target keyword, to be used for subsequent processing；? In step S3 ', classification determines that equipment 1 determines preferred searching order routing information corresponding to the target keyword；In step In S4 ', classification determines that equipment 1 according to the preferred searching order routing information, adjusts the searching order of the target keyword Routing information；If search sequence matches with the target keyword, in step S5 ', after classification determines equipment 1 according to adjustment The target keyword searching order routing information, search result corresponding to the search sequence is supplied to described look into Ask user corresponding to sequence.Here, to correspond to step with shown in Fig. 3 respectively same or similar by step S1 ' and step S2 ', so Place repeats no more, and is incorporated herein by reference.

Specifically, in step S3 ', classification determines that equipment 1 determines preferred search row corresponding to the target keyword Sequence routing information.Here, in step S3 ', classification determine the mode of preferred searching order routing information described in equipment 1 include but It is not limited to following at least any one:

1) by the sequence of each common search with reference to corresponding to keyword included by target critical term clustering path Information, as the preferred searching order routing information.For example, it is assumed that classification determines that equipment 1 determines target in step S1 ' Target critical term clustering belonging to keyword goal-query is keyword clustering cluster1, then in step S3 ', classification is true Locking equipment 1 public with reference to corresponding to keyword, that is, sample keyword I, III and VI can be searched each included by cluster1 Rope sequence routing information such as S_A → S_C → S_D → S_E → S_G, as the preferred searching order routing information.

2) to each searching order routing information with reference to corresponding to keyword included by the target critical term clustering Processing for statistical analysis, with the determination preferred searching order routing information, as counted the search row for showing that keyword is passed by The frequency is met the searching order path of predetermined threshold by sequence path frequency information, or using high frequency searching order path as institute State preferred searching order routing information.For example, connecting example, in step S3 ', classification determines that equipment 1 can be by cluster1 packet The each searching order routing information place for statistical analysis with reference to corresponding to keyword i.e. sample keyword I, III and VI included Reason, by the frequency meet predetermined threshold for example frequency of occurrence meet 2 searching order path such as S_A → S_C → S_D → S_E → S_G → S_F, as the preferred searching order routing information.

In step S4 ', classification determines that equipment 1 according to the preferred searching order routing information, adjusts the target and closes The searching order routing information of keyword.For example, it is assumed that classification determines the preferred search row that equipment 1 determines in step S3 ' Sequence routing information is S_A → S_C → S_D → S_E → S_G, then in step S4 ', classification determines equipment 1 according to the preferred search Sort routing information, by such as adjustment algorithm or machine learning model such as SVM model, adjusts target keyword goal- The searching order routing information of target keyword goal-query is such as adjusted to preferential by the searching order routing information of query Execute the preferred searching order routing information S_A → S_C → S_D → S_E → S_G.

If search sequence matches with the target keyword, in step S5 ', after classification determines equipment 1 according to adjustment The target keyword searching order routing information, search result corresponding to the search sequence is supplied to described look into Ask user corresponding to sequence.Specifically, in step S5 ', classification determines that equipment 1 obtains search sequence first；Then, judge Whether the search sequence matches with the target keyword, if matching, in step S5 ', classification determines equipment 1 according to tune Search result corresponding to the search sequence is supplied to institute by the searching order routing information of the target keyword after whole State user corresponding to search sequence.Here, the matched meaning includes search sequence and the target keyword complete one It causes, search sequence is contained in the target keyword.

Specifically, in step S5 ', classification determines equipment 1 first by the dynamic web page techniques such as ASP, JSP, Huo Zhetong The application programming interfaces (API) of search engine offer are provided, the search sequence that user is inputted by user equipment is obtained.For example, if Search user B inputs keyword " fresh flower " by its PC equipment in search engine search column, by "enter" key", in step S5 ' In, classification determines that equipment 1 by dynamic web page techniques such as ASP, JSP or PHP, can get search user B input Keyword " fresh flower ".It will be understood by those skilled in the art that the mode of above-mentioned acquisition search sequence is only for example, other it is existing or The mode for the acquisition search sequence being likely to occur from now on is such as applicable to the present invention, should also be included in the scope of the present invention with It is interior, and be incorporated herein by reference.

Then, in step S5 ', classification determines the target keyword that equipment 1 obtains in step S 1 ' according to it, In such a way that text compares, judge whether the search sequence matches with the target keyword.

If matching, in step S5 ', classification determines equipment 1 according to the searching order of the target keyword adjusted Search result corresponding to the search sequence is supplied to user corresponding to the search sequence by routing information.For example, connecing Upper example, it is assumed that in step S 1 ', classification determines the target keyword such as " fresh flower " of the acquisition of equipment 1, " fresh flower express delivery ", Then in step S5 ', classification determines that equipment 1 judges search sequence " fresh flower " and target keyword such as " fresh flower ", " fresh flower express delivery " Match, then, in step S5 ', classification determines search of the equipment 1 by search engine according to the target keyword adjusted Sort search result corresponding to routing information such as " fresh flower Baidu discussion bar ", " fresh flower picture materials day is off line ", as inquiry Search result corresponding to sequence " fresh flower ", and pass through the dynamic web page techniques such as ASP, JSP or PHP or other agreements Communication mode, such as http or https communication protocol is supplied to user corresponding to the search sequence i.e. user B, such as should The user equipment of user is browsed for user.

It should be noted that the present invention can be carried out in the assembly of software and/or software and hardware, for example, can adopt With specific integrated circuit (ASIC), general purpose computer or any other realized similar to hardware device.In one embodiment In, software program of the invention can be executed to implement the above steps or functions by processor.Similarly, of the invention Software program (including relevant data structure) can be stored in computer readable recording medium, for example, RAM memory, Magnetic or optical driver or floppy disc and similar devices.In addition, some of the steps or functions of the present invention may be implemented in hardware, example Such as, as the circuit cooperated with processor thereby executing each step or function.

In addition, a part of the invention can be applied to computer program product, such as computer program instructions, when its quilt When computer executes, by the operation of the computer, it can call or provide according to the method for the present invention and/or technical solution. And the program instruction of method of the invention is called, it is possibly stored in fixed or moveable recording medium, and/or pass through Broadcast or the data flow in other signal-bearing mediums and transmitted, and/or be stored according to described program instruction operation In the working storage of computer equipment.Here, according to one embodiment of present invention including a device, which includes using Memory in storage computer program instructions and processor for executing program instructions, wherein when the computer program refers to When enabling by processor execution, method and/or skill of the device operation based on aforementioned multiple embodiments according to the present invention are triggered Art scheme.

It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie In the case where without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims Variation is included in the present invention.Any reference signs in the claims should not be construed as limiting the involved claims.This Outside, it is clear that one word of " comprising " does not exclude other units or steps, and odd number is not excluded for plural number.That states in device claim is multiple Unit or device can also be implemented through software or hardware by a unit or device.The first, the second equal words are used to table Show title, and does not indicate any particular order.

Claims

1. a kind of for determining relevance of searches class method for distinguishing corresponding to target keyword, wherein this method includes following Step:

A determines that the target is closed from one or more keyword clusterings according to the searching order routing information of target keyword Target critical term clustering belonging to keyword, wherein described search sequence routing information is for showing that search engine is determining candidate The sort algorithm code path information passed through in search results ranking information process；

B determines relevance of searches classification corresponding to the target keyword, after being used for according to the target critical term clustering Continuous processing, wherein the meaning of relevance of searches refers to the matching degree of keyword and search result, described search correlation classification Including in high correlation classification, low correlation classification, uncorrelated classification, cheating keyword categories any one of at least.

2. according to the method described in claim 1, wherein, the keyword clustering includes for characterizing the keyword clustering Class searching order routing information；

Wherein, the step a includes:

Described in corresponding to the searching order routing information of the target keyword and one or more of keyword clusterings Class searching order routing information is compared, and is searched for the searching order routing information of the determination target keyword and the class The smallest edit distance of sequence routing information；

According to the smallest edit distance, determine that the searching order routing information of the target keyword and class search are arranged The sequence similarity of paths of sequence routing information；

According to the sequence similarity of paths, the target critical term clustering is determined.

3. method according to claim 1 or 2, wherein this method further include:

X determines preferred searching order routing information corresponding to the target keyword；

According to the preferred searching order routing information, the searching order routing information of the target keyword is adjusted；

Wherein, this method further include:

If search sequence matches with the target keyword, according to the searching order road of the target keyword adjusted Search result corresponding to the search sequence is supplied to user corresponding to the search sequence by diameter information.

4. according to the method described in claim 3, wherein, the step x includes:

Each sequence of the common search with reference to corresponding to keyword path letter according to included by the target critical term clustering Breath, determines the preferred searching order routing information.

5. according to the method described in claim 3, wherein, the step x includes:

It unites to each searching order routing information with reference to corresponding to keyword included by the target critical term clustering Analysis processing is counted, with the determination preferred searching order routing information.

6. method according to claim 1 or 2, wherein this method further include:

Y determines keyword set to be optimized corresponding to the keyword clustering；

According to common search sequence corresponding to the one or more keyword to be optimized that the keyword set to be optimized includes Routing information determines the Optimizing Search sequence routing information of one or more of keywords to be optimized, with described for adjusting The searching order routing information of one or more keywords to be optimized.

7. according to the method described in claim 6, wherein, the step y includes:

By the actual search results relevant information of all keywords included by the keyword clustering and system index information into Row compares, and the keyword set to be optimized is determined from all keywords.

8. method according to claim 1 or 2, wherein the step a includes:

One or more keywords to be measured to be processed are obtained, using as the target keyword；

According to the searching order routing information of the target keyword, the mesh is determined from one or more keyword clusterings Mark target critical term clustering belonging to keyword；

Wherein, the step b includes:

According to the target critical term clustering, relevance of searches classification corresponding to the target keyword is determined；

According to described search correlation classification, Screening Treatment is carried out to the target keyword.

9. a kind of classification for determining relevance of searches classification corresponding to target keyword determines equipment, wherein the category Determine that equipment includes:

Determining device is clustered, for the searching order routing information according to target keyword, from one or more keyword clusterings Target critical term clustering belonging to the middle determination target keyword, wherein described search sequence routing information is for showing to search Index holds up the sort algorithm code path information passed through in determining candidate search sort result information process；

Classification determining device, for determining search phase corresponding to the target keyword according to the target critical term clustering Closing property classification, to be used for subsequent processing, wherein the meaning of relevance of searches refers to the matching degree of keyword and search result, Described search correlation classification include high correlation classification, low correlation classification, uncorrelated classification, cheating keyword categories in extremely It is any one of few.

10. classification according to claim 9 determines equipment, wherein the keyword clustering includes for characterizing the pass The class searching order routing information of keyword cluster；

Wherein, the cluster determining device includes:

Comparing unit, for by the searching order routing information of the target keyword and one or more of keyword clusterings The corresponding class searching order routing information is compared, with the searching order routing information of the determination target keyword With the smallest edit distance of the class searching order routing information；

Similarity determining unit, for determining the searching order path of the target keyword according to the smallest edit distance The sequence similarity of paths of information and the class searching order routing information；

Determination unit is clustered, for determining the target critical term clustering according to the sequence similarity of paths.

11. classification according to claim 9 or 10 determines equipment, wherein the category determines equipment further include:

Preferred path determining device, for determining preferred searching order routing information corresponding to the target keyword；

Device is adjusted, for adjusting the searching order road of the target keyword according to the preferred searching order routing information Diameter information；

Wherein, the category determines equipment further include:

Device is provided, if matching for search sequence and the target keyword, according to the target keyword adjusted Searching order routing information, search result corresponding to the search sequence is supplied to use corresponding to the search sequence Family.

12. classification according to claim 11 determines equipment, wherein the preferred path determining device is used for:

13. classification according to claim 11 determines equipment, wherein the preferred path determining device is used for:

14. classification according to claim 9 or 10 determines equipment, wherein the category determines equipment further include:

Gather determining device, for determining keyword set to be optimized corresponding to the keyword clustering；

Path optimizing determining device, the one or more keyword to be optimized for including according to the keyword set to be optimized Corresponding common search sequence routing information determines the Optimizing Search sequence path of one or more of keywords to be optimized Information, with the searching order routing information for adjusting one or more of keywords to be optimized.

15. classification according to claim 14 determines equipment, wherein the set determining device is used for:

16. classification according to claim 9 or 10 determines equipment, wherein the cluster determining device is used for:

Wherein, the classification determining device is used for:

17. a kind of for determining the search engine of relevance of searches classification corresponding to target keyword, wherein the search engine Equipment is determined including the classification as described in any one of claim 9 to 16.

18. a kind of for determining the search engine plug-in unit of relevance of searches classification corresponding to target keyword, wherein the search Engine plug-in unit includes that the classification as described in any one of claim 9 to 16 determines equipment.