JP5987953B2 - Device for providing web content and collecting access information - Google Patents

Description

  The present invention relates to an apparatus, a method, and a program for providing web content and collecting access information based on a user's web access.

  There are a wide variety of contents provided by web pages. For this reason, it is desired to select and appropriately provide content that matches the user's preference. For this purpose, it is important to grasp the content and keywords that match the user's preference. Even when providing the same content, it is also important to construct a web page design that suits the user's preference.

  For this reason, a technique is known in which a situation in which a client is actually viewing one content is measured and a usage rate of each content is calculated (Patent Document 1). The usage rate is calculated from the browser size of the client, the size of the content, and the number of clicks on the client. Alternatively, the usage rate is calculated from the amount of movement of the scroll bar in the browser of the client and the size of the browser.

  In addition, a technique for acquiring text that is a target of a specific operation from a corresponding portion of web content displayed on a web browser is known (Patent Document 2). In this technique, the acquired text is transmitted to another system that uses the text. For example, a system or search engine that conducts a web audience rating receives this acquired text and uses it to provide information about an object that the user is interested in.

  In addition, when the user is viewing an arbitrary page on the WWW, a range displayed by the user or a partial area on the web page traced with the mouse is specified, and text information included therein is extracted as a search key. A technique is known (Patent Document 3). The extracted text information is transmitted to the information processing server device. The information processing server device searches for advertisements using information received from the text extraction unit of interest and user preference information stored in advance in the preference information database. A page for displaying an advertisement is created using the search result. The created advertisement display page is transmitted to the advertisement display section of the advertisement providing browser, and the advertisement is displayed.

  Further, a technique for recording a keyword (user keyword) viewed by the user from content browsed by the user is known (Patent Document 4). In this technology, evaluation scores for each user keyword are accumulated in a keyword database, and contents are displayed on the computer terminals of individual users in descending order of evaluation scores based on the evaluation scores of the user keywords on the keyword database.

International Publication No. 2004/097654 JP 2001-188792 A JP 2004-348241 A JP 2003-308339 A

  However, the user often accesses the web page due to factors other than the user's preference, such as ease of clicking based on the layout of the web page and the appearance of the web page. Therefore, it cannot be said that the above-described technology sufficiently provides information and keyword extraction that meet the user's preference.

  The present embodiment has been made to solve such a technical problem. The keyword extraction and web content access information suitable for the user's preference are collected, and the user's preference is provided. An object of the present invention is to provide an apparatus, method, and program for performing keyword extraction and the like.

  According to an aspect of the present embodiment, an instruction count counting unit that counts the number of user instructions for each link of a web page, and a predicted value of an instruction ratio for each link area are calculated from the web page. A link attention degree specifying section for specifying the attention degree of each link based on the instruction ratio prediction value calculating section, the number of instructions, and the prediction value of the instruction ratio, and a link having a high link attention degree. There is provided a server device comprising: a layout changing unit that changes a layout to be replaced with a link at a position where a predicted value of a ratio of an instruction to a link area is higher than a link having a low degree of attention.

  According to one aspect of the present embodiment, it is possible to collect keyword extraction and web content access information suitable for a user's preference, provide information in consideration of the user's preference, and extract a keyword.

It is a figure which shows the structure of the system of one Embodiment of this invention. It is a figure which shows the hardware constitutions of a server apparatus. It is a figure which shows the example of the click frequency and click ratio for every link of a web page. It is a figure which shows the example which divides | segments a web page and calculates the predicted value of a click ratio. It is a figure which shows the example of the calculation result of the predicted value of the click ratio for every link. It is a figure which shows the example of the method of specifying the attention degree of a link. It is a figure which shows the example of the method of extracting a keyword. It is a figure which shows the example of the method of specifying the importance of a keyword. It is a figure which shows the example of normalization of the link attention degree in the case of handling a some page. It is a figure which shows the example of a process of a web page layout change It is a figure which shows the example of the result of a change of a web page layout. It is a figure which shows an example of the functional block diagram in each embodiment. It is a figure which shows the flow of the method as one Embodiment.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following embodiments, the word “click” will be described as an example, but “click” is an example of “instruction”. “Instruction” means, for example, “click” in the case of a mouse and may be called “tap” in the case of a touch panel. Alternatively, on the web page, the tab key is used to move the focus to a link or button, and when the Enter key is pressed in a state where the focus is moved to a desired link, processing similar to that performed by clicking with the mouse is performed. Therefore, it should be noted that the present invention is not limited to clicking with a mouse, but covers any “instructions” using a user interface.

  FIG. 1 shows a configuration of a system 100 according to an embodiment of the present invention. The system 100 includes a web server device 120 to which a storage device A122 is connected, a web content management server device 140 to which a storage device B142 is connected, a client 131, a client 132, and a network 110. The web server device 120, the web content management server device 130, the client 131, and the client 132 are connected to the network 110 via lines 111 to 114, respectively.

  The web server device 120 provides a web page to the client 131 or 132 via the network 110 using the web content 124 stored in the storage device A 122. In addition, the web server device 120 stores various events relating to access to web content in the access log 126. This access log 126 may be provided to the web content management server device 140 or the like.

  The web content management server device 140 may calculate the degree of attention of a plurality of links in the web content 124 published by the web server device 120 and store the calculated degree of attention as link attention degree data 144 in the storage device B142. . Further, the web content management server device 140 calculates the importance of keywords existing in the web content 124 and the keywords in the content indicated by the links in the web content, and stores them as keyword importance data 145 in the storage device B 142. May be stored. The web content management server device 140 may use information in the access log 126 and the like in order to perform the above calculation. Further, in the storage device 142, the link click rate prediction value 146 of the web content 124 and the layout of the web content customized for each user may be stored as the web layout data 147. Details of the various data and processing described above will be described later.

  Note that the web server device 120 and the web content management server device 140 are not necessarily separate server devices. The various functions described above may be implemented in one server device. Further, in the present specification, some of the functions and data described as existing in the web server device 120 may exist in the web content management server device 140. The reverse is also true. Further, both servers may hold a copy of the data.

  The client 131 includes a device having a function of connecting to a network, such as a personal computer, a mobile phone, a multi-function phone, and a tablet computer.

  FIG. 2 shows a hardware configuration 200 of the server apparatuses 120 and 140 of this embodiment. The server device includes a CPU 210, a drive 220 that can read and write a machine-readable medium 225 such as a CD-ROM, a dynamic storage device 230 such as a hard disk, a RAM 232, a ROM 234, an I / O 236, and a communication device 238.

  The functions of the various embodiments described below may be virtually realized by an appropriate program being read and executed by the CPU 210.

  Further, the present invention relating to a program may be stored in a machine-readable recording medium. Examples of the machine-readable recording medium include a magnetic recording medium, an optical disk, a magneto-optical recording medium, and a semiconductor memory. Magnetic recording media include HDDs, flexible disks (FD), magnetic tapes (MT) and the like. Examples of the optical disk include a DVD (Digital Versatile Disc), a DVD-RAM, a CD-ROM (Compact Disc-Read Only Memory), and a CD-R (Recordable) / RW (ReWriteable). Magneto-optical recording media include MO (Magneto-Optical disk).

<Identification of link attention>
FIG. 3 shows examples of web page links, the number of clicks measured for each link, and the click ratio. FIG. 3A shows the layout of the web page 300, and links 1 to 8 are pasted in the respective areas. For example, when the user clicks the area 303 of the link 3, the browser displays the page of the URL assigned to the link 3 (for example, a hyperlink) on the display. In addition to simply displaying the URL embedded in the link, the link may activate a script or the like to activate display of a moving image or a desired application.

  FIG. 3B shows the number of clicks of each area in the web page. The area 313 of the link 3 indicates that the number of clicks is 12233. As the number of clicks, the number of clicks for all users who browse the web page, the number of clicks of a specific user set (for example, male users), the number of clicks of a specific user, or the like can be acquired. Information that a specific link is clicked can be acquired by analyzing the access log of the web page 300 and each linked web page. Alternatively, a script that collects an event when each link is clicked may be described in the web page 300, and the event information may be transmitted to the web content management server device 140, for example. The user can be specified by inputting a user ID and a password to display the page. In addition, in order to specify the user layer (one or more users who make up the set), the user is required to input information (gender, date of birth, years of computer use, type of industry, etc.) required when registering the user ID and password. This can be done by prompting. In addition, once a user ID and password are entered by using a cookie placed on the client machine, it is possible to specify an individual even if the ID and password are not entered. It is.

  FIG. 3C shows the click ratio of each area of the web page. For example, the ratio of the click of the area 323 corresponding to the link 3 to the number of clicks of the entire web page is 21%. The click ratio can be calculated by dividing 12233, which is the number of clicks in the area of link 3, by the number of clicks on the entire web page. The total ratio of clicks in each area may be 100%. In the example of FIG. 3, the ratio of the number of clicks of each region is calculated based on the number of clicks of the entire page, but the present invention is not limited to this. For example, attention may be paid to a specific user, and the click ratio of each region may be calculated by assigning a larger weight to the number of clicks of that specific user than the number of clicks of other users.

  FIG. 4 shows an example for dividing the web page and calculating the predicted value of the click ratio. FIG. 4A shows an example in which a web page is divided into 35 equal parts. The reason for dividing the web page is to simplify the calculation of the predicted value of the click ratio. The number of divisions and the division method may be arbitrarily changed. In this specification, each of the divided areas is referred to as a “square” (note that it is referred to as a square here, but does not have to be a square). On the other hand, an area constituting each link is referred to as an “area” (for example, an area 403) and is distinguished from the square 410.

  FIG. 4B shows the ease of clicking as a percentage using Fitz's UI law. Fitz's UI law means that an object that is far from the current position of the mouse pointer and is small in size is generally difficult to click. In the example of FIG. 4B, the square 410 slightly above the center of the page is set to be most easily clicked, and a value of 100% is assigned to the square 410. Then, as the distance from the square 410 increases, the value given to each square is decreased according to the movement distance of the mouse pointer. These values relating to the ease of clicking may be set in advance as fixed values.

  In FIG. 4C, a predicted value having a high click ratio is given to a square having a high contrast ratio between the background color of the page and the square color. The reason for this is based on a human behavior pattern in which the higher the contrast ratio, the more noticeable to the user, and thus the user's click is likely to be induced.

  For example, when the background color is white, 0% is set when the square color is white (RGB: 0xFFF), and 100% is set when the square color is black (RGB: 0x000). For example, the color occupying the widest part of each square is adopted as the color of the square. For example, if the color of the square is 0x800, a value of 50% may be given to the square.

  FIG. 5 shows an example of the calculation result of the predicted value of the click ratio for each link. In this example, the degree of ease of clicking obtained in FIGS. 4B and 4C is multiplied for each square. Next, the numerical values of the squares overlapping each link area are summed to obtain a predicted value of the provisional click ratio of each link area. If the link area and the square do not completely overlap, the value assigned to the square is divided by the ratio of the overlapping and non-overlapping parts of the link area and the square, and the value of the overlapping square area May be calculated and added to the value of the link area. Then, the value of each link area is normalized by multiplying the value of each link area by a constant so that the total number of the values of each link area is 100%. The result is shown in FIG. FIG. 5 is an example showing the predicted value of the click ratio based on the contrast distribution of the web page and the ease of movement of the mouse cursor. This predicted value can be regarded as a predicted value of the click ratio that is not based on the user's preference by the keyword related to the link. In the above-described embodiment, the predicted value of the click ratio is calculated based on the contrast distribution of the web page and the ease of movement of the mouse cursor. However, the present invention is not limited to this.

  The user who viewed the web page clicks on the link based on the user's preference from words or meaningful pictures displayed on the web page, and the contrast distribution of the web page or the movement of the mouse cursor. There are cases where a link is clicked based on ease. In order to extract the user's preferences from the measured number of clicks, it is possible to influence the click measurement values that are not based on the user's preferences, such as the contrast distribution of the web page and the ease of movement of the mouse cursor. It is important to remove as much as possible.

  Therefore, for example, by subtracting the predicted value of the click ratio in FIG. 5 from the actual click ratio in FIG. 3C, the value of the degree of attention of the link that is truly closer to the click ratio based on the user's preference is calculated. can do. In this embodiment, the click ratio based on the user's preference is estimated and calculated by subtracting the predicted value of the click ratio in FIG. 5 from the actual click ratio in FIG. This value is hereinafter referred to as link attention. The degree of attention of a link can be used as an index of the degree to which the user is paying attention to the link based on the user's preference. The attention degree of the link may be calculated by the following formula.

Here, F _m is the degree of attention of the m th link when the number of all links is M, n _m is the number of clicks of the m th link, and α _m is the number of clicks of the m th link. This is the predicted value of the click rate. β is a constant satisfying 0 <β.

  In the above embodiment, the subtraction is simply performed (β = 1). However, the value of β may be changed, and the subtraction may be performed by applying a necessary weight to the predicted value of the click ratio. Alternatively, other calculation methods may be used as appropriate. In addition, the value of β may be different between the case where the attention level of each link with respect to the entire accessed user is calculated and the case where the attention level of the link with respect to a specific single user is calculated.

  FIG. 6 shows an example of a technique for calculating the link attention degree from the actual access ratio for each link and the predicted value of the access ratio. FIG. 6A shows the actual click ratio measured for each link described in FIG. FIG. 6B shows the predicted value of the click ratio described in FIG. FIG. 6C shows the degree of attention of the link when a value obtained by subtracting the value of FIG. 6B from the value of FIG. 6A is adopted for each link. As the arithmetic method, the above formula (1) may be used, or other arithmetic may be adopted.

  For example, the actual click ratio of the area 611 in FIG. 6A corresponding to the link 1 is 20%, and the predicted value of the click ratio is 3% as shown in FIG. 6B. Using these numerical values, as shown in FIG. 6C, the importance of the link in the region 631 is 17%, and it can be seen that the web page 630 shows the highest degree of attention. On the other hand, the area 613 corresponding to the link 3 has a high numerical value with an actual click rate of 30%. However, since the predicted value of the click ratio of the area 623 is 26%, it indicates that the link attention degree of the area 633 is specified as 4%.

  Based on the link attention degree shown in FIG. 6C, the attention degree of the content or the like indicated by the link can be recognized. Further, as described above, by specifying the population (for example, male users), it is possible to know the degree of attention of the link related to a specific user group. Moreover, the link attention degree of a specific one user can be specified by measuring the specific one user.

  Based on the specified link attention degree, the link being noticed may be moved to a position where it is easier to click. It is also possible to set a layout (my page) for a specific user. Details of the layout change will be described later.

<Identification of keyword importance>
FIG. 7 is a diagram illustrating an example of a technique for extracting keywords. A link 1 (701) is displayed on the displayed web page. Hereinafter, processing for storing a keyword group related to the link 1 (701) in the link keyword table 720 will be described. The link 1 (701) is expressed as “PC security measure” on the web page 700, and “http: //ABC.def.hg...” Is embedded as URL information of the hyperlink. When this hyperlink is clicked, the link destination page 730 of the URL is displayed on the display. The web content management server device or the keyword extracting unit 710 puts the URL “http: //ABC.def.” In the URL entry field (722) corresponding to “Link 1” in the link ID field (721) of the link keyword table. hg ... "is entered. Then, the keyword extraction unit 710 may extract keywords from the text information of the link 1 (701) and information including the ortho attribute and store them in the keyword group entry field (723). If the link 1 is composed of information other than text, such as a photograph, information such as an ortho attribute may be used. Further, the keyword extraction unit 710 may analyze the text information 731 and the meta tag information 732 of the linked page 730, extract the keywords, and store them in the keyword group entry field (723).

  In order to extract keywords, in the case of Japanese, morphological analysis may be performed, and algorithms such as utilization conversion and narrowing of keyword candidates using an unnecessary word list may be used.

  As described above, the keyword group corresponding to each link of the web page 700 is accumulated in the link keyword table 720. Note that this link keyword table is used by a keyword importance specifying unit described later, and importance is given to each keyword.

  FIG. 8A shows an example of a technique for specifying the importance of a keyword. The link attention level specifying unit 810 specifies the link attention level for each link ID, and outputs link attention level data. One example of specifying the link attention degree has already been described with reference to FIG.

  Further, the keyword extraction unit 710 extracts a keyword for each link. An example of the processing of the keyword extraction unit 710 has already been described with reference to FIG.

  The keyword importance level specifying unit 830 receives link attention level data 630 from the link attention level specifying unit 810 and data of the link keyword table 720 from the keyword extracting unit 710.

  First, the keyword identification unit 830 uses the link ID to associate a link attention level with each keyword and performs a process of assigning the link attention level to the keyword (831). Next, the keywords are aggregated by adding together the link attention degrees related to the same keyword (832). Next, the keywords are sorted in descending order of the sum of the associated link attention degrees (833).

  Through the above processing, keyword importance data in which keywords are arranged in order of importance is generated and output. This data may be used by a layout changing unit described later. The keyword importance data can be obtained as keyword importance data for a specific user, a user (for example, a male) belonging to a specific set, or all users who have accessed the web page. Needless to say, link attention data can be acquired for various users in the same manner.

  FIG. 8B shows an example of normalizing the link attention level when handling a plurality of pages. This figure shows an example when the link attention degree data 631 of the web page A and the link attention degree data 632 of the web page B are integrated to create the link attention degree data 633. Assume that the total number of clicks on web page A is 1000 and the total number of clicks on web page B is 100. In this case, in order to match the link attention degree of the web page A, normalization is performed by multiplying the link attention degree of the web page B by the ratio 100/1000 of the total number of clicks of the web page A and the web page B. Also good. By performing this operation, the link attention level data of both the web page A and the web page B can be integrated and used for later processing.

<Change of web page layout>
Next, an embodiment in which the layout of a web page is changed using the above-described link attention level data and keyword importance level data will be described with reference to the drawings. In changing the layout of the web page, it is not necessary to execute the processing described below in real time when accessing the web page of a specific user. That is, the web content management server apparatus may perform processing sequentially using log information and event information, and accumulate the processing results of the web page layout information for later use.

  FIG. 9A shows an example of processing related to a change of the web page layout related to a specific user. As already described, the link attention level specifying unit 810 and the keyword importance level specifying unit 830 can output information related to various users. In the embodiment in FIG. 9A, the link attention level specifying unit 810 may output the link attention level data 941 of a specific user and the link attention level data 942 of a plurality of users belonging to the set to the layout changing unit 950. Further, the keyword importance level specifying unit 830 may output the keyword importance level data 943 of a specific user and the keyword importance level data 944 of a plurality of users belonging to the set to the layout change 950. In FIG. 9A, there are two data lines 941 and 942 from the link attention level specifying unit 810, but in reality, one line may be used. This is because the two data may be transmitted in series. This also applies to the keyword importance specifying unit 830.

  The layout changing unit 950 may receive these inputs by SW1 to SW4 having a switch function, and may select only necessary inputs by receiving setting input by a specific user, for example. For example, by turning off SW1 and SW3, it is possible not to consider data relating to a specific user. In this case, the layout of the web page of a specific user can be a layout that reflects the preferences of a plurality of users belonging to the set. On the contrary, if SW2 and SW4 are turned off, it is possible to have a web page layout that strongly reflects the preference relating to a specific user.

  Next, for each input data 941 to 944, a threshold value setting unit 951 may set a threshold value for each input data based on, for example, user input settings. This threshold is limited to link attention data having an attention level equal to or higher than a specific value, or only to keywords having an importance level equal to or higher than a specific value, and is used for subsequent layout change processing. May be. Alternatively, information relating to a specific user may be used as a threshold for checking whether the number of times a web page is viewed, the number of clicks, etc. is a certain value or more (in this case, the number of clicks is set as a threshold). It may be given to the part 951 (not shown). If these values are below a certain value, it means that the data parameter is small. In this case, the acquired data may not be used, and the layout may not be changed.

  Thereafter, the processing order of each input data may be determined by the processing order setting unit 952 based on, for example, user input settings. For example, a case is assumed where data of users belonging to a set is processed first, and then data of a specific user is processed. In this case, since the data of the specific user is processed later, the specific user's preference will be more strongly reflected in the layout.

  Note that if the above setting is not performed by the user, a default value may be set. Alternatively, an operator who manages the web page may set. In the process of layout change 953, the relationship between the link and the keyword stored in the link keyword table may be used.

  Also, a link may be stocked, and a web page may be constructed by selecting a page that matches the user's preference, including the stocked link. Alternatively, the search engine may be made to search for a keyword having high importance, and the search result may be displayed in an area on the web page.

  The layout data may be stored together with a user ID (or user set ID) of a specific user, a keyword preferred by a specific user (or specific user set), and the like (960).

  FIG. 9B shows a web page in which the layout of the web page shown in FIG. 3A is changed by the above-described layout change process. For example, the link 1 has a high link attention level of 17% as shown in FIG. 6C, whereas the link 3 is 4% as shown in FIG. 6C. Has low link attention. As a result, in FIG. 9B, the link of the area 901 and the area 903 is switched, and the layout of the web page is changed so that the link 1 is located in the area 903. As illustrated in FIG. 6B, the region 903 is a region where the predicted value of the click ratio is high and is easy to click. In this area 903, the link 1 having the highest link attention degree on the web page is arranged. By changing the layout, a user-friendly My Web page can be constructed. And this my web page should just be provided to a corresponding user.

If there is an area whose layout is to be fixed, the layout changing unit may designate the area so that the layout cannot be changed.
<System configuration>
FIG. 10 shows an example of a functional block diagram in each of the embodiments described above. Log information and event information from the web server device 120 are input to the click count unit 1010 of the web content server device 140, and the number of clicks for each link is counted. This count is sent to the link attention level specifying unit 810.

  The click ratio predicted value calculation unit 1020 calculates the predicted value of the click ratio of each link using the layout information from the layout change unit 950. This predicted value is sent to the link attention level identification unit 810.

  The link attention level specifying unit 810 receives these inputs and specifies the attention level of each link. Then, the link attention level data specified by the link attention level specifying unit 810 is output and input to the layout changing unit 950. The link attention level data may be used for web design or the like as important data when building a web page.

  Note that the link attention level normalization unit 811 performs processing for integrating link attention level data obtained from, for example, a plurality of web pages. Specifically, for example, the processing shown in FIG. 8B is performed as described above, and the value of the link attention level data is adjusted using the total number of clicks of each web page, so that the links existing in a plurality of web pages Integrate link attention. The integrated data is returned to the link attention level identifying unit 810.

  The keyword extraction unit 710 extracts the keyword of each link by executing the process shown in FIG. 7 using the web layout information and the actual web page data, and stores it in the link keyword table 720.

  The keyword importance specifying unit 830 specifies the importance of each keyword using information in the link keyword table 720 and link attention degree data. An example of this processing has been described with reference to FIG. 8A.

  The layout changing unit 950 changes the layout of the web page using link attention level data, keyword importance level data, information in the link keyword table 720, and the like. An example of this process is shown in FIG. 9A. The layout changing unit 950 may store web layout data (1) and web layout data (2) storing layout information and the like.

  The web page layout information changed by the layout changing unit 950 is used by the web server device 120, the click ratio predicted value calculation unit 1020, the keyword extraction unit 710, and the like.

  FIG. 11 shows a flow of a method as an embodiment. Note that the method according to the appended claims can be carried out by changing the order of the steps as long as there is no contradiction. Further, it goes without saying that a method in which the order of the steps is changed belongs to the technical scope of the invention of the method according to the appended claims.

  In step 1102, the number of user clicks for each link on the web page is counted.

  In step 1104, a predicted value of the ratio of clicks for each link is calculated. In this calculation, the calculation may be performed based on the contrast distribution of the web page and the ease of movement of the mouse cursor.

  In step 1106, the attention level of each link is specified. The degree of attention may be specified from the actual number of clicks of each link on the web page and the predicted value of the click ratio. The attention level of each link may be output to the outside.

  In step 1108, keywords for each link may be extracted. The keyword may be extracted from the text information of the link and the ortho attribute. In addition, keywords may be extracted by analyzing text information and meta tag information of a linked page.

  In step 1110, the importance of each keyword is identified. This keyword importance level information may be output to the outside.

  In step 1112, the layout of the web page may be changed. The layout change may be executed based on at least one of the attention level of each link and the importance level of each keyword.

  It should be noted that real-time processing is not necessarily required for each processing related to acquisition of link attention and keyword importance and change of web page layout described in this embodiment. In other words, the web content management server device 140 may perform processing at a necessary timing or interval using information including log information and event information accompanying access to the web page, and store the processing result. Then, link attention level data, keyword importance level data, web page layout information, and the like may be read from the storage device and used at a necessary timing.

  Each embodiment has been described above. However, it should be noted that each embodiment is not exclusive, and components of each embodiment can be selected and appropriately combined as necessary. .

Regarding the above embodiment, the following additional notes are disclosed.
(Appendix 1)
An instruction count unit that counts the number of user instructions for each link on the web page;
From the web page, an instruction ratio predicted value calculation unit that calculates a predicted value of an instruction ratio for each link;
A link attention level identifying unit that identifies a degree of attention of each link based on the number of times of instruction and a predicted value of the ratio of the instructions;
The server apparatus characterized by having.
(Appendix 2)
A keyword extractor for extracting one or more keywords related to each link;
A keyword importance level identifying unit that identifies the importance level of each keyword based on the attention level and each keyword corresponding to each link;
The server device according to supplementary note 1, further comprising:
(Appendix 3)
The link attention level specifying unit

ここで、
Ｆ_ｍは、全てのリンクの数をＭ個とした場合のｍ番目のリンクの注目度、
ｎ_ｍは、前記ｍ番目のリンクの指示回数、
α_ｍは、前記ｍ番目のリンクの指示の比率の予測値、
βは０＜βを満足する定数、
によってリンクの注目度を算出することを特徴とする付記１又は２に記載のサーバ装置。
（付記４）
前記指示の比率の予測値は、前記各リンクの、面積、位置、及び背景とのコントラスト比のうち少なくともいずれか一つに基づき決定される、付記１乃至３のうちいずれか１項に記載のサーバ装置。
（付記５）
前記各キーワードは、前記リンクを構成する文字列を含む情報から、及び／又は、前記リンクが指し示すコンテンツから、抽出することを特徴とする付記１乃至４のうちいずれか１項に記載のサーバ装置。
（付記６）
複数のウエブページの各々の指示回数に応じてリンクの注目度を重み付けすることによって、前記複数のウエブページに存在する前記リンクの各々の注目度を正規化するリンク注目度正規化部、
を更に有することを特徴とする付記１乃至５のうちいずれか１項記載のサーバ装置。
（付記７）
前記ユーザは、全てのユーザ、女性のユーザ、男性のユーザ、特定の年齢層のユーザ、特定の時間帯に前記ウエブページにアクセスするユーザのうちのいずれか一つである、付記１乃至６のうちいずれか１項記載のサーバ装置。
（付記８）
コンピュータにより実行される方法であって、
ウエブページの各リンクに対するユーザの指示回数をカウントし、
前記ウエブページから、前記各リンクに対する、指示の比率の予測値を算出し、
前記指示回数と、前記指示の比率の予測値とに基づいて、前記各リンクの注目度を特定することを特徴とする方法。
（付記９）
前記各リンクに関連する１つ以上のキーワードを抽出し、
前記各リンクに対応する、前記注目度と、前記各キーワードとに基づいて、前記各キーワードの重要度を特定することを特徴とする付記８に記載の方法。
（付記１０）
前記リンク注目度を特定する処理は、

here,
F _m is the degree of attention of the m-th link when the number of all links is M,
n _m is the number of instructions of the m-th link,
α _m is a predicted value of the ratio of the indication of the m-th link,
β is a constant satisfying 0 <β,
3. The server device according to appendix 1 or 2, wherein the attention level of the link is calculated by:
(Appendix 4)
The predicted value of the ratio of the indication is determined based on at least one of the area, the position, and the contrast ratio of the background of each link, according to any one of supplementary notes 1 to 3. Server device.
(Appendix 5)
5. The server device according to claim 1, wherein each keyword is extracted from information including a character string constituting the link and / or from content indicated by the link. .
(Appendix 6)
A link attention degree normalization unit that normalizes the attention degree of each of the links existing in the plurality of web pages by weighting the attention degree of the link according to the number of times of instruction of each of the plurality of web pages;
The server device according to any one of appendices 1 to 5, further comprising:
(Appendix 7)
The user is any one of all users, female users, male users, users of a specific age group, and users who access the web page at a specific time period. The server apparatus of any one of them.
(Appendix 8)
A method performed by a computer,
Count the number of user instructions for each link on the web page,
From the web page, calculate a predicted value of the ratio of instructions for each link,
A method of identifying the degree of attention of each link based on the number of times of instructions and a predicted value of the ratio of the instructions.
(Appendix 9)
Extracting one or more keywords related to each link;
9. The method according to claim 8, wherein the importance level of each keyword is specified based on the attention level and the keywords corresponding to the links.
(Appendix 10)
The process of specifying the link attention level is as follows:

ここで、
Ｆ_ｍは、全てのリンクの数をＭ個とした場合のｍ番目のリンクの注目度、
ｎ_ｍは、前記ｍ番目のリンクの指示回数、
α_ｍは、前記ｍ番目のリンクの指示の比率の予測値、
βは０＜βを満足する定数、
によってリンクの注目度を算出することを特徴とする付記８又は９に記載の方法。
（付記１１）
前記指示の比率の予測値は、前記各リンクの、面積、位置、及び背景とのコントラスト比のうち少なくともいずれか一つに基づき決定される、付記８乃至１０のうちいずれか１項に記載の方法。
（付記１２）
前記各キーワードは、前記リンクを構成する文字列を含む情報から、及び／又は、前記リンクが指し示すコンテンツから、抽出することを特徴とする付記８乃至１１のうちいずれか１項に記載の方法。
（付記１３）
複数のウエブページの各々の指示回数に応じてリンクの注目度を重み付けすることによって、前記複数のウエブページに存在する前記リンクの各々の注目度を正規化することを特徴とする付記８乃至１２のうちいずれか１項記載の方法。
（付記１４）
前記ユーザは、全てのユーザ、女性のユーザ、男性のユーザ、特定の年齢層のユーザ、特定の時間帯に前記ウエブページにアクセスするユーザのうちのいずれか一つである、付記８乃至１３のうちいずれか１項記載の方法。
（付記１５）
ウエブページの各リンクに対するユーザの指示回数をカウントし、
前記ウエブページから、前記各リンクに対する、指示の比率の予測値を算出し、
前記指示回数と、前記指示の比率の予測値とに基づいて、前記各リンクの注目度を特定する、
処理をコンピュータに実行させるプログラム。
（付記１６）
前記各リンクに関連する１つ以上のキーワードを抽出し、
前記各リンクに対応する、前記注目度と、前記各キーワードとに基づいて、前記各キーワードの重要度を特定する、
処理を更にコンピュータに実行させることを特徴とする付記１５に記載のプログラム。

here,
F _m is the degree of attention of the m-th link when the number of all links is M,
n _m is the number of instructions of the m-th link,
α _m is a predicted value of the ratio of the indication of the m-th link,
β is a constant satisfying 0 <β,
10. The method according to appendix 8 or 9, wherein the attention degree of the link is calculated by:
(Appendix 11)
The predicted value of the indication ratio is determined based on at least one of an area, a position, and a contrast ratio of each link, according to any one of appendices 8 to 10. Method.
(Appendix 12)
The method according to any one of appendices 8 to 11, wherein each keyword is extracted from information including a character string constituting the link and / or from content indicated by the link.
(Appendix 13)
Appendices 8 to 12, wherein the attention degree of each of the links existing on the plurality of web pages is normalized by weighting the attention degree of the link according to the number of times each of the plurality of web pages is designated. The method of any one of these.
(Appendix 14)
The user is any one of all users, female users, male users, users of a specific age group, and users who access the web page at a specific time period. The method of any one of them.
(Appendix 15)
Count the number of user instructions for each link on the web page,
From the web page, calculate a predicted value of the ratio of instructions for each link,
Based on the number of instructions and the predicted value of the ratio of the instructions, the degree of attention of each link is specified.
A program that causes a computer to execute processing.
(Appendix 16)
Extracting one or more keywords related to each link;
Identifying importance of each keyword based on the degree of attention and each keyword corresponding to each link;
The program according to appendix 15, further causing the computer to execute processing.

110 Network 120 Web server device 122 Storage device A
124 Web content 126 Access log 131 Client 140 Web content management server device 142 Storage device B

Claims (8)

An instruction count unit that counts the number of user instructions for each link on the web page;
An indication ratio in which the web page is divided into a plurality of areas, and an instruction ratio prediction value for each link area is calculated from a predicted value of an instruction ratio for each area and a position of each area and each link area A predicted value calculation unit;
A link attention level identifying unit that identifies a degree of attention of each link based on the number of instructions;
A layout changing unit that performs a layout change to replace a link with a high link attention level with a link at a position where the predicted value of the ratio of instructions to the link area is higher than a link with a low link attention level;
The server apparatus characterized by having. An instruction count unit that counts the number of user instructions for each link on the web page;
From the web page, an instruction ratio prediction value calculation unit that calculates an instruction ratio prediction value for each link area;
A link attention level identifying unit that identifies a degree of attention of each link based on the number of instructions;
A layout changing unit that performs a layout change to replace a link with a high link attention level with a link at a position where the predicted value of the ratio of instructions to the link area is higher than a link with a low link attention level;
A keyword extractor for extracting one or more keywords related to each link;
A keyword importance level identifying unit that identifies the importance level of each keyword based on the attention level and each keyword corresponding to each link;
The server apparatus characterized by having. The link attention level specifying unit
3. The server device according to claim 1, wherein a degree of attention of each link is specified based on the number of instructions and a predicted value of the ratio of the instructions. Each keyword, from the information including the character string constituting the link, and / or server device according to claim 1 or 3, characterized in that the content which the link points is extracted. 5. The server device according to claim 2 , wherein the predicted value of the indication ratio is determined based on at least one of an area, a position, and a contrast ratio of the area of each link. A link attention degree normalization unit that normalizes the attention degree of each of the links existing in the plurality of web pages by weighting the attention degree of the link according to the number of times of instruction of each of the plurality of web pages;
The server device according to any one of claims 1 to 5 , further comprising: Counting the instruction number of the user for each link c Ebupeji,
The web page is divided into a plurality of areas, and from the predicted value of the ratio of the instruction for each area and the position of the area of each area and each link, the predicted value of the instruction ratio for the area of each link is calculated,
Based on the number of instructions, specify the degree of attention of each link,
Change the layout to replace a link with a high degree of link attention with a link with a higher predicted value of the ratio of instructions to the link area than a link with a low degree of link attention
A program that causes a computer to execute processing . Count the number of user instructions for each link on the web page,
From the web page, calculate the predicted value of the ratio of instructions for the area of each link,
Based on the number of instructions, specify the degree of attention of each link,
Link prominence is a high link, have rows layout change to replace the predicted value of the ratio of instructions to the region of the link than the lower link prominence links and link high position,
Extracting one or more keywords related to each link;
Identifying importance of each keyword based on the degree of attention and each keyword corresponding to each link;
A program that causes a computer to execute processing.

Images