JP7579729B2 - Recommendation system, recommendation method, and program for recommending products, services, or actions - Google Patents

Description

The present invention relates to a system that recommends products, etc. to users.

Conventionally, recommended products, advertisements, coupons, etc. are delivered to users based on their purchase history on EC (Electronic Commerce) sites and their smartphone GPS behavior history, and although the user can guess to some extent why a particular product is being recommended, the user still receives the information passively. For this reason, a system is known in which users are not passive, but instead directly register their interests with a recommendation service provider.

For example, Patent Document 1 discloses a product recommendation service system that utilizes human sociality. According to this system, a recommendation service provider provides users with an interest information management function, and the users use this function to directly or indirectly register their own interest information with the recommendation service provider. When recommending products, the recommendation service provider uses not only product-related information and user behavior information, but also the user's interest information to extract recommended products and present them to the user together with the interest information. The interest information is then linked to other users, allowing the other users to update their own interest information. The store's location information is then used to send the identified recommendation information to the mobile terminal. For recommended targets on media, authentication is performed and the recommended targets are identified, and recommendation information is also presented.

JP 2017-58893 A

However, even systems like the one in Patent Document 1 above do not allow users to self-analyze their own behavioral tendencies to find out why a particular product or advertisement is being shown to them.

In view of the above problems, the present invention aims to provide a system that can perform self-analysis of the purchasing trends, behavioral trends, etc. that are the basis for the products, services, or actions that are recommended.

To solve the above problems, the present invention provides the following solutions:

(1) A recommendation system that recommends products, services, or actions (products, etc.) to a user, comprising: a purchase history, etc. storage means that stores the purchase history and/or action history of the user; an evaluation information storage means that stores the user's evaluation information on the purchase history and/or action history; a recommendation information generation means that generates recommendation information on the recommended products, etc. by referring to the purchase history and/or the action history stored in the purchase history, etc. storage means and the information stored in the evaluation information storage means, and presents the recommendation information to a user terminal; The system is characterized by comprising: a rating information registration means for registering the user's rating information and rating information on the user's actual purchases and/or actions in the rating information storage means; a similar user purchase and action information presentation means for presenting purchase and/or action information of similar users who have similar purchasing and/or action tendencies to the user; and a comparison result registration means for presenting the purchase and/or action information of the user and the purchase and/or action information of the similar users extracted from the rating information registered in the rating information storage means to the user terminal and registering the comparison results made by the user in the rating information storage means.

(2) In the configuration described in (1) above, the purchase history, etc. storage means is further characterized in that it stores the user's search history and/or favorite registration information.

(3) In the configuration described in (1) or (2) above, the device further includes an evaluation input request timing setting means for setting the timing of requesting the user to input the evaluation information when the user purchases or executes a recommended product, etc., depending on the characteristics of the product, etc., the characteristics of the user, or the user's behavioral tendency.

(4) In the configuration described in any one of (1) to (3) above, the evaluation information is characterized in that it includes information that the user received a recommendation but did not execute it.

(5) In the configuration described in any one of (1) to (4) above, the evaluation information includes an evaluation value input by the user for the product or the like and a reason for the evaluation.

(6) In the configuration described in any one of (1) to (5) above, the similar user purchase/behavior information presentation means is characterized in that it presents evaluation information of similar users and evaluation information of general users for the same comparison target side by side.

(7) In the configuration described in any one of (1) to (6) above, the evaluation information storage means is characterized in that it includes the user's evaluation information on the recommendation for each source of the recommendation.

(8) A recommendation method for recommending a product, service, or behavior (product, etc.) to a user, characterized in that the computer executes the steps of: storing the purchase history and/or behavior history of the user; storing the user's evaluation information on the purchase history and/or behavior history; generating recommendation information on the recommended product, etc., by referring to the purchase history and/or behavior history and the evaluation information, and presenting the information to the user terminal; registering the user's evaluation information on the recommendation information presented to the user terminal, and evaluation information on the user's actual purchases and/or behaviors; presenting purchase and/or behavior information of similar users who have similar purchasing and/or behavioral tendencies to the user; and presenting the purchase and/or behavior information of the user and the purchase and/or behavior information of the similar users extracted from the registered evaluation information to the user terminal, and registering the comparison results made by the user.

(9) A program for recommending products, services, or actions (products, etc.) to a user, the program being characterized by causing a computer to execute each of the steps described in (8) above.

The present invention provides a recommendation system that can perform self-analysis of the purchasing behavior, behavioral trends, etc. that are the basis for the recommended products, services, or actions.

FIG. 1 is a diagram for explaining an image of a recommendation system according to an embodiment of the present invention. FIG. 1 is a diagram showing a functional configuration of a recommendation system according to an embodiment of the present invention. FIG. 11 is a diagram showing processing on the user terminal side according to an embodiment of the present invention. FIG. 4 is a diagram showing a process of a recommendation information generating means according to an embodiment of the present invention. 13A and 13B are diagrams illustrating an example of a recommendation screen generated by the recommendation system and an example of a rating registration screen for the recommendation. 13A and 13B are diagrams illustrating an example of a recommendation screen for an action and an evaluation registration screen for a recommendation. FIG. 13 is a diagram showing an example of a subsequent evaluation screen for a recommendation. FIG. 13 is a diagram showing a comparison of evaluations of recommendations by general users and similar users.

Below, a detailed description of a form for carrying out the present invention (hereinafter, an embodiment) will be given with reference to the attached drawings. In the following figures, the same elements are given the same numbers or symbols throughout the description of the embodiment. Also, in the diagrams of the functional configuration, the arrows between the functional blocks indicate the direction of data flow or the direction of processing flow.

(Image of embodiment of the invention)
1 is a diagram for explaining an image of a recommendation system according to an embodiment of the present invention. Below, (1) to (4) in the diagram will be explained.

When a user purchases or acts, in addition to the purchase information and behavior information, the user can register his/her own satisfaction evaluation as evaluation information for the purchase or behavior (purchase and/or behavior: the same below). Here, instead of simple evaluations or evaluations such as other people's word-of-mouth, for example, reasons such as "Because of XX, I should have bought B instead of A" or "Because of △△, I should have gone to D instead of C" can also be registered. In other words, simply registering evaluations such as satisfaction/dissatisfaction and necessity for recommended information is passive, but by actively registering impressions and evaluations of things purchased or places visited, it is possible to know tendencies of oneself that one was not aware of. For example, one can know one's own tendencies such as only buying black clothes or going to XX on Saturdays.

When a user wants to make a purchase or take an action, that information is obtained based on the registered information. When a user wants to buy something or take an action, that information can be obtained based on the information registered about themselves. In (1), an evaluation of actual actions (such as buying XX or going to △△ at what time) is registered, while in (2), information about what you are going to do or want to try is obtained and an evaluation of that information is registered. Therefore, it is possible to have a function for making recommendations based on the data in (1) and (2). As a result, you can know the reason why a suggestion was made. Also, by doing this, if you think "I want to do something different from usual," you can receive unusual suggestions (best, second best, normal, things that you normally wouldn't do).

Here, the fact that a recommendation (suggestion) was received but was not carried out can also be recorded in (1) and the reason entered in (2). This can be used for the next suggestion. For example, if the reason (satisfaction level) registered is that the user wanted to go on a family trip but was busy at work and was unable to go, a suggestion could be made to avoid busy periods at work next time. In addition, if the user registered a reason (satisfaction level) that the user went to an amusement park and wanted to take their child on a roller coaster but was unable to do so due to a height restriction of 150 cm, a suggestion could be made to buy clothes for the child that are 150 cm tall next time.

Obtaining purchase and behavioral history of similar users When a user wants to purchase something or take some action, not only the information registered by the user but also other users with similar purchase and behavioral histories can be extracted from big data on the purchase and behavioral history of all users, making it possible to obtain information such as recommended products and destinations derived from those trends.

Registering the results of comparing your registered information with the tendencies of similar users. After comparing the information in (2) and (3) above, register the results (e.g., whether you chose (2) or (3)) again as evaluation information of your own purchases and actions.

As an example of a recommendation, if the user has evaluated "Last year, I went to see the autumn leaves at XX, but unfortunately the weather was not so good," the system will suggest a day this year when the weather is expected to be good. Also, if the user has evaluated "I was stuck in traffic on the way home and it was terrible," the system will suggest a place that is relatively uncrowded.

(Recommendation system functional configuration)
FIG. 2 is a diagram showing the functional configuration of a recommendation system according to an embodiment of the present invention. In the recommendation system 100 (hereinafter referred to as the present system) of this embodiment, a business such as a mail order site provides users with recommendation information on recommended products and recommended services via direct mail, the business's website, advertisements, etc. For this reason, as shown in the figure, the present system is a system connected to a user terminal 10 via a network, and includes a recommendation information generating means 101, a purchase history storage means 102 for storing purchase history, behavior history, search history, and favorite registration information of products (objects) or services, a product/service information database 103, an evaluation information registration means 104, an evaluation information storage means 105, a similar user purchase/behavior information presentation means 106, purchase/behavior history big data 107, a comparison result registration means 108, and an evaluation input request timing setting means 109. However, the purchase/behavior information big data 107 is usually used by connecting to an external database. Below, each functional block will be described in order.

The recommendation information generating means 101 refers to the user's own purchase history, behavior history, search history, favorite registration information, etc., to generate recommendation information including products, services, or actions (hereinafter referred to as products, etc.) recommended to the user, and presents it to the user terminal 10. Alternatively, when the user wants to purchase something or take an action, the means 101 has the user input keywords for that purpose, searches for recommended products, etc. from the product/service information database 103, generates recommendation information, and presents it to the user terminal 10. In either case, the means 101 refers to evaluation information registered in the evaluation information storage means 105 described below.

The purchase history storage means 102 is a database that stores the user's purchase history, behavior history, search history, favorite registration information, etc., as described above. An "action" is generally defined as "actually doing something with a certain purpose," so "purchase," "search," and "registration to favorite information" are also actions in a broad sense, but in the following, it refers to anything the user does, regardless of actions such as purchase, search, and registration to favorite information. For example, walking, jogging, resting, and sleeping are "actions" that are performed regardless of the user's purchase, search, etc. Therefore, the "action history" stored in the purchase history storage means 102 may include the history of the above-mentioned "actions." The action history may include various histories recorded in the user's smartphone and histories such as location data and health data acquired from wearable devices such as Apple Watch (registered trademark).

The product/service information database 103 is a database that stores detailed information and reviews of all products, etc. handled on the websites (mail order sites, travel sites, specialty store sites, etc.) of the business operator that operates this system. If the product, etc. desired by the user is not handled on the business operator's website, the recommendation information generating means 101 may search a general search site, and if a corresponding product, etc. is found, the information on that product, etc. may be presented as recommendation information.

The evaluation information registration means 104 receives the user's evaluation information on the recommendation information presented on the user terminal and the evaluation information on the user's actual purchase/behavior, and registers it in the evaluation information storage means 105. The evaluation information here includes the user's own evaluation value (degree of satisfaction or dissatisfaction) for that purchase/behavior, and information describing the reason for that evaluation. The evaluation value is a numerical value (evaluation score), a symbol (evaluation symbol), etc., while the reason can be an ordinary sentence (including a standard phrase).

The evaluation information storage means 105 stores the evaluation value and reason for each user's purchase/behavior information (including date and time information, information on the purchased product, etc., the place of purchase or behavior, and information on the site that sent the recommendation, etc.). When the same product, etc. is purchased multiple times or the same behavior is performed, an evaluation is added to each purchase/behavior. In other words, the evaluations for the same evaluation target are recorded in chronological order and used as a reference when making the next recommendation.

The similar user purchase/behavior information presentation means 106 refers to the purchase/behavior history big data 107 of all users, extracts similar users whose purchase/behavior history is similar to the user's own purchase/behavior history from the purchase/behavior history of all users, and obtains the purchase/behavior information of the similar users. By presenting such purchase/behavior information of similar users to the user terminal 10, information that cannot be obtained from the user's own purchase/behavior information alone can be obtained and used as a reference when making decisions.

The big data 107 of the purchase and behavior history of all users is a database that stores the purchase and behavior history of all users, and is usually an external database that is connected to and linked with this system.

The comparison result registration means 108 receives the results of the comparison and consideration by the user between the recommendation information presented by the recommendation information generation means 101 and the information of similar users presented by the similar user purchase/behavior information presentation means 106 and stores the results of the comparison in the evaluation information storage means 105. In this way, the evaluation information storage means 105 can grow the evaluation information itself as evaluation information that refers to the purchase/behavior information of similar users in addition to the evaluation information of the user's own purchase/behavior information.

The evaluation input request timing setting means 109 checks the evaluation information registered in the evaluation information storage means 105, and for any evaluation that has not been entered, sets the timing for requesting the user to enter an evaluation. Typically, a user enters evaluation information after purchasing a product or performing an action, but the timing for requesting the input can be set appropriately depending on the characteristics of the user and the nature of the purchased product or action.

For example, the user may be prompted to enter a rating when the GPS information of the user terminal 10 indicates that the user has taken a route different from their usual commute route, or when the user has purchased something that they would not normally buy. The timing of the request for rating may be different for users who tend to frequently enter ratings and users who tend to only enter rough ratings. Since the timing at which ratings can be given may differ depending on the product or service, the timing of the request for rating may be changed depending on the product purchased. For example, the timing for requesting a rating may be set to immediately after purchase, two weeks later, or several months later.

The above-mentioned functional configuration of the present system is merely an example, and one functional block (database and functional processing unit) may be divided, or multiple functional blocks may be combined into one functional block. Each functional processing unit is realized by a computer program executed by a CPU (Central Processing Unit) built into the device, which reads a computer program stored in a storage device such as a ROM (Read Only Memory), flash memory, SSD (Solid State Drive), or hard disk. That is, each functional processing unit is realized by the computer program reading and writing necessary data such as tables from a database (DB; Data Base) stored in the storage device or a storage area on the memory, and in some cases controlling related hardware (e.g., an input/output device, a display device, a communication interface device). In addition, the database (DB) in the embodiment of the present invention may be a commercial database, but it also means a simple collection of tables and files, and the internal structure of the database itself is not important.

(Processing flow)
Fig. 3 is a diagram showing the processing on the user terminal side according to the embodiment of the present invention. Fig. 4 is a diagram showing the processing of the recommendation information generating means according to the embodiment of the present invention. In the processing flow diagrams (flowcharts) of Fig. 3 and Fig. 4, the processing order of each step may be changed as long as the relationship between the input and output of each step is not lost.

In the process on the user terminal side in FIG. 3, first in step S10 it is determined whether the user wants to buy something or take some action (desire to purchase/take some action). If there is no such desire, the process proceeds to step S11, where it is checked whether a request to input a rating has been received from the system. A request to input a rating is a notification of a request for input sent from the system when the user has already purchased some product, etc., but has not yet input any rating information from the user. However, even if there is no request to input a rating, the user can always input or modify a rating of a purchased product, etc., from the purchase history page, etc.

If it is determined in step S11 that a request to enter a rating has not been received, the process returns to step S10. If a request to enter a rating has been received, the process proceeds to step S12, where the rating for that purchase is registered in the system and the process ends.

In step S10, if the user wishes to make a purchase or take an action, the process proceeds to step S13, where information on recommended purchases and actions related to the desired purchase or action is obtained based on the user's own registered information. At this time, if there is nothing related to the user's registered information, the desired purchase or action information is newly registered (rating information may not be entered at this point). At this time, purchased and unpurchased items may be managed separately.

Next, in step S14, information on recommended purchases and actions regarding the desired purchases and actions is obtained based on the purchases and actions of similar users who have similar purchasing and behavioral tendencies to the user.

Then, in step S15, the purchase and behavior information of the user and similar users is compared, and the comparison result is registered in the evaluation information storage means 105 in step S16. In this way, the user can receive recommendation information that references not only the purchase and behavior information registered by the user, but also the purchase and behavior information of similar users.

Next, the processing on the system side of FIG. 4 will be described. First, in step S20, it is determined whether purchase/action request information has been received from the user terminal.

If purchase/action request information is received, the process proceeds to step S21, where user registration information is obtained from the evaluation information storage means 105.

Next, in step S22, it is determined whether there is any registered information related to the purchase/action desire. If there is no related registered information, the process proceeds to step S24. If there is related registered information, in step S23, recommended purchase/product information based on the user's registered information is presented to the user terminal.

Then, in step S24, purchase and behavior information of similar users is obtained from the big data. At this time, the user's own purchase and behavior trends may be analyzed in advance together with the user's profile information and registered in the purchase and behavior evaluation storage means 102. For example, purchase and behavior trends may be determined such that a user who is presented with a category of recommended products as shown in FIG. 5 described below is interested in health care (health) and tends to purchase a lot of outdoor products.

Next, in step S25, recommended purchase/behavior information based on the purchases/behavior of similar users is presented to the user terminal. The recommended purchase/behavior information presented in steps S23 and S25 is compared on the user side, and the comparison result is registered in the evaluation information storage means 105.

Next, in step S26, the timing of the evaluation input request is set. The timing of the evaluation input request refers to the timing at which the user is asked to input an evaluation for the registered purchase/behavior, and the simplest timing is set after a certain period of time has passed (for example, usually one to two weeks) after the user purchases a product, etc. This process is performed by the evaluation input request timing setting means 109. Here, the timing of the evaluation input request may be set to the optimal timing as described above depending on the characteristics of the user or the characteristics of the purchased product, etc.

If no purchase/action request has been received in step S20, the process moves to step S30, where it is determined whether or not a rating registration has been received.

If a rating registration has been received in step S30, the process proceeds to step S31, where the received rating information is registered in the rating information storage means 105, and the process ends.

If no evaluation registration has been received in step S30, the process proceeds to step S32, where it is determined whether it is time to request an evaluation.

If it is determined in step S32 that it is the optimal timing to make a request for evaluation, then in step S33, a request for evaluation is sent. For example, a message such as "Are you satisfied with the XX product you purchased on XX date? If so, please enter your rating and the reason for your rating" may be sent by email or displayed when the user opens the business's website.

(Screen example)
5 is a diagram showing an example of a recommendation screen generated by the recommendation system and an evaluation registration screen for the recommendation. The recommendation presentation screen 200 shown in the figure is an example of a screen displayed when a user presses a button for a recommended product on an online shopping site or clicks a link for a recommended product in direct mail. In this example, recommended products on the online shopping site used by the user are displayed by category based on the user's past purchase history and products added to "favorites."

When the user selects, for example, category 201, "Hotel Accommodation Vouchers" from the recommended products on recommendation presentation screen 200, rating registration screen 210 is displayed, and recommended hotels 211 are displayed in a predetermined order (recommended order, highest user rating order, price order, most reviews order, etc.). If there are too many recommended hotels 211, the user may be prompted to input narrowing conditions. Note that while "recommended actions" in category 202 are not displayed in the recommended information of normal online shopping sites, this system can display actions (not necessarily actions that require compensation) recommended based on the user's product purchase history, action history, profile information, etc. Examples of recommendation presentation screens for actions are further explained in FIG. 6.

Here, the system does not simply present a list of recommended hotels, but also allows the user to register rating information for the presented recommendations. This rating information may include a rating value and the reason. The rating value may be a numerical rating point or a rating symbol such as the number of stars or ◎, 〇, △, ×. Of course, inputting the user's rating information is optional, but the input rating information is registered, and if the same hotel is recommended later, the user's past rating information is also displayed. Information on whether or not the user actually stayed at the recommended hotel may also be recorded. Furthermore, it may be possible to register the reasons why the hotel was selected or not selected.

Doing so allows the user to evaluate the recommendation itself, which contributes to improving the quality of recommendations. In particular, if it is unclear why a particular recommendation was made, this may be added to the evaluation information. The evaluation registration screen 210 may also display the user's past accommodation history 213. Here, if there are any hotels for which a rating has not yet been entered, the user may be prompted to enter a rating. Conversely, a recommended hotel list 211 may be displayed from the accommodation history 213.

Figure 6 shows an example of a recommendation screen for behavior and an evaluation registration screen for recommendations. The recommendation presentation screen 300 shown in the figure is an example of a screen that is displayed when a user presses a recommended behavior button on a recommended behavior site or clicks a recommended behavior link in direct mail. In this example, recommended actions on the recommended behavior site used by the user are displayed by category based on the user's past behavior history, search history, and actions added to "favorites." As mentioned above, the behavior history may include various histories recorded on the user's smartphone, and history such as location data and health data obtained from wearable devices such as Apple Watch (registered trademark).

Recommendations for actions can also be generated from the purchase history of goods or services on mail-order or store sites, but such recommended actions are mostly related to the goods or services purchased and often require a payment (for example, hiking may be recommended after buying sneakers). For this reason, it is rare to obtain a history of actions that do not require a payment from mail-order or store sites, so it is advisable for the user to register desired actions in advance as "favorite actions" on specialized sites or video sites.

The recommendation presentation screen 300 shown in FIG. 6 shows recommended actions categorized based on the information registered in this "favorite actions." Based on the user's search history on specialized sites, video sites, and the like, the screen presents recommendations for actions that do not necessarily require compensation. Actions that do not necessarily require compensation include, for example, recommended actions that are good for health, enjoying nature, playing with children, listening to music, spending time with pets, volunteering, and the like, as well as desires such as wanting to cure stiff shoulders, making donations, and earning more money.

The rating registration screen 310 is almost the same as that shown in FIG. 5, but in the rating 312 for the recommendation, the user is asked to check the actions that they have actually performed, and the checked actions are displayed in the behavior history 313.

Figure 7 shows an example of a subsequent evaluation screen for a recommendation. The illustrated evaluation screen 300 for a recommendation displays, for each recommendation, the providing site, the recommendation list, whether the recommendation was followed, and the user's evaluation and reason. For example, recommendation 310 indicates that a recommended hotel list was sent from X travel site on a certain date. Also, recommendation 320 indicates that a recommended home appliance list was sent from Y home appliance mail order site on a certain date.

In the example of recommendation 310, the user ratings and number of reviews are the ratings of all users who have registered ratings and reviews. And "Whether or not they stayed", "Rating", and "Reason" are information entered by the user who received the recommendation. In this example, the user has never stayed at any hotel and has no rating for the accommodation, so all information is "-" (not entered). Meanwhile, in the example of recommendation 320, "No" in "Whether or not they purchased" indicates that they have no plans to purchase even after receiving the recommendation. And "Yes" in "Whether or not they purchased" indicates that they already had the product or purchased it after receiving the recommendation.

In this way, a user can register multiple recommendation sources and accumulate evaluation information, allowing the user to evaluate the sites that receive recommendations. In this way, the user can judge the reliability of the recommendation source site itself and its suitability for the user, and select recommended sites. This is because it is not realistic to receive recommendations from all sites. Furthermore, by collecting user evaluations, sites can learn about the characteristics and tendencies of users, as well as improve the quality of recommendations.

Figure 8 is a diagram showing a comparison of ratings of general users and similar users for recommendations. A user can compare ratings based on the information acquired by the user with ratings of similar users and register the results before and after the comparison, but can also compare rating information of general users who are not similar users. The general/similar user comparison screen 400 shown in the figure shows a general user rating screen 410, a similar user rating screen 420, and the user's own rating screen 430 (before and after comparison). The user's own rating is the rating that the user made before and after comparing his or her own rating with ratings of general users and/or similar users. Here, a case is shown in which the user has received a recommendation of a recommended hotel, but the user himself or herself has never stayed at any of the hotels.

If the user has already rated the hotel beforehand based on hotel information or the like, the user's own rating before comparison is compared with the ratings of general users and/or similar users, and the user's own ratings before and after the comparison are registered. This is because a user's own rating may change depending on the ratings of general users and similar users. Although not shown in the figure, the rating of the site that sent the recommendation may also be added to the comparison. In this way, the user can judge the reliability of the site itself that sent the recommendation and its suitability for the user.

In the example of Figure 8, for simplicity's sake, only three hotels, AAA Hotel, BBB Hotel, and CCC Hotel, are shown, so the hotels on the general user rating screen 410, the similar user rating screen 420, and the user's own rating screen before and after comparison 430 are all the same. Of course, it goes without saying that the same hotel will be rated differently by each user group.

In general, it is considered that hotels rated by general users (e.g., top 10 hotels) and hotels rated by similar users (also top 10 hotels) often do not match, and in such cases, the user may use only hotels that are commonly rated by both user groups as the target of his/her own comparison. For example, if AAA Hotel, BBB Hotel, and CCC Hotel are recommended among general users, and AAA Hotel, DDD Hotel, and EEE Hotel are recommended among similar users, the user may specify that only AAA Hotel be registered in the comparison results, or that all of the top 10 hotels in each category be registered in the comparison results. Alternatively, if it is considered that the evaluations of similar users are often more useful to the user himself/herself than the evaluations of general users, the user may add not only the commonly recommended AAA Hotel, but also DDD Hotel and EEE Hotel, which are only recommended by similar users, to the registration targets of the user's own comparison results.

Note that the above explanation is for the case where the user is searching for a hotel themselves, but even if the user has already stayed at a specific hotel, they may make their own rating by referring to the ratings of general users and similar users. Conversely, the user may first make their own rating and then refer to the ratings of similar users. It is up to the user whether or not to change their own rating based on this.

It is also possible to add products highly rated by similar users to the user's own recommendation list. That is, xx users who are particularly similar to the similar users may be selected, and common products with high ratings among those users may be extracted, or users who have had common experiences may be extracted, and added to the user's own recommendation list. Specifically, for example, if the product category is books, and the highly rated book among 50 people who are particularly similar is "Miyamoto Musashi's Book of Five Rings," which the user himself has read, and many of those similar users have read "Moshidora (If High School Girls Read XXX Management)," "Moshidora" may be added to the "recommended books" for the user. Usually, when recommending books, it is common to recommend books related to the user's past reading history or books that the user has read, but it is possible to recommend unexpected books that the user has no interest in and have no relation to books he or she has read in the past, based on the evaluation results of similar users. "The Book of Five Rings" is a serious book that explains military strategy, so it may seem surprising that a seemingly light-hearted book like "If Dora Were Really Here?" would be recommended from it, but the two have something in common that is relevant to the business of management.

Similarly, the evaluation results of similar users can be used as a reference for recommendations of actions that do not involve purchases. For example, if a photo spot is recommended to a user who likes photography, the user can learn that people with similar sensibilities behave differently from the user, making the recommendation useful. Below are some concrete examples.

(Example 1) If a recommendation such as "Here's a nearby place where you can take photos of tulips right now!" is made to someone who likes taking photos of flowers, it will be welcome information for the person who received the recommendation (especially if winter tulips are currently in season). Also, if someone who likes flowers but doesn't have their eye on tulips at the moment receives such a recommendation while taking photos of plum blossoms on Saturday, it will motivate them to go on Sunday. It is even better if the same recommendation is presented with a grace period to allow the person to take action in the near future.

(Example 2) If a recommendation such as "You can take beautiful photos of factory night views from a boat!" is made to a person who usually only takes photos of natural scenery during the day, daytime natural scenery and factories at night have completely opposite impressions, but if you group them together in the sense of beautiful subjects, factory night views can also be considered as a subject, and the fact that they are taken from a boat might encourage them to try out their photography skills, making this a useful recommendation to test their skills. In this way, when someone with the same sensibilities begins to become interested in things that are different from your own, you can discover recommendations that make them think, "That might be good."

(Effects of the embodiment)
According to this system, it is possible to provide a recommendation system capable of self-analyzing the purchasing trends, behavioral trends, etc. that are the basis for the products etc. that are recommended.
in particular,
(1) Since the system is provided with an evaluation information storage means for storing a user's purchase history, behavior history, search history, favorite registration information, and user evaluation information on the purchase history or behavior history, when making a new recommendation, the user's evaluation of previous recommendations can be referenced. In addition, since the system presents purchase/behavior information of similar users, the user's own purchase/behavior can be compared with the purchase/behavior evaluations of similar users to analyze the differences.
(2) The information sources used to generate recommendations can include not only purchase history and behavioral history, but also search information and favorite registration information, making it possible to make recommendations over a wider range.
(3) When a user purchases a recommended product, etc., the timing for requesting the user to enter evaluation information can be set depending on the characteristics of the product, etc., the characteristics of the user, or the characteristics of the action, allowing the user to enter the evaluation at the optimal timing and preventing the user from forgetting to enter the evaluation altogether.
(4) The evaluation information may include information about recommendations that the user received but did not follow, so that the evaluation information can be used as a reference when making new recommendations.
(5) The evaluation information includes the evaluation value input by the user for the product or the like and the reason for the evaluation, and is therefore useful information when making new recommendations.
(6) When presenting purchase and behavior information of similar users, evaluation information of similar users and evaluation information of general users for the same comparison subject are presented side by side, so that the evaluations of general users and similar users can be compared and the differences analyzed.
(7) The evaluation information storage means includes user evaluation information on the recommendations for each source of the recommendation, so that the user can grasp the differences in recommendations for each site, their reliability, suitability for the user, and the like.

Although the present invention has been described above using the embodiments, it goes without saying that the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be clear to those skilled in the art that various modifications and improvements can be made to the above embodiments. Furthermore, it is clear from the claims that forms incorporating such modifications or improvements can also be included in the technical scope of the present invention.

In the above embodiment, the present invention has been described as a recommendation system, but it can also be considered as a method invention (recommendation method) or a computer program invention (program).

REFERENCE SIGNS LIST 10 User terminal 100 Recommendation system 101 Recommendation information generation means 102 Purchase history, etc. storage means 103 Product/service information database 104 Evaluation information registration means 105 Evaluation information storage means 106 Similar user purchase/behavior information presentation means 107 Purchase/behavior history big data of all users 108 Comparison result registration means 109 Evaluation input request timing setting means 200 Recommendation presentation screen 300 Evaluation screen for recommendation 400 General/similar user comparison screen

Claims (9)

A recommendation system that recommends products, services, or actions (products, etc.) to a user,
A purchase history storage means for storing the purchase history and/or behavior history of the user;
evaluation information storage means for storing evaluation information of the user with respect to the purchase history and/or the behavior history;
a recommendation information generating means for generating recommendation information of a recommended product or the like by referring to the purchase history and/or the behavior history stored in the purchase history etc. storing means and information stored in the evaluation information storing means, and presenting the recommendation information to a user terminal;
an evaluation information registering means for registering, in the evaluation information storage means, evaluation information of the user with respect to the recommendation information presented to the user terminal and evaluation information with respect to the actual purchase and/or behavior of the user;
a similar user purchase and behavior information presentation means for presenting purchase and/or behavior information of similar users who have similar purchasing and/or behavioral tendencies to the user;
a comparison result registration means for presenting the purchase and/or behavior information of the user and the purchase and/or behavior information of the similar users extracted from the evaluation information registered in the evaluation information storage means to the user terminal, and registering a comparison result made by the user in the evaluation information storage means;
A recommendation system comprising: The recommendation system according to claim 1, characterized in that the purchase history storage means further stores the user's search history and/or favorite registration information. The recommendation system according to claim 1 or 2, further comprising an evaluation input request timing setting means for setting the timing of requesting the user to input the evaluation information when the user purchases or executes a recommended product, etc., depending on the characteristics of the product, etc., the characteristics of the user, or the user's behavioral tendency. The recommendation system according to any one of claims 1 to 3, characterized in that the evaluation information includes information that the user received a recommendation but did not execute it. The recommendation system according to any one of claims 1 to 4, characterized in that the evaluation information includes an evaluation value input by the user for the product or the like and a reason for the evaluation. The recommendation system according to any one of claims 1 to 5, characterized in that the similar user purchase/behavior information presentation means presents evaluation information of similar users and evaluation information of general users for the same comparison target side by side. The recommendation system according to any one of claims 1 to 6, characterized in that the evaluation information storage means includes the user's evaluation information on the recommendation for each source of the recommendation. A method for recommending a product, service, or action (product, etc.) to a user,
storing a purchase history and/or behavior history of the user;
storing evaluation information of the user with respect to the purchase history and/or the behavior history;
generating recommendation information of recommended products or the like by referring to the purchase history and/or behavior history and the evaluation information, and presenting the recommendation information to a user terminal;
registering evaluation information of the user regarding the recommendation information presented to the user terminal, and evaluation information regarding actual purchases and/or actions of the user;
presenting purchase and/or behavior information of similar users who have similar purchasing and/or behavioral tendencies to the user;
a step of presenting the purchase and/or behavior information of the user extracted from the registered evaluation information and the purchase and/or behavior information of the similar users to the user terminal and registering a comparison result made by the user;
The recommendation method is characterized in that the above steps are executed by a computer. A program for recommending products, services, or actions (products, etc.) to a user, the program being characterized by causing a computer to execute each of the steps described in claim 8.