Method for determining importance of web pages

Abstract

The application discloses a method for determining the importance of a webpage, comprising the following steps: converting the webpage and the relationship between the webpage into a graph structure; obtaining the estimation of the probability quantity received from a target node t and finally left in a node k after the probability quantity propagation process is completed; updating the node k into another node in the graph structure; repeating the step of obtaining the probability quantity received from the target node t and finally left in the node k of the node k in the graph structure until the estimation of the probability quantity received from the target node t and finally left in the node k of all the nodes k in the graph structure is obtained; obtaining the estimation of the PageRank score of the target node t according to the probability quantity received from the target node t and finally left in the node k of all the nodes in the graph structure, and determining the importance of the webpage corresponding to the target node t. The method for determining the importance of the webpage can more efficiently complete the calculation of the PageRank of a specified webpage.

Description

TECHNICAL FIELD

[0001] The present application relates to the computer technology field, and particularly to a method for determining the importance of a webpage. BACKGROUND

[0002] With the rapid growth of information on the World Wide Web, the rapid development of search engine manufacturers such as Baidu and Google, and the wide penetration and universal application of information retrieval technology in social development, how to efficiently calculate the importance of a specified webpage on the World Wide Web has become one of the problems that search engine manufacturers, universities, research institutes and other types of institutions focus on. For the calculation of the importance score of a webpage, the existing search engines represented by Baidu and Google generally use PageRank, a webpage importance measurement index, to quantify the importance of each webpage on the World Wide Web. The calculation goal can be specified as calculating the PageRank score of a specified webpage on the World Wide Web.

[0003] However, due to the huge scale of information stored on the World Wide Web today, the existing method for calculating the PageRank score of a specified webpage on the World Wide Web has low running efficiency and is difficult to support large-scale World Wide Web data.

[0004] The information disclosed in this BACKGROUND section is only intended to increase an understanding of the general background of the application and should not be construed as a recognition of the prior art of any form. SUMMARY

[0005] The purpose of the present application is to provide a method for determining the importance of a webpage, which can more accurately calculate the PageRank of a specified webpage while consuming less time than existing methods, and thus can more efficiently measure the importance of the webpage.

[0006] To achieve the above purpose, the present application provides a method for determining the importance of a webpage, comprising: converting a webpage and the relationship between webpages into a graph structure, wherein the graph structure G includes n nodes k corresponding to the webpages and multiple edges corresponding to the relationship between the webpages, and a target webpage corresponds to a target node t in the graph structure;

[0007] obtaining an estimate of the probability quantity from the target node t and finally remaining in the node k after the probability quantity propagation process is completed wherein the node k is any node in the graph structure;

[0008] updating the node k to another node of the n nodes in the graph structure;

[0009] repeating the step of obtaining the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k until the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k is obtained;

[0010] obtaining the estimate of the PageRank score of the target node t according to the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k

[0011] obtaining the estimate of the PageRank score of the target node t according to the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k determining the importance of the target webpage corresponding to the target node t.

[0012] In a preferred embodiment, the estimate of the PageRank score of the target node t is obtained according to the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k during the propagation of the amount of probability comprising:

[0013] obtaining the estimate according to Formula I the Formula I is:

[0014]

[0015] wherein d t is the number of neighbors of the target node t on the graph structure, d k is the number of neighbors of the node k on the graph structure.

[0016] In a preferred embodiment, the estimate of the amount of probability received by each node k in the graph structure from the target node t and finally remaining in the node k after the propagation of the amount of probability is completed before, further comprising

[0017] Step S101, obtaining all nodes receiving the propagated probability at time i-1, wherein i is an integer in the range of [1, L] and the initial value of i is 1, and L is the farthest step length of the propagation of the amount of probability;

[0018] Step S102, selecting one of all nodes receiving the propagated probability at time i-1 as a sampling node u at time i;

[0019] Step S103, obtaining the sampling node u at time i and the propagated probability received by the sampling node u at time i from the sampling node at time i-1 at time i-1;

[0020] Step S104, according to the preset condition and the propagation probability received by the sampling node u at time i from the sampling node at time i-1, all neighbor nodes v of the sampling node u at time i or part of the neighbor nodes w of the neighbor nodes v at time i are taken as the sampling nodes at time i+1, and the propagation probability of the sampling node u at time i to the sampling node at time i+1 is calculated;

[0021] Step S105, according to the probability of the sampling node u at time i to the sampling node at time i+1, the estimation of the probability amount of the sampling node at time i+1 received from the sampling node at time i in the probability amount propagation process and finally remaining in the sampling node at time i+1 is updated;

[0022] Step S106, another node in all nodes receiving the propagation probability at time i-1 is selected as the sampling node at time i; steps S103-S105 are repeated until each node receiving the propagation probability at time i-1 is selected as the sampling node at time i, and the current time i is updated to i+1;

[0023] Step S107, the steps of steps S101-S106 are repeated until the current time i is updated to L and stopped.

[0024] In a preferred embodiment, the preset condition includes a first preset condition, and the propagation probability of the sampling node u at time i to the sampling node at time i+1 includes:

[0025] If the sampling node u at time i satisfies the first preset condition, all neighbor nodes v of the sampling node u at time i are taken as the sampling nodes at time i+1, wherein the first preset condition is: (1-α)· Wherein θ is an estimation error parameter specified according to an actual application scenario, and the value range is θ∈(0,1);

[0026] The probability of the sampling node at time i to the sampling node at time i+1 The formula two is:

[0027]

[0028] Wherein, is the propagation probability received by the sampling node u at time i from the sampling node at time i-1, d u is the number of neighbor nodes of the sampling node u on the graph structure.

[0029] In a preferred embodiment, the preset conditions include a first preset condition and a second preset condition. The step of determining all neighboring nodes v of sampling node u at time i, or a subset w of all neighboring nodes v, as sampling nodes at time i+1, based on the preset conditions and the propagation probability received by sampling node u at time i-1 from sampling node u at time i-1, and calculating the propagation probability from sampling node u at time i to sampling node u at time i+1, further includes:

[0030] If the sampling node u at time i does not satisfy the first preset condition but satisfies the second preset condition, then sample the neighbor node w from all neighbor nodes v of the sampling node u at time i as the sampling node at time i+1. The second preset condition is:

[0031] The probability that sampling node u at time i is transmitted to sampling node u at time i+1 Calculated using Formula 3, which is:

[0032]

[0033] In a preferred embodiment, updating the estimate of the probability received by the sampling node at time i from the sampling node at time i and ultimately remaining at the sampling node at time i+1 during the probability propagation process, based on the probability that the sampling node u at time i transmits to the sampling node at time i+1, includes:

[0034] Estimation of the probability received by the sampling node at time i+1 during the probability propagation process, based on Formula 4, and the probability that the sampling node at time i+1 ultimately remains at the sampling node at time i+1. The updated formula four is as follows:

[0035]

[0036] The initial value before the probability propagation process begins is 0; The initial value before the probability propagation process begins is α.

[0037] In a preferred embodiment, updating the probability of the sampling node at time i+1 receiving data from the sampling node at time i during probability propagation, and ultimately remaining at the sampling node at time i+1, based on the probability that the sampling node u at time i transmits to the sampling node at time i+1, includes:

[0038] According to Formula Five, the estimate π of the probability mass of the sampling node at time i+1 received from the sampling node at time i and finally left in the sampling node at time i+1 during the probability mass propagation process t (w) updating, Formula Five is:

[0039]

[0040] The initial value before the start of the probability mass propagation process is 0; The initial value before the start of the probability mass propagation process is α.

[0041] In a preferred embodiment, the sampling of the neighbor node w from all neighbor nodes v of the node u includes:

[0042] Establishing a neighbor set N(u) storing all neighbor nodes v of the sampling node u at the current time;

[0043] Selecting one node in the neighbor set N(u) as a node to be sampled;

[0044] Repeating the step of selecting one node in the list as a node to be sampled until a preset condition is met;

[0045] Selecting all the nodes to be sampled as the sampled neighbor nodes w.

[0046] In a preferred embodiment, the step of selecting one node in the neighbor set N(u) as a node to be sampled includes:

[0047] According to Formula Six, determining the position of the node to be sampled in the neighbor set N(u) of the node u Wherein,

[0048] Formula Six is:

[0049]

[0050] Wherein, rg is a random number subject to a geometric distribution ;

[0051] According to the position of the node to be sampled in the neighbor set N(u) of the node u Selecting the node to be sampled;

[0052] Wherein, before the start of sampling The preset condition is

[0053] In a preferred embodiment, the random number subject to a geometric distribution is generated by the following steps:

[0054] A uniform random number rand in the range of (0, 1) is generated, and a random number rg obeying a geometric distribution is calculated according to Formula Seven:

[0055]

[0056] wherein, is the probability of the sampling node at time i delivering the probability quantity θ to the sampling node at time i+1,

[0057] Compared with the prior art, the method for determining the importance of a webpage provided by the embodiment can more accurately complete the calculation of the ranking of a specified webpage, and further can more efficiently measure the importance of the webpage, under the premise of consuming less time than the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 is a flowchart of the method for determining the importance of a webpage according to an embodiment of the present application.

[0059] Figure 2 is a schematic diagram of the delivery of a probability quantity according to an embodiment of the present application. DETAILED DESCRIPTION

[0060] The specific embodiments of the present application are described in detail below with reference to the accompanying drawings, but it should be understood that the scope of protection of the present application is not limited by the specific embodiments.

[0061] Unless otherwise explicitly stated, throughout the specification and claims, the term "comprise" or its variants such as "comprises" or "comprising" will be understood to encompass the stated element or elements or components, but not to preclude the presence or addition of one or more other elements or components.

[0062] Search engines represented by Baidu and Google are essentially information retrieval systems, which can assist users in efficiently retrieving information stored in the Internet, and the retrieval results are generally presented in the form of webpages. In order to improve the information retrieval experience of users, the mainstream search engine manufacturers generally calculate an importance score for the webpages on the World Wide Web, and when the user performs information retrieval, the relevant webpages with higher importance scores are preferentially displayed, thereby ensuring the quality of the information retrieval results.

[0063] ​As for the method for measuring the importance of a webpage, the existing search engines represented by Baidu and Google generally use PageRank, a webpage importance measurement index, to quantify the importance of webpages on the World Wide Web. The definition of PageRank is as follows: webpages on the World Wide Web generally have a link relationship. Through the link relationship between webpages, a user can easily jump from one webpage to another. If a webpage u is linked by many webpages (i.e., many webpages have a jump link pointing to the webpage), or if the webpage u is linked by a webpage known to be very important (for example, a webpage with an official endorsement, such as a university website, is generally considered to be of high importance, and the content on the webpage is also of high credibility), the importance of the webpage u is high. In particular, in the actual use of PageRank to measure the importance of a webpage, the existing method generally calculates the PageRank score of each webpage, and then measures the importance of the webpage based on the size of the PageRank score.

[0064] However, the existing method for calculating the PageRank score of a webpage has low running efficiency and is difficult to support large-scale World Wide Web data. In order to improve the running efficiency, the existing method generally focuses on the PageRank score of a specified webpage on the World Wide Web, and only calculates an estimated value of the PageRank score of the specified webpage rather than an accurate value. However, the estimated value of the PageRank score of the specified webpage calculated by the existing method in a short time is of low quality and has a large gap compared with the true value, which is difficult to meet the needs of actual application scenarios. Based on the above, the embodiment provides a method for calculating the PageRank of a target webpage, which can more efficiently calculate the PageRank score of the webpage and further efficiently measure the importance of the webpage.

[0065] More specifically, if all webpages in the World Wide Web (or search engine) and the link relationship (Link) between the webpages are converted into a graph structure G, the graph structure G includes nodes corresponding to the webpages and edges corresponding to the link relationship between the webpages, the object of the embodiment is to calculate an estimated value of the PageRank score of a specified node (i.e., a target node t) on the graph G It is worth noting that the calculation process of the estimated value of the PageRank score of the node t is corresponding to a probability quantity propagation process, and the estimated value of the PageRank score of the node t is equal to a weighted average of the probability quantities received by each node u on the graph from the node t and finally remaining in the node u in the probability quantity propagation process. Let denote the estimated value of the probability quantity received by any node u on the graph from the node t and finally remaining in the node u in the probability quantity propagation process, and the estimated value of the PageRank score of the node t is equal to a weighted average of obtained by each node u on the graph after the probability quantity propagation process ends. ​

[0066] The core of the embodiment "probability quantity propagation process" is that initially, 1 unit of probability quantity is placed at the target node t, and then at each time, each node transfers part of its current received probability quantity to its neighbor nodes and keeps another part for itself. The "time" here does not correspond to a specific time, but is only used to distinguish the number of rounds of propagation of the probability quantity on the graph structure.

[0067] As shown in the flowchart of the method for determining the importance of a webpage according to the preferred embodiment of the present application, the method comprises steps S1-S6. Figure 1

[0068] In step S1, the webpages and the relationships between the webpages are converted into a graph structure, wherein the graph structure G includes n nodes k corresponding to the webpages and a plurality of edges corresponding to the relationships between the webpages, and a target webpage corresponds to a target node t in the graph structure.

[0069] It should be noted that in step S1, all the webpages in the World Wide Web or the search engine and the link relationships (Links) between the webpages are converted into a graph structure G, which includes nodes corresponding to the webpages and edges corresponding to the link relationships between the webpages. In particular, the embodiment is dedicated to calculating the PageRank score of a specified webpage in the World Wide Web, which is converted into a node t in the graph structure G, and the node t is referred to as a target node. The graph structure G includes n nodes and m edges. The PageRank score of the target node t in the graph structure G can reflect the importance of the target user on the social network, and the larger the PageRank score π(t) of the target node t, the higher the PageRank of the target webpage among all the webpages in the World Wide Web, and the content of the webpage with high importance is generally more reliable.

[0070] The webpages in the embodiment refer to all the webpages in the World Wide Web, and the relationships between the webpages refer to the link relationships between the webpages. For example, all the webpages in the Baidu search engine and the link relationship network of the webpages.

[0071] Specifically, all the webpages in the Baidu search engine are converted into graph nodes, and the link relationships between the webpages are converted into edges in the graph structure. If there is a link relationship between a webpage u and a webpage v, a non-directed edge (denoted as edge (u, v)) is established from the node u to the node v in the graph structure. We refer to all the nodes having an edge relationship with the node u as the neighbor nodes of the node u, refer to the set of all the neighbor nodes of the node u as the neighbor set N(u) of the node u, and refer to the number of nodes having an edge relationship with the node u in the graph structure as the neighbor number of the node u, denoted as d u .

[0072] In step S2, the estimate of the probability quantity received from the target node t and finally left in the node k after the completion of the probability quantity propagation process is obtained​ wherein the node k is any node in the graph structure.

[0073] Step S3, updating the node k to another node of the n nodes in the graph structure.

[0074] Step S4, repeating the step of obtaining the probability amount received by the node k from the target node t and finally left in the node k in the graph structure until obtaining the estimation of the probability amount received by all the nodes k from the target node t and finally left in the node k in the graph structure.

[0075] Step S5, obtaining the estimation of the PageRank score of the target node t according to the probability amount received by all the nodes k from the target node t and finally left in the node k in the graph structure.

[0076] Step S6, determining the importance of the target node t according to the estimation of the PageRank score of the target node t. determining the importance of the target node t corresponding to the target webpage.

[0077] determining the importance of the target webpage corresponding to the target node t according to the estimation of the PageRank score of the target node t corresponding to the target webpage. determining the importance of the target webpage corresponding to the target node t, and further measuring the importance of the target webpage.

[0078] Therefore, the method for determining the importance of the webpage provided by the embodiment can more accurately complete the estimation of the PageRank of a specified webpage under the premise of consuming less time than the prior art, and further can more efficiently measure the importance of the webpage.

[0079] In one embodiment, the step S5 comprises: calculating according to Formula One. The Formula One is:

[0080]

[0081] wherein d t is the number of nodes having an edge relationship with the target node t on the graph structure, also referred to as the number of neighbors of the target node t; d k is the number of nodes having an edge relationship with the node k on the graph structure, also referred to as the number of neighbors of the node k.

[0082] Before the step S2, further comprising steps S101-S107.

[0083] Step S101, obtaining all the nodes receiving the propagation probability at time i-1, wherein i is an integer in the range of [1, L] and the initial value of i is 1, and L is the farthest step length of the propagation of the probability amount.

[0084] Wherein, i is the initial value 1, time i-1 is time 0, the sampling node of time 0 is the target node t, and the sampling node of time 0 is only one, and the nodes receiving the propagation probability at time i-1 are the neighbor nodes of the sampling node of time i-1.

[0085] Step S102, one of the nodes receiving the propagation probability at time i-1 is selected as the sampling node u of time i.

[0086] Step S103, the sampling node u of time i and the propagation probability from the sampling node of time i-1 received by the sampling node u of time i at time i-1 are obtained.

[0087] Step S104, according to the preset condition and the propagation probability from the sampling node of time i-1 (t, i is 1) received by the sampling node u of time i at time i-1, all the neighbor nodes v of the sampling node u of time i or part of the neighbor nodes w are selected as the sampling nodes of the next time (i+1), and the propagation probability of the sampling node u of time i to the sampling nodes of the next time (i+1) is calculated. Or

[0088] Step S105, the estimation of the probability amount received by the sampling nodes of the next time (i+1) from the sampling node u of time i in the probability amount propagation process and finally remaining in the sampling nodes of the next time (i+1) is updated according to the propagation probability of the sampling node u of time i to the sampling nodes v or w of the next time (i+1).

[0089] It should be noted that time i can be understood as the current time, time i-1 is the time before the current time i, and time i+1 is the time after the current time i.

[0090] Step S106, another node receiving the propagation probability at time i-1 is selected as the sampling node of time i, and steps S103-S105 are repeated until each node receiving the propagation probability at time i-1 is selected as the sampling node of time i, and the current time i is updated to i+1.

[0091] Specifically, another one of all nodes receiving the propagation probability at time i-1 is selected as the sampling node at time i; the sampling node at the current time i and the propagation probability received by the sampling node at the last time (time i-1) from the sampling node at the last time (time i-1) are repeatedly obtained; according to a preset condition and the propagation probability received by the sampling node at the current time i from the sampling node at the last time, all neighbor nodes or part of the neighbor nodes of the sampling node at the current time i are taken as the sampling node at the next time, and the probability of the sampling node at the current time i being transmitted to the sampling node at the next time is calculated; the step of updating the estimation of the probability amount received by the sampling node at the next time from the sampling node at the current time and finally remaining in the sampling node at the next time in the probability amount propagation process according to the probability of the sampling node at the current time being transmitted to the sampling node at the next time is repeated until each node receiving the propagation probability at time i-1 is selected as the sampling node at time i, and the current time i is updated to i+1.

[0092] In step S107, the steps of steps S101-S106 are repeated until the time i is updated to L and then stopped.

[0093] The specific value of L can be: L=log 1-α (0.05α / n).

[0094] Specifically, the step of repeatedly obtaining all nodes receiving the propagation probability at time i-1 as all sampling nodes at time i; selecting one of all nodes receiving the propagation probability at time i-1 as a sampling node at time i; obtaining the sampling node at current time i and the propagation probability received by the sampling node at the last time (time i-1) from the sampling node at the last time (time i-1); according to a preset condition and the propagation probability received by the sampling node at current time i from the sampling node at the last time, taking all neighbor nodes of the sampling node at current time i or part of the nodes in all neighbor nodes as sampling nodes at the next time, and calculating the probability of the sampling node at current time i transferring to the sampling node at the next time; updating the estimation of the probability amount received by the sampling node at the next time from the sampling node at current time i and finally remaining in the sampling node at the next time in the probability amount propagation process according to the probability of the sampling node at current time transferring to the sampling node at the next time; selecting another of all nodes receiving the propagation probability at time i-1 as a sampling node at time i; repeatedly obtaining the sampling node at current time i and the propagation probability received by the sampling node at the last time (time i-1) from the sampling node at the last time (time i-1); according to a preset condition and the propagation probability received by the sampling node at current time i from the sampling node at the last time, taking all neighbor nodes of the sampling node at current time i or part of the nodes in all neighbor nodes as sampling nodes at the next time, and calculating the probability of the sampling node at current time i transferring to the sampling node at the next time; updating the estimation of the probability amount received by the sampling node at the next time from the sampling node at current time i and finally remaining in the sampling node at the next time in the probability amount propagation process according to the probability of the sampling node at current time transferring to the sampling node at the next time; until each node receiving the propagation probability at time i-1 is selected as a sampling node at time i, the step of updating current time i to i+1 until current time i is updated to L and then stopping.

[0095] Figure 2 FIG. 1 is a schematic diagram of probability amount propagation according to an embodiment of the present application. Figure 2 An exemplary sequence of probability amount propagation is described. Specifically, at time i-1, for example, time 0, node t obtains 1 unit of probability amount, and the node receiving the probability amount at time 0 is only node t, and node t is a sampling node at time 0. At time i, for example, time 1, node t transmits the obtained probability amount to its neighbor node u Figure 2 The example given in FIG. 1 is transmitted to u1, u2, and u1, u2 are sampling nodes at time 1. At time i+1, for example, time 2, u receiving the probability at time i, for example, time 1, transmits the probability amount obtained at time 1 to its neighbor Figure 2The given example is node v, w) v, w is the sampling node at time 2. The above description is only exemplary. In the probability propagation process, the probability quantity can be propagated to all neighbor nodes, or one or more of all neighbor nodes, Figure 2 The propagation mode is not limited.

[0096] The starting time (i.e. time 0) allocates 1 unit of probability quantity at the target node t; use to represent the probability quantity from node t at time i propagated to node u. Therefore, For all other nodes u∈V on the graph structure, (i.e. the probability quantity at node t at time 0 propagated to other nodes on the graph is 0).

[0097] At time i, all nodes u that received the probability quantity at time i-1 propagate a part of the probability quantity received by the node u at time i-1 (i.e. ) to all its neighbor nodes v or sampled neighbor nodes w; the node v or node w that receives the probability quantity at time i updates the estimate of the probability quantity received from the sampling node at time i in the probability quantity propagation process and finally left in the node v or node w.

[0098] It is worth noting that only node t is allocated 1 unit of probability quantity at time 0. Therefore, at time 1, only node t needs to propagate a part of the probability quantity obtained by the node t at time 0 (i.e. ) to all its neighbor nodes v or sampled neighbor nodes w.

[0099] In one embodiment, the preset condition includes a first preset condition, and step S104 includes:

[0100] If the sampling node u at the current time i satisfies the first preset condition, all neighbor nodes v of the sampling node u at the current time i are taken as the sampling node at the next time, wherein the first preset condition is: Wherein, θ is an estimation error parameter specified according to the actual application scenario, and the value range is θ∈(0,1);

[0101] The probability of the sampling node at the current time i propagated to the sampling node at time i+1 The formula two is calculated by formula two, and the formula two is:

[0102]

[0103] Wherein, α is a probability propagation attenuation coefficient, and the value range is α∈(0,1). The initial value before the start of the probability quantity propagation process is 0.

[0104] In the case that the sampling node u at time i satisfies the first preset condition, step S105 comprises: updating the probability quantity of the sampling node v at time i+1 received from the sampling node at time i in the probability quantity propagation process and finally remaining in the sampling node at time i+1 according to formula four The formula four is:

[0105]

[0106] If the node v is not the target node t, The initial value before the start of the probability quantity propagation process is 0. The initial value before the start of the probability quantity propagation process is α.

[0107] In one embodiment, the preset conditions comprise the first preset condition and the second preset condition, and step S104 further comprises:

[0108] If the sampling node u at time i does not satisfy the first preset condition but satisfies the second preset condition, sampling a neighbor node w from all neighbor nodes v of the sampling node u at the current time i as the sampling node at time i+1, and the second preset condition is:

[0109] The probability of the sampling node at time i being transmitted to the sampling node at time i+1 The formula three is calculated by formula three:

[0110]

[0111] In the case that the sampling node u at the current time i does not satisfy the first preset condition but satisfies the second preset condition, step S105 comprises: updating the estimation π of the probability quantity of the sampling node at time i+1 received from the sampling node at time i in the probability quantity propagation process and finally remaining in the sampling node at time i+1 according to formula five t (w) The formula five is:

[0112]

[0113] If the node w is not the target node t, The initial value before the start of the probability quantity propagation process is 0. The initial value before the start of the probability quantity propagation process is α.

[0114] In one embodiment, the sampling of the neighbor node w from all neighbor nodes v of the node u in step 104 can comprise:

[0115] Establishing a neighbor set N(u) storing all neighbor nodes v of the sampling node u at the current time;

[0116] selecting one node in the neighbor set N(u) of the sampling node u as a node to be sampled;

[0117] repeating the step of selecting one node in the list as a node to be sampled until a preset condition is met;

[0118] all the selected nodes to be sampled as the sampled neighbor nodes w.

[0119] In one embodiment, the step of selecting one node in the neighbor set N(u) of the sampling node u as a node to be sampled comprises: determining the position of the node to be sampled in the neighbor set N(u) of the sampling node u according to Formula Six and determining the position of the node to be sampled in the neighbor set N(u) of the sampling node u according to the position of the node to be sampled in the neighbor set N(u) of the sampling node u at the current time selecting the node to be sampled.

[0120] The Formula Six is:

[0121]

[0122] wherein rg is a random number subject to a geometric distribution before the sampling starts

[0123] In one embodiment, the preset condition is

[0124] The random number subject to the geometric distribution may be generated by the following steps:

[0125] generating a uniform random number rand in the range of (0, 1) and calculating a random number rg subject to a geometric distribution according to Formula Seven:

[0126]

[0127] wherein, is the probability of the sampling node at the current time i passing the probability amount θ of the sampling node to the next time,

[0128] when The method of the present embodiment can obtain an estimated value of the PageRank score of the target node t in the time of and can ensure that the probability that the relative error between the estimated value of the PageRank score of the target node t calculated by the method and the accurate value π(t) is not more than 0.1 is not less than 2 / 3, i.e. ​The probability of establishment is not less than 2 / 3. Alpha is a probability propagation attenuation coefficient, and alpha is in the range of alpha∈(0, 1). As a comparison, the optimal algorithm before the present method can only guarantee that the relative error between the estimated value of the target node t PageRank score and the accurate value π(t) is not more than 0.1 within the time of t , where d t is the degree of the target node t, m is the number of edges in the graph G (i.e. the number of links between web pages in the World Wide Web), and n is the number of nodes in the graph G (i.e. the number of web pages in the World Wide Web).

[0129] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer usable program code.

[0130] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus generate a means for implementing the functions specified in the flow Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0131] These computer program instructions can also be stored in a computer-readable memory capable of causing the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction means, which implements the functions specified in the flow Figure 1 one or more flows and / or blocks. Figure 1 one or more flows and / or blocks.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a means for implementing the functions specified in the flow Figure 1one or more processes and / or blocks Figure 1 the steps of the functions specified in the one or more blocks.

[0133] The foregoing description of specific exemplary embodiments of the application has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the application to the precise form disclosed, and various modifications and variations are possible in light of the above teachings. It is intended that the application embrace all alternatives, modifications and variations as can befall within the scope of the description and claims. It is intended that the scope of the application be defined by the claims appended hereto, properly interpreted in accordance with the doctrine of equivalents.

Claims

1. A method of determining the importance of a web page, characterized by, Comprise: Convert the web page and the relationship between the web pages into a graph structure, wherein the graph structure G includes n nodes k corresponding to the web pages and a plurality of edges corresponding to the relationship between the web pages, and the target web page corresponds to the target node t in the graph structure; obtaining an estimate of the probability mass received from the target node t and eventually left at node k after the probability mass propagation process is completed wherein the node k is any node in the graph structure; Update the node k to another node of the n nodes in the graph structure; Repeat the step of obtaining the estimation of the probability amount received by the node k from the target node t and finally remaining in the node k in the graph structure until the estimation of the probability amount received by all nodes k from the target node t and finally remaining in the node k in the graph structure is obtained; Based on the estimate of the amount of probability received by all the nodes k in the graph from the node t and eventually left at the node k, the estimate of the PageRank score of the target node t is obtained based on the estimated PageRank score of the target node t determining the importance of the target webpage corresponding to the target node t; wherein the estimate of the PageRank score of node t is obtained from the estimate of the probability mass received by all nodes k in the graph structure from node t during the propagation of the probability mass and ultimately remaining at node k comprising: According to Formula One The Formula One is: where d t is the number of neighbors of target node t on the graph structure, d k is the number of neighbors of node k on the graph structure.

2. The determination method according to claim 1, characterized in that, said acquisition of an estimate of the probability mass received from the target node t and finally left at the node k after the probability mass propagation process is completed before, also comprising Step S101, obtain all nodes receiving propagation probability at time i-1, wherein i is an integer in the range of [1, L] and the initial value of i is 1, and L is the farthest step length of probability amount propagation; Step S102, select one of all nodes receiving propagation probability at time i-1 as a sampling node u at time i; Step S103, obtain the sampling node u at time i and the propagation probability received by the sampling node u at time i from the sampling node at time i-1 at time i-1; Step S104, according to the preset condition and the propagation probability received by the sampling node u at time i from the sampling node at time i-1 at time i-1, all neighbor nodes v or part of the nodes w in all neighbor nodes v of the sampling node u at time i are selected as the sampling node at time i+1, and the propagation probability of the sampling node u at time i transmitted to the sampling node at time i+1 is calculated; Step S105, update the estimation of the probability amount received by the sampling node at time i+1 from the sampling node at time i and finally remaining in the sampling node at time i+1 in the process of probability amount propagation according to the probability of the sampling node u at time i transmitted to the sampling node at time i+1; Step S106, select another node in all nodes receiving propagation probability at time i-1 as a sampling node at time i; repeat steps S103-S105 until each node receiving propagation probability at time i-1 is selected as a sampling node at time i, and update the current time i to i+1; Step S107, repeat the steps of steps S101-S106 until the current time i is updated to L and then stop.

3. The determination method according to claim 2, characterized in that, The preset condition includes a first preset condition, and the propagation probability of the sampling node u at time i transmitted to the sampling node at time i+1 is calculated according to the preset condition and the propagation probability received by the sampling node u at time i from the sampling node at time i-1 at time i-1, all neighbor nodes v or part of the nodes w in all neighbor nodes v of the sampling node u at time i are selected as the sampling node at time i+1, and the propagation probability of the sampling node u at time i transmitted to the sampling node at time i+1 is calculated. If the sampling node u at time i satisfies the first preset condition, all neighbor nodes v of the sampling node u at time i are taken as the sampling nodes at time i+1, wherein the first preset condition is: (1-α)· wherein θ is an estimation error parameter specified according to an actual application scenario, and the value range is θ∈(0,1). Probability of a sampling node at time i to pass to a sampling node at time i+1 The calculation is made by Equation Two, which is: wherein, is the propagation probability from the sampling node at time i - 1 to the sampling node u at time i - 1, d u is the number of neighbors of the sampling node u on the graph structure.

4. The determination method according to claim 3, characterized in that, The preset condition comprises a first preset condition and a second preset condition, and the preset condition and the propagation probability of the sampling node u at the moment i received from the sampling node at the moment i-1 at the moment i-1 are used to take all neighbor nodes v of the sampling node u at the moment i or part of the nodes w in all neighbor nodes v as the sampling node at the moment i+1, and the propagation probability of the sampling node u at the moment i transmitted to the sampling node at the moment i+1 is calculated. If the sampling node i at time i does not satisfy the first preset condition but satisfies the second preset condition, a neighbor node w is sampled from all neighbor nodes v of the sampling node i at time i as the sampling node at time i+1, and the second preset condition is that: Probability of a sampling node u at time i to pass to a sampling node at time i+1 The calculation is made by equation three, which is:

5. The determination method according to claim 3, characterized in that, The estimation of the probability amount of the sampling node at the moment i+1 received from the sampling node at the moment i and finally remaining in the sampling node at the moment i+1 in the probability amount propagation process is updated according to the probability of the sampling node u at the moment i transmitted to the sampling node at the moment i+1. An estimate of the probability mass received by a sample node at time i+1 from sample nodes at time i, and ultimately remaining with the sample node at time i+1, during the propagation of probability mass, according to Equation Four An update is made, according to Equation Four, which is: The initial value before the probability quantity propagation process starts is 0; The initial value before the probability quantity propagation process starts is α.

6. The determination method according to claim 4, characterized in that, The estimation of the probability amount of the sampling node at the moment i+1 received from the sampling node at the moment i and finally remaining in the sampling node at the moment i+1 in the probability amount propagation process is updated according to the probability of the sampling node i at the moment i transmitted to the sampling node at the moment i+1. According to Equation Five, the estimate π of the probability mass received by the sampling nodes at time i+1 from the sampling nodes at time i, and ultimately remaining with the sampling nodes at time i+1, during the propagation of the probability mass is t (w) performing an update, said Equation Five being: The initial value before the probability quantity propagation process starts is 0; The initial value before the probability quantity propagation process starts is α.

7. The determination method according to claim 4, characterized in that, The neighbor node w is sampled from all neighbor nodes v of the node u, comprising: A neighbor set N(u) storing all neighbor nodes v of the sampling node u at the current moment is established; A node in the neighbor set N(u) is selected as a node to be sampled; The step of selecting a node in the list as a node to be sampled is repeated until the preset condition is met; All the selected nodes to be sampled are taken as the sampled neighbor nodes w.

8. The determination method according to claim 7, characterized in that, The node in the neighbor set N(u) is selected as the node to be sampled, comprising: The position of the to-be-sampled neighbor node of the sampling node u at time i in the neighbor set N(u) of the node u is determined according to Formula Six wherein, Formula six is: where rgis a random number subject to a geometric distribution with parameter p. the position of the node to be sampled according to the sampling node u at time i in the neighbor set N(u) of node u selecting a node to be sampled; Wherein, before the sampling starts The preset condition is 9. The determination method according to claim 8, characterized in that, The random numbers of the geometric distribution are generated by the following steps: A uniform random number rand in the range (0, 1) is generated, and a random number rg following a geometric distribution is calculated according to Equation Seven : rand = -ln(1 - rg) wherein, is the probability that a sampling node at time i passes on the quantity of probability θ to a sampling node at time i + 1,

Images