JP2022012066A - Recommendation system and method - Google Patents

Abstract

To provide a system and method for recommending an item to a user dynamically.SOLUTION: A recommendation system 10 includes: a recommendation candidate acquisition unit 104 which acquires a first item which is at least one of a recommendation candidate determined for each item and a recommendation candidate determined for each user, and executes search with a search query to acquire a second item; a score addition unit 107 which adds scores to the first item and the second item; and a distribution unit 109 which distributes the item with a higher score added by the score addition unit 107.SELECTED DRAWING: Figure 2

Description

The present invention relates to a system and a method of recommending an item to a user.

Conventionally, a recommendation system (also referred to as a recommendation system or a recommendation engine) that recommends an object (hereinafter, also referred to as an item) that the user may be interested in to the user is known. As a method of recommender system, we recommend items related to the item that the user is browsing (also called content-based filtering), and recommend items based on the operation history such as browsing of people who have similar tastes to the user. A method (also called collaborative filtering) is known (see Patent Document 1). Further, a hybrid recommender system that combines these methods is known.

Japanese Unexamined Patent Publication No. 2018-181135

However, in the conventional hybrid recommender system, a predetermined item (that is, an item or an item to be recommended for each user) is recommended to the user. Therefore, when recommending an item, it is not easy to dynamically change the item recommended to the user.

Therefore, an object of the present invention is to construct a system that can dynamically recommend an item to a user.

The recommendation system according to the embodiment of the present invention acquires and searches the first item, which is at least one of the recommendation candidate defined for each target item and the recommendation candidate defined for each target user. A recommendation candidate acquisition unit that executes a search using a query to acquire a second item, a scoring unit that assigns a score to the first item and the second item, and a scoring unit that assigns a score to the first item and the second item. It is equipped with a distribution unit that distributes items with the highest score.

According to the present invention, items can be dynamically recommended to the user.

It is a block diagram of the whole which concerns on one Embodiment of this invention. It is a functional block diagram of the recommender system which concerns on one Embodiment of this invention. This is an example of a search index according to an embodiment of the present invention. It is a figure for demonstrating the data stored in the search index storage part which concerns on one Embodiment of this invention. It is a figure for demonstrating the data stored in the score storage part which concerns on one Embodiment of this invention. It is a figure for demonstrating the data stored in the item axis recommendation candidate storage part which concerns on one Embodiment of this invention. It is a figure for demonstrating the data stored in the user axis recommendation candidate storage part which concerns on one Embodiment of this invention. It is a sequence diagram of the construction of the search query and the item recommendation (mixing with the item axis recommendation candidate) processing which concerns on one Embodiment of this invention. It is a sequence diagram of the construction of the search query and the item recommendation (mixing with the user axis recommendation candidate) processing which concerns on one Embodiment of this invention. It is a flowchart of the search query construction process which concerns on one Embodiment of this invention. It is a flowchart of the item recommendation (mixing with the item axis recommendation candidate) process which concerns on one Embodiment of this invention. It is a flowchart of the item recommendation (mixing with a user axis recommendation candidate) process which concerns on one Embodiment of this invention. It is a hardware block diagram of the recommender system, the media operator terminal, and the user terminal which concerns on one Embodiment of this invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<Explanation of terms>
The terms used herein will be described.
-"Items" are objects handled by recommender systems. For example, the item is a product if it is an EC (electronic commerce) site and a video if it is a video distribution site, but the item is not limited to these.
-A "media operator" is an operator of a website (for example, an EC site, a video distribution site, etc., also called a media) that introduces a recommender system.
-A "user" is a person who uses a website (for example, an EC site, a video distribution site, etc., also called a media) that introduces a recommender system.

<Overall composition>
FIG. 1 is an overall configuration diagram according to an embodiment of the present invention. As shown in FIG. 1, the recommender system 10 is communicably connected to the media operator terminal 20 and the user terminal 30 via an arbitrary network. Each will be described below.

The recommendation system 10 is a system that recommends items to the user 31. Specifically, the recommendation system 10 is a target item (an item that is a basis for selecting a recommended item. For example, a user can use a website (for example, an EC site, a video distribution site, etc.) that introduces a recommendation system. Recommendation candidates defined for each item being viewed) and target users (users who are the basis for selecting recommended items. For example, websites that introduce recommender systems (for example, EC sites, video distribution sites). Etc.) can be recommended to the user 31 by adding the item searched by using the search query to at least one of the recommendation candidates determined for each user). The recommender system 10 comprises one or more computers. Later, the recommendation system 10 will be described in detail with reference to FIG. 2.

The media operator terminal 20 is a terminal operated by the media operator 21. The media operator 21 can specify a condition for constructing a search query by using the media operator terminal 20. For example, the media operator terminal 20 is a personal computer, a tablet, a smartphone, or the like.

The user terminal 30 is a terminal operated by the user 31. The user 31 can browse the items recommended by the recommender system 10 on a website (for example, an EC site, a video distribution site, etc.) in which the recommender system 10 is introduced by using the user terminal 30. For example, the user terminal 30 is a personal computer, a tablet, a smartphone, or the like.

FIG. 2 is a functional block diagram of a recommender system 10 according to an embodiment of the present invention. As shown in FIG. 2, the recommendation system 10 includes an item axis recommendation candidate storage unit 101, a user axis recommendation candidate storage unit 102, a search index storage unit 103, a recommendation candidate acquisition unit 104, a search query construction unit 105, and a search. The query storage unit 106, the score giving unit 107, the score storage unit 108, and the distribution unit 109 can be provided. Further, the recommender system 10 can function as a recommendation candidate acquisition unit 104, a search query construction unit 105, a score giving unit 107, and a distribution unit 109 by executing a program. Each will be described below.

First, the data managed by the recommender system 10 will be described with reference to FIGS. 3 to 7.

FIG. 3 is an example of a search index according to an embodiment of the present invention. In one embodiment of the present invention, an index for searching for an item is managed by a key-value method. In the example of FIG. 3, the key is the item "id (identifier)", "name (name)", "description (description)", "price (price)", "maker (manufacturer)", "category_id (category)". (Identifier) ”,“ profit_ratio (profit margin) ”,“ stock_quantity (stock quantity) ”.

FIG. 4 is a diagram for explaining data stored in the search index storage unit 103 according to the embodiment of the present invention. In the search index storage unit 103, for each item (item in which "name (name)" is a, b, c, d, e in the example of FIG. 4), the key of the search index (in the example of FIG. 4, the search index storage unit 103 is used. The values of "name (name)", "category_id (category identifier)", "price (price)", and "stock_quantity (stock quantity)") are stored.

FIG. 5 is a diagram for explaining data stored in the score storage unit 108 according to the embodiment of the present invention. In the score storage unit 108, each item (in the example of FIG. 5, "name (name)" is a, b, c, d, e) for each user (in the case of user A in the example of FIG. 5). The information necessary to calculate the score of the item) is stored. Alternatively, the score may be calculated and stored in advance. It is assumed that the larger the score value, the higher the possibility that the user 31 is interested in the item.

For example, the score stored in the score storage unit 108 may be calculated based on the user's attributes (age, gender, etc.), or based on the user's behavior history (purchase history, browsing history, etc.). It may be calculated, or it may be calculated based on the user attribute and the action history.

FIG. 6 is a diagram for explaining data stored in the item axis recommendation candidate storage unit 101 according to the embodiment of the present invention. The item axis recommendation candidate storage unit 101 is filled with "target items (items that are the basis for selecting recommended items. For example, a user on a website that introduces a recommendation system (for example, an EC site, a video distribution site, etc.)". (Items being browsed by) "and a predetermined number of" recommendation candidates "are stored in association with each other in advance. For example, the "recommendation candidate" is an item having similar characteristics to the "target item", an item related to the "target item", and the like.

Each nomination candidate has a pre-calculated score. For example, the score indicates the degree to which the characteristics of the "target item" and the "recommendation candidate" are similar, the degree to which the "target item" and the "recommendation candidate" are related, and the like. In the item axis recommendation candidate storage unit 101, an item having a large score value (that is, an item having similar characteristics to the target item and an item related to the target item) is stored as a recommendation candidate.

FIG. 7 is a diagram for explaining data stored in the user axis recommendation candidate storage unit 102 according to the embodiment of the present invention. In the user axis recommendation candidate storage unit 102, "a target user (a user who is a source for selecting a recommended item. For example, a website for introducing a recommender system (for example, an EC site, a video distribution site, etc.) is browsed." A predetermined number of "recommendation candidates" are associated with each other in advance and stored. For example, the "recommendation candidate" is an item based on the operation history of a person whose taste (for example, operation history or attribute) is similar to that of the "target user" (for example, an item that the person has viewed). A person having similar tastes to the "target user" may include the target user himself / herself.

Each nomination candidate has a pre-calculated score. For example, the score indicates the degree to which the tastes of the "target user" and the "person whose taste is similar to that of the target user" are similar. In the user axis recommendation candidate storage unit 102, an item having a large score value (that is, an item based on the operation history of a person having similar tastes to the target user) is stored as a recommendation candidate.

Returning to the description of FIG. The recommendation candidate acquisition unit 104 acquires an item (also referred to as a first item) that is at least one of a recommendation candidate defined for each target item and a recommendation candidate defined for each target user. Further, the recommendation candidate acquisition unit 104 executes a search using a search query to acquire an item (also referred to as a second item). Hereinafter, it will be described in detail.

<< Acquisition of item axis recommendation candidates >>
The recommendation candidate acquisition unit 104 is an item that is viewed on a website (for example, an EC site, a video distribution site, etc.) that introduces the recommendation system 10 from among the recommendation candidates stored in the item axis recommendation candidate storage unit 101. Get the "recommendation candidate" when (that is, the item displayed by the website) is the "target item".

As described with reference to FIG. 6, the item axis recommendation candidate storage unit 101 stores the target item and a predetermined number of recommendation candidates in advance in association with each other.

<< Acquisition of user axis recommendation candidates >>
The recommendation candidate acquisition unit 104 is a user who is browsing a website (for example, an EC site, a video distribution site, etc.) that introduces the recommendation system 10 from among the recommendation candidates stored in the user axis recommendation candidate storage unit 102. Acquires a "recommendation candidate" when 31 is a "target user". For example, a user 31 who is browsing a website can be specified by using a cookie or the like.

As described with reference to FIG. 7, the user axis recommendation candidate storage unit 102 stores the target user and a predetermined number of recommendation candidates in advance in association with each other.

<< Execution of search >>
The recommendation candidate acquisition unit 104 executes a search using a search query to acquire an item. Specifically, the recommendation candidate acquisition unit 104 is an item that satisfies the condition of the search query based on the search query stored in the search query storage unit 106 and the search index stored in the search index storage unit 103. To extract.

As described with reference to FIGS. 3 and 4, the search index storage unit 103 stores the search index key and its value for each item.

The search query storage unit 106 stores a search query constructed by the search query construction unit 105 in response to an instruction from the media operator terminal 20.

The search query construction unit 105 constructs a search query in response to an instruction from the media operator terminal 20. Further, the search query construction unit 105 stores the constructed search query in the search query storage unit 106. The search query will be described in detail below.

<< Search query based on the conditions specified by the media operator >>
The search query construction unit 105 can use conditions such as keywords received from the media operator terminal 20 to construct a search query.

For example, the media operator 21 can specify conditions related to basic information such as an item name (product name, etc.), an item description (product description, etc.), an item price, an item manufacturer, and an item category.

Further, for example, the media operator 21 can specify conditions related to the sales of the current month for each item / category and the degree of achievement of the sales target for each item / category.

Further, for example, the media operator 21 can specify conditions relating to the gross profit of the item and the profit margin of the item.

Further, for example, the media operator 21 can specify conditions related to information related to physical distribution such as the number of stocks of each item distribution center (DC). For example, the search query construction unit 105 can construct a search query for extracting an item having an excessive inventory in a DC close to the delivery destination to the user 31.

Further, for example, the media operator 21 can specify a condition regarding a tag attached to an image of an item.

For example, the search query construction unit 105 has an item whose "price (price)" is "1000" or more among items whose "" category_id (category identifier) "is" 5001 "based on the designation of the media operator 21. You can build a search query that says "find an item" or "" description "contains the characters" ○○○ "and" stock_quantity "is" 1 or more "". ..

In this way, by using the search query constructed based on the conditions such as the keywords specified by the media operator 21, the items that the media operator 21 wants to recommend can be mixed and recommended. That is, the item desired by the media operator 21 can be dynamically recommended when the media operator 21 desires.

<< Search query based on search keywords entered by the user in the past >>
The search query construction unit 105 is a keyword (that is, a product or a video) that the user 31 has entered for a search in the past (for example, the latest) on a website (for example, an EC site, a video distribution site, etc.) that introduces the recommendation system 10. The keyword you entered to search for) can be used to build a search query. For example, if the user 31 has entered the keyword "XXX" for a search in the past (for example, most recently), the search query construction unit 105 asks the user 31 to "name (name)". Alternatively, you can build a search query that says "find items that contain the characters" XXX "in" description "."

In this way, by using a search query constructed based on search keywords that the user 31 has entered in the past (for example, most recently), items that are more likely to be of interest to the user 31 are recommended. be able to. That is, it is possible to dynamically recommend items that the user 31 is likely to be interested in.

The score giving unit 107 assigns a score to the recommendation candidate acquired by the recommendation candidate acquisition unit 104 and the item searched by the recommendation candidate acquisition unit 104. Specifically, the score giving unit 107 gives a score by referring to the score of the user 31 stored in the score storage unit 108. Hereinafter, it will be described in detail.

<< Giving scores to item axis recommendation candidates >>
The score giving unit 107 assigns the score of the user 31 stored in the score storage unit 108 to the recommendation candidate acquired by the recommendation candidate acquisition unit 104. Specifically, the score giving unit 107 is not a score stored in the item axis recommendation candidate storage unit 101 but a user 31 stored in the score storage unit 108 in the recommendation candidate acquired by the recommendation candidate acquisition unit 104. Give a score of.

For example, in the examples of FIGS. 5 and 6, the score of "d", which is a recommendation candidate when the "target item" is "a", is "0.7", and the score of "c" is "0.6". "It is. These scores are used as the scores of the user 31 stored in the score storage unit 108 (in the example of FIG. 5, the score of "d" is "0.5" and the score of "c" is "0.3"). It will be re-rolled.

<< Giving scores to user axis recommendation candidates >>
The score giving unit 107 assigns the score of the user 31 stored in the score storage unit 108 to the recommendation candidate acquired by the recommendation candidate acquisition unit 104. Specifically, the score giving unit 107 is not a score stored in the user axis recommendation candidate storage unit 102 but a user 31 stored in the score storage unit 108 in the recommendation candidate acquired by the recommendation candidate acquisition unit 104. Give a score of.

For example, in the examples of FIGS. 5 and 7, when the "target user" is "user A", the score of "a", which is a recommendation candidate, is "0.7", and the score of "d" is "0.". 5 ". These scores are used as the scores of the user 31 stored in the score storage unit 108 (in the example of FIG. 5, the score of "a" is "0.7" and the score of "d" is "0.5"). It will be re-rolled. The score value stored in the user axis recommendation candidate storage unit 102 and the score value stored in the score storage unit 108 may be the same value or different values. good.

<< Giving a score to the searched item >>
The score giving unit 107 assigns the score of the user 31 stored in the score storage unit 108 to the item searched by the recommendation candidate acquisition unit 104. For example, in the example of FIG. 5, if the searched item is "b", a score "0.6" is given to the item "b".

In this way, a plurality of items that are candidates for items recommended to the user (that is, the recommendation candidates in the item axis recommendation candidate storage unit 101, the recommendation candidates in the user axis recommendation candidate storage unit 102, and the searched items). Is given a score (that is, a score in the score storage unit 108) indicating how appropriate it is for the user to display these items. Therefore, it becomes possible to handle a plurality of items that are candidates for recommended items in a unified manner.

As described with reference to FIG. 5, the score storage unit 108 stores information necessary for calculating the score of each item for each user. Alternatively, the score may be calculated and stored in advance.

It should be noted that the storage area and the calculation time can be reduced by storing the information necessary for calculating the score in the score storage unit 108 and calculating the score at the time of distribution of the item.

The distribution unit 109 recommends the item. Specifically, the distribution unit 109 distributes a predetermined number of items having a higher score given by the score giving unit 107 to the user terminal 30 (for example, displayed on the screen of the user terminal 30).

<Processing method>
Hereinafter, the processing method will be described with reference to FIGS. 8 to 12.

FIG. 8 is a sequence diagram of a search query construction and item recommendation (mixing with item axis recommendation candidates) processing according to an embodiment of the present invention. The explanation is divided into <construction of search query> and <recommendation of items>.

<Building search queries>
In step 11 (S11), the recommender system 10 acquires conditions such as keywords used for constructing a search query from the media operator terminal 20.

In step 12 (S12), the recommender system 10 is a keyword (that is, a product or a video) that the user 31 has previously input for a search on a website (for example, an EC site, a video distribution site, etc.) in which the recommender system 10 is introduced. The keyword entered to search for) is acquired as a keyword used to construct a search query.

The order of S11 and S12 does not matter.

In step 13 (S13), the recommender system 10 constructs a search query using the keywords and the like acquired in S11 and S12.

<Item recommendation>
In step 14 (S14), it is assumed that the user 31 browses an item on a website (for example, an EC site, a video distribution site, etc.) in which a recommendation system 10 is introduced by using a user terminal 30.

In step 15 (S15), the recommender system 10 is a "recommendation candidate" when the item being viewed in S14 is a "target item" among the recommendation candidates stored in the item axis recommendation candidate storage unit 101. To get.

In step 16 (S16), the recommender system 10 executes a search using a search query to acquire an item. Specifically, the recommendation system 10 extracts items satisfying the conditions of the search query based on the search query stored in the search query storage unit 106 and the search index stored in the search index storage unit 103. do.

In this way, the "recommendation candidate" associated with the predetermined "target item" and the item searched by the search query can be mixed and recommended to the user 31.

In step 17 (S17), the recommender system 10 assigns scores to the recommendation candidates of S15 and the items of S16. Specifically, the recommender system 10 refers to the score of the user 31 stored in the score storage unit 108 and assigns a score.

In step 18 (S18), the recommender system 10 recommends an item. Specifically, the recommender system 10 distributes a predetermined number of items having a higher score given in S17 to the user terminal 30 (for example, displayed on the screen of the user terminal 30).

FIG. 9 is a sequence diagram of a search query construction and item recommendation (mixing with user axis recommendation candidates) processing according to an embodiment of the present invention. The explanation is divided into <construction of search query> and <recommendation of items>.

<Building search queries>
In step 21 (S21), the recommender system 10 acquires conditions such as keywords used for constructing a search query from the media operator terminal 20.

In step 22 (S22), the recommender system 10 is a keyword (that is, a product or a video) that the user 31 has previously input for a search on a website (for example, an EC site, a video distribution site, etc.) in which the recommender system 10 is introduced. The keyword entered to search for) is acquired as a keyword used to construct a search query.

The order of S21 and S22 does not matter.

In step 23 (S23), the recommender system 10 constructs a search query using the keywords and the like acquired in S21 and S22.

<Item recommendation>
In step 24 (S24), it is assumed that the user 31 uses the user terminal 30 to browse a website (for example, an EC site, a video distribution site, etc.) in which the recommendation system 10 is introduced.

In step 25 (S25), the recommender system 10 is a "recommendation candidate" when the user 31 browsing in S24 is a "target user" from among the recommendation candidates stored in the user axis recommendation candidate storage unit 102. "Get.

In step 26 (S26), the recommender system 10 executes a search using a search query to acquire an item. Specifically, the recommendation system 10 extracts items satisfying the conditions of the search query based on the search query stored in the search query storage unit 106 and the search index stored in the search index storage unit 103. do.

In this way, the "recommendation candidate" associated with the predetermined "target user" and the item searched by the search query can be mixed and recommended to the user 31.

In step 27 (S27), the recommender system 10 assigns scores to the recommendation candidates of S25 and the items of S26. Specifically, the recommender system 10 refers to the score of the user 31 stored in the score storage unit 108 and assigns a score.

In step 28 (S28), the recommender system 10 recommends an item. Specifically, the recommender system 10 distributes a predetermined number of items having a higher score given in S27 to the user terminal 30 (for example, displayed on the screen of the user terminal 30).

In addition, the recommender system 10 mixes the "recommendation candidate" associated with the "target item", the "recommendation candidate" associated with the "target user", and the item searched by the search query. It can also be configured (that is, a combination of FIGS. 8 and 9).

FIG. 10 is a flowchart of the search query construction process according to the embodiment of the present invention.

In step 101 (S101), the search query construction unit 105 of the recommender system 10 acquires keywords and the like used for constructing the search query. For example, the search query construction unit 105 can acquire conditions such as keywords used for constructing a search query from the media operator terminal 20. For example, the search query construction unit 105 searches for keywords (that is, products and videos) that the user 31 has previously input for a search on a website (for example, an EC site, a video distribution site, etc.) that introduces the recommendation system 10. The keyword entered for this purpose) can be acquired as a keyword used for constructing a search query.

In step 102 (S102), the search query construction unit 105 constructs a search query using the keywords and the like acquired in S101. Then, the search query construction unit 105 stores the constructed search query in the search query storage unit 106.

FIG. 11 is a flowchart of an item recommendation (mixing with an item axis recommendation candidate) process according to an embodiment of the present invention.

In step 201 (S201), the recommendation candidate acquisition unit 104 of the recommender system 10 identifies the item being viewed on the user terminal 30. Specifically, the recommendation candidate acquisition unit 104 is an item viewed by the user 31 on a website (for example, an EC site, a video distribution site, etc.) in which the recommendation system 10 is introduced (that is, an item displayed by the website). ).

In step 202 (S202), the recommendation candidate acquisition unit 104 acquires the recommendation candidates stored in the item axis recommendation candidate storage unit 101. Specifically, the recommendation candidate acquisition unit 104 acquires the "recommendation candidate" when the item specified in S201 is the "target item".

In step 203 (S203), the recommendation candidate acquisition unit 104 executes a search using a search query to acquire an item. Specifically, the recommendation candidate acquisition unit 104 is an item that satisfies the condition of the search query based on the search query stored in the search query storage unit 106 and the search index stored in the search index storage unit 103. To extract.

In step 204 (S204), the score giving unit 107 of the recommender system 10 assigns a score to the recommendation candidate of S202 and the item of S203. Specifically, the score giving unit 107 gives a score by referring to the score of the user 31 stored in the score storage unit 108.

In step 205 (S205), the distribution unit 109 of the recommender system 10 recommends an item. Specifically, the distribution unit 109 distributes a predetermined number of items having a higher score given in S204 to the user terminal 30 (for example, displayed on the screen of the user terminal 30).

FIG. 12 is a flowchart of an item recommendation (mixing with a user axis recommendation candidate) process according to an embodiment of the present invention.

In step 301 (S301), the recommendation candidate acquisition unit 104 of the recommender system 10 identifies the user 31. Specifically, the recommendation candidate acquisition unit 104 uses a cookie or the like to identify a user 31 who is browsing a website (for example, an EC site, a video distribution site, etc.) in which a recommendation system 10 is introduced.

In step 302 (S302), the recommendation candidate acquisition unit 104 acquires the recommendation candidates stored in the user axis recommendation candidate storage unit 102. Specifically, the recommendation candidate acquisition unit 104 acquires the "recommendation candidate" when the user 31 specified in S301 is the "target user".

In step 303 (S303), the recommendation candidate acquisition unit 104 executes a search using a search query to acquire an item. Specifically, the recommendation candidate acquisition unit 104 is an item that satisfies the condition of the search query based on the search query stored in the search query storage unit 106 and the search index stored in the search index storage unit 103. To extract.

In step 304 (S304), the score giving unit 107 of the recommender system 10 assigns a score to the recommendation candidate of S302 and the item of S303. Specifically, the score giving unit 107 gives a score by referring to the score of the user 31 stored in the score storage unit 108.

In step 305 (S305), the distribution unit 109 of the recommender system 10 recommends an item. Specifically, the distribution unit 109 distributes a predetermined number of items having a higher score given in S304 to the user terminal 30 (for example, displayed on the screen of the user terminal 30).

In addition, the recommender system 10 mixes the "recommendation candidate" associated with the "target item", the "recommendation candidate" associated with the "target user", and the item searched by the search query. It can also be configured (that is, a combination of FIGS. 11 and 12).

<Effect>
In one embodiment of the present invention, a recommendation candidate is recommended to a user by adding an item searched by using a search query to at least one of the recommendation candidate defined for each target item and the recommendation candidate defined for each target user. It is possible to dynamically change the item to be used. In other words, you can flexibly generate candidate items to recommend.

<Hardware configuration>
FIG. 13 is a hardware configuration diagram of a recommender system 10, a media operator terminal 20, and a user terminal 30 according to an embodiment of the present invention. The recommendation system 10, the media operator terminal 20, and the user terminal 30 have a CPU (Central Processing Unit) 1001, a ROM (Read Only Memory) 1002, and a RAM (Random Access Memory) 1003. The CPU 1001, ROM 1002, and RAM 1003 form a so-called computer.

Further, the recommendation system 10, the media operator terminal 20, and the user terminal 30 may have an auxiliary storage device 1004, a display device 1005, an operation device 1006, an I / F (Interface) device 1007, and a drive device 1008.

The hardware of the recommender system 10, the media operator terminal 20, and the user terminal 30 are connected to each other via the bus B.

The CPU 1001 is an arithmetic device that executes various programs installed in the auxiliary storage device 1004.

ROM 1002 is a non-volatile memory. The ROM 1002 functions as a main storage device for storing various programs, data, and the like necessary for the CPU 1001 to execute various programs installed in the auxiliary storage device 1004. Specifically, the ROM 1002 functions as a main storage device for storing boot programs such as BIOS (Basic Input / Output System) and EFI (Extensible Firmware Interface).

The RAM 1003 is a volatile memory such as a DRAM (Dynamic Random Access Memory) or a SRAM (Static Random Access Memory). The RAM 1003 functions as a main storage device that provides a work area to be expanded when various programs installed in the auxiliary storage device 1004 are executed by the CPU 1001.

The auxiliary storage device 1004 is an auxiliary storage device that stores various programs and information used when various programs are executed.

The display device 1005 is a display device that displays the internal state of the recommendation system 10, the media operator terminal 20, the user terminal 30, and the like.

The operation device 1006 is an input device in which a person who operates the recommendation system 10, the media operator terminal 20, and the user terminal 30 inputs various instructions to the recommendation system 10, the media operator terminal 20, and the user terminal 30.

The I / F device 1007 is a communication device for connecting to a network and communicating with other devices.

The drive device 1008 is a device for setting the recording medium 1009. The recording medium 1009 referred to here includes a medium such as a CD-ROM, a flexible disk, a magneto-optical disk, or the like that optically, electrically, or magnetically records information. Further, the recording medium 1009 may include a semiconductor memory for electrically recording information such as an EPROM (Erasable Programmable Read Only Memory) and a flash memory.

The various programs installed in the auxiliary storage device 1004 are installed, for example, by setting the distributed recording medium 1009 in the drive device 1008 and reading the various programs recorded in the recording medium 1009 by the drive device 1008. Will be done. Alternatively, various programs installed in the auxiliary storage device 1004 may be installed by being downloaded from the network via the I / F device 1007.

Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above-mentioned specific embodiments, and various modifications are made within the scope of the gist of the present invention described in the claims.・ Can be changed.

10 Recommendation system 20 Media operator terminal 21 Media operator 30 User terminal 31 User 101 Item axis recommendation candidate storage unit 102 User axis recommendation candidate storage unit 103 Search index storage unit 104 Recommendation candidate acquisition unit 105 Search query construction unit 106 Search query storage Unit 107 Score granting unit 108 Score storage unit 109 Distribution unit 1001 CPU
1002 ROM
1003 RAM
1004 Auxiliary storage device 1005 Display device 1006 Operation device 1007 I / F device 1008 Drive device 1009 Recording medium

Claims (10)

A system that recommends items to users
Acquire the first item, which is at least one of the recommendation candidate defined for each target item and the recommendation candidate defined for each target user.
A recommendation candidate acquisition unit that executes a search using a search query and acquires a second item,
A scoring unit that gives a score to the first item and the second item,
A recommendation system including a distribution unit that distributes items with a higher score given by the score-giving unit. It also has a score storage unit that stores the score of each item for each user.
The recommendation system according to claim 1, wherein the score giving unit assigns a score of the user stored in the score storage unit. The recommendation system according to claim 1 or 2, wherein the search query is a query constructed based on the conditions acquired from the media operator terminal. The recommendation system according to claim 3, wherein the search query is a query for extracting items based on the number of items in stock. The recommendation system according to claim 3, wherein the search query is a query for extracting an item based on at least one of the gross profit and the profit margin of the item. The recommendation system according to claim 3, wherein the search query is a query for extracting an item based on at least one of sales of the item and achievement of a sales target. The recommendation system according to claim 3, wherein the search query is a query for extracting an item based on at least one of the name of the item, the description of the item, and the price of the item. The recommendation system according to claim 3, wherein the search query is a query for extracting an item based on a tag attached to an image of the item. The recommendation system according to claim 1 or 2, wherein the search query is a query constructed based on a search keyword that the user has entered for a search in the past. This is the method performed by the system that recommends items to users.
Acquire the first item, which is at least one of the recommendation candidate defined for each target item and the recommendation candidate defined for each target user.
Steps to perform a search using a search query to get a second item,
The step of giving a score to the first item and the second item, and
A method including the step of delivering the item with the highest score given.

Images