CN103399859A - Web page authority value calculating method, device and system - Google Patents
Web page authority value calculating method, device and system Download PDFInfo
- Publication number
- CN103399859A CN103399859A CN2013102747560A CN201310274756A CN103399859A CN 103399859 A CN103399859 A CN 103399859A CN 2013102747560 A CN2013102747560 A CN 2013102747560A CN 201310274756 A CN201310274756 A CN 201310274756A CN 103399859 A CN103399859 A CN 103399859A
- Authority
- CN
- China
- Prior art keywords
- authority
- value
- webpage
- fragment
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention discloses a web page authority value calculating method, device and system. The web page authority value calculating method comprises the following steps that an adjacent matrix of a web page is cut into N matrix sections, wherein N is an integer larger than 1 and not larger than the number of a plurality of servers; the N matrix sections are respectively distributed to the corresponding servers, authority value calculation is conducted, and N authority value sections are obtained; the N authority value sections are combined, and the final authority value of the web page is obtained. Therefore, according to the web page authority value calculating method, device and system, the calculation work of the web page authority value is distributed to the servers to be completed, then, calculation results of all the servers are combined, the final authority value of the web page is obtained, so that the defect that the authority values of a great number of web pages are difficult to calculate in the prior art can be overcome.
Description
Technical field
The present invention relates to internet arena, be specifically related to a kind of computing method, Apparatus and system of webpage authority value.
Background technology
In the Webpage search technology, webpage authority value is an important index, has influence on the sort order of Search Results.Generally, what webpages are authority's value of webpage can be defined as and point to it., when having more webpage while pointing to it, represent that the popularity of this webpage is higher, the prestige of going beyond one's commission.
In the prior art, the mode of calculating webpage authority value is as follows:
A, B, C, four webpages of D are for example arranged, use adjacency matrix M to be expressed as follows the points relationship of these four webpages:
Wherein, the ordinate of above-mentioned adjacency matrix represents the promoter of points relationship, and row-coordinate represents the sensing person of webpage points relationship.When wherein element was 0, expression link was pointed to, and when element was 1, expression had link and points to.For example the second row of matrix, represent and have which webpage to point to the B webpage,, according to the points relationship in adjacency matrix, to only have the A webpage to point to the B webpage.
Simultaneously, before the authority who carries out webpage is worth calculating, also can set initial authority's value R according to the popularity of each webpage
i, corresponding to above-mentioned matrix M, initial authority's value of webpage A, B, C, D is respectively R
A0, R
B0, R
C0, R
D0Therefore, can obtain a vectorial V of initial authority's value
0, i.e. [R
A0R
B0R
C0R
D0].
If it is V that the authority that all webpages are finally calculated is worth the vector of composition
1, with matrix and vector, can be expressed as formula 1:
M*V
0+ V
0=V
1 Formula 1
And can be expressed as intuitively formula 2:
Wherein, the R that calculates
A1, R
B1, R
C1, R
D1It is exactly the final authority's value of each webpage.
But, the inventor finds, there are the following problems in the calculating of authority's value of existing webpage: the authority who carries out webpage, be worth while calculating, the calculating parameters such as adjacency matrix all are kept in the internal memory of same server, but present webpage quantity is very huge, the memory size of individual server, calculate very difficult so carry out the full dose of webpage authority value head and shoulders above.
Summary of the invention
In view of the above problems, the invention provides a kind of computing method, Apparatus and system of webpage authority value, in order to overcome the problems referred to above or address the above problem at least in part.
According to one aspect of the present invention, a kind of webpage authority is provided the computing method of value, be applied to utilize the environment of authority's value of multiple servers Distributed Calculation webpage, the method comprises:
The adjacency matrix of webpage is cut into N part matrix fragment, and wherein, N is the integer greater than 1, and is not more than the quantity of multiple servers;
N part matrix fragment is respectively allocated to corresponding server carries out the calculating of authority's value, obtain N part authority value fragment;
Merge N part authority value fragment, obtain final authority's value of webpage.
Alternatively, before the adjacency matrix with webpage cuts into N part matrix fragment, also comprise:
According to the link points relationship between webpage, for webpage is set up adjacency matrix.
Alternatively,, according to the link points relationship between webpage, for webpage, set up adjacency matrix, comprising:
Webpage is sorted, and according to webpage, put in order webpage respectively as the summit of matrix, and set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element;
Link points relationship between the webpage that identifies according to summit carries out assignment to each element in adjacency matrix:
When webpage corresponding to element column points to the webpage that this element is expert at corresponding, with this row element assignment, be 1;
When webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.
Alternatively, the adjacency matrix of webpage is cut into N part matrix fragment, comprising:
Obtain the line number of adjacency matrix;
When line number is not more than the quantity of multiple servers, according to the row order, every row of adjacency matrix is split out successively and is used as a matrix fragment, obtain N part matrix fragment;
, according to every part of matrix fragment, comprise at least two row and according to row ordered pair adjacency matrix, cut during greater than the quantity of multiple servers when line number, and making the umber N of the matrix fragment that cutting obtains be not more than the quantity of multiple servers.
Alternatively, N part matrix fragment is respectively allocated to corresponding server, comprises:
Fractionation time ordered pair N part matrix fragment according to adjacency matrix is carried out respectively mark, and N part matrix fragment is distributed to different servers.
Alternatively, merge N part authority value fragment, obtain final authority's value of webpage, comprising:
Obtain N part authority value fragment, and read the sign of every part of fractionation order in authority's value fragment, the authority that will obtain according to the fractionation order is worth fragment and merges, and obtains the final authoritative value of webpage.
Alternatively, every station server step of carrying out the calculating of authority's value to being dispensed to local matrix fragment comprises:
Every row element in this matrix fragment and preset initial vector are done dot-product operation, obtain primary vector, wherein, initial vector comprises initial authority's value of webpage;
Take out initial authority's value corresponding to every row element in initial vector, as secondary vector;
, with the element in primary vector and the corresponding addition of the element in secondary vector, obtain the corresponding authority of this matrix fragment value fragment.
Alternatively, before the server that N part matrix fragment is respectively allocated to correspondence carries out the calculating of authority's value, also comprise:
For webpage arranges respectively initial authority's value, and obtain initial vector;
Initial vector is sent to respectively multiple servers.
Alternatively, before the server that N part matrix fragment is respectively allocated to correspondence carries out the calculating of authority's value, also comprise:
For any station server,
Judge whether this server possesses enough internal memories and carry out the calculating of authority's value to being dispensed to local matrix fragment;
When the machine low memory, increase the quantity of server; And
According to the quantity of the server after increasing, adjacency matrix is carried out cutting again;
Repeat the judgement of server memory capacity and the increase operation of number of servers, until the machine internal memory of server enough carries out the calculating of authority's value to being dispensed to local matrix fragment.
Alternatively, after obtaining final authority's value of webpage, also comprise:
According to final authority's value of webpage, the webpage of showing in Webpage searching result is sorted; Perhaps,
Recommend webpage according to final authority's value of webpage to the user.
According to one aspect of the present invention, a kind of webpage authority is provided the calculation element of value, be applied to utilize the environment of authority's value of multiple servers Distributed Calculation webpage, this device comprises:
Cutting module, be configured to the adjacency matrix of webpage is cut into N part matrix fragment, and wherein, N is the integer greater than 1, and is not more than the quantity of multiple servers;
Distribution module, be configured to that N part matrix fragment is respectively allocated to corresponding server and carry out the calculating of authority's value, obtains N part authority value fragment;
Merge module, be configured to merge N part authority value fragment, obtain final authority's value of webpage.
Alternatively, this device also comprises:
Set up module, before being configured to the adjacency matrix of webpage be cut into N part matrix fragment at cutting module, according to the link points relationship between webpage, for webpage is set up adjacency matrix.
Alternatively, the above-mentioned block configuration of setting up becomes to set up as follows adjacency matrix:
Webpage is sorted, and according to webpage, put in order webpage respectively as the summit of matrix, and set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element;
Link points relationship between the webpage that identifies according to summit carries out assignment to each element in adjacency matrix:
When webpage corresponding to element column points to the webpage that this element is expert at corresponding, with this row element assignment, be 1;
When webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.
Alternatively, above-mentioned cutting module is configured to as follows the adjacency matrix of webpage be cut into N part matrix fragment:
Obtain the line number of adjacency matrix;
When line number is not more than the quantity of multiple servers, according to the row order, every row of adjacency matrix is split out successively and is used as a matrix fragment, obtain N part matrix fragment;
, according to every part of matrix fragment, comprise at least two row and according to row ordered pair adjacency matrix, cut during greater than the quantity of multiple servers when line number, and making the umber N of the matrix fragment that cutting obtains be not more than the quantity of multiple servers.
Alternatively, above-mentioned distribution module also is configured to carry out respectively mark according to fractionation time ordered pair N part matrix fragment of adjacency matrix, and N part matrix fragment is distributed to different servers.
Alternatively, above-mentioned merging module also is configured to obtain N part authority value fragment, and reads the sign of every part of fractionation order in authority's value fragment, and the authority that will obtain according to the fractionation order is worth fragment and merges, and obtains the final authoritative value of webpage.
Alternatively, every part of authority's value fragment that above-mentioned merging module merges calculates in the following manner by every station server:
Every row element in this matrix fragment and preset initial vector are done dot-product operation, obtain primary vector, wherein, initial vector comprises initial authority's value of webpage;
Take out initial authority's value corresponding to every row element in initial vector, as secondary vector;
, with the element in primary vector and the corresponding addition of the element in secondary vector, obtain the corresponding authority of this matrix fragment value fragment.
Alternatively, this device also comprises:
Module is set, is configured in distribution module, N part matrix fragment to be respectively allocated to before corresponding server carries out the calculating of authority value, for webpage arranges respectively initial authoritative value, and obtain initial vector;
Sending module, be configured to initial vector is sent to respectively multiple servers.
Alternatively, this device also comprises:
Judge module, be configured in distribution module, N part matrix fragment to be respectively allocated to before corresponding server carries out the calculating of authority value,, for any station server, judge whether this server possesses enough internal memories and carry out the calculating of authority's value to being dispensed to local matrix fragment;
Increase module, be configured to when judge module is judged the machine low memory, increase the quantity of server; And
Cutting module, also be configured to according to the quantity of the server after increasing, adjacency matrix be cut again;
Repeated trigger judge module and the performed operation of increase module, until the machine internal memory of server enough carries out the calculating of authority's value to being dispensed to local matrix fragment.
Alternatively, this device also comprises:
Application module, be configured to according to final authority's value of webpage, the webpage of showing in Webpage searching result be sorted after the merging module obtains final authority's value of webpage; Perhaps,
Recommend webpage according to final authority's value of webpage to the user.
According to one aspect of the present invention, a kind of computing system of webpage authority value is provided, comprise the calculation element of above-mentioned webpage authority value, also comprise the multiple servers that webpage authority value is carried out Distributed Calculation.
The invention provides a kind of computing method, Apparatus and system of webpage authority value, by the present invention, make the separate unit server in the situation that can't complete the calculating of webpage authority value, the evaluation work of webpage authority value can be distributed to multiple servers completes, again the result of calculation of every station server is merged afterwards, obtain the final authority of webpage value.As seen, method provided by the invention, Apparatus and system, overcome the defect of the authority's value that is difficult to calculate huge quantity webpage in the prior art.In addition, adopt the present invention, the calculated amount of webpage authority value can also be distributed to multiple servers equably, the load of server that one side is balanced, on the other hand,, because multiple servers participates in calculating simultaneously, also improved the computing velocity of webpage authority value.
Above-mentioned explanation is only the general introduction of technical solution of the present invention, for can clearer understanding technological means of the present invention, and can be implemented according to the content of instructions, and for above and other objects of the present invention, feature and advantage can be become apparent, below especially exemplified by the specific embodiment of the present invention.
Description of drawings
By reading hereinafter detailed description of the preferred embodiment, various other advantage and benefits will become cheer and bright for those of ordinary skills.Accompanying drawing only is used for the purpose of preferred implementation is shown, and does not think limitation of the present invention.And in whole accompanying drawing, with identical reference symbol, represent identical parts.In the accompanying drawings:
Fig. 1 shows a kind of according to an embodiment of the invention computing system structural drawing of webpage authority value;
Fig. 2 shows the link points relationship schematic diagram between webpage according to an embodiment of the invention;
Fig. 3 shows the corresponding relation schematic diagram of matrix fragment and multiple servers according to an embodiment of the invention;
Fig. 4 shows a kind of according to an embodiment of the invention computing method process flow diagram of webpage authority value;
Fig. 5 shows a kind of according to an embodiment of the invention method flow diagram of setting up adjacency matrix; And
Fig. 6 shows a kind of according to an embodiment of the invention server and the matrix fragment is calculated the method flow diagram of authority's value fragment.
Embodiment
Exemplary embodiment of the present disclosure is described below with reference to accompanying drawings in more detail.Although shown exemplary embodiment of the present disclosure in accompanying drawing, yet should be appreciated that and can realize the disclosure and the embodiment that should do not set forth limits here with various forms.On the contrary, it is in order thoroughly to understand the disclosure that these embodiment are provided, and can with the scope of the present disclosure complete convey to those skilled in the art.
Mention in prior art,, due to becoming increasingly abundant of Internet resources, cause webpage quantity very huge, when webpage authority value is calculated, huge calculated amount is the memory size of individual server head and shoulders above, carries out the full dose of webpage authority value and calculates very difficult.
For addressing the above problem, the embodiment of the present invention provides a kind of computing system of webpage authority value.Fig. 1 has shown the computing system structural drawing according to a kind of webpage authority value of the embodiment of the present invention.As shown in Figure 1, this calculation element 100 is coupled with the multiple servers 200 that carries out Distributed Calculation, when above-mentioned calculation element 100 need to carry out the calculating of authority's value to webpage, calculation task can be cut into some parts, and send to respectively each server in multiple servers 200, calculate respectively part authority value fragment by each server,, finally again by the result of calculation of calculation element 100 comprehensive each servers, obtain final authority's value of webpage.
This shows, the computing system that adopts the embodiment of the present invention to provide, can work as the calculated amount of webpage authority value when larger, utilize the multiple servers distributed earth to calculate webpage authority value, realize the calculating to authority's value of huge quantity webpage, also improved the computing velocity of webpage authority value simultaneously.
Above stated the inventive concept of the embodiment of the present invention, now respectively the each several part assembly of this system architecture and the function of corresponding each several part have been specifically introduced.
The above-mentioned calculation element of mentioning 100, can be applicable to utilize the environment of authority's value of multiple servers Distributed Calculation webpage.When the user searches for webpage according to element-specific (as the keyword of user's input) by search engine, sort to webpage according to the webpage authority value that calculation element 100 provides this moment, and be user's display web page.Lower mask body is introduced this calculation element 100.Referring to Fig. 1, at first, set up module 140 by calculation element 100, obtain related web page, and the link points relationship between definite webpage, should link again points relationship and be created as the adjacency matrix form, and the adjacency matrix that will be created as is given with the cutting module 110 of its coupling and carried out follow-up processing.
Adjacency matrix and webpage can shine upon, and it is different according to the parameter of webpage that it sets up mode, and the multiple different mode of setting up can be arranged.Alternatively, in the present embodiment, set up module 140 when setting up adjacency matrix, at first the webpage of authority's value to be calculated sorted, put in order webpage respectively as the summit of matrix according to webpage, and set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element.Being worth is explanatorily that the webpage in the present embodiment puts in order and can be random order, at this, the purpose that webpage is sorted is, webpage during to subsequent calculations sequentially limits, and when calculating webpage authority value, its order is also the order that this moment, webpage was arranged.Subsequently, the link points relationship of setting up between the webpage that module 140 can identify according to summit carries out assignment to each element in adjacency matrix.In order to learn points relationship between webpage by element value more intuitively, can work as webpage corresponding to element column while pointing to the webpage that this element is expert at corresponding, be 1 with this row element assignment, and when webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.
The process of setting up adjacency matrix below by an instantiation to setting up module 140 is described.In specific embodiment of the present invention, suppose that the webpage of authority's value to be calculated comprises webpage 1, webpage 2, webpage 3, webpage 4, its link points relationship is as shown in Figure 2.Alternatively, setting up module 140 at first arranges above-mentioned webpage according to the order of webpage 1, webpage 2, webpage 3, webpage 4, certainly can also arrange according to other order, putting in order of the present embodiment is only exemplary, do not limit implementation of the present invention.Afterwards, set up module 140 and the above-mentioned webpage that arranges is set to respectively the summit of matrix, and the middle element that is matrix according to link points relationship shown in Figure 2 carries out assignment, the adjacency matrix M that finally obtains is as follows:
Concrete introduction has been carried out in above-mentioned operation of setting up adjacency matrix to setting up module 140, below then introduces and sets up after module 140 sends to cutting module 110 with adjacency matrix, the processing operation of 110 pairs of adjacency matrix of cutting module.In this example, cutting module 110 can cut into N part matrix fragment with the adjacency matrix of setting up module 140 foundation, and by distribution module 120, the server that N part matrix fragment is respectively allocated to correspondence is carried out the calculating of authority's value, thereby obtains N part authority value fragment.
Further, the umber N that cutting module 110 cuts into adjacency matrix should be greater than 1, and is not more than the quantity of multiple servers 200.Be not more than the quantity of server due to N, therefore, can meet the calculating of all matrix fragments that cutting module 110 is cut into.In addition, when the matrix fragment is cut, should be taken into account the line number of matrix fragment and the relation between current multiple servers 200 quantity that participate in the calculating of authority's value, in order to cut out comparatively suitable matrix fragment for server, when both quantitative relation not simultaneously, the mode of taking when the adjacency matrix of webpage is cut is also different.Below respectively to both quantitative relation not simultaneously, the different cutting modes of taking are introduced.
At first, introduce the first situation, namely the line number of adjacency matrix is not more than the situation of the quantity of multiple servers 200.At this moment, because the quantity of server is comparatively sufficient, cutting module 110 can split out every row of adjacency matrix and be used as a matrix fragment successively according to the capable order of adjacency matrix, obtains N part matrix fragment.This cutting mode makes every station server only need the delegation of matrix fragment is calculated, and has alleviated the calculated amount of every station server, participates in the time of again due to multiple servers, has also accelerated simultaneously the computing velocity of webpage authority value.
Secondly, introduce the second situation, corresponding with above-mentioned the first situation, the second situation is the quantity of the line number of adjacency matrix greater than multiple servers 200.At this moment,, if still adopt cutting mode in above-mentioned the first situation, can cause due to the lazy weight of server some matrix fragment without corresponding server, it to be calculated, cause the calculating of matrix fragment imperfect.So for the second situation, in order to guarantee the complete computation to the matrix fragment, cutting module 110 can comprise at least two row and according to row ordered pair adjacency matrix, cut according to every part of matrix fragment, and make the umber N that cuts the matrix fragment that obtains be not more than the quantity of multiple servers 200.
Due in practice, generally speaking, the quantity of webpage usually can be much larger than the situation of server, namely usually in practice the second situation frequently occur.In order more clearly to describe the cutting process of 110 pairs of adjacency matrix of cutting module, the present invention describes the second situation by following instantiation.
In specific embodiment of the present invention, the line number of supposing adjacency matrix is 40000, and the quantity that participates in the multiple servers 200 of authority's value calculating is 3, at this moment, when cutting module 110 cuts at the adjacency matrix to webpage, adjacency matrix can be cut into 3 parts.Alternatively, cutting module 110 can be with the front 13334 row integral body of matrix as a matrix fragment, with the centre of matrix 13333 row and rear 13333 row respectively as a matrix fragment.Certainly, in addition, can also take other cutting mode that adjacency matrix is cut into 3 parts, at this, no longer be illustrated.In the present embodiment, matrix is divided into 3 parts can, so that all server can both participate in calculating webpage authority value, accelerate the computing velocity of webpage authority value.And, when the calculated amount of adjacency matrix hour, can also be the umber that is less than number of servers with the adjacency matrix cutting, to reduce the utilization to server resource.In the present embodiment, matrix is averaged to cut apart can make the calculated amount of every station server obtain equilibrium.But the amount of ram of working as certain server is larger, the amount of ram of certain server hour, also can cut adjacency matrix unequally, be about to comprise the more matrix fragment of line number corresponding to the larger server of amount of ram, make the resource of server be utilized fully.
In foregoing description, also mentioned a kind of specific implementation as the present embodiment, cutting module 110 can be respectively allocated to corresponding server with N part matrix fragment by distribution module 120 and carry out the calculating of authority's value, thereby obtains N part authority value fragment.But, in order to guarantee that server can have enough memory headrooms to calculate the matrix fragment, the present embodiment also provides one preferred embodiment, referring to Fig. 1, this calculation element 100 has also comprised judge module 170, itself and cutting module 110 are coupled, and are used to every part of matrix fragment selected server, and judge whether every station server possesses enough internal memories its corresponding matrix fragment is carried out the calculating of authority's value., when the whole servers of judgement all have enough internal memories when its corresponding matrix fragment is calculated, by distribution module 120, N part matrix fragment is respectively allocated to corresponding server and carries out the calculating of authority value.
Decision operation for judge module 170, also there is another kind of situation in judged result, be that the low memory of at least one station server is to calculate the matrix fragment, at this moment, can increase by increase module 180 shown in Figure 1 the quantity of the server that participates in calculating authority value, and again pass through cutting module 110 according to the number of servers after increasing, adjacency matrix is cut again.In order to guarantee that every station server all has enough internal memories and calculates being dispensed to local matrix fragment, after again adjacency matrix being cut, the increase operation that again triggers the decision operation of judge module 170 and increase module 180, circulate with this,, until all have enough internal memories when its corresponding matrix fragment is calculated when the whole servers of judge module 170 judgement, by distribution module 120, all matrix fragments are respectively allocated to corresponding server and carry out the calculating of authority value.
In one embodiment of the invention,, for convenient operation to judge module 170 is described, also provide Fig. 3, be used for showing the corresponding relation of matrix fragment and multiple servers 200.As shown in Figure 3, have 3 parts of matrix fragments (the matrix line number that every part of matrix fragment comprises is 4 row) and need to carry out authority's value and calculate, wherein, judge module 170 can be slice of matrix section 1 selected server 1,, for matrix fragment 2 selected servers 2, be matrix fragment 3 selected servers 3.After having selected server, judge respectively whether server 1, server 2 and server 3 have enough internal memories its corresponding matrix fragment is calculated.
In addition, cutting module 110 further can be according to the number of servers after increasing, the operation that adjacency matrix is cut again, in one embodiment of the invention, as example, the cutting operation again of cutting module 110 is specifically introduced during take server 3 low memory shown in Figure 3.Alternatively,, in order to reduce the resource occupation to the server that participates in the calculating of authority's value, can first select to increase a station server, i.e. server in Fig. 34 by increasing module 180.In addition, because server 1 is all enough with the internal memory of server 2, in order to reduce the calculated amount of being brought by the matrix cutting, cutting module 110 can directly (for example cut into 2 parts with the corresponding matrix fragment of server 3, every part of matrix line number that comprises is 2 row), a copy of it corresponding with service device 3, the server 4 that another part correspondence is newly-increased, while also being not enough to the matrix fragment is calculated after increasing server 4, then continuation increases the quantity of server.In addition, when also having the low memory of other servers except server 3 in the server that participates in the calculating of authority's value, take method same as described above again to cut adjacency matrix, do not repeat them here.
Mention in foregoing description, the server that obtains participating in the calculating of authority's value when judge module 170 judgement all has enough internal memories its corresponding matrix fragment is calculated, by distribution module 120, all matrix fragments are respectively allocated to corresponding server and carry out the calculating of authority's value, below, specifically introduce distribution module 120.In the present embodiment, which part matrix fragment what calculate for clear and definite each server is, distribution module 120 can be carried out respectively mark according to fractionation time all matrix fragments of ordered pair of adjacency matrix, and the matrix fragment after mark is distributed to by judge module 170 and is the selected server of every part of matrix fragment.
, in order specifically to introduce the batch operation of distribution module 120, in one embodiment of the invention, also as an example of Fig. 3 example, describe.Wherein, distribution module 120 can be according to the fractionation order of adjacency matrix, for the matrix fragment marks, mark sequence number " 1 " as the matrix fragment 1 in as shown in Fig. 3, matrix fragment 2 is marked sequence number " 2 ", matrix fragment 3 is marked sequence number " 3 ", the matrix fragment 1,2,3 after marking is sent to respectively server 1,2,3.
Below, then introduce after distribution module 120 is dispensed to multiple servers 200 with the matrix fragment concrete operations of every station server.The present embodiment is introduced as an example of the station server in multiple servers 200 example.Wherein, server after receiving the matrix fragment, at first to its calculating of carrying out authority's value, calculate complete after, then the merging module 130 in Fig. 1 is returned to result of calculation.The present embodiment is specifically introduced the operation of server from calculating authority's value fragment and returning to authoritative value fragment two aspects respectively.
At first, introduce server and calculate the process of authority's value fragment.When the matrix fragment was calculated, server was first done dot-product operation with the every row element in this matrix fragment and preset initial vector, with the vector that obtains as primary vector.Server also takes out initial authority's value corresponding to every row element in initial vector, with it as secondary vector.After obtaining primary vector and secondary vector, the corresponding addition of the element in the primary vector that server will obtain and the element in secondary vector, obtain the corresponding authority of this matrix fragment value fragment.The above-mentioned initial vector that relates to is to be set in advance by the module 150 that arranges shown in Figure 1, and sends to by sending module 160 in every station server of the multiple servers 200 that participates in the calculating of authority's value.Wherein, initial vector is the one-dimensional vector of initial authoritative value that comprises the webpage of authority's to be calculated value.Alternatively, module 150 is set can be according to the popularity of webpage (for example, web page access amount etc.) initial authority's value of webpage is set, its popularity is higher, for the initial authority value of its setting larger, certainly can also be that webpage is set initial authority's value according to other parameters, the present embodiment no longer illustrates.
Calculate the process of authority's value fragment in order more clearly to describe server, in one embodiment of the invention, also as an example of above-mentioned matrix M example, be specifically described.For example, the matrix fragment that server receives comprises the first row and the second row of adjacency matrix M, is M
n, the initial vector of webpage authority value is V
0At this moment, the server process that calculates authority's value fragment realizes as follows.
1, calculate primary vector V by computing formula 3
aThe above-mentioned formula of mentioning 3 is: M
n* V
0=V
a, parameter value wherein is as follows:
V
0=[R
10R
20R
30R
40]; After bringing the above-mentioned parameter value into, above-mentioned formula 3 can also be expressed as:
2, take out webpage 1 in initial vector and be worth with the initial authority of webpage 2, the secondary vector V that obtains
b, wherein,
Taking out webpage 1 and be worth with the initial authority of webpage 2 in initial vector, is the first row and the second row corresponding webpage 1 of difference and the webpage 2 due to adjacency matrix.
3, with primary vector V obtained above
aIn element and secondary vector V
bIn the corresponding addition of element, obtain authority's value fragment V
01Wherein, can calculate authority's value fragment V by computing formula 4
01, formula 4 is: V
01=V
a+ V
bAs calculated, the authority who obtains is worth fragment and is
Wherein, R
11Value be R
10+ R
20, R
21Value be R
20
After the understanding server calculated the operation of authority's value fragment, next step was introduced server and returns to the step of authority's value fragment.When foregoing description is mentioned to distribution module 120 the matrix fragment is dispensed to server, can carry out respectively mark according to fractionation time all matrix fragments of ordered pair of adjacency matrix.At this, when server returns to authority's value fragment, too for authority's value fragment stamp with the matrix fragment in identical mark, this just makes calculation element 100 authority to be worth fragment according to mark to carry out correct merging, obtain finally authoritative value fragment.
Finally, the authority that every station server calculates self is worth fragment and returns to merging module 130, by merging module 130, carries out the merging of authority's value fragment.The function that below is combined module 130 introduction of giving an example.In one embodiment of the invention, for example, whole authority's value fragments of obtaining are fragment V
01With fragment V
02Wherein, fragment V
01Order identification be " 1 ", fragment V
02Order identification be " 2 ", merge module 130 by the sign order with fragment V
01With fragment V
02Merge, obtain authority's value fragment V
1
In the present embodiment, suppose
At this moment, with fragment V
01With fragment V
02Merge the vectorial V of final authority's value that obtains
1=[R
11R
21R
31R
41].
Foregoing has been introduced the structure of this calculation element 100 and the effect of each several part particularly.But, also it should be noted that in the present embodiment, after obtaining final authority's value of webpage, can also realize the concrete application to authority's value, at this moment, referring to Fig. 1, also comprise application module 190 in this calculation element 100, itself and above-mentioned merging module 130 are coupled, and can merge after module 130 obtains complete webpage authority value, and the Search Results after webpage being sorted and will sort according to the height of authority's value shows.If viewing area is limited or the user has certain situations such as demand, a kind of preferred processing mode is: according to the height of webpage authority value, to the user, recommend authority's value sequence default number webpage the preceding., to user's display web page, make the user can view the higher webpage of authority's value according to the height of authority value, facilitate the user to find quickly valuable webpage, promoted user's experience.
In sum, the system that the embodiment of the present invention provides, realized when the separate unit server in the situation that can't complete the calculating of webpage authority value, the evaluation work of webpage authority value can be distributed to multiple servers completes, again the result of calculation of every station server is merged, obtain the final authority of webpage value.Overcome the defect of the authority's value that is difficult to calculate huge quantity webpage in the prior art.In addition, calculation element 100 can also be distributed to multiple servers equably with the calculated amount of webpage authority value, and the load of server that one side is balanced on the other hand,, because multiple servers participates in calculating simultaneously, has also improved the computing velocity of webpage authority value.
Being worth is explanatorily that the above-mentioned calculation element that provides 100, be only preferred embodiment on framework, the practical structures of calculation element 100 do not caused restriction.Can also there be a plurality of other device (module) or interfaces in calculation element 100, determine as the case may be.
Be directed to the above-mentioned calculation element of mentioning 100, the embodiment of the present invention also provides a kind of webpage authority the computing method of value, and the method is applied to utilize the environment of authority's value of multiple servers Distributed Calculation webpage.As preferred embodiment a kind of, method provided by the present invention may be implemented in the search engine server of webpage, but, this search engine server only is a better embodiment of the inventive method to the application that webpage carries out the calculating of authority's value, is not limited to the range of application of the inventive method.The method that below the present embodiment is provided is specifically introduced.
Fig. 4 shows a kind of computing method process flow diagram of webpage authority value.Referring to Fig. 4, this method comprises that step S402 is to step S414.In the present embodiment, by search engine server, carry out the method.The computing method flow process of webpage authority value originates in step S402, at first, and according to the link points relationship between webpage, for webpage is set up adjacency matrix.Wherein, the webpage in the present embodiment comprises the web search system collections of web pages that search obtains according to certain element-specific (as the keyword of user's input), and with the webpage in collections of web pages, the direct extended web set that links points relationship is arranged.Wherein, the extended web set comprises webpage and the webpage pointed of the webpage in collections of web pages that points to collections of web pages.The concrete mode of setting up adjacency matrix in step S402 can be with reference to the description that provides in figure 5.
Concrete, at first, execution step S501, sort to webpage, and according to webpage, put in order webpage respectively as the summit of matrix, set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element.Enter subsequently step S502, in step S502, can be that set according to step S501, the link points relationship between webpage that summit identifies carries out assignment to each element in adjacency matrix.The rule of the assignment of step S502 can be: when webpage corresponding to element column points to the webpage that this element is expert at corresponding, be 1 with this row element assignment, when webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.
After step S402 has completed the foundation operation of adjacency matrix, enter step S404, in step S404, the adjacency matrix of webpage is cut into N part matrix fragment.In the present embodiment, N is the integer greater than 1, and is not more than many quantity that participate in the server of authority's value calculating., due to the quantity of the no more than server of N, therefore, can meet all matrix fragments that cut into are calculated respectively.In addition, when the matrix fragment is cut, should be taken into account the line number of matrix fragment and the relation between the current number of servers that participates in the calculating of authority's value, when both quantitative relation not simultaneously, in order to cut out comparatively suitable matrix fragment for server, so the mode of taking when the adjacency matrix of webpage is cut also may be different, specifically according to actual conditions, adopt corresponding cutting mode.It specifically can be divided into following two kinds of situations.
The first situation, when the line number of adjacency matrix is not more than the quantity of multiple servers, because the quantity of server is comparatively sufficient, the cutting mode of taking can be: the capable order according to adjacency matrix splits out every row of adjacency matrix successively and is used as a matrix fragment, obtains N part matrix fragment.This cutting mode makes every station server only need the delegation of matrix fragment is calculated, and has alleviated the calculated amount of every station server, participates in the time of again due to multiple servers, has also accelerated simultaneously the computing velocity of webpage authority value.
Corresponding with above-mentioned the first situation, in the second situation, when the line number of adjacency matrix during greater than the quantity of multiple servers, the cutting mode of taking can be: can comprise at least two row and according to row ordered pair adjacency matrix, cut according to every part of matrix fragment, and make the umber N that cuts the matrix fragment that obtains be not more than the quantity of multiple servers, guaranteed the complete computation to the matrix fragment.
In the present embodiment, in order to guarantee that every station server all has enough internal memories and calculates being dispensed to local matrix fragment.After execution of step S404, by step S406, judge whether every station server all possesses enough internal memories the matrix fragment of step S404 cutting is carried out the calculating of authority's value.The purpose of execution step S406 is, while having avoided server to calculate the matrix fragment, and the dyscalculia that causes due to low memory.In the present embodiment, step S406 need to first carry out the server that authority's value is calculated for every part of matrix fragment is selected, judges afterwards whether every station server all possesses enough internal memories its corresponding matrix fragment is carried out the calculating of authority's value.The server that obtains participating in the calculating of authority's value when step S406 judgement all has enough internal memories its corresponding matrix fragment is calculated, and performs step S410.When judgement obtains the low memory of at least one station server in server, execution step S408.
Above mention, when judging the low memory of at least one station server in step S406, enter step S408.In step S408, increase the quantity that participates in calculating the authoritative server that is worth, and according to the number of servers after increasing, adjacency matrix is cut again.Alternatively,, in order to reduce the resource occupation to the server that participates in the calculating of authority's value, first select to increase a station server.After the number of servers according to newly increasing is cut adjacency matrix again, in order to guarantee that every station server all has enough internal memories and calculates being dispensed to local matrix fragment, also need the decision operation of repeated execution of steps S406, when still being not enough to the matrix fragment is calculated, continue to trigger step S408 and increase the quantity of server, until the every station server of step S406 judgement all has enough internal memories while to being dispensed to local matrix fragment, calculating, stop performing step S406 and S408, and directly perform step S410.
Above also mention, the server that obtains participating in the calculating of authority's value when step S406 judgement all has enough internal memories when the matrix fragment of this locality is calculated, and enters step S410.In step S410, all matrix fragments that cutting is obtained are respectively allocated to corresponding server.In the present embodiment, which part matrix fragment what calculate for clear and definite each server is, step S410 can carry out respectively mark according to fractionation time all matrix fragments of ordered pair of adjacency matrix, and all the matrix fragments after mark are distributed to its corresponding server.
For every station server, after it received and is sent to local matrix fragment, execution step S412, namely carried out the calculating of authority value to the matrix fragment that receives, and obtains authoritative value fragment, and the authority that will calculate is worth fragment and returns.The present embodiment is specifically introduced step S412 from calculating authority's value fragment and returning to authority value fragment two aspects respectively.
Calculate the operation of authority's value fragment for server in step S412, can be with reference to the detailed description that provides in figure 6.At first, execution step S601, do dot-product operation with the every row element in this matrix fragment and preset initial vector, obtains primary vector.The initial vector of mentioning in step S601 is to send in advance to participate in multiple servers that authority's value calculates, and initial vector is the one-dimensional vector of initial authoritative value that comprises the webpage of authority's value to be calculated.Alternatively, the initial authoritative value of webpage can arrange according to the popularity (for example, web page access amount etc.) of this webpage, and its popularity is higher, for the initial authority value of its setting larger.Next, execution step S602, take out initial authority's value corresponding to every row element, as secondary vector in initial vector.Finally, execution step S603,, with the element in primary vector and the corresponding addition of the element in secondary vector, obtain the corresponding authority of this matrix fragment value fragment.Need to prove, the execution of above-mentioned steps S601 and step S602 there is no the sequencing relation.
The above-mentioned operation that server is calculated authority's value fragment is introduced, and below continues to introduce server and returns to the operation of authoritative value fragment.Alternatively, in order to distinguish the authority that different server calculates, be worth fragment, step S412 can be authoritative value fragment and mark when every station server returns to authority's value fragment, and wherein, this mark is identical with the mark of the matrix fragment that receives.
After receiving authority that multiple servers returns and being worth fragment, execution step S414, namely merge the whole authority's value fragments that receive, and obtains the final authoritative value of webpage.In the present embodiment, step S414 merges all methods of authority's value fragment and can comprise: obtain whole authority's value fragments, and reading every part of sign in authority's value fragment, the authority that will obtain according to the sign order is worth fragment and merges, and obtains the final authoritative value of webpage.Comprise sequence notation because authority is worth in fragment, guaranteed that step S414 merges integrality and the correctness of the final authority of the webpage value that obtains.
Foregoing has been introduced the concrete grammar that adopts when calculating authority's value for above-mentioned calculation element 100 particularly.After calculating authority's value by said method, the Search Results after can sorting to webpage and will sort according to the height of webpage authority value shows; Perhaps, recommend authority's value sequence default number webpage the preceding according to the height of webpage authority value to the user., to user's display web page, make the user can view the higher webpage of authority's value according to the height of authority value, facilitate the user to find quickly valuable webpage, promoted user's experience.
In sum, the computing method of a kind of webpage authority value that the embodiment of the present invention provides, can be in the situation that the separate unit server can't be completed the calculating of webpage authority value, the evaluation work of webpage authority value is distributed to multiple servers to be completed, again the result of calculation of every station server is merged, obtain the final authority of webpage value.Adopt method that the embodiment of the present invention provides can overcome in prior art to be difficult to the defect of authority's value of calculating huge quantity webpage.
Further, the method that adopts the present embodiment to provide, can also distribute to multiple servers equably with the calculated amount of webpage authority value, the load of server that one side is balanced, on the other hand,, because multiple servers participates in calculating simultaneously, also improved the computing velocity of webpage authority value.
A3, method described according to A2, it is characterized in that, described according to the link points relationship between described webpage, for described webpage is set up described adjacency matrix, comprise: described webpage is sorted, and according to webpage, put in order described webpage respectively as the summit of matrix, and set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element; Link points relationship between the webpage that identifies according to summit carries out assignment to each element in described adjacency matrix: when webpage corresponding to element column points to the webpage that this element is expert at corresponding, with this row element assignment, be 1; When webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.
A4, method described according to A1 to 3 any one, is characterized in that, described adjacency matrix with described webpage cuts into N part matrix fragment, comprising: the line number of obtaining described adjacency matrix; When described line number is not more than the quantity of described multiple servers, according to the row order, every row of described adjacency matrix is split out successively and is used as a matrix fragment, obtain N part matrix fragment; , according to every part of matrix fragment, comprise at least two row and according to the described adjacency matrix of row ordered pair, cut during greater than the quantity of described multiple servers when described line number, and making the umber N of the matrix fragment that cutting obtains be not more than the quantity of described multiple servers.A5, method described according to A4, it is characterized in that, the described server that described N part matrix fragment is respectively allocated to correspondence comprises: the fractionation time described N part of the ordered pair matrix fragment according to described adjacency matrix is carried out respectively mark, and described N part matrix fragment is distributed to different servers.A6, method described according to A5, it is characterized in that, the described N part authority of described merging value fragment, obtain final authority's value of described webpage, comprise: obtain described N part authority value fragment, and reading the sign of every part of fractionation order in authority's value fragment, the authority that will obtain according to described fractionation order is worth fragment and merges, and obtains the final authoritative value of described webpage.A8, method described according to A7, is characterized in that, before the described server that described N part matrix fragment is respectively allocated to correspondence carries out the calculating of authority's value, also comprises: for described webpage arranges respectively initial authority's value, and obtain described initial vector; Described initial vector is sent to respectively described multiple servers.B13, device described according to B12, it is characterized in that, the described block configuration of setting up becomes to set up as follows described adjacency matrix: described webpage is sorted, and according to webpage, put in order described webpage respectively as the summit of matrix, and set the be expert at points relationship that links of corresponding webpage of webpage corresponding to wherein this element column of each element representation and this element; Link points relationship between the webpage that identifies according to summit carries out assignment to each element in described adjacency matrix: when webpage corresponding to element column points to the webpage that this element is expert at corresponding, with this row element assignment, be 1; When webpage corresponding to element column do not point to the webpage that this element is expert at corresponding, with this row element assignment, be 0.B14, device described according to B11 to 13 any one, is characterized in that, described cutting module is configured to as follows the adjacency matrix of described webpage be cut into N part matrix fragment: the line number of obtaining described adjacency matrix; When described line number is not more than the quantity of described multiple servers, according to the row order, every row of described adjacency matrix is split out successively and is used as a matrix fragment, obtain N part matrix fragment; , according to every part of matrix fragment, comprise at least two row and according to the described adjacency matrix of row ordered pair, cut during greater than the quantity of described multiple servers when described line number, and making the umber N of the matrix fragment that cutting obtains be not more than the quantity of described multiple servers.B15, device described according to B14, is characterized in that, described distribution module also is configured to carry out respectively mark according to the fractionation time described N part of ordered pair matrix fragment of described adjacency matrix, and described N part matrix fragment is distributed to different servers.B16, device described according to B15, it is characterized in that, described merging module also is configured to obtain described N part authority value fragment, and read the sign of every part of fractionation order in authority's value fragment, the authority that will obtain according to described fractionation order is worth fragment and merges, and obtains final authority's value of described webpage.B18, device described according to B17, it is characterized in that, described device also comprises: module is set, be configured in described distribution module, described N part matrix fragment to be respectively allocated to before corresponding server carries out the calculating of authority value, for described webpage arranges respectively initial authority's value, and obtain described initial vector; Sending module, be configured to described initial vector is sent to respectively described multiple servers.B20, device described according to B11 to 19 any one, it is characterized in that, described device also comprises: application module, be configured to according to final authority's value of described webpage, the described webpage of showing in Webpage searching result be sorted after described merging module obtains final authority's value of described webpage; Perhaps, recommend described webpage according to final authority's value of described webpage to the user.
Intrinsic not relevant to any certain computer, virtual system or miscellaneous equipment with demonstration at this algorithm that provides.Various general-purpose systems also can with based on using together with this teaching.According to top description, it is apparent constructing the desired structure of this type systematic.In addition, the present invention is not also for any certain programmed language.Should be understood that and can utilize various programming languages to realize content of the present invention described here, and the top description that language-specific is done is in order to disclose preferred forms of the present invention.
In the instructions that provides herein, a large amount of details have been described.Yet, can understand, embodiments of the invention can be in the situation that do not have these details to put into practice.In some instances, be not shown specifically known method, structure and technology, so that not fuzzy understanding of this description.
Similarly, be to be understood that, in order to simplify the disclosure and to help to understand one or more in each inventive aspect, in the description to exemplary embodiment of the present invention, each feature of the present invention is grouped together in single embodiment, figure or the description to it sometimes in the above.Yet the method for the disclosure should be construed to the following intention of reflection: namely the present invention for required protection requires the more feature of feature of clearly putting down in writing than institute in each claim.Or rather, as following claims reflected, inventive aspect was to be less than all features of the disclosed single embodiment in front.Therefore, follow claims of embodiment and incorporate clearly thus this embodiment into, wherein each claim itself is as independent embodiment of the present invention.
Those skilled in the art are appreciated that and can adaptively change and they are arranged in one or more equipment different from this embodiment the module in the equipment in embodiment.Can be combined into a module or unit or assembly to the module in embodiment or unit or assembly, and can put them into a plurality of submodules or subelement or sub-component in addition.At least some in such feature and/or process or unit are mutually repelling, and can adopt any combination to disclosed all features in this instructions (comprising claim, summary and the accompanying drawing followed) and so all processes or the unit of disclosed any method or equipment make up.Unless clearly statement in addition, in this instructions (comprising claim, summary and the accompanying drawing followed) disclosed each feature can be by providing identical, be equal to or the alternative features of similar purpose replaces.
In addition, those skilled in the art can understand, although embodiment more described herein comprise some feature rather than further feature included in other embodiment, the combination of the feature of different embodiment mean be in scope of the present invention within and form different embodiment.For example, in the following claims, the one of any of embodiment required for protection can be used with array mode arbitrarily.
All parts embodiment of the present invention can realize with hardware, perhaps with the software module of moving on one or more processor, realizes, perhaps the combination with them realizes.It will be understood by those of skill in the art that and can use in practice microprocessor or digital signal processor (DSP) to realize some or repertoire according to some or whole parts in the calculation element of the webpage of embodiment of the present invention authority value.The present invention can also be embodied as be used to part or all equipment or the device program (for example, computer program and computer program) of carrying out method as described herein.The program of the present invention that realizes like this can be stored on computer-readable medium, perhaps can have the form of one or more signal.Such signal can be downloaded and obtain from internet website, perhaps provides on carrier signal, perhaps with any other form, provides.
It should be noted above-described embodiment the present invention will be described rather than limit the invention, and those skilled in the art can design alternative embodiment in the situation that do not break away from the scope of claims.In the claims, any reference symbol between bracket should be configured to limitations on claims.Word " comprises " not to be got rid of existence and is not listed in element or step in claim.Being positioned at word " " before element or " one " does not get rid of and has a plurality of such elements.The present invention can realize by means of the hardware that includes some different elements and by means of the computing machine of suitably programming.In having enumerated the unit claim of some devices, several in these devices can be to carry out imbody by same hardware branch.The use of word first, second and C grade does not represent any order.Can be title with these word explanations.
Claims (10)
1. the computing method of webpage authority value, be applied to utilize the environment of authority's value of multiple servers Distributed Calculation webpage, comprising:
The adjacency matrix of described webpage is cut into N part matrix fragment, and wherein, described N is the integer greater than 1, and is not more than the quantity of described multiple servers;
Described N part matrix fragment is respectively allocated to corresponding server carries out the calculating of authority's value, obtain N part authority value fragment;
Merge described N part authority value fragment, obtain final authority's value of described webpage.
2. method according to claim 1, is characterized in that, before described adjacency matrix with described webpage cuts into N part matrix fragment, also comprises:
According to the link points relationship between described webpage, for described webpage is set up described adjacency matrix.
3. the described method of according to claim 1 to 2 any one, is characterized in that, every station server carries out the calculating of authority's value to being dispensed to local matrix fragment step comprises:
Every row element in this matrix fragment and preset initial vector are done dot-product operation, obtain primary vector, wherein, described initial vector comprises initial authority's value of described webpage;
Take out initial authority's value corresponding to described every row element in described initial vector, as secondary vector;
, with the corresponding addition of the element in the element in described primary vector and described secondary vector, obtain the corresponding authority of this matrix fragment value fragment.
4. the described method of according to claim 1 to 3 any one, is characterized in that, before the described server that described N part matrix fragment is respectively allocated to correspondence carries out the calculating of authority's value, also comprises:
For any station server,
Judge whether this server possesses enough internal memories and carry out the calculating of authority's value to being dispensed to local matrix fragment;
When the machine low memory, increase the quantity of server; And
According to the quantity of the server after increasing, described adjacency matrix is carried out cutting again;
Repeat the judgement of server memory capacity and the increase operation of number of servers, until the machine internal memory of server enough carries out the calculating of authority's value to being dispensed to local matrix fragment.
5. the described method of according to claim 1 to 4 any one, is characterized in that, after the described final authority value that obtains described webpage, also comprises:
According to final authority's value of described webpage, the described webpage of showing in Webpage searching result is sorted; Perhaps,
Recommend described webpage according to final authority's value of described webpage to the user.
6. the calculation element of webpage authority value, be applied to utilize the environment of authority's value of multiple servers Distributed Calculation webpage, comprising:
Cutting module, be configured to the adjacency matrix of described webpage is cut into N part matrix fragment, and wherein, described N is the integer greater than 1, and is not more than the quantity of described multiple servers;
Distribution module, be configured to that described N part matrix fragment is respectively allocated to corresponding server and carry out the calculating of authority's value, obtains N part authority value fragment;
Merge module, be configured to merge described N part authority value fragment, obtain final authority's value of described webpage.
7. device according to claim 6, is characterized in that, described device also comprises:
Set up module, before being configured to the adjacency matrix of described webpage be cut into N part matrix fragment at described cutting module, according to the link points relationship between described webpage, for described webpage is set up described adjacency matrix.
8. the described device of according to claim 6 to 7 any one, is characterized in that, every part of authority's value fragment that described merging module merges calculates in the following manner by every station server:
Every row element in this matrix fragment and preset initial vector are done dot-product operation, obtain primary vector, wherein, described initial vector comprises initial authority's value of described webpage;
Take out initial authority's value corresponding to described every row element in described initial vector, as secondary vector;
, with the corresponding addition of the element in the element in described primary vector and described secondary vector, obtain the corresponding authority of this matrix fragment value fragment.
9. the described device of according to claim 6 to 8 any one, is characterized in that, described device also comprises:
Judge module, be configured in described distribution module, described N part matrix fragment to be respectively allocated to before corresponding server carries out the calculating of authority value,, for any station server, judge whether this server possesses enough internal memories and carry out the calculating of authority's value to being dispensed to local matrix fragment;
Increase module, be configured to when described judge module is judged the machine low memory, increase the quantity of server; And
Described cutting module, also be configured to according to the quantity of the server after increasing, described adjacency matrix be cut again;
The described judge module of repeated trigger and the performed operation of described increase module, until the machine internal memory of server enough carries out the calculating of authority's value to being dispensed to local matrix fragment.
10. the computing system of webpage authority value, comprise and the calculation element of the described webpage authority of claim 6 to 9 any one value also comprise the multiple servers that the authoritative value of webpage is carried out Distributed Calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102747560A CN103399859A (en) | 2013-07-02 | 2013-07-02 | Web page authority value calculating method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013102747560A CN103399859A (en) | 2013-07-02 | 2013-07-02 | Web page authority value calculating method, device and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103399859A true CN103399859A (en) | 2013-11-20 |
Family
ID=49563489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013102747560A Pending CN103399859A (en) | 2013-07-02 | 2013-07-02 | Web page authority value calculating method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103399859A (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1737803A (en) * | 2004-07-30 | 2006-02-22 | 微软公司 | Efficiently ranking web pages via matrix index manipulation and improved caching |
-
2013
- 2013-07-02 CN CN2013102747560A patent/CN103399859A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1737803A (en) * | 2004-07-30 | 2006-02-22 | 微软公司 | Efficiently ranking web pages via matrix index manipulation and improved caching |
Non-Patent Citations (2)
| Title |
|---|
| MD. NAZRUL ISLAM等: "An Empirical Distributed Matrix Multiplication Algorithm to Reduce Time Complexity", 《PROCEEDING OF THE INTERNATIONAL MULTICONFERENCE OF ENGINEERS AND COMPUTER SCIENTISTS》 * |
| 吴军: "《数学之美》", 30 June 2012, 人民邮电出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103491205B (en) | The method for pushing of a kind of correlated resources address based on video search and device | |
| CN104951099A (en) | Method and device for showing candidate items based on input method | |
| CN106339756A (en) | Training data generation method and device and searching method and device | |
| CN104462611A (en) | Modeling method, ranking method, modeling device and ranking device for information ranking model | |
| CN103425687A (en) | Retrieval method and system based on queries | |
| CN106156082A (en) | A kind of body alignment schemes and device | |
| CN103870607A (en) | Sequencing method and device of search results of multiple search engines | |
| CN108062303A (en) | The recognition methods of refuse messages and device | |
| CN111949306B (en) | Pushing method and system supporting fragmented learning of open-source project | |
| CN104063458A (en) | Method and device for providing corresponding solution for terminal fault problem | |
| CN104881504A (en) | Information search method and device | |
| CN106815265A (en) | The searching method and device of judgement document | |
| CN104317931A (en) | Webpage title determining method and device | |
| CN103885767B (en) | System and method used for geographical area correlated websites | |
| CN102567565A (en) | Cable parameter processing method and system utilizing same | |
| CN105159884A (en) | Method and device for establishing industry dictionary and industry identification method and device | |
| CN103870563B (en) | It is determined that the method and apparatus of the theme distribution of given text | |
| CN110427574B (en) | Route similarity determination method, device, equipment and medium | |
| CN104182518A (en) | Collaborative filtering recommendation method and device | |
| CN105335363A (en) | Object pushing method and system | |
| CN102331987A (en) | Patent data mining system and method | |
| US8463799B2 (en) | System and method for consolidating search engine results | |
| CN111382250A (en) | Question text matching method and device, computer equipment and storage medium | |
| CN104933149A (en) | Information searching method and information searching device | |
| CN104462556A (en) | Method and device for recommending question and answer page related questions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20131120 |