KR20010109744A - Ranking appraisement method for website - Google Patents

Abstract

The present invention relates to a website ranking method. In order to evaluate the ranking of the website, the present invention employs a statistical method to evaluate not only the various items related to the access of the website but also these items in aggregate. In gathering information for evaluation, web surfing related information, for example, access, is simultaneously performed from an evaluation member (30,30 ') who actively surfs the web and a registered member (40,40') that is the target of web surfing. You will be provided with the number, dwell time, page views, connection speed and reliability. On the other hand, in evaluating the rankings of the websites 47, 50 and 50 ', a sample group is extracted from the evaluation members 30 and 30' by using stratification extraction and allocation extraction, and for each item, the non-member websites 50 , 50 ') is extracted by the probability proportional sampling method, and the ranking of each website (47, 50, 50') is used for each item. In addition, the value of each item is synthesized using principal component analysis to evaluate the ranking of each website (47, 50, 50 '). According to the present invention, it is possible to more accurately rank the respective websites 47, 50, and 50 'present.

Description

Ranking appraisement method for website}

The present invention relates to a website ranking method, and more particularly, web surfing related information delivered from an evaluation member who actively surfs a website and a registered member which is a web surfing target in a statistical method. How to evaluate your website's ranking.

There are no institutions on the Internet that reliably evaluate websites around the world, and there are websites that rank ranking preferences for websites, but simply re-extend existing related website links based on access. As a result, there is no clear evaluation based on statistical reliability. In other words, it is not meaningful to survey the ranking of websites based on simple number of visits to unspecified Internet users.

In general, large Internet web sites publish their own statistics based on the number of Internet users' access and pageviews over a certain period of time, but this is because they provide information about their own web sites. have. In particular, considering that the pricing of advertisements on the Internet is mainly based on the number of members and the number of connections, it is very meaningful to investigate the preferences of websites in advertisements on the Internet.

There are two main methods of evaluating website rankings. One of them is to analyze the log file of a website and evaluate the website based only on the number of visits to the website. However, since this method simply evaluates the ranking of websites based on the number of visitors without any information about the visitors, it is not really valuable as a material for more diversified marketing, and a log file must be provided on the website. There is a problem that the log files provided must be analyzed.

The other is to install log transfer software on the sample users' computers and use the web surfing information of the transferred users. Since this method does not know the exact access information of the site, but depends only on the statistically extracted sample, there is a problem that results vary considerably for each research institute according to the sampling method.

Accordingly, an object of the present invention is to solve the problems of the prior art as described above, and to provide a more accurate website ranking according to various evaluation criteria.

1 is a block diagram showing the configuration of a system for implementing a website ranking method according to the present invention.

Figure 2 is a flow chart illustrating a process of registering as an evaluation member and a registered member in the embodiment of the present invention.

3 is a flow chart showing that the evaluation member in the embodiment of the present invention to obtain the web surfing-related information while accessing the web to access the Internet.

Figure 4 is a flow chart showing an evaluation member hooking web surfing-related information in the embodiment of the present invention and exchange mutual information between the evaluation site.

5 is an explanatory diagram for explaining a discriminant analysis method for determining the characteristics of a website in an embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

10: Evaluation Site 12: Server

14: Member Information Database 15: Capture Information Database

16: Processing Information Database 20: Information Requester

30,30 ': Evaluation member 32: Information provider

34: User Program 35: Personal Database

40,40 ': Registered Member 41: Server

42: Information Provider 47: Website

50,50 ': Website

According to a feature of the present invention for achieving the object as described above, the present invention comprises the steps of registering an evaluation member actively web surfing on the network and a registered member that is the target of web surfing on the server, Web surfing related information The server provides the information provider to obtain and transmit the information to install in the system of the evaluation member and registered member, and the server delivers web surfing-related information through the information provider of the evaluation member and registered member connected to the network And receiving the web surfing related information received by the server in a database, and processing the databased web surfing related information in a statistical method at the server to determine the ranking of the web site. It is composed.

The web surfing-related information includes the number of accesses, residence time, page view, access speed and website reliability of each website of the evaluation member, and the number and access times of users accessing the website of the registered member. , Page views and connection speeds.

The determining of the ranking of the website may include extracting a predetermined sample group from the evaluation member using a stratification extraction method or an allocation extraction method for the website classified by the discriminant analysis method, and the probability proportionality from the sample group. And extracting web surfing related information such as the number of visits, residence time, page view, etc. of the non-member website using sampling method.

And analyzing the extracted web surfing related information by a principal component analysis method to determine the ranking of the web site.

Based on the number of evaluation members corresponding to a predetermined sample group, the server extracts the total number of Internet users corresponding to a predetermined sample group accessing a specific site by using a capture-recapture extraction method, and determines the tendency of the Internet user. It is done.

In the present invention, the ranking is evaluated by classifying the characteristics of the website by discriminant analysis.

According to the present invention as described above there is an advantage that it is possible to more accurately evaluate the ranking for each specific item for each website.

Hereinafter, a preferred embodiment of the website ranking method according to the present invention as described above will be described in detail with reference to the accompanying drawings.

1 shows a system for implementing a website ranking method according to the present invention. As shown in the drawing, the evaluation site 10, evaluation members 30 and 30 ', registered members 40 and 40', and non-member websites 50 and 50 'are connected to the Internet 1, which is a network. .

The evaluation site 10 is provided with a server 12, which is connected to the member information database 14, the capture information database 15 and the processing information database 16 and the like. The server 12 exchanges and processes various information with these connected databases.

The evaluation members 30 and 30 'are individual users who use the Internet 1 and do not have a separate website. The system of the evaluation members 30, 30 'is provided with a web browser 31 for surfing the Internet 1, and the information provider 32 is provided within or separately from the web browser 31 itself. The information provider 32 sends and receives various information with the web browser 31 and the OS of the system, and hooks the web surfing related information to the server 12 of the evaluation site 10. to be. Here, the web surfing related information includes the number of connections, residence time, page view, connection speed, and reliability of the specific website 47, 50, 50 '. In addition, the information provider 32 hooks all the messages (for example, a keyboard message, a mouse message, a timer message, etc.) transmitted and received from the Windows and the web browser 31 as the OS and the cookies of the web browser 31. Can also be sent. The information provider 32 also has a function of displaying in real time the information provided by the server 11 of the evaluation site 10.

The information provider 32 as described above is preferably installed in the web browser 31 in the form of a panel so as to provide the necessary information. The information provider 32 may be provided in an iconized state on the web browser 31.

Meanwhile, a user program 34 provided by the evaluation site 10 may be installed in the system of the evaluation members 30 and 30 '. The user program 34 provides useful functions (for example, wall paper management and firewall management) for the purpose of attracting the evaluation members 30 and 30 '. The user program 34 may have an information collection function similar to that of the information provider 32 to transmit the collected information to the information provider 32, or may be integrated with the information provider 32. .

In addition, the system of the evaluation members 30 and 30 'is provided with a personal database 35 to store various information of the evaluation members 30 and 30'. For example, personal information of the evaluation members 30 and 30 'or an identifier for identifying the evaluation members 30 and 30' may be stored.

The registered members 40 and 40 'are members who have a website 47, for example, a hub site, a portal site, or a site providing various contents. The server 41 controlling the website 47 of the registered members 40 and 40 'is provided with an information provider 42. The information provider 42 is responsible for delivering information related to the users accessing the website 47 to the server 11 of the evaluation site 10. Meanwhile, the evaluation members 30 and 30 'may be registered members 40 and 40' at the same time.

In FIG. 1, reference numeral 20 denotes an information requester who wants to use the information evaluated in the evaluation site 10, and 50 and 50 'are various websites connected to the Internet 1, in particular, as a member of the evaluation site 10. Refers to a website that is not subscribed.

When the website ranking method according to the invention in the above configuration is briefly described as follows.

First, a web obtained by recruiting evaluation members 30, 30 'or registered members 40, 40' and actively surfing websites 47, 50, 50 '. Information related to the registered member (40, 40 ') by surfing-related information and the general user or evaluation member (30, 30') accesses the registered member (40, 40 ')' s website (47). The information that can be obtained is collected and temporarily stored in the capture information database 15, and among them, the statistical value is collected and the server 12 processes and stores the processed information database 16.

Here, various statistical methods are performed in processing the information stored in the capture information database 15.

First, a sample group is extracted from the evaluation members 30 and 30 ', and the stratification extraction method or the allocation extraction method is used here. That is, the specific information is extracted from the member information database 14 by referring to the data about the evaluation members 30 and 30 ', and for example, a specific sample group such as a Korean 20s man can be extracted.

In addition, the web surfing-related information collected by the sample group extracted as described above, for example, information such as the number of times, residence time, page view, etc. for a specific website (47, 50, 50 ') is selected, and the probability By using the proportional sampling method, the actual surfing related information about the websites 50 and 50 'is estimated. In this case, not all of the web surfing related information collected by the sample group and stored in the capture information database is used, but only statistically meaningful ones are used.

Using the estimated web surfing-related information as described above, the ranking for each item, for example, the number of accesses, residence time, and page views, is determined for each website 50, 50 '.

In determining the ranking of a website, it is not a data that is suitable for a specific purpose desired by determining with one element, for example, the number of connections. Therefore, the rank of each website 47, 50, 50 'is determined by using Principal Component Analysis among the multivariate analysis.

Now, the ranking of the website through the above process will be described in more detail.

Referring to FIG. 2, the evaluation members 30 and 30 ′ and the registered members 40 and 40 ′ are registered, and the information providers 32 and 42 and the user program 34 are described. First, when accessing the evaluation site 10 through the Internet (step 101), the terms and conditions relating to membership. The basics are presented. (Step 102) The mobile terminal is subjected to the step 103 of understanding the terms of the agreement and determining whether they agree with them.

If the terms and conditions are not agreed, the access is terminated (step 104). If the terms and conditions are agreed to, the user is provided with a window for inputting a predetermined ID and password to confirm whether the user is a member (step 105).

If you are a member, you can download the latest information provider (32, 42) or user program (34), and update your membership information if necessary. (Step 106) If you are not a member, enter your personal details in the prescribed form. And (step 107), the evaluation member (30, 30 ') (step 108) or registered members (40, 40') (step 109) is to go through the step of determining. In the case of the registered members 40 and 40 ', various information required as registered members are entered (step 110), the information manager 32 is downloaded (step 111), and the connection is terminated. In this case, the information in step 110 refers to the URL of the website 47 of the registered members 40 and 40 'and a detailed description of the website 47.

In the case of the evaluation members 40 and 40 ', the information manager 32 and the user program 34 are downloaded. (112) Of course, the user program 34 is downloaded by the evaluation members 40 and 40'. You can choose whether to receive.

In addition, in the case of the evaluation members (30, 30 '), the latest information update (step 106) proceeds to the next step to change the program setting (step 113) and ends the membership registration (step 115). A database 35 is created to store basic data for exchanging data with the server 12 of the evaluation site 10 of the evaluation members 40 and 40 '. In this state, the evaluation members 30 and 30 'may collect information of various websites 47, 50 and 50' while surfing the web.

Next, the information manager 32 and the user program 34 in the system of the evaluation members 30 and 30 'will be described with reference to FIG. 3.

When the evaluation members 30 and 30 ', which are users, start using the computer, the information manager 32 and the user program 34 are operated (step 203). It is determined whether access to the Internet 1 is established. (Step 204) If no connection to the Internet 1 is established, the user program 34 performs a predetermined function. (Step 205) In this state, the user program 34 continues. The connection to the Internet 1 is waiting.

On the other hand, when access to the Internet 1 is made, it is determined whether the member is entered by inputting an ID and a password (step 207). At this time, if the password is forgotten, it can be transmitted to the member by e-mail.

In step 207, the system asks whether it is a self-evaluating member (30, 30 ') using the Internet (1) in the system. This is to induce a subscription unless the evaluation members 30, 30 'are used by a plurality of users jointly.

Once determined to be a member, the information provider 32 or the user program 34 is upgraded (step 208). Of course, if there are no upgrades, this step just passes.

Next, as the evaluation members 30 and 30 'surf the Internet 1, information is collected by the information provider 32 (step 220). The collected web surfing-related information is stored in the information provider ( 32 is transmitted to the server 12 of the evaluation site (10). At this time, the information in the processing information database 16 of the server 12 is also delivered to the information provider 32 in real time, so that the information related to the website 47, 50, 50 'which is currently connected is evaluated. The information hooked by the information provider 32 in the process of transmitting and receiving information between the server 12 and the information provider 32 is stored in the personal database (30, 30 '). Is stored temporarily.

On the other hand, the information delivered by the information provider 32 to the server 12 is stored in the capture information database 15. The capture information database 15 temporarily stores various types of information transmitted from the evaluation members 30 and 30 'or the registered members 40 and 40'. The server 12 selects and processes statistically valuable information among the temporarily stored information and stores it in the processing information database 16 (step 240).

Meanwhile, information stored in the processing information database 16 is transmitted to the information provider 32 for the websites 47, 50, and 50 'visited by the evaluation members 30 and 30' while surfing the web. (Step 250)

If it is determined in step 207 that the member is not a member, a process for joining as a member is performed. First, it asks whether or not to join as a member (step 209). If it is not desired to join as a member, it is terminated so that the user can surf the Internet 1 without being a member.

And if you want to join as a member, enter the details of the membership required (step 210). Next, it is determined whether to join only the evaluation members (30, 30 ') (step 211). If you want to join only (30,30 '), the program is updated (step 212) and the process goes to step 220.

If the user wants to join the registered members 40 and 40 'in step 211, the process proceeds to step 214, and the information necessary for the registration of the registered members 40 and 40' is entered. Then, it downloads and installs the information manager 42 necessary for the website 47 of the registered members 40 and 40 'and ends (step 215).

On the other hand, in the case of being selected as an excellent member among the registered members (40, 40 '), the certification mark and source code provided by the evaluation site 10 will be provided on the website 47 of the registered members (40, 40'). .

4 illustrates a process in which an evaluation member hooks information using an information provider while surfing the web, and exchanges information with a server 12 of the evaluation site 10. That is, step 220 to step 250 of FIG. 3 will be described in more detail.

In the information collection step, information of the evaluation members 40 and 40 'is first collected (step 301). That is, the information provider 32 or the user program 34 reads the registry information of the evaluation members 40 and 40'. For example, it collects OS, CPU, resolution, memory, currently used evaluation members (40, 40 '), hardware information, software applications, environment settings, and the like.

Next, the web browser 31 and the OS system are monitored to check messages or URL changes in real time (step 302).

The host name and the path name are separated from the URL obtained in the step (step 303), or the IP is obtained from the URL or the URL is obtained from the IP (step 304). Count the time to stay at (step 305).

In addition, meaningful retention times are obtained with the movement of the mouse. In other words, it is determined whether the mouse or the pointing device has moved (step 306). If the mouse moves, the dwell time is counted. If the mouse does not move for a predetermined time (for example, one minute), the dwell time count is stopped. In step 307, the mouse waits for the movement of the mouse. If the movement is detected (step 308), the dwell time is counted (step 307).

Next, the process of evaluating the reliability of the websites 47, 50, and 50 'is performed (step 310). Present and score by item. Evaluation items include professionalism, design, completeness, and membership reliability of the website (47, 50, 50 ').

On the other hand, the website 47, 50, 50 'currently connected to the processing information database 16 of the evaluation site 10 is provided in reverse to provide various information to the evaluation members 30, 30'. (Step 312)

The web surfing related information hooked by the information provider 32 is transmitted to the server 12 of the evaluation site 10 (step 313).

In this case, steps 312 and 313 may be repeatedly performed in real time or may be repeatedly performed for a predetermined time.

On the other hand, the information provider 42 of the registered members 40, 40 'is related to the web surfing associated with all users (evaluation members 30, 30') and all the users including non-members accessing the website 47. Information is provided to the server 12.

Now, the web surfing related information provided by the information providers 32 and 42 is processed as described above to evaluate the ranking of each website 47, 50, and 50 'in more detail.

First, in order to rank the websites 47, 50 and 50 'by processing the information provided by the evaluation members 30 and 30', a specific sample group must be extracted from the evaluation members 30 and 30 '. In this case, the stratified extraction method and the allocation extraction method may be used, respectively, or a combination thereof. For example, to determine the ranking of a website in relation to a man in his twenties, a sample group is created by extracting a man in his twenties from among the evaluation members.

Next, by using the web surfing-related information provided by the sample group extracted as described above, each piece of information (e.g., the number of connections, residence time, page view, etc.) of each website (47, 50, 50 ') Estimate the actual amount of.

Here, using the probability proportional sampling method, estimating the actual number of visits, residence time and page view of each website 50, 50 'will be described with reference to Table 1.

Of the total 100 users of the Internet, the amount of information 1600 times Member Information Actual amount of information Member Site ABCDE 10 times 20 times 30 times 40 times 200 1200 2500 4200 5400 25000 Non-member site F 1300 ???

A, B, C, D, E, the registered member (40, 40 ') website 47, A, B, C, D, E, F for the top few percentages of statistically significant websites If we know the exact number of visits to, then we determine the number of visits to the non-registered member's website (50,50 ') F

Once all the log information (Act B, C, D, E, which is the website 47 having the information provider 42 among the registered members 40, 40 'of the evaluation site 10, is already available (the corresponding web) The number of connections to the site 47, residence time, page views, etc.). For example, as shown in Table 1, for example, the total number of accesses to site A is 1200, the number of evaluation members (30,30 ') to site A is 10, the total number of sites B is 2500 and 20 members of the evaluation member (30,30 '), 4,200 visitors to the C site, 30 members of the evaluation member (30,30'), and 5,400 users of the D site. It is assumed that the number of evaluation members 30 and 30 'is 40, the total number of accesses to the E-site is 25000 and the number of evaluation members 30 and 30' is 200. At this time, the actual number of connections of non-member F site is extracted. Here, the number of evaluation members 30 and 30 'connected to the F site is 1,300 times.

In this case, use the following equation to find the actual number of connections to site F.

Where τ is the total number of connections (30,30 ') of the evaluation members (30,30') for a certain period, X _i is A, B, C, D, E, F is the number of accesses to the site, N is the total number of Internet user accesses during the period, x is the number of Internet user accesses to the registered member website 47, F is the number of evaluation member accesses to the F website. to be.

Substituting Equation (2) into Equation (1), the total number of actual connections (N) is obtained by 167,166 persons in the probability proportional sampling method. If you multiply the total number of connections, you get 128,600. In this way it is possible to estimate the number of connections of non-member websites 50,50 '. Of course, the same method can be applied not only to the number of connections, but also to page views and residence time.

First, according to the above probability proportional sampling method, information about the number of accesses, page views, and residence time is provided not only for the website 47 of the registered member 40, 40 'but also the general website 50, 50'. In this way, a ranking of each item for each website 47, 50, 50 'can be obtained.

On the other hand, in order to obtain the ranking of the website 47, 50, 50 'for the above items, the ranking can be determined by classifying according to the characteristics of the website (47, 50, 50'). Used here is Discriminant Analysis.

Discriminant analysis can be used to find which independent variable has a large influence on the difference between the mean values, to classify a given individual, or to show whether there is a difference between means for two or more previously determined groups. Problems in classifying plants, discrimination between bankruptcy companies and healthy companies, and distinguishing honor students from failing students in terms of entrance grades, IQs, girlfriends, parents' income, college grades, etc. And the like, in which the present inventors applied this to distinguish the nature of numerous websites. As shown in FIG. 5, if two distribution functions are provided and a value X ₀ of an entity is given, the question of which distribution function is close to which of the two distribution functions is a website is the first population ( For example, it is in line with the question of which website can be classified as a shopping mall) or a second group (for example, a game mall). At this time, a discrimination function is flowed through a discriminant analysis and an object is classified based on the discriminant function. Thus, substituting the X ₀ value into two distribution functions yields a probability value of the first population larger than that of the second population, so that the entity can be classified as belonging to the first population.

On the other hand, in determining the ranking of each website 47, 50, 50 ', for example, each of these items is not determined by ranking only for one item such as the number of visits, residence time and page view. On the basis of all, ranking of websites 47, 50, and 50 'is as follows.

In this case, each of the items has different criteria for the value. For example, the number of connections is an integer unit, the dwell time is an hour unit, the page view is an integer unit, and the reliability is a percentage. Therefore, it is necessary to convert the value of each of these items into one value to rank the websites 47, 50, and 50 '. To this end, the present inventors used Principal Componen Analysis among multivariate analysis methods.

The server 12 of the evaluation site 10 extracts and uses statistically significant information from the information of the capture information database 15. In order to remove the influence of the unit of each item, the data is normalized with the mean as 0 and the variance as 1, and then the correlation matrix indicating the degree of correlation between the variables is obtained. The main component (1st component, 2nd component, 3rd component) which can fully explain inside is selected. By evaluating the correlation between the variables and the effect on the cumulative contribution rate, the principal component scores are obtained from the principal component relations equation, and the overall ranking of each website (47, 50, 50 ') is evaluated.

Here, the principal component analysis method will be briefly described. Principal component analysis generally refers to a comprehensive characteristic, and means a common component inherent in many items (explanatory variables) as a means of sizing. Through the linear change technique of a large number of explanatory variables, a few new independent variables are created to identify the main influences on the response variables and to discover what the main influences are. That is, it is possible to identify many items (variables) as explanatory variables, which explanatory variables have the most influence on response variables, which factors act as '+' factors, and which factors act as '-' factors. Statistical technique.

The evaluation items include connection statistics, dwell time statistics, page view statistics, connection speed statistics, membership statistics (only among corporate members among registered members), and reliability (credit level) statistics. Since it is different from each other, it is impossible to make a direct comparison, so the principal component analysis method using the correlation matrix method (using a cumulative contribution rate of 60% or more and using an eigenvalue of 1.0 or more) is used.

Related equations are as follows.

Where Y is the principal component, a _i is the response variable (dependent variable), and X _i is the explanatory variable (evaluation item).

On the other hand, the world's Internet, country, or region's actual Internet users, or age, interest, gender, occupation, religion, etc., by statistically estimating or identifying trends (travel direction or time taken) Use Capture-Recapture Sampling.

However, using this method may result in the creation of new Internet users, whether members of the Internet or non-members, can withdraw their membership or suspend their use of the Internet, and that each Internet user has the same probability of visiting a site of interest. none. However, this situation will be overcome by simplifying the problem through statistical analysis.

In order to estimate the number of Internet users of each item, the number of members identified in the average number of Internet users who visited each website for a certain period of time in the top few percent of the number of members corresponding to each item is represented by each item. It can be obtained as spring that is equal to the average ratio. The website can be divided into a global site and a local site, and the criteria for classification are limited to a few percent of the ratio of foreign visitors / total visitors or foreign visitors / local visitors, or to actual Internet users in each country or region. You can limit it to a few percent or less of the website visitor rate.

Under these prerequisites, among the total internet users, statistically meaningful users among the users who have registered as members of this website are classified accordingly.

And after a certain period of time, the number of members corresponding to each item is equal to the average number of Internet users who visited each website within a certain period of time, and the average of the members whose visits were confirmed. To the limit or confidence interval (eg 99%, 95%).

Finally, the size of the population corresponding to the item you want to know is calculated by the following equation.

Where N is the size of the population corresponding to each item, M is the number of members corresponding to each item, n is the average number of Internet users who visited during a certain period, and m is the average of the members of each item identified for a certain period.

According to the website ranking method according to the present invention as described above in detail, it is possible to statistically summarize the ranking of various websites connected to the Internet, more accurate and objective website ranking information for Internet users Can be provided.

Claims (6)

Registering an evaluation member actively performing web surfing on the network and a registered member who is the target of web surfing on the server; Providing, by the server, an information provider for acquiring and transmitting web surfing related information to the evaluation member and the registered member's system; Receiving, by the server, web surfing related information through an information provider of the evaluation member and the registered member connected to the network; Recording the web surfing related information received by the server in a database; And determining the rank of the web site by processing the databased web surfing related information in a statistical method at the server. According to claim 1, The web surfing related information includes the number of access to each website of the evaluation member, residence time, page view, access speed and website reliability, the user who accesses the website of the registered member Web site ranking method comprising the number of times of stay, the time of stay, page views and speed of connection. The method of claim 1, wherein ranking the website comprises: Extracting a predetermined sample group from the evaluation members using the stratification extraction method or the allocation extraction method for websites classified by the discriminant analysis method; And extracting web surfing related information such as the number of times of non-member website access, residence time, page view, etc. using the probability proportional sampling method from the sample group. 4. The method of claim 3, further comprising the step of comprehensively analyzing the extracted web surfing related information by a principal component analysis method to determine the ranking of the web site. 4. The method of claim 3, wherein the server extracts the total number of Internet users corresponding to a predetermined sample group accessing a specific site by using the capture-recapture extraction method based on the number of evaluation members corresponding to the predetermined sample group. Website ranking method characterized by determining the trend. The method according to any one of claims 1 to 5, characterized in that the characteristics of the websites are classified by discriminant analysis.