WO2013044559A1

WO2013044559A1 - Method and system for recommending website and network server

Info

Publication number: WO2013044559A1
Application number: PCT/CN2011/083678
Authority: WO
Inventors: 吴军; 王欣; 金键
Original assignee: 中国科学院计算机网络信息中心
Priority date: 2011-09-26
Filing date: 2011-12-08
Publication date: 2013-04-04
Also published as: CN102316166B; CN102316166A

Abstract

The present invention provides a method and a system for recommending a website and a network server. The method comprises: a network server acquiring characteristic information corresponding to websites accessed by a user separately in a plurality of preset time intervals according to locally-stored Internet access information; performing cluster analysis on the websites according to the characteristic information to acquire a plurality of website clusters; and if it is determined that the websites comprise a first website corresponding to a website address when a network access request comprising a website address and sent by a user terminal is received, determining a website to be recommended to the user according to the characteristic information corresponding to the websites in a website cluster where the first website is positioned, embedding the website address of the recommended website into a network access response, and returning the network access response to the user terminal. The network server can be used for recommending more websites to the network access user based on user network access behaviors corresponding to the websites, thereby enabling the user to acquire more information of interests.

Description

Website recommendation method and system and web server

TECHNICAL FIELD The present invention relates to communication technologies, and in particular, to a website recommendation method and system, and a network server. Background technique

With the development of electronic information technology, the Internet has changed people's lifestyles. For example, people can use the Internet to get books, movies, music, and even products that they are interested in. Therefore, the Internet has brought us efficient and convenient life. People have become accustomed to using computers, mobile phones and other Internet-enabled devices to learn, entertain, and shop by browsing the web pages they are interested in to meet their multi-faceted needs.

People can use the network to obtain rich information for learning and entertainment more efficiently. Specifically, the web server will recommend the same type of related website to the user for reference according to the type of website visited by the user, for example, the user accesses information technology. For a type of website, the web server will recommend other websites in the information technology type for users to refer to; the web server stores the type of website that the user frequently visits and obtains related website recommendations to the user, so that the user can obtain more interested parties. News.

However, the network server in the prior art only obtains the relevant website recommendation to the user for reference according to the type of the website accessed by the user, so that the information obtained by the user is limited, and has certain limitations. Summary of the invention

In view of the above-mentioned deficiencies of the prior art, embodiments of the present invention provide a website recommendation method and system, and a network server.

An embodiment of the present invention provides a website recommendation method, including:

The network server obtains the feature information corresponding to the website accessed by the user according to the locally stored Internet access information in a preset plurality of time periods;

The network server performs cluster analysis on the website according to the feature information to obtain more a website cluster, in order to determine whether the website includes a first website corresponding to the website address when receiving a network access request including a web address sent by the user terminal, and if so, according to the website group of the website group where the first website is located The corresponding feature information determines a website recommended to the user, and embeds the website address of the recommended website into the network access response and returns it to the user terminal.

An embodiment of the present invention provides a network server, including:

The first obtaining module is configured to obtain feature information corresponding to the website accessed by the user according to the locally stored Internet access information in a preset plurality of time periods;

a second acquiring module, configured to perform cluster analysis on the website according to the feature information to obtain a plurality of website clusters;

a determining module, configured to: when receiving a network access request that includes a web address sent by the user terminal, determining whether the website includes a first website corresponding to the web address;

a processing module, configured to: if it is determined that the website includes a first website corresponding to the website address, determine a website recommended to the user according to the feature information corresponding to the website in the website cluster where the first website is located, and The URL of the recommended website is embedded in the network access response and returned to the user terminal.

The embodiment of the invention provides a website recommendation system, which comprises the above network server and user terminal.

The website recommendation method and system and the network server provided by the embodiment of the present invention obtain the feature information corresponding to the website visited by the user in the preset multiple time periods according to the locally stored Internet access information, according to the feature information. The website performs cluster analysis to obtain a plurality of website clusters. When receiving the network access request including the website sent by the user terminal, it is determined whether the website analyzed by the cluster includes the first website corresponding to the website address. If yes, the website recommended by the user is determined according to the feature information corresponding to the website in the website cluster where the first website is located, and the website address of the recommended website is embedded in the network access response and returned to the user terminal, thereby realizing more network services. The website, so that users can get more information of interest. BRIEF DESCRIPTION OF DRAWINGS FIG. 1 is a flowchart of an embodiment of a website recommendation method according to the present invention;

2 is a flowchart of another embodiment of a website recommendation method according to the present invention; 3 is a schematic structural diagram of an embodiment of a network server according to the present invention;

FIG. 4 is a schematic structural diagram of an embodiment of a website recommendation system according to the present invention. detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is a partial embodiment of the invention, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

FIG. 1 is a flowchart of an embodiment of a website recommendation method according to the present invention. As shown in FIG. 1, the method includes:

Step 100: The network server obtains feature information corresponding to the website accessed by the user according to the locally stored online information in a preset plurality of time periods;

The user can send a network access request to the network server for network access through a user terminal having a network function such as a mobile phone or a computer, and the network server can store the online information of the user who accesses the network for a period of time according to a preset refresh time. It can be understood that the refresh time of the web server in this embodiment is set according to specific application requirements, for example, three days or one week. The online information of the user stored by the network server specifically includes: an IP address of the user terminal, a website visited each time, and a corresponding start time and end time.

The network server obtains the feature information corresponding to the website accessed by the user according to the locally stored Internet access information in a preset plurality of time periods. It should be noted that the feature information in this embodiment reflects the website visited by the user. The behavior information of the corresponding user accessing the website in the preset different time period, the characteristic information may specifically include at least one of a frequency feature, a variance feature, and an entropy feature that the website is accessed by the user in each preset time period. Characteristics, wherein the frequency characteristics reflect the frequency of the website being accessed by the user in each preset time period; the variance characteristic reflects the variance of the number of times the website is accessed by the user in each preset time period, and is used to measure the website The degree of change in the number of times the user is accessed during each preset time period; the entropy feature reflects the entropy of the IP address of the user accessed by the website during each preset time period, which is used to measure the stability of the user of the website, for example It is said that during the period from 8:00 am to 10:00, the website A has been visited 5 times, IP1 has visited 1 time, and IP2 has visited 3 times. IP3 visited once, the user's IP address The entropy is: -((l/5)log(l/5)+(3/5)log(3/5)+(l/5)log(l/5)).

The multiple time periods preset in this embodiment may be preset in the network server according to specific application situations. For example, if multiple preset time periods are 8:00 ~ 10:00, 10:00 every day. ~ 12:00 , 18:00 - 21 :00 ^ 21 : 00 ~ 24:00 , that is , the web server collects statistics according to the stored Internet information in each set time period and is accessed by the user in each time period . Characteristic information corresponding to each website. In a specific implementation process, the network server performs analog-to-digital conversion on at least one of the acquired feature information, such as: a frequency feature, a variance feature, and an entropy feature, or weights a digital quantity of several of the features to obtain a corresponding Feature information. It should be noted that the feature information in this embodiment is not limited to the above-mentioned several features, and may be adjusted according to the obtained specific Internet information to obtain other feature information. The specific processing process is as above, and is no longer Narration. In order to explain the meaning of the feature information more clearly, as shown in Table 1, Table 1 shows the feature information corresponding to the website visited by the user in a preset plurality of time periods, and the feature information is at each preset. The value obtained by performing analog-to-digital conversion weighting on the frequency, variance, and entropy features of each website visited by the user during the time period.

Table 1

Step 101: The network server performs cluster analysis on the website according to the feature information to obtain a plurality of website clusters.

The network server performs cluster analysis on all websites to obtain a plurality of website clusters according to the feature information corresponding to the website accessed by the user acquired in a plurality of preset time periods. Cluster analysis

(Cluster Analysis) Also known as group analysis, it is a process of classifying data into different classes or clusters, so objects in the same cluster have great similarities, and objects between different clusters have Great difference. The calculation methods of cluster analysis mainly include partitioning methods, hierarchical methods, density-based methods, grid-based methods and model-based methods. Model-based methods). The specific implementation process of each clustering method belongs to the prior art, in order to more clearly illustrate the process of clustering analysis, based on the K-means in the splitting method and the probabilistic latent semantic model in the model-based method (Probabilistic Latent Semantic Analysis, PLSA) is given as an example for specific explanation, and the rest of the clustering methods are no longer described.

1 : The algorithm for introducing K-means is as follows:

Step (1): When the website cluster k=2 is taken as an example, two websites are randomly selected from the website A to the website Z as the initial center of mass (the center of the category), assuming that the website A and the website B are selected; step (2): For each of the remaining websites, calculate the distance to each centroid based on the feature information, compare the distance between each website to website A and website B, and select the distance to divide into a class; the rest of the website, and so on, In the end, all the websites can be divided into two classes with the center of website A and website B as the center of mass;

Step (3): Recalculate the centroid for each class. The calculation method is to average the weights of each website. After calculating the new centroid of each class, calculate the distance to each centroid for all websites, and so on. Until the center of mass no longer changes.

Step (4): For each class, calculate the mean square error within the class, that is, the distance from all the sites in the class to the centroid, and compare their mean square error, the trend should be gradually reduced, when the mean square error value drops significantly to no Then the significantly degraded K value can be used as the final K, which is the number of website clusters.

2: The algorithm for introducing the PLSA model is as follows:

Step (1): Define a likelihood function L(D, T) = f _j f _j w; log P(d _i , t _J ) ; Step (2): Determine the parameter value by using an iterative calculation of the EM algorithm, and the EM algorithm The specific steps are as follows:

Step E uses the old parameters to calculate the posterior probability of the latent factor variable. The formula is as follows:

The M step obtains a new parameter by maximizing the expected function of the likelihood function, and the formula is as follows: Q : £ί i ( lc.g ^p ( ^j ' ) Step (3 ) : The formula for updating each parameter during the maximization process is as follows:

Step (4): Repeat the above E step and always monotonically increase during this process. When the maximum value is reached, the parameter value is determined and the update process is stopped.

Step 102: Determine whether the website includes a first website corresponding to the website address, and if yes, determine a website recommended to the user according to the feature information corresponding to the website in the website cluster where the first website is located, and The URL of the website is embedded into the network access response and returned to the user terminal.

When receiving the network access request including the web address sent by the user terminal, the web server queries the clustered website according to the web address to determine whether the first website corresponding to the web address is included. If it is determined that the first website is included in the website that is clustered, the first website is also subjected to cluster analysis, and the website cluster obtained in step 101 is queried according to the website and the website cluster where the first website is located is determined. The performance URL of the website visited by the user, wherein the website address includes a domain name or an IP address, and the domain name and the IP address can be converted by the domain name server to determine the website visited by the user. Based on the above, it can be known that the websites in the website cluster have similarities based on the user's access behavior to the website. Obtaining feature information corresponding to the website in the website cluster where the first website is located, and determining the website recommended to the user according to the set recommendation rule according to the feature information. For example, the first website may be removed from the website cluster where the first website is located. A certain number of websites are randomly selected and recommended to the user, because the user access behavior corresponding to the first website accessed by the user is similar to the user access behavior corresponding to the remaining websites in the website cluster, and the embodiment may be based on the website pair. The user access behavior should recommend to the user a website that the user may be interested in. It should be noted that the recommendation rule is specifically set according to a specific application scenario, and the specific recommendation rule is not limited in this embodiment.

The web server will embed the URL of the website recommended by the user into the network access response and return it to the user terminal. The website address includes a domain name and/or an IP address. If the website address of the website recommended by the network server is determined to be an IP address according to the online information, the network server can directly embed the IP address into the network access response and return it to the user. The terminal may also send a domain name anti-query request including an IP address to the domain name server. The domain name server returns the domain name corresponding to the IP address to the network server through the PTR type domain name resolution, and the network server embeds the IP address of the website and the corresponding domain name into the terminal. The network access response is returned to the user terminal for reference by the user, and the domain name is returned to the user terminal, which is convenient for the user to memorize and write, thereby making the user more convenient to retrieve and access the recommended website. If it is determined according to the online information that the website URL recommended by the user is a domain name, the network server may directly embed the domain name into the network access response and return it to the user terminal, or may send a domain name query request including the domain name to the domain name server, and the domain name server passes the A. The type domain name resolution returns an IP address corresponding to the domain name to the network server, and the network server embeds the IP address of the website and the corresponding domain name into the network access response, returns the user terminal for reference, and returns an IP address to the user terminal, thereby To enable users to search and access the recommended website more directly, there is no need to initiate a domain name query request to the domain name server.

The website recommendation method provided by the embodiment obtains the feature information corresponding to the website visited by the user in the preset multiple time periods according to the locally stored Internet information, and performs cluster analysis on the website according to the feature information. When receiving the network access request including the web address sent by the user terminal, the plurality of website clusters determine whether the website analyzed by the cluster includes the first website corresponding to the website address. If yes, the website recommended by the user is determined according to the feature information corresponding to the website in the website cluster where the first website is located, and the website address of the recommended website is embedded in the network access response and returned to the user terminal, so that the web server can be based on the website. The corresponding user network access behavior recommends more websites to users who make network access, so that users can obtain more information of interest.

2 is a flowchart of another embodiment of a website recommendation method according to the present invention. As shown in FIG. 2, the party The law includes:

Step 200: The network server obtains feature information corresponding to the website accessed by the user according to the locally stored online information in a preset plurality of time periods;

Step 201: The network server performs cluster analysis on the website according to the feature information to obtain a plurality of website clusters.

Step 202: The network server, when receiving a network access request that includes a web address sent by the user terminal, determines whether the website includes a first website corresponding to the website address, and if not, broadcasts the website address to the remaining network servers. And the online information query request of the plurality of time periods, if the online information of the first website returned by the remaining network servers in the multiple time periods is received, acquiring the first information according to the online information The characteristic letter corresponding to a website

When receiving the network access request including the web address sent by the user terminal, the network server queries the clustered website according to the web address to determine whether the first website corresponding to the web address is included. If it is determined that the first website is not included in the website that is clustered, the first website is not accessed by the user through the web server in each preset time period, that is, the user is in each preset time period. The website accessed through the web server does not include the first website.

The network server broadcasts the website address including the first website and the online information query request of the preset time period to the remaining network servers in the Internet system, and the remaining network servers query the request according to the received online information, and each network server is based on the first website. The web address is queried from the online information stored in the preset time period of the local storage to include the online information of the first website, and if the web server can receive the first website returned by the remaining web servers in each preset time period The information of the Internet access is obtained according to the information about the Internet access of the first website. For the process of obtaining the specific feature information, refer to step 100 in the first embodiment, and details are not described herein.

Step 203: The network server acquires corresponding aggregation profile information according to the feature information corresponding to the website in each website cluster, and determines the first part by using a similarity measure according to the feature information corresponding to the first website and the aggregated contour information. a cluster of websites to which a website belongs;

The network server obtains the corresponding aggregated contour information according to the feature information of the website in each website cluster obtained in the above step 201, and the aggregated contour information, that is, the average weight of the feature information corresponding to the website in each website cluster; The network server performs the similarity measurement according to the feature information of the first website and the acquired aggregated contour information. It is noted that the method of the similarity measure is, for example, a Pearson correlation coefficient or a cosine coefficient, and the like, which is not specifically limited in this embodiment. Obtaining a matching score of the feature information corresponding to the first website and each of the aggregated contour information by using the similarity measure to determine a website cluster to which the first website belongs, and the matching score is larger, indicating that the similarity between the first website and the website in the website cluster is more High, the site cluster that selects the maximum matching score is determined to be the cluster of sites to which the first site belongs.

In step 204, the website recommended by the user is determined according to the feature information corresponding to the website in the website cluster where the first website is located, and the website address of the recommended website is embedded in the network access response and returned to the user terminal.

Obtaining feature information of the website in the cluster of the website where the first website is located, and determining the website recommended to the user according to the set recommendation rule according to the feature information, specifically, the feature information of the remaining websites in the website cluster where the first website is located Performing a weighted average to obtain the recommended scores of the remaining websites, and determining the recommended websites to the users according to the recommended recommendation criteria of each of the remaining websites, for example, according to the recommendation scores of each website, from the high to the low until the preset As far as the number of recommended websites is concerned, the selected website will be used as a website recommended to users. The network server embeds the URL of the recommended website into the network access response and returns it to the user terminal for the user to refer to. For the specific process, refer to the foregoing embodiment, and details are not described herein.

For the specific implementation process of step 201 and step 202 in this embodiment, refer to the embodiment shown in FIG. 1, and details are not described herein again.

The website recommendation method provided by the embodiment obtains the feature information corresponding to the website visited by the user in the preset multiple time periods according to the locally stored Internet information, and performs cluster analysis on the website according to the feature information. When receiving a network access request including a web address sent by the user terminal, if it is determined that the website that is clustered does not include the first website corresponding to the web address, the website cluster performs a broadcast query to the remaining web servers, if received The network information of the first website returned by the remaining web servers determines the website cluster where the first website is located, and determines the website recommended to the user according to the characteristic information of the website in the website cluster where the first website is located, and the website URL of the recommended website Embedded in the network access response is returned to the user terminal, enabling the user to recommend more websites, so that the user can obtain more information of interest.

One of ordinary skill in the art can understand that all or part of the steps of the above method embodiments are implemented. The foregoing program may be stored in a computer readable storage medium, and the program is executed to perform the steps including the foregoing method embodiments; and the foregoing storage medium includes: a ROM, A variety of media that can store program code, such as RAM, disk, or optical disk.

3 is a schematic structural diagram of an embodiment of a network server according to the present invention. As shown in FIG. 3, the network server includes: a first obtaining module 11, a second obtaining module 12, a determining module 13, and a processing module 14, wherein the first obtaining The module 1 is configured to acquire the feature information corresponding to the website accessed by the user according to the locally stored Internet access information in a preset plurality of time periods. The second obtaining module 12 is configured to perform cluster analysis on the website according to the feature information to obtain multiple The determining module 13 is configured to determine whether the website includes the first website corresponding to the website address when receiving the network access request that is sent by the user terminal, and the processing module 14 is configured to: when determining that the website includes the first website corresponding to the website address The website determines the website recommended to the user according to the feature information corresponding to the website in the website cluster where the first website is located, and embeds the website address of the recommended website into the network access response and returns it to the user terminal.

For the embodiment shown in FIG. 3, the second obtaining module 12 may perform cluster analysis on the website according to the feature information by a splitting method, a hierarchical method, a density-based method, a grid-based method, and a model-based method.

For the functions and processing procedures of the modules in the network server provided in this embodiment, refer to the method embodiment shown in FIG. 1 , and the implementation principle and technical effects are similar, and details are not described herein again.

Based on the embodiment shown in FIG. 3, the processing module 14 is further configured to: if it is determined that the website does not include the first website corresponding to the website address, broadcast the online information query request including the website address and the plurality of time periods to the remaining network servers. And if the first website returned by the remaining web servers receives the online information in the multiple time periods, the feature information corresponding to the first website is obtained according to the online information; and the corresponding aggregation is obtained according to the feature information corresponding to the website in each website cluster. And contour information, and determining, by the similarity measure, the website cluster to which the first website belongs according to the feature information and the aggregated contour information corresponding to the first website.

For the functions and processing procedures of the modules in the network server provided in this embodiment, refer to the method embodiment shown in FIG. 2, and the implementation principle and technical effects are similar, and details are not described herein again.

4 is a schematic structural diagram of an embodiment of a website recommendation system according to the present invention. As shown in FIG. 4, the system includes: a network server 1 and a user terminal 2, wherein the network server 1 can be The network server provided by the embodiment of the present invention, the user terminal 2 is the user terminal involved in the embodiment of the present invention, and the functions and processing procedures of the devices in the website recommendation system provided by this embodiment may be referred to the foregoing method and apparatus embodiment. The implementation principle and technical effect are similar, and will not be described here.

It should be noted that the above embodiments are only for explaining the technical solutions of the present invention, and are not intended to be limiting; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those skilled in the art that: The technical solutions described in the foregoing embodiments are modified, or some of the technical features are equivalently replaced. The modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

Claim

A website recommendation method, comprising:

The network server performs cluster analysis on the website according to the feature information to obtain a plurality of website clusters, so as to determine, when receiving a network access request including a web address sent by the user terminal, whether the website includes a URL corresponding to the website address. a first website, if yes, determining a website recommended to the user according to the feature information corresponding to the website in the website cluster where the first website is located, and embedding the website address of the recommended website into the network access response and returning to the user terminal.

The website recommendation method according to claim 1, wherein the method further comprises: if it is determined that the website does not include the first website corresponding to the website address, the method further comprises:

The network server broadcasts an online information query request including the website address and the plurality of time periods to the remaining network servers, if the first website returned by the remaining network servers is received within the plurality of time periods The online information is obtained, and the feature information corresponding to the first website is obtained according to the online information;

And the network server obtains corresponding aggregation profile information according to the feature information corresponding to the website in each website cluster, and determines, according to the feature information corresponding to the first website and the aggregated contour information, the first website belongs by using a similarity measure Site cluster.

The website recommendation method according to claim 1, wherein the clustering analysis of the website according to the feature information comprises:

The website is clustered and analyzed according to the feature information by a split method, a hierarchical method, a density-based method, a grid-based method, and a model-based method.

The website recommendation method according to claim 1, wherein the determining the website recommended to the user according to the feature information corresponding to the website in the website cluster where the first website is located comprises: according to the first website The feature information corresponding to the remaining websites in the website cluster obtains the recommended scores of the remaining websites;

A website recommended to the user is determined according to the recommended recommendation criteria according to the recommended score.

The website recommendation method according to any one of claims 1 to 4, wherein the feature information comprises: a frequency feature, a variance feature, and an entropy feature of the website accessed by the user in each preset time period. At least one feature.

The website recommendation method according to any one of claims 1 to 4, wherein the website address comprises: a domain name and/or an IP address.

7. A network server, comprising:

The network server according to claim 7, wherein the processing module is further configured to:

If it is determined that the website does not include the first website corresponding to the website address, the network information query request including the website address and the multiple time periods is broadcasted to the remaining network servers, and if the remaining network server is received, The online information of the first website in the multiple time periods is obtained, and the feature information corresponding to the first website is obtained according to the online information;

Corresponding aggregated profile information is obtained according to the feature information corresponding to the website in each website cluster, and the website cluster to which the first website belongs is determined by the similarity measure according to the feature information corresponding to the first website and the aggregated outline information.

The network server according to claim 8, wherein the second obtaining module is specifically configured to:

A website recommendation system, comprising: the network server according to any one of claims 7 or 8 or 9, and a user terminal.