JP6839640B2 - Accessor characteristic estimation system - Google Patents

Description

The present invention relates to an accessor characteristic estimation system that estimates the characteristics of a person who has accessed a web page (hereinafter referred to as "accessor").

Patent Documents 1 and 2 disclose, for example, a system that infers the characteristics of an accessor (for example, the name of the company to which the user belongs) and obtains useful information for business based on the result. In each case, the characteristics are estimated based on the IP address of the accessor.

However, it is not always certain to guess the characteristics based on the IP address. For example, when a dynamic IP address is used, the IP address of one accessor may fluctuate with each access, or the IP address of one may be used by accessors having different characteristics. Further, for example, an accessor who uses a notebook computer may access from an IP address whose characteristics cannot be inferred, such as a public WIFI station.

Japanese Unexamined Patent Publication No. 2016-024530 Japanese Unexamined Patent Publication No. 2016-118957

An object of the present invention is to provide an accessor characteristic estimation system that estimates the characteristics of an accessor with high accuracy.

The accessor characteristic estimation system of the present invention
It is an accessor characteristic estimation system that infers the characteristics of the organization to which the accessor who accessed via the Internet belongs.
A characteristic estimation unit that is attached to the program that executes the access and estimates the characteristics based on a cookie having an ID,
A related update unit that updates the ID-characteristic related data, and
ID-characteristic-related data that associates the ID with the characteristic and is referred to by the characteristic estimation unit,
It includes access log data generated for each access and accumulating access logs including information on the IP address of the access source and the ID and access time of the cookie detected in the access.
When the characteristic associated with the ID exists in the ID-characteristic-related data, the characteristic estimation unit estimates that the characteristic is the characteristic.
The related update unit (1) has the same IP address (stable IP address) in the access log data, and has an ID associated with the same characteristic (stable characteristic) in the ID-characteristic related data. Access and second access are included, and (2) a third access that is at a time between the first access and the second access and has the stable IP address is included. , (3) When the ID of the cookie related to the third access is not associated with a characteristic different from the stable characteristic in the ID-characteristic-related data, the ID related to the third access is referred to as the stable characteristic. It is characterized in that the ID-characteristic related data is updated so as to be associated with.

According to this feature, high-precision ID-characteristic-related data can be constructed, and based on this, the characteristic can be inferred from the cookie ID (representing a computer to which the cookie is attached) with high accuracy. That is, accurate ID-characteristic-related data can be constructed by associating the ID related to the third access with the stable characteristic. On the other hand, the ID related to the third access may be associated with a characteristic other than the stable characteristic. In such a case, it is preferable to determine whether to associate with the already associated characteristic or the stable characteristic by a statistical method such as the reliability of the association. If such a situation does not exist, the ID related to the third access is associated with the stability characteristic.

The accessor characteristic estimation system of the present invention
The related update unit has an IP address that is different from the stable IP address and has a time between the first access and the second access to the access log data, and the first access or the first access. ID of 2 Access Only when the 4th access having the same ID is not included, the ID related to the 3rd access is associated with the ID of the 1st access and the 2nd access. It is characterized in that the ID-characteristic-related data is updated so as to be associated with the characteristic.

According to this feature, when the computer that made the first access or the second access accesses from a different IP address, it is treated as having low reliability of the relationship between the ID and the characteristic of the cookie. It is possible to reduce the possibility of updating the ID-characteristic-related data that is erroneously associated with the accessor's characteristics in the third access.

The accessor characteristic estimation system of the present invention
The related update unit assigns the ID related to the third access only when the difference between the time of the first access and the time of the second access in the access log data is equal to or less than a predetermined value. It is characterized in that the ID-characteristic-related data is updated so as to be associated with the characteristic associated with the ID of the first access and the second access.

According to this feature, the estimation regarding the third access is performed only when the time of the first access and the time of the second access are different from each other by a predetermined value or less. A dynamic IP address was used, after which the IP address was assigned to a person with other characteristics after the first access, while a third access was again assigned to a person with the characteristics associated with the ID. When the second access is performed later, the possibility of updating the ID-characteristic-related data that is erroneously associated with the characteristics of the accessor in the third access can be reduced. Therefore, it is preferable that the "predetermined value" is a short time during which the dynamic IP address is not assigned twice.

The accessor characteristic estimation system of the present invention
The characteristic estimation unit is the same as the characteristic given to the accessor with respect to the most recent other access from the IP address of the access source when the characteristic associated with the ID does not exist in the ID-characteristic-related data. It is characterized by presuming that it is a characteristic.

According to this feature, if the characteristic associated with the ID does not exist, the characteristic can be inferred based on the IP address.

The accessor characteristic estimation system of the present invention
The characteristic guessing unit has the same characteristics as those given to the accessor with respect to the other most recent access only when the access for estimating the characteristic of the accessor is performed before the end of the session of the other most recent access. It is characterized by inferring that.

According to this feature, it is possible to guarantee that the dynamic IP address is not reassigned for the IP address for the estimation of the characteristic based on the IP address.

The assessor characteristic estimation system of the present invention
When an access is received from the application software that does not have the ID-bearing cookie, the application software is attached with the ID-bearing cookie.

According to this feature, it is possible to attach a cookie when first accessed and then use the cookie. Here, it is considered that the "application software" is an Internet browser in many cases, but software such as mail processing that handles html can also be attached with a cookie and is included in the application software.

The assessor characteristic estimation system of the present invention
Equipped with an initial setting section
The initial setting unit is
It has a URL associated with the characteristic, and by accessing the URL, the cookie is attached to the application software of the terminal.
The ID-characteristic-related data is characterized in that the ID of the cookie is associated with the characteristic.

According to this feature, a cookie associated with a characteristic can be attached by having a person with a known characteristic access the URL associated with the characteristic.

The accessor characteristic estimation system of the present invention
It runs on a server that provides two or more homepages,
It is characterized in that the access log is generated for an accessor who has accessed any of the two or more homepages.

According to this feature, many access logs can be generated, and the estimation accuracy of the characteristics can be improved.

According to the present invention, it is possible to provide an accessor characteristic estimation system that estimates the characteristics of an accessor with high accuracy.

FIG. 1 is a diagram showing a configuration of an accessor characteristic estimation system and its usage status. FIG. 2 is a diagram showing ID-characteristic related data. FIG. 3 is a diagram showing access log data. FIG. 4 is a flowchart showing the processing of the access receiving unit. FIG. 5 is a flowchart showing the processing of the characteristic estimation unit. FIG. 6 is a flowchart showing the processing (modification example) of the characteristic estimation unit. FIG. 7 is a diagram showing access log data (modification example).

Hereinafter, examples of the present invention will be described.

(System configuration)
FIG. 1 is a diagram showing a configuration of an accessor characteristic estimation system and its usage status. The server 1 provides a plurality of homepages 11, and includes an accessor characteristic estimation system 2. The terminal 3 accesses the homepage 11 via the network 4. It should be noted that the homepage 11 may be held by the server 1 (the computer operating as) or by another computer via the network 4 (even if the server 1 operates as the name server). It may be both. Further, as long as the data can be accessed by the URL, it does not necessarily have to be the home page of the Internet.

The accessor characteristic estimation system 2 includes an access reception unit 21, a cookie granting unit 22, a characteristic estimation unit 23, a related update unit 24, and an initial setting unit 25, and holds an ID-characteristic related data 26 and an access log data 27. There is.

The access receiving unit 21 is a part that first receives access to the homepage 11 on the server 1. The access receiving unit 21 adds the access log of the access log data 27 each time it is accessed. The details of the processing of the access receiving unit 21 will be described later.

The cookie giving unit 22 receives the notification from the access receiving unit 21, and when there is an access from the terminal 3 that does not have the cookie 32, the cookie giving unit 22 gives the cookie 32 to the terminal 3.

The characteristic estimation unit 23 receives a notification from the access reception unit 21, estimates the characteristics of the accessor based on the ID-characteristic-related data 26, and outputs the characteristics as a characteristic value. The details of the processing of the characteristic estimation unit 23 will be described later.

The related update unit 24 updates the ID-characteristic related data 26 based on the access log data 27. The details of the processing of the related update unit 24 will be described later.

The initial setting unit 25 includes a cookie giving homepage 251. The cookie-giving homepage 251 grants the cookie 32 to the terminal 3, and the ID of the cookie to be given is associated with a characteristic known in advance in the ID-characteristic-related data.

The ID-characteristic-related data 26 is data representing the correspondence between the ID of the cookie and the characteristic. FIG. 2 is a diagram showing ID-characteristic related data. The correspondence between the cookie ID and the characteristics is shown. A plurality of IDs (A1, A2) may correspond to a characteristic (Company A) of 1.

The access log data 27 is data that accumulates access logs generated by the access receiving unit 21 each time an access is made. FIG. 3 is a diagram showing access log data. Each line in the figure is an access log of 1, and the access log contains the IP address of the access source, the ID of the cookie, and the time when the access is received.

When the terminal 3 accesses the homepage 11 via the network 4, the access receiving unit 21 processes the access.

The terminal 3 includes the application software 31 and can hold the cookie 32 attached to the application software 31.

(Operation when receiving access)
When the terminal 3 accesses the homepage 11 via the network 4, the access receiving unit 21 is activated. FIG. 4 is a flowchart showing the processing of the access receiving unit.

The access receiving unit 21 first confirms whether or not the cookie 32 exists in the access source terminal (step 21A). If the cookie 32 does not exist, the cookie granting unit 22 is activated (step 21B).

The activated cookie giving unit 22 gives a cookie to the application software 31 of the terminal 3. The cookie can be added by the usual method using the HTTP header. The cookie given has an ID for distinguishing it from other cookies. For example, a unique character string can be set as the Name attribute.

The access receiving unit 21 then activates the characteristic estimation unit 23 (step 21C). The characteristic estimation unit 23 estimates the characteristic and outputs it as a characteristic value.

Finally, the access receiving unit 21 provides the terminal 3 with access to the home page 11. A home page for providing access is determined according to the URL received from the terminal 3, and the home page (html file) is transmitted (step 21D).

The characteristic value can be used by the homepage 11 by passing it as an argument to the homepage 11, storing it in a predetermined file, or using other methods. The homepage 11 can select, for example, an advertisement to be displayed.

The above processing can be performed regardless of whether the application software 31 is a browser or other software that processes html.

(Guessing characteristics)
Hereinafter, the operation of the characteristic estimation unit 23 will be described. FIG. 5 is a flowchart showing the processing of the characteristic estimation unit. The characteristic estimation unit 23 operates based on the ID of the cookie 32 existing in the access source terminal and the access source IP address.

First, the characteristic estimation unit 23 refers to the ID-characteristic-related data 26 and confirms whether or not the characteristic associated with the ID of the cookie 32 exists (step 23A). If there is a characteristic associated with the ID of the cookie 32, that characteristic is output as an inferred characteristic value (step 23B).

When the characteristic associated with the ID of the cookie 32 does not exist, the characteristic estimation unit 23 refers to the access log data 27 and determines whether or not the access log satisfies a predetermined requirement (step 23C).

Here, the "predetermined requirement" is (1) a first having the same IP address (stable IP address) in the access log data 27 and having an ID associated with the same characteristic in the ID-characteristic-related data 26. Access and second access are included, and (2) the third access, which is the time between the first access and the second access and has a stable IP address, is currently being processed. Is to be. When the access log satisfies a predetermined requirement, the characteristic estimation unit 23 outputs the characteristics associated with the first access and the second access as estimated characteristic values (step 23D).

The ID-characteristic-related data 26 shown in FIG. 2 and the access log data 27 shown in FIG. 3 will be described as examples. It is assumed that the access related to the access log 27Q is being processed now.

The access log 27P has a cookie ID of "A1" and is associated with the characteristic of "Company A" in the ID-characteristic-related data 26. Further, the access log 27S has a cookie ID of "A2" and is associated with the characteristic of "Company A" in the ID-characteristic-related data 26. Assuming that the access related to the access log 27P is the first access, the access related to the access log 27S is the second access, and the access related to the access log 27Q currently being processed is the third access, the IP address is "*". Since it is "*. ***. 1", it satisfies the predetermined requirements. That is, it is estimated that the characteristic of the access related to the access log 27Q is "Company A".

Here, the IP address of the access log 27R is also "**. ***. 1". However, the access log 27R has a cookie ID of "B" and is associated with the characteristic of "Company B" in the ID-characteristic-related data 26. That is, the access logs 27P and 27R are associated with different characteristics. Therefore, it is not presumed that the characteristic of the access related to the access log 27Q that satisfies the predetermined requirements is "Company A".

That is, the access log 27P and the access log 27R do not serve as a basis for estimating the characteristics because the IP address is "**. ***. 1". Only when the access log 27S exists, the characteristic is "Company A" because the IP address is "**. ***. 1" (The access log 27R is accidental, for example, by a visitor's notebook computer or the like. It can be inferred. The measures to be taken when the access log 27R is not accidental will be described later.

On the other hand, at the time of receiving the access related to the access log 27Q, the access related to the access log 27S has not been made yet, and it is determined that the access log does not satisfy the predetermined requirements even if the access log data 27 is referred to. It is thought that it will be done. Since the requirement includes a second access at a time after the access currently being processed, the second access is made within a very short time from receiving the access until the characteristic estimation unit 23 is started. Except for, it will be judged that the prescribed requirements are not met at the time of receiving the access.

Therefore, the characteristic estimation unit 23 estimates the characteristics of the access that does not satisfy the predetermined requirements as follows. First, there is a past access from the same IP address as the access for which the characteristic is to be inferred, and the characteristic can be inferred (the cookie ID is associated with some characteristic in the ID-characteristic related data 26). Confirm whether or not to do so (step 23E). If there is no such past access, the characteristics cannot be inferred (step 23H).

If there are such past accesses, it is reasonable to infer that they have the same characteristics as the most recent one (the one that was made at the time closest to the access for which the characteristic is to be inferred). This is because when a dynamic IP address is assigned to one company, there is no reassignment in a short time, and it is considered that the latest access and the access whose characteristics are to be inferred are assigned to the same company.

In order to confirm that the IP address has not been reassigned between the latest access and the access whose characteristics are to be inferred, the characteristic estimation unit 23 continues (ends) the session of the latest access. Confirm that there is no such thing (step 23F). This is because the dynamic IP address is not reassigned during the duration of the session. The characteristic estimation unit 23 estimates that the characteristics are the same as the latest access when the session is continuous (step 23G), and makes it impossible to estimate the characteristics when the session is not continuous (step 23H). ..

Here, step 23F may be omitted. That is, regardless of the continuation of the session, it may be inferred that the characteristics are the same as those of the latest access whose characteristics can be inferred. There may be cases where the method of using the guess results can tolerate erroneous guesses (for example, when the guess results are used for selecting advertisements), and it may be acceptable as long as a certain system can be obtained.

There is a range of ways to define a session as to whether or not the session continues (for example, if the session ends due to a time-out of a certain period of time, it is at the discretion of each vendor how to determine the fixed time). Strict judgment is difficult. It is preferable to use the same definition as the vendor that assigns the dynamic IP address, but as described above, any design may be used as long as it can be misguessed within an acceptable range.

Hereinafter, measures will be described for the point that it is judged that the predetermined requirements are not satisfied at the time of receiving the access. FIG. 6 is a flowchart showing the processing (modification example) of the characteristic estimation unit. If the access log does not meet the predetermined requirement, the characteristic estimation unit 23 determines whether or not the access log satisfies the predetermined requirement after waiting (step 23I) instead of the above-mentioned processes of steps 23E and 23G. May be re-judged. When there is an access related to the access log 27S, an accurate guess is made assuming that a predetermined requirement is satisfied. The timing of re-judgment may be, for example, once per minute or when the access log data 27 is updated, and may be arbitrarily designed.

In addition, in order to prevent the re-judgment from continuing indefinitely, for example, the re-judgment will be terminated after 10 minutes, the re-judgment will be terminated at the end of the session, or other methods will be terminated.

The above "predetermined requirements" can be made even more stringent with additional requirements.

FIG. 7 is a diagram showing access log data. In the access log data 27, the time of the access log 27Q is later than that shown in FIG. 3, and the access log data 27 is the access log 27Q'. The time of the access log 27S is delayed and becomes the access log 27S'. Further, the access log 27X is used instead of the access log 27R.

Here, assuming that the access related to the access log 27P is the first access, the access related to the access log 27S'is the second access, and the access related to the access log 27Q'currently being processed is the third access, the access log 27Q'satisfies the above requirements.

However, in the access log data 27 shown in FIG. 7, as a result of using the dynamic IP address, the IP address "**. ***. 1" was given to Company A on October 1st, and October 2nd. It is presumed that the IP address "**. ***. 3" was assigned on the day and the IP address "**. ***. 1" was assigned on October 3. Then, it is considered that the IP address "**. ***. 1" at the time of access related to the access log 27Q'is assigned to a company other than company A.

As an additional requirement, for example, an IP address that is at a time between the first access and the second access and is different from the stable IP address and the same as the ID of the first access or the second access. It may be an additional requirement that the fourth access with the ID is not included.

According to this, it is possible to detect that another IP address is assigned on the way by the fourth access and reduce the erroneous estimation related to the third access. In FIG. 7, the access log 27X is the fourth access, and the access related to the access log 27Q'is not presumed to have the characteristic "Company A".

Further, as an additional requirement, for example, the difference between the time of the first access and the time of the second access is less than or equal to a predetermined value and is stable because it is the time between the first access and the second access. It may be an additional requirement that the IP address different from the IP address and the fourth access having the same ID as the ID of the first access or the second access are not included.

According to this, by setting a predetermined value to a value smaller than the time when the IP address is assigned twice, for example, one day, the case where another IP address is assigned in the middle is excluded, and the third It is possible to reduce the misestimation related to the access of. This is to deal with the case where the above-mentioned access log 27R is not accidental. In FIG. 7, the difference between the time of the access log 27P (first access) and the time of the access log 27S'(second access) is 2 days or more, and the access related to the access log 27Q'has a characteristic of ". It is not presumed to be "Company A".

Generally speaking, the setting of the above additional requirements can be considered to process the "reliability" related to the ID-characteristic related data 26 and the access log data 27. Additional requirements may not be fixedly set and may be determined by statistical processing or the like.

(Operation other than when receiving access)
The above-mentioned "first access and second access having the same IP address (stable IP address) in the access log data 27 and having an ID associated with the same characteristic in the ID-characteristic-related data 26 are Guess the characteristics on the condition that the third access, which is included, at the time between the first and second access, and has a stable IP address, is the access currently being processed. For the "do" process, the second access may occur after the accessor leaves the home page 11. In this case, the characteristic estimation unit 23 cannot estimate the characteristics.

However, the ID of the cookie 32 according to the third access should be associated with the characteristics relating to the first access and the second access.

Therefore, the related update unit 24 updates the ID-characteristic related data 26.

The accessor characteristic estimation system periodically (for example, once an hour) activates the related update unit 24. The related update unit 24 checks whether or not there is an access satisfying a predetermined requirement in the access log, and if so, the cookie ID of the access log and the characteristics estimated by satisfying the predetermined requirement. The information associated with is added to the ID-characteristic related data 26.

As a result, the predetermined requirements are examined later and the ID-characteristic related data 26 is updated. Even if the characteristic estimation unit 23 cannot make an estimate in the first access, it is surely inferred in the later access.

(Pre-setting)
The estimation of the characteristic described above requires that the ID of at least one cookie 32 is associated with the characteristic and the information exists in the ID-characteristic-related data 26. When the operation of the accessor characteristic estimation system 2 is started, it is a problem to generate such a cookie.

Of course, the operator has information about the characteristics, and it is possible to create the ID-characteristic-related data 26 by himself / herself. However, it can be efficiently generated by the initial setting unit 25.

The initial setting unit 25 has as many cookie-giving homepages (URLs) 251 associated with the characteristics as there are characteristics. When the user (future accessor) accesses the cookie giving homepage 251, the cookie 32 is given to the application software 31 of the terminal 3. The cookie granting is the same as that of the cookie granting unit 22.

When the cookie 32 is given by the cookie-giving homepage 251 (one of them if there is more than one), the initial setting unit 25 provides information that associates the ID of the cookie 32 with the characteristics associated with the cookie-giving homepage 251. It is added to the ID-characteristic related data 26.

That is, each homepage of the cookie-giving homepage 251 is accessed only by the user of the characteristic associated with the homepage, so that the ID is associated with the characteristic and the cookie 32 in which the information exists in the ID-characteristic-related data 26 is generated. To.

Each home page can be accessed only by users with the characteristics associated with it, and the access can be guided, for example, as follows.

An email is sent to an email address whose characteristics are specified (for example, an email address having a URL of a known company after "@"). According to scraping etc., such an e-mail address can be legally obtained. By operating on the mail, the cookie-giving homepage 251 associated with the characteristic is accessed.

Of course, instead of using e-mail, access to the cookie-giving homepage 251 may be guided by other methods such as sending a document.

(Notes)
The accessor characteristic estimation system 2 does not need to target access to only one homepage. By targeting access to a large number of homepages, the number of access logs will increase. This increases the likelihood that the characteristics can be inferred and increases the accuracy of the inferred characteristics.

Therefore, it is preferable that the accessor characteristic estimation system 2 operates on a server that provides a large number of homepages, and generates an access log for an accessor who has accessed any of the homepages.

(Summary)
As described in detail above, according to the accessor characteristic guessing system 2 of this embodiment, the guessing by the cookie 32 and the guessing at the IP address can be used at the same time, and the guessing corresponding to the dynamic IP address can be performed with high accuracy. The characteristics of the accessor can be inferred.

It is an accessor characteristic estimation system that estimates the characteristics of the accessor with high accuracy. It can be used by many homepage management companies.

1 Server 11 Homepage 2 Accessor characteristic estimation system 21 Access receiver 22 Cookie granting unit 23 Characteristic estimation unit 24 Related update unit 25 Initial setting unit 26 ID-Characteristic related data 27 Access log data 31 Application software 32 Cookie 4 Network

Claims (8)

It is an accessor characteristic estimation system that infers the characteristics of the organization to which the accessor who accessed via the Internet belongs.
A characteristic estimation unit that is attached to the program that executes the access and estimates the characteristics based on a cookie having an ID,
A related update unit that updates the ID-characteristic related data, and
ID-characteristic-related data that associates the ID with the characteristic and is referred to by the characteristic estimation unit,
It includes access log data generated for each access and accumulating access logs including information on the IP address of the access source and the ID and access time of the cookie detected in the access.
When the characteristic associated with the ID exists in the ID-characteristic-related data, the characteristic estimation unit estimates that the characteristic is the characteristic.
The related update unit (1) has the same IP address (stable IP address) in the access log data, and has an ID associated with the same characteristic (stable characteristic) in the ID-characteristic related data. Access and second access are included, and (2) a third access that is at a time between the first access and the second access and has the stable IP address is included. , (3) When the ID of the cookie related to the third access is not associated with a characteristic different from the stable characteristic in the ID-characteristic-related data, the ID related to the third access is referred to as the stable characteristic. An accessor characteristic estimation system, characterized in that the ID-characteristic-related data is updated so as to be associated with. The related update unit has an IP address that is different from the stable IP address and has a time between the first access and the second access to the access log data, and the first access or the first access. ID of 2 Access Only when the 4th access having the same ID is not included, the ID related to the 3rd access is associated with the ID of the 1st access and the 2nd access. The accessor characteristic estimation system according to claim 1, wherein the ID-characteristic-related data is updated so as to be associated with the characteristic. The related update unit assigns the ID related to the third access only when the difference between the time of the first access and the time of the second access in the access log data is equal to or less than a predetermined value. The accessor characteristic estimation according to claim 1 or 2, wherein the ID-characteristic-related data is updated so as to be associated with the characteristics associated with the ID of the first access and the second access. system. The characteristic estimation unit is the same as the characteristic given to the accessor with respect to the most recent other access from the IP address of the access source when the characteristic associated with the ID does not exist in the ID-characteristic-related data. The accessor characteristic estimation system according to any one of claims 1 to 3, wherein the characteristic is estimated. The characteristic guessing unit has the same characteristics as those given to the accessor with respect to the other most recent access only when the access for estimating the characteristic of the accessor is performed before the end of the session of the other most recent access. The accessor characteristic estimation system according to claim 4, wherein the accessor characteristic estimation system is characterized in that. The invention according to any one of claims 1 to 5, wherein the cookie having an ID is attached to the application software when an access is received from the application software having the ID and not having the cookie. Accessor characteristic estimation system. Equipped with an initial setting section
The initial setting unit is
It has a URL associated with the characteristic, and by accessing the URL, the cookie is attached to the application software of the terminal.
The accessor characteristic estimation system according to any one of claims 1 to 6, wherein the ID of the cookie is associated with the characteristic in the ID-characteristic-related data. It runs on a server that provides two or more homepages,
The accessor characteristic estimation system according to any one of claims 1 to 7, wherein an access log is generated for an accessor who has accessed any of the two or more homepages.