JP5451673B2 - Search ranking generation apparatus and method - Google Patents

Description

  The present invention relates to a search ranking generation apparatus and method for generating a ranking of a Web page search system that searches Web pages on the Internet.

  2. Description of the Related Art Conventionally, a technology related to online shopping and online auction in which shopping is performed using a terminal device that can be connected to a web server via a communication network is known (Patent Document 1). Each of the websites (hereinafter referred to as “product sales sites”) that conduct such online shopping and online auctions handles a wide variety of products. Therefore, a user who wants to purchase a product first relates to the product in the search system. It is common to perform a search based on a search query (for example, a product name).

  In this regard, in a conventional search system, a search result with a high number of browsing (clicks) is presumed to be information of general interest and displayed preferentially (for example, see Patent Document 2) or input from the user. The order of display as a search result is determined based on the relationship between the search query and the Web page.

JP-A-11-219389 Japanese Patent Laid-Open No. 2005-190065

  However, although the conventional search system can present a search result ranking a Web page that is highly likely to be clicked by the user, the possibility that the user actually purchases a product on the clicked Web page is Not necessarily correlated.

  Therefore, an object of the present invention is to provide a search ranking generation apparatus and method that can rank Web pages that handle products that are highly likely to be purchased by users in the top of search results.

  (1) A search ranking generation device that generates a ranking of a Web page search system that searches a Web page on the Internet, and responds to an input of a search query for a product from a user terminal in response to the search query. In addition, a search query log database that stores operation history information indicating the operation history of the user terminal including user identification information and a time stamp, and a transition destination web in response to a transition operation from the user terminal to the web page being accepted. A click log database that stores the operation history information in association with a page, and in response to receiving a purchase operation of a product from a user terminal, the operation history information and the product Purchase information log database for storing product attributes and the search query Based on the operation history information and the product attributes stored in the search database, the click log database, and the purchase information log database, a score indicating a high possibility of accepting a purchase operation in association with a search query is displayed. Score calculating means for calculating for each product attribute, and ranking generating means for generating a ranking of the Web page for the search query based on the score calculated by the score calculating means and the product attribute included in the Web page from which the product is purchased. A search ranking generation device comprising:

  According to the search ranking generation device of (1), since the score is generated by referring to the information stored in the purchase information log database storing the purchase record and the ranking is generated, the user may execute the purchase operation. Web pages that handle expensive products can be ranked higher in the search results.

Here, in the search ranking generation device, in addition to the purchase information log database, a search query log database that stores a search query received from the user terminal, and a Web page transition operation (for example, a link click operation) received from the user terminal. A click log database to be stored, and each database is linked by operation history information indicating an operation history of the user terminal. Thereby, the user's operation history can be specified, and it is determined whether the product purchased by the user is based on a series of operations starting from a search operation or based on an operation other than the series of operations. be able to. Then, the search ranking generation device refers to the product attributes of products purchased based on a series of operations starting from the search operation, and evaluates a Web page that sells products with product attributes that are likely to be purchased by the user Is calculated for each search query.
Thus, according to the search ranking generation device, not only the search ranking is generated with reference to the purchase results, but also for each search query, what product attribute contributes to the user's willingness to purchase is specified. Web pages that handle products that are likely to be purchased by the user can be ranked higher in the search results.

  “Product attributes” refers to various information attached to products such as product type, product name, product manufacturer, exhibitor, number of exhibitors, seller evaluation, number of bids, remaining bid time, shipping date and time, price, etc. In addition, the administrator of the search ranking generation apparatus of the present invention arbitrarily sets.

  (2) The score calculation means associates the search query with the operation history based on the operation history information stored in the search query log database, the click log database, and the purchase information log database, and provides the data as training data. The search ranking generation device according to (1), comprising: log analysis means; and learning means for performing machine learning using the provided training data as a feed and calculating the score for each search query.

  According to the search ranking generation device of (2), the learning means performs machine learning using the information associated by the log analysis means as training data, and calculates a significant product attribute score for each search query. Accordingly, significant product attributes can be automatically specified for each search query, and Web pages that handle products that are likely to be purchased by the user can be ranked in the top of the search results.

  (3) A method of generating a ranking of a Web page search system that searches for a Web page on the Internet, which is executed by a computer, in response to receiving an input of a product search query from a user terminal. In response to the step of storing operation history information indicating the operation history of the user terminal including user identification information and a time stamp in association with the search query log database, and receiving a transition operation from the user terminal to the web page, The operation history information is stored in the click log database in association with the transition destination Web page, and the operation is performed in association with the Web page from which the product is purchased in response to receiving the purchase operation of the product from the user terminal. Purchasing information log database with historical information and product attributes The possibility of accepting a purchase operation in association with a search query based on the operation history information and the product attribute stored in the search query log database, the click log database, and the purchase information log database Calculating a web page ranking for a search query based on the calculated score and the product attribute included in the web page from which the product was purchased; Including methods.

  According to the method (3), the same effect as the search ranking generation device (1) can be obtained.

  According to the present invention, it is possible to rank Web pages that handle products that are highly likely to be purchased by the user at the top of the search results.

It is a block diagram which shows the function structure of the search ranking production | generation apparatus of this invention. It is a figure which shows a search query log database. It is a figure which shows a click log database. It is a figure which shows a purchase information log database. It is a figure which shows an association database. It is a figure which shows a score management database. It is a figure which shows the example of a series of operation triggered by search operation. It is a flowchart which shows a ranking production | generation process. It is a block diagram which shows the function structure of the search ranking production | generation apparatus of another embodiment.

  Hereinafter, an embodiment of a search ranking generation device 1 of the present invention will be described with reference to the drawings. The search ranking generation device 1 of the present invention determines a weighted ranking based on the product attributes of the products handled by the Web page (product introduction page) in the search system, that is, the products having the product attributes that contribute to the user's willingness to purchase. A ranking that gives a certain advantage to a Web page to be handled is determined.

[Configuration of Search Ranking Generating Device 1]
With reference to FIG. 1, the structure of the search ranking production | generation apparatus 1 of this invention is demonstrated.
The search ranking generation device 1 is communicably connected to the user terminal 30, and includes a search query reception unit 11, a click reception unit 12, a purchase request reception unit 13, a score calculation unit 14, a ranking generation unit 15, a search The query log database 21, the click log database 22, the purchase information log database 23, the score management database 24, and the product page database 25 are configured.

  The search query receiving unit 11 receives an input of a search query from the user terminal 30 and stores the received search query in the search query log database 21. Here, the search query receiving means 11 receives user identification information (bcookie) for identifying the user (specifically, the user terminal 30), a time stamp indicating the date and time when the input of the search query is received, together with the search query. And stored in the search query log database 21.

  As shown in FIG. 2, the search query log database 21 stores user identification information, a time stamp, and the like in association with the search query received by the search query receiving unit 11. Here, the user identification information, the time stamp, and the like are information used to specify a series of operations of the user terminal 30 triggered by a search operation as described later, and are hereinafter referred to as “operation history information”. The “operation history information” is not limited to only the user identification information and the time stamp, and other information necessary for specifying a series of operations of the user terminal 30 may be used.

  The click receiving unit 12 receives a web page transition operation from the user terminal 30. Here, the “transition operation” refers to a click operation on an anchor text to which a link is set, an operation for inputting a link destination URL into the address bar, or the like. When accepting a transition operation from the user terminal 30, the click accepting unit 12 displays operation history information such as user identification information, a time stamp, and referrer information indicating information of a link source web page (current web page) as a transition destination. The URL is stored in the click log database 22 in association with the URL of the Web page.

  As shown in FIG. 3, the click log database 22 stores operation history information in association with the URL of the Web page that is the transition destination of the transition operation received by the click receiving unit 12. Here, the referrer information is stored in the click log database 22 as operation history information, and is used to specify a series of operations of the user terminal 30 triggered by a search operation. That is, for example, when the referrer information is information indicating a search result screen, it is a transition operation for the Web page presented as the search result, and is a series of operations of the user terminal 30 triggered by the search operation. Can be identified.

  The purchase request accepting unit 13 accepts a purchase operation of a product in the product sales site from the user terminal 30, and corresponds the operation history information and the product attribute of the purchased product to the URL of the product introduction page that introduces the purchased product. At the same time, it is stored in the purchase information log database 23. Here, the purchase operation in the present embodiment includes an operation of clicking a bid button in addition to an operation of clicking a purchase button. That is, the merchandise sales site of the present embodiment includes an online shopping site and an online auction site. However, in the case of an online shopping site, the click of the purchase button corresponds to a purchase operation, and the online sales site. In this case, clicking the bid button corresponds to the purchase operation.

As shown in FIG. 4, the purchase information log database 23 is associated with the URL of the product introduction page that introduces the product purchased by the purchase operation accepted by the purchase request accepting unit 13, and stores the operation history information and the purchased product information. Store product attributes. Here, “product attribute” means various items for specifying a product such as product type, product name, product manufacturer, exhibitor, number of exhibits, seller evaluation, number of bids, remaining bid time, shipping date and time, price, etc. Arbitrary information set by the administrator of the search ranking generation apparatus 1 of the present invention can be used.
The purchase information log database 23 stores referrer information as operation history information, and is used to specify a series of operations of the user terminal 30 triggered by a search operation.

The score calculation means 14 is a score indicating the acceptability of a purchase operation in association with a search query based on operation history information and product attributes stored in the search query log database 21, click log database 22, and purchase information log database 23. Is calculated for each product attribute of the product. That is, the score calculation means 14 calculates a score for evaluating a product attribute that contributes to the user's willingness to purchase for each search query.
In order to calculate such a score, the score calculation unit 14 includes a log analysis unit 141, an association database 142, and a learning unit 143.

The log analysis unit 141 analyzes the operation history information stored in the search query log database 21, the click log database 22, and the purchase information log database 23, and identifies a series of operation histories triggered by the search operation. More specifically, the log analysis unit 141 specifies whether or not a product has been purchased with the search as a trigger.
As an example, the log analysis unit 141 refers to the referrer information and the time stamp of the operation history information having the same user identification information, identifies a series of operation histories of the user terminal 30, and whether the product has been purchased in response to the search. Specify whether or not.

The range of the operation history triggered by the search operation can be arbitrarily set, and will be specifically described with reference to FIG.
7A is a search result page 100, FIG. 7B is a product introduction page 200, and FIG. 7C is a product introduction page 300.

The search result page 100 is a search result page that presents a search result based on the search query “bicycle” in the search system. In FIG. 7A, “bicycle A” “bicycle B” “bicycle C” “ Electric bicycle A "etc. are presented.
The product introduction page 200 is a product introduction page that introduces / sells “Bicycle A”. In addition to the product attribute of “Bicycle A” and the bid button of “Bicycle A”, the recommended product related to “Bicycle A” is displayed. Present. In the product introduction page 200 of FIG. 7B, “bicycle B”, “bicycle C”, “bicycle D”, and the like are presented as recommended products related to “bicycle A”.
The product introduction page 300 is a product introduction page that introduces / sells “Bicycle B”. In addition to the product attribute of “Bicycle B” and the bid button of “Bicycle B”, recommended products related to “Bicycle B” are displayed. Present.

  In FIG. 7, the purchase operation A or the purchase operation B is received from the user terminal 30. With reference to these purchase operation A and purchase operation B, the range of the operation history triggered by the search operation will be described.

  The purchase operation A is a purchase operation for “bicycle A” by clicking a bid button on the product introduction page 200 that is transitioned by clicking the search result “bicycle A” presented on the search result page 100. That is, the purchase operation A is a purchase operation on the product introduction page 200 that has directly transitioned from the search result page 100 by the search operation, and is included in the operation history triggered by the search operation.

The purchase operation B is a product that has been further transitioned by clicking the recommended product “Bicycle B” on the product introduction page 200 that has been transitioned by clicking the search result “Bicycle A” presented on the search result page 100. This is a purchase operation for “bicycle B” by clicking a bid button on the introduction page 300. That is, the purchase operation B is a purchase operation on the product introduction page 300 that has indirectly transited from the search result page 100 by the search operation via the product introduction page 200.
Such a purchase operation on the product introduction page that has indirectly shifted from the search result page may or may not be included as an operation history triggered by the search operation.
For example, when the number of page transitions (determined from the referrer information) from the search result page to the product introduction page that accepted the purchase operation is within a predetermined number, it may be included in the operation history triggered by the search operation. If it is larger, it may not be included in the operation history triggered by the search operation. In addition, when the time from accepting the search operation to accepting the purchase operation (determined from the time stamp) is within a predetermined time, it may be included in the operation history triggered by the search operation. It may not be included in the operation history triggered by the search operation.
Moreover, it is good also as including as operation history triggered by search operation based on the relevance of a search query and the goods of a goods introduction page, and good also as not including. For example, the product bid by the purchase operation B in FIG. 7 is “Bicycle B” and has a certain degree of relevance with the search query “Bicycle”, and therefore included as an operation history triggered by the search operation. And On the other hand, when the purchase operation C for “dry battery Z” is performed repeatedly from the search result page by the search query “bicycle”, the product “dry battery Z” and the search query “bicycle” are constant. Since there is no relevance as described above, the purchase operation C may not be included in the operation history triggered by the search operation.

  When the log analysis unit 141 identifies an operation history triggered by a search operation, the log analysis unit 141 associates the operation history identified for each search query with the product attributes of the products to be introduced / sold on the product introduction page included in the operation history. And stored in the association database 142.

The association database 142 stores an operation history, product attributes, and the like in association with each search query as shown in FIG.
Here, the “operation history” is specified for each product introduction page, whether or not there is a transition to the product introduction page (that is, whether or not a link to the product introduction page has been clicked), and purchase on the product introduction page. At least the presence or absence of operation is included. That is, the “operation history” includes information that has not changed to the product introduction page and information that the product has not been purchased (bidding) on the product introduction page. For example, referring to the purchase operation B in FIG. 7, in the purchase operation B, “bicycle A” “bicycle B” is clicked with respect to the search query “bicycle”, and “bicycle C” “bicycle D” “electric bicycle” is clicked. “A” is not clicked. Therefore, in the association database 142, the presence / absence of a click on the product introduction page introducing “bicycle A” and “bicycle B” becomes “O”, and “bicycle C” “bicycle D” “electric bicycle A” is introduced. The presence / absence of a click on the product introduction page becomes “x”. Further, in the purchase operation B, a bid for “bicycle B” is performed, but a bid for “bicycle A” is not performed. Therefore, the presence / absence of purchase for the product introduction page introducing “bicycle B” is “◯”, and the presence / absence of purchase for the product introduction page introducing “bicycle A” is “x”.

  The association database 142 preferably stores not only the product attributes of products actually purchased (bidding) but also the product attributes of products not purchased (bid) in association with the search query. This is because merchandise attributes that the user has not purchased (bidding) can also be used to identify merchandise attributes that contribute to the user's willingness to purchase. As an example, if the product purchased (bid) at the price “9000 yen” is not purchased (bid) at the price “10000 yen”, the product is purchased when the price is “less than 10,000 yen”. It can be specified that the product is easy to be sold.

  The learning unit 143 receives the information stored in the association database 142 as training data, and calculates a score for evaluating a product attribute that contributes to the user's purchase intention for each search query. The score for each search query calculated by the learning unit 143 is stored in the score management database 24.

As shown in FIG. 6, the score management database 24 stores a score for evaluating a product attribute that contributes to the user's purchase intention for each search query. An example of scores stored in the score management database 24 will be described with reference to FIG.
For example, in the case of the search query “bicycle”, a score is stored such that the product attribute “price” has a great influence, and the product attribute “product type (bicycle, electric bicycle, etc.)” has a moderate influence. The Further, in the case of the search query “antique”, a score is stored such that the product attribute “exhibitor” and “number of evaluations” have a great influence, and the product attribute “price” has little influence. In the case of the search query “AA battery”, the product attributes “product type (AA battery, single battery, etc.)” and “price” have a great influence, and the product attributes “product name” “product manufacturer” Scores such as having little effect are stored.
Note that the score management database 24 shown in FIG. 6 is for facilitating the understanding of the description, and actually stores a predetermined calculation formula for calculating a score for each search query. For example, a score “100” is added to the search query “AA battery” when the product type is “AA battery”, and a score “0” is added when the product type is “single battery”. Score "100" is added when it is "0 to 100 yen", score "50" is added when it is "101 to 200 yen", and score "10" is added when it is "201 to 999 yen" A calculation formula that reflects the degree of influence shown in FIG. 6 is stored for each search query.

The ranking generation unit 15 refers to the score management database 24 and the product page database 25 and generates a ranking of Web pages to be presented as a search result for each search query. Here, the product page database 25 (not shown) stores the product attributes of the products introduced / sold on the product introduction page for each product introduction page in the product sales site.
The ranking generation means 15 performs a predetermined calculation based on the score (calculation formula) stored in the score management database 24 for the product attribute stored in the product page database 25, and ranks the Web page for each search query. Generate.
The ranking generated by the ranking generation unit 15 may be applied only to the search system in the product sales site, or may be applied to a search system used for the entire network.

[Hardware Configuration of Search Ranking Generating Device 1]
The hardware of the search ranking generation device 1 described above can be configured by one or a plurality of general computers. For example, a general computer includes a central processing unit (CPU) as a control unit, and a memory (RAM, ROM), a hard disk (HDD), and an optical disk (CD, DVD, etc.) as a storage unit, and a network communication device. Various wired and wireless LAN devices as display devices, for example, various displays such as liquid crystal displays and plasma displays, and as input devices, for example, keyboards and pointing devices (mouse, tracking ball, etc.) Are connected by a bus line. In such a general computer, the CPU controls the search ranking generation device 1 in an integrated manner, reads and executes various programs as appropriate, and cooperates with the above-described hardware, thereby providing various functions according to the present invention. Is realized.

[Processing of search ranking generation apparatus 1]
Next, the ranking generation process of the search ranking generation device 1 will be described with reference to FIG. The search ranking generation device 1 executes ranking generation processing at a predetermined timing.

S1: First, the score calculation unit 14 (log analysis unit 141) searches the search query log database 21, the click log database 22, and the purchase information log database 23 that store various logs on the network in real time. Extract various information.
S2: Subsequently, the score calculation unit 14 (log analysis unit 141) specifies a series of operation histories triggered by the search operation based on the operation history information, associates them with the search query, and stores them in the association database 142. At this time, the score calculation unit 14 (log analysis unit 141) stores the product attribute of the product introduced / sold on the product introduction page included in the operation history in the association database 142 in association with the search query.

S3: Subsequently, the score calculation means 14 (learning means 143) performs machine learning using the information stored in the association database 142 as training data, and determines the degree of influence of the product attribute that contributes to the user's purchase motivation for each search query. Calculate and store in the score management database 24.
S4: Subsequently, the ranking generation unit 15 refers to the score management database 24 and the product page database 25, and generates a ranking of Web pages to be presented as a search result for each search query.

[Effect of the search ranking generation device 1]
According to the search ranking generation device 1 as described above, since the ranking is generated with reference to the purchase information log (purchase record) stored in the purchase information log database 23, a product that is likely to be purchased by the user in the search system. The product introduction page that introduces can be ranked at the top.

At this time, the search ranking generation device 1 refers to the product attributes of the purchased (or not) products as well as the purchase results, and what product attributes contribute to the user's willingness to purchase for each search query. The ranking is generated after specifying. As a result, the search ranking generation device 1 introduces / sells a product with a product attribute that is predicted to be purchased by the user, rather than simply ranking a product introduction page that introduces / sells a product with a purchase record. The product introduction page to be ranked can be ranked higher, and even a product with no sales record can be appropriately ranked higher in the search results.
As a result, in the search ranking generation device 1, in the search system, a product sales site that introduces / sells a product desired by the user can be appropriately displayed on the top, and the convenience of the user can be improved.

Note that the operation history triggered by the search operation for generating the ranking may be only the operation history triggered by the search operation in the search system in the product sales site, or the search in the search system used for the entire network. The entire operation history triggered by the operation may be used.
At this time, when using the entire operation history triggered by the search operation in the search system used for the entire network, the rate at which the product was purchased triggered by the search operation, or the transition to the product sales site triggered by the search operation It is preferable to calculate a score for each search query in consideration of the ratio and the like. As a result, a score that reflects the user's search intention (that is, whether the user intends to purchase a product or is simply for information search) can be calculated for each search query. As a result, if the search query has a high possibility of purchasing a product, a ranking with a higher weight by the product attribute can be generated, and if the search query has a low possibility of purchase, the weight by the product attribute is increased. A reduced ranking can be generated.

  As mentioned above, although embodiment of this invention was described, this invention is not restricted to embodiment mentioned above. The effects described in the embodiments of the present invention are only the most preferable effects resulting from the present invention, and the effects of the present invention are limited to those described in the embodiments of the present invention. is not.

In the above embodiment, the score for each search query is calculated by machine learning using the information analyzed by the log analysis unit 141 as training data. However, the calculation of the score for each search query is not limited to machine learning. Absent. For example, the score for each search query may be calculated by rule-based determination. FIG. 9 is a block diagram showing a functional configuration of the search ranking generation apparatus 1A of another embodiment that calculates a score for each search query based on such rule-based determination.
The search ranking generation device 1A according to another embodiment includes a rule determination unit 143A instead of the learning unit 143. The rule determination unit 143A calculates a score for each search query based on information (operation history and product attributes) analyzed by the log analysis unit 141 and stored in the association database 142 according to a predetermined rule. As an example, the number of purchases for each product name (purchased or not) is extracted, and the purchase number of products with the highest number of purchases is 80% of the total number of product introduction page displays (clicks) including the product name. If it exceeds, the score that the product name has a big influence on the search query is calculated, and the number of purchases of the top number of purchases by the number of purchases is the number of product introduction pages that contain the product name (number of clicks) If less than 40% of the total, a score is calculated such that the product name has a small effect on the search query.
In addition, the rule determination unit 143A extracts the number of purchases (whether or not purchased) for each product name, and the number of non-purchases of a predetermined number of non-purchased items is the number of displayed product introduction pages including the product name (clicks) The score is calculated so that the product name has a great influence (negative influence) on the search query even when the total number exceeds 80%.
Thus, even if the score is calculated according to a predetermined rule for the information stored in the association database 142, the score management database 24 shown in FIG. Can be generated.

  In addition, the present invention calculates a score for evaluating a product attribute that contributes to a user's willingness to purchase for each search query, and generates a ranking. The information referred to in the above embodiment (for example, search log) The score may be calculated with reference to information other than operation history information and product attributes. For example, the score may be calculated with reference to the ranking displayed as the search result. In other words, the product attributes of products that did not accept purchase operations despite high rankings, and the product attributes of products that accepted purchase operations despite low rankings, have characteristics that contribute more strongly to the user's willingness to purchase. The score may be calculated after performing a predetermined weighting based on the ranking of the search result of the product introduction page that has received (or not received) the purchase operation.

DESCRIPTION OF SYMBOLS 1 Search ranking production | generation apparatus 11 Search query reception means 12 Click reception means 13 Purchase request reception means 14 Score calculation means 141 Log analysis means 142 Association database 143 Learning means 15 Ranking generation means 21 Search query log database 22 Click log database 23 Purchase information log database 24 score management database 25 product page database 30 user terminal

Claims (3)

A search ranking generation device for generating a ranking of a Web page search system for searching for Web pages on the Internet,
A search query log database for storing operation history information indicating an operation history of the user terminal including user identification information and a time stamp in association with the search query in response to receiving an input of a product search query from the user terminal; ,
A click log database that stores the operation history information in association with a transition destination web page in response to accepting a transition operation from the user terminal to the web page;
A purchase information log database that stores the operation history information and the product attributes of the product in association with the Web page from which the product is purchased in response to receiving the purchase operation of the product from the user terminal;
Based on the operation history information and the product attributes stored in the search query log database, the click log database, and the purchase information log database, a purchase operation is accepted in association with the search query stored in the search query log database. A score calculating means for calculating a score indicating a high possibility for each product attribute of the product , and storing the search query and the score for each product attribute in the score management database in association with each other ;
It said score stored in said score management database, and product introduction based on the product attribute included in the page, ranking generating means for generating a ranking of the product introduction page search every query,
A search ranking generation device comprising: The score calculation means includes
Log analysis means for associating a search query with an operation history based on the operation history information stored in the search query log database, the click log database and the purchase information log database, and providing as training data;
Learning means for performing machine learning using the provided training data as a feed, and calculating the score for each search query;
The search ranking generation device according to claim 1 provided with. A method of generating a ranking of a Web page search system that is executed by a computer and searches a Web page on the Internet,
In response to receiving an input of a product search query from the user terminal, operation history information indicating the operation history of the user terminal including user identification information and a time stamp is stored in the search query log database in association with the search query. Steps,
Storing the operation history information in the click log database in association with the transition destination web page in response to accepting the transition operation from the user terminal to the web page;
Storing the operation history information and the product attribute of the product in a purchase information log database in association with the Web page from which the product was purchased in response to receiving the purchase operation of the product from the user terminal;
Based on the operation history information and the product attributes stored in the search query log database, the click log database, and the purchase information log database, a purchase operation is accepted in association with the search query stored in the search query log database. Calculating a score indicating a high possibility for each product attribute of the product , and storing the search query and the score for each product attribute in the score management database in association with each other ;
It said score stored in said score management database, comprising the steps of: based on the product attributes contained in the product introduction page, generates a ranking of the product introduction page search every query,
Including methods.

Images