JP6087467B2 - Product recommendation service method based on relative comparison, recommendation service device and program therefor - Google Patents

Description

  The present invention relates to a product recommendation service method based on relative comparison, a recommendation service device for the same, and a program. More specifically, the user determines the ranking of products through relative comparison according to his / her preference, and between users. Further, the present invention relates to a method for determining a user with a similar preference through cross comparison with the ranking of products and recommending a product based on the ranking of products determined by the user with a similar preference.

  Abundant information is shared on the Internet, and users select products with reference to evaluations and recommendations on products that others have experienced or possessed among a variety of information.

  As an example, when a product is searched for on the search service, evaluations and recommendations for various products are presented as search results, and the user selects a product suitable for himself based on such information.

  However, this situation is abused, and the reliability of evaluations and recommendations for products declines due to disruption of advertising information, etc., and people have different personalities. On the other hand, when a product is selected by evaluation or recommendation for the product, there is a problem that a product that is substantially unsuitable for one's taste is selected.

  In addition, there are many methods for recommending products by displaying product ratings at a level such as the number of stars. However, in the case of ratings using the number of stars, there is no comparison target, so only one product is displayed. On the other hand, because the user evaluates the feeling, even though the same user evaluates the same product, the feeling at that time is reflected differently, and many evaluations cause different results. Therefore, there is a limit that it is difficult to evaluate consistently, and there is a problem that comparison is not possible between products having the same number of stars.

Korean Patent Publication No. 10-2012-0101188

  The present invention is for solving the problems of the prior art as described above, and solves the problem that the reliability of evaluation and recommendation for a product presented on the Internet due to advertisement information etc. decreases. Is for.

  In particular, because each person has a different personality, the taste for the favorite product is also different, and if you select a product that trusts the evaluation and recommendation for the product, the product that is not actually suitable for your taste is selected. This is to solve the problem.

  In addition, in the case of evaluation for products using the number of stars, etc., there are limits to which consistent evaluation is difficult even though the same user evaluates the same product, and it has the same number of stars Since there is a problem that cannot be compared between products, it is intended to provide a recommendation service that enables relative comparison and evaluation using a new method.

  The recommended service device according to the present invention capable of achieving the technical problem is based on selection of a user's preferred product by a relative comparison with respect to a plurality of products in accordance with each user's preferences, A product ranking setting unit for setting a ranking, a product designation receiving unit for accepting designation of a product that the user has never experienced, a pre-verification / evaluation unit for requesting prior verification and evaluation of the designated product, The cross-comparison of the ranking for the product set by the specific user and the ranking for the product set by one or more other users is made to determine other users who have similar preferences to that of the specific user. Based on rankings for products set by a similar user determination unit and other users having a similar preference to that of the specific user A product recommendation unit that recommends products to a regular user, the product ranking setting unit receives a plurality of product information input from the user, generates a product list, and includes a plurality of items included in the product list Based on the user's preference product selection through a relative comparison according to the user's preference for the product, the ranking for the plurality of products is set, a product ranking list is generated, and the pre-verification / evaluation unit specifies Request for prior verification / evaluation of whether or not the specified product related to the user suits the taste before the specific user experiences the specified product, and the similar user judgment unit is a product corresponding to the specific user A similarity is calculated by cross-comparison between a ranking list and a product ranking list corresponding to one or more other users among the plurality of users, and the one or more other users are calculated based on the similarity One or more other users who have a preference similar to that of the specific user among the users are determined, and in the calculation of the similarity by the similar user determination unit, a product corresponding to the specific user For each of the ranking list and the product ranking list corresponding to the other user, a plurality of products are classified into a plurality of groups according to a sequential order or a non-sequential order, and correspond to the specific user. A group of product ranking lists and a group of product ranking lists corresponding to the other users are cross-compared for each group, and the similarity of the other users to the specific user is determined based on the number of the same products for each group. It is characterized by calculating.

  A product recommendation service method according to the present invention is a product recommendation service method by a recommendation service device including a product ranking setting unit, a similar user determination unit, and a product recommendation unit, wherein the product ranking setting unit includes a plurality of usages. The user has never experienced the step of setting the ranking for the product for each user based on the user's preference product selection by relative comparison with respect to a plurality of products according to each user's taste, and the product designation receiving unit A step of accepting designation of a product to be worried, a step of requesting prior verification and evaluation for evaluating in advance whether or not the specified user is suitable for the preference before the specific user experiences the specified product, The similar user judgment unit ranks the product set by the specific user and the rank set by one or more other users. A step of judging cross-comparison and determining other users having a preference similar to that of the specific user, and a product recommendation unit set by another user having a preference similar to that of the specific user Recommending a product to the specific user based on a ranking for the product, and the product ranking setting unit receives a plurality of product information from the user to generate a product list, and the product Based on the user's preference product selection through a relative comparison according to the user's preference for a plurality of products included in the list, the ranking for the plurality of products is set, a product ranking list is generated, and the advance The verification / evaluation unit requests prior verification / evaluation as to whether it suits the taste before the specific user experiences the designated product related to the specific user among the plurality of users, The similar user determination unit calculates a similarity by cross-comparing a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more other users among the plurality of users. Determining one or more other users having a preference similar to that of the specific user among the one or more other users based on the similarity, and the similar user judging unit In the calculation of the similarity, a plurality of products are sequentially ordered or non-sequentially ordered for each of the product ranking list corresponding to the specific user and the product ranking list corresponding to the other user. The product ranking list group corresponding to the specific user and the product ranking list group corresponding to the other user are cross-compared for each group, The similarity of the other user to the specific user is calculated based on the number of the same product for each group.

  The program of the present invention is a product ranking setting unit that sets a ranking for a product for each user based on a user's preferred product selection based on a relative comparison with respect to a plurality of products according to each user's preferences. A product designation receiving unit that accepts designation of a product that the user has never experienced, a pre-verification / evaluation unit that requests pre-verification and evaluation of the designated product, and a product set by a specific user A similar user determination unit that cross-compares a ranking for a product set by one or more other users and determines other users having a similar preference to the particular user's preference, and Recommend a product to the specific user based on a ranking for a product set by another user having a similar preference to the specific user The product ranking setting unit generates a product list upon receiving a plurality of product information input from the user, and depends on the user's preference for the plurality of products included in the product list. Based on the user's preferred product selection through relative comparison, ranking is set for the plurality of products, and a product ranking list is generated. The specific user requests prior verification / evaluation as to whether or not it suits the taste before experiencing the designated product, and the similar user determination unit includes the product ranking list corresponding to the specific user and the plurality of users. A similarity is calculated by cross-comparison with a product ranking list corresponding to one or more other users, and the identification among the one or more other users is performed based on the similarity One or more other users having a preference similar to that of the user are determined, and in the calculation of the similarity by the similar user determination unit, a product ranking list corresponding to the specific user and the other user For each product ranking list corresponding to a user, a plurality of products are classified into a plurality of groups according to a sequential order or a non-sequential order, and a product ranking list group corresponding to the specific user The product ranking list group corresponding to the other user is cross-compared for each group, and the similarity of the other user to the specific user is calculated based on the number of the same product for each group.

According to the present invention, by recommending a product based on an evaluation for a product set by another user having a taste similar to the user's taste, it is possible to make a product recommendation that the user can satisfy. it can.

  In addition, since the product is recommended based on the evaluation of a large number of other users who have similar preferences to the user who requested the product recommendation, it is possible to eliminate advertisement information and increase the reliability of the recommendation. Can do.

  In addition, by overcoming the limitations of simple evaluation for products that use the number of stars, etc., and making recommendations through comparative evaluation among a large number of products, users can select products more effectively. .

  In addition, before a specific user uses a product, it can be verified and evaluated in advance whether it is close to his / her preference.

1 is a diagram illustrating an embodiment of a recommendation service system for providing a product recommendation service according to the present invention. 1 is a configuration diagram of an embodiment of a recommendation service device for providing a product recommendation service according to the present invention. 5 is a schematic flowchart for an embodiment of a product recommendation service method based on relative comparison according to the present invention. It is the figure which showed embodiment with respect to the product ranking setting process of the recommendation service method which concerns on this invention. It is the figure which showed embodiment with respect to the similar preference user judgment process of the recommendation service method which concerns on this invention. It is the figure which showed the embodiment with respect to the product recommendation process of the recommendation service method which concerns on this invention. It is the figure which showed the embodiment with respect to the recommendation merchandise information of the recommendation service method which concerns on this invention. It is a figure explaining the goods recommendation process of the recommendation service method which concerns on the example of improvement of embodiment of this invention. It is a figure explaining the goods recommendation process of the recommendation service method which concerns on the example of improvement of embodiment of this invention. It is a block diagram of the recommendation service apparatus which concerns on the example of improvement of embodiment of this invention. It is a flowchart which shows the process sequence of the example of the improvement. It is a figure which shows the example of a screen display. It is a figure which shows the example of a screen display.

  In order to illustrate the present invention and the operational advantages of the present invention and the objects achieved by the practice of the present invention, a preferred embodiment of the present invention will now be illustrated and described with reference thereto.

  First, the terminology used in this application is merely used to describe a particular embodiment, and is not intended to limit the invention, and the singular expression is clearly different in context. Multiple expressions can be included unless implied. Also, in this application, terms such as "include" or "have" indicate that there are features, numbers, steps, operations, components, parts, or combinations thereof as described in the specification. It is intended to pre-exist the existence or additional possibilities of one or more other features or numbers, steps, actions, components, parts or combinations thereof. Should be understood.

  In the description of the present invention, if it is determined that a specific description of a related known configuration or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

  According to the present invention, a user determines a product ranking through a relative comparison according to his / her preference, determines a user with a similar preference through a cross-comparison with respect to the product ranking between users, and a product determined by a user with a similar preference A method for recommending products based on the ranking of

  In the present invention, the product applied to the recommendation service may be a form or intangible information, but may include various items such as a smartphone, a car, a restaurant, Locations such as cafes can also be included, and content such as movies and music can also be included.

  FIG. 1 illustrates an embodiment of a recommendation service system for providing a product recommendation service according to the present invention.

  The recommended service device 100 holds product ranking list information determined by each user according to each user's preferences, and is based on the user-specific product ranking list information. Other users who have similar preferences to the user's preferences are grasped, and a product recommendation service is provided to the specific user based on product ranking information determined by other users of similar preferences.

  A large number of users 10a, 10b,... 10n can receive recommended services by connecting to the recommended service device 100 with their own communication terminals, such as smartphones, PCs, notebook computers, and the like. Various generally known communication terminals can be applied.

  The user can connect to the recommended service device 100 through the network 30 by using a communication terminal that the user owns, and the network 30 is a general Internet line, which is both a wired network and a wireless network. Can be included. In addition, when the recommended service according to the present invention is provided in a limited range such as a specific company, the network network can be a restricted area or a virtual private network or an intranet network accessible only by an authenticated person.

  Further, the recommendation service device 100 can provide a recommendation service to the users 10a, 10b,... 10n in conjunction with a general search service server 50. For example, the users 10a, 10b,. It is possible to request a recommendation for a product by connecting to the search service server 50 that is normally used. In this case, the search service server 50 recommends a product considering the user's preference to the recommendation service device 100. Upon request, the recommendation service device 100 can provide the search service server 50 with a product recommendation result that takes into account the user's preferences.

  FIG. 2 is a block diagram illustrating an embodiment of a recommendation service apparatus for providing a product recommendation service according to the present invention.

  The recommendation service device 100 may generally include a ranking setting unit 110, a similar user determination unit 130, a product recommendation unit 150, and the like, and may further include a user information DB 170.

  The ranking setting unit 110 generates a user-specific product list for products that each user has experienced or possessed. When each user selects a preferred product through a relative comparison according to his / her preference, the ranking setting unit 110 displays the product list on the product list. A ranking for each product is determined, and a product ranking list for each user is generated by matching the product list with the corresponding ranking for the product. Here, the ranking setting unit 110 inputs a product list that the user has experienced or possessed, and uses a service page or tool that allows the user terminal to determine a preferred product through a relative comparison with each product according to his / her preference. It can also be provided.

  Preferably, the product ranking list generated by the ranking setting unit 110 can be stored in the user information DB 170 corresponding to the identification information of the corresponding user. Note that the ranking setting unit 110 may omit the process of setting the ranking for products that the user registered in advance does not like or dislikes, and directly place them in the lower group. According to this, when the number of products registered in the product ranking list is small, it is distinguished by tagging so that pre-registered unfavorable and disliked products are not mixed with favorite products and become recommended products. It becomes possible.

  The similar user determination unit 130 determines similar users based on the user-specific product ranking list generated by the ranking setting unit 110. As an example, the similar user determination unit 130 determines the product ranking list corresponding to the specific user and other users. Cross-comparison with the corresponding product ranking list is performed to determine other users who have similar preferences to that of the specific user. At this time, the similar user determination unit 130 can calculate the similarity based on the cross-comparison of the product ranking list, and can select one other user who exhibits the highest similarity or a preset range. It is also possible to select many other users who are within the similarity of

  The product recommendation unit 150 confirms one or more other users who have similar preferences to the preferences of the corresponding user determined by the similar user determination unit 130, and other users having similar preferences have decided. Based on the product ranking list, recommendation information for recommending products that the corresponding user does not have experience or possess is generated and provided.

  The recommended service device 100 has such a configuration, and the recommended service according to the present invention is provided by the function of each configuration. Here, the recommended service device 100 may be embodied by one device including each configuration. Alternatively, each configuration may be configured by each individual device or individual module and selectively combined, and the recommended service device 100 may be implemented by a group of each configuration. In addition, a kind of server capable of performing a storage function, a calculation function, and the like is applied to the recommendation service device 100, and a computer program for performing each stage of the product recommendation service method according to the present invention described later is recorded in the recommendation service device. It can happen.

  The product recommendation service based on the relative comparison considering the user's preference can be provided through the above-described recommendation service device according to the present invention, but the method for providing the product recommendation service based on the relative comparison according to the present invention This will be described through the embodiment.

  FIG. 3 is a schematic flowchart for an embodiment of a product recommendation service method based on relative comparison according to the present invention.

  Since the product recommendation service method according to the present invention is implemented by the above-described recommendation service apparatus according to the present invention, it will be described with reference to the recommendation service apparatus.

  The recommended service method according to the present invention includes a product ranking setting process (S100) for setting a ranking for a product through relative comparison according to a user's preference, a product ranking determined by a specific user who requests a recommendation service, Cross-comparison with the rankings of products determined by other users to determine one or more other users who have similar preferences to the specific user's preferences (S200), In addition, a product recommendation process (S300) of recommending a product to the specific user based on a ranking for a product determined by another user having a similar preference may be included.

  The respective processes presented by the recommended service method according to the present invention will be described in more detail through the following detailed embodiments.

  First, the product ranking setting process (S100) will be described with reference to an embodiment of the product ranking setting process in the recommended service method according to the present invention shown in FIG. 4 together with FIG.

  First, the ranking setting unit 110 of the recommended service device 100 generates a product list (S120) in response to an input of product information that the user has experienced or possessed (S110). FIG. 4 is an embodiment for setting a ranking for a restaurant. The ranking setting unit 110 of the recommendation service device 100 receives restaurant information that the user A has experienced from the user A, and collects the restaurant information as shown in FIG. As shown in a), a restaurant list 210 is created in which the restaurant information 215 experienced by the user A is collected.

  Then, the ranking setting unit 110 of the recommendation service device 100 selects a product that the user prefers through a relative comparison between the products included in the product list according to the user's preference, and sets a ranking for the product based on the selected product. (S130), and a product ranking list is generated (S140).

  As shown in FIG. 4B, the user A is compared with the restaurant 1 (221) and the restaurant 2 (223) on the restaurant list 210 of the user A according to the user A's preference. The restaurant 2 (225) that is relatively preferred can be selected, and the ranking setting unit 110 of the recommendation service device 100 repeatedly performs such a relative comparison, and the restaurant setting list 210 experienced by the user according to the taste of the user. A ranking is set based on the selection for the upper restaurant, and a restaurant ranking list is generated by collecting the rankings. Here, as shown in FIG. 4C, the restaurant ranking list can be generated as a restaurant ranking list 230 in which all restaurants 215 on the restaurant list 210 are listed 235 and the ranking 231 is set. .

  The recommended service device 100 can provide a service page and application for inputting product information and setting rankings to the user's terminal for user product input and ranking setting.

  Through such a product ranking setting process (S100), rankings for products are set through relative comparisons between products according to user preferences, and the product ranking setting process (S100) is repeatedly performed for each of a number of users. A user-specific product ranking list for each of a large number of users can be generated.

  The product ranking list for each user generated in this way can be stored in the user information DB 170 corresponding to each user.

  In addition, the user can update the overall product ranking list by adjusting the preference selection of the corresponding product by re-experience with existing products existing in the product ranking list. It is also possible to update the product ranking list by newly adding a ranking for the newly owned product to the existing product ranking list.

  Second, the similar preference user determination process (S200) will be described with reference to an embodiment of the similar preference user determination process in the recommended service method according to the present invention shown in FIG. 5 together with FIG.

  In a state where the product ranking list for a large number of users is held through the product ranking setting process (S100), the similar user determination unit 130 of the recommendation service device 100 performs the identification by the product recommendation request (S210) from a specific user. The product ranking list corresponding to the user and the product ranking list corresponding to at least one other user are cross-compared (S220), and the degree of preference for the specific user and other users is calculated. (S230). Then, based on the calculated similarity, another user having a preference similar to that of the specific user is determined (S240).

  For example, as shown in FIG. 5A, the restaurant corresponding to the user N from the restaurant ranking list 310a in which the ranking is set in a sequential order for each of the plurality of restaurants corresponding to the user A. When the user A requests restaurant recommendation while the ranking list 310n is held, the restaurant corresponding to the user N is selected from the restaurant ranking list 310a corresponding to the user A and the restaurant ranking list 310b corresponding to the user B. The ranking list 310n is cross-compared to calculate the similarity of each other user to the user A.

  At this time, the similarity is calculated in consideration of only the restaurant ranking list including the same number of products as the products included in the product ranking list corresponding to the user A among the users B to N. You can also. The similarity may be calculated in consideration of not only the number of the same products registered in the product ranking list but also the ratio of the same products. For example, a person with a very high ratio of the same product may not be selected as a comparison target because there is a high possibility that information on a new product is not included or is low.

  In connection with the similarity calculation, referring to the example shown in FIG. 5 (b), the restaurant ranking list in which the ranking is set in a sequential order for each of a plurality of restaurants, In FIG. 5B, the restaurants included in the restaurant ranking list corresponding to the user A are classified into groups 320a and 330a by ten according to the order of order. Similarly, ten restaurants included in the restaurant ranking list corresponding to the user B are also classified into groups 320b and 330b.

  Then, the similarity of the user B to the user A is calculated through the cross comparison by restaurant group included in the restaurant ranking list of the user A and the user B who requested the restaurant recommendation. The ranking group 320a is cross-compared with the upper ranking group 320b and the lower ranking group 330b corresponding to the user B, and the lower ranking group 330a corresponding to the user A is compared with the upper ranking group 320b and the lower ranking group 330b corresponding to the user B. Cross comparison is made for each ranking group 330b.

  For example, as a grasp of the similar preference between the user A and the user B, the restaurant 323a included in the higher ranking group 320a corresponding to the user A and the restaurant 323b included in the upper ranking group 320b corresponding to the user B And the number of restaurants that belong to the same higher ranking group as in restaurant 1 (325a, 325b). Then, the restaurant 333a included in the lower ranking group 330a corresponding to the user A and the restaurant 333b included in the lower ranking group 330b corresponding to the user B are compared, and a restaurant 12 (335a, 335b) is obtained. Identifies the number of restaurants that belong to the same lower ranking group.

  Further, as a grasp of the opposite preference between the user A and the user B, the restaurant 333a included in the higher ranking group 320a corresponding to the user A and the restaurant 333b included in the lower ranking group 330b corresponding to the user B And the number of restaurants belonging to the opposite ranking group, such as restaurant 2 (327a, 337b), is grasped. Similarly, the restaurant 333a included in the lower ranking group 330a corresponding to the user A is compared with the restaurant 323b included in the upper ranking group 320b corresponding to the user B, and the restaurant belonging to the opposite ranking group Know the number.

  Based on such comparison, the similarity of the user B to the user A is calculated, but the restaurant belonging to the same higher ranking group is likely to have similar preferences of the user A and the user B. Therefore, a higher weighting value is applied to the number of restaurants belonging to the upper ranking group, and the tastes of the user A and the user B may be somewhat similar in restaurants belonging to the same lower ranking group. Therefore, an intermediate weight is applied to the number of restaurants belonging to the lower ranking group, and restaurants belonging to the opposite ranking group may have different preferences for user A and user B. Therefore, a lower weight value is applied to calculate the degree of similarity.

  As an example, the similarity of the user B to the user A can be calculated by the following [Equation 1].

        SAB = (α × H + β × L + γ × R) / T; (α> β> γ) [Formula 1]

  Here, SAB is the similarity of user B to user A, α is a weighted value for products belonging to the same higher ranking group, β is a weighted value for products belonging to the same lower ranking group, and γ is Weights for products belonging to the opposite ranking group, H is the number of products belonging to the same upper ranking group, L is the number of products belonging to the lower ranking group, and R is to the opposite ranking group The number of products, T, represents the number of products obtained by adding H, L and R.

  Such a process is repeatedly performed, and the similarity with respect to the user A is calculated from the user B to the user N shown in FIG.

  Then, another user having a high degree of similarity with respect to the user A is selected. At this time, only one other user having the highest degree of similarity can be selected, or a preset numerical value can be selected. Many other users with similarities up to can be selected.

  Third, the product recommendation process (S300) will be described with reference to the product recommendation process in the recommendation service method according to the present invention shown in FIG. 6 together with FIG.

  When one or more similar preference users having a preference similar to that of the specific user who requested the recommendation of the product through the similar preference user determination process (S200) are selected, the product recommendation unit 150 of the recommendation service device 100 is selected. Extracts a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more selected similar preference users (S310).

  Next, the product recommendation unit 150 of the recommendation service device 100 compares the product ranking list of the specific user with the product ranking list of the similar preference user (S320), and the product ranking list corresponding to the similar preference user. The products not included in the product ranking list corresponding to the specific user are extracted as new products (S330).

  Then, the product recommendation unit 150 of the recommendation service device 100 determines a recommended product based on the ranking for the new product set by the similar preference user (S340), and provides recommended information including the recommended product to the specific user. (S350).

  For example, when only one other user who has the highest degree of similarity to a specific user who requested product recommendation is selected as a similar preference user, a new product is selected from the ranking list corresponding to one similar preference user. The new product is extracted and selected as a recommended product in the ascending ranking up to the preset number of new products or new products.

  Furthermore, a case where a large number of other users up to a numerical value for which similarity is already set is selected as a similar preference user will be described with reference to the embodiment illustrated in FIG.

  FIG. 6 shows an example in which recommended restaurants are extracted based on restaurant ranking settings of a number of similar preference users in association with the restaurant recommendation described above.

  First, a restaurant ranking list corresponding to user B, user D,..., User C selected as user A and similar preference users is extracted, and the restaurant ranking corresponding to user A is selected from each restaurant ranking list. A restaurant that is not included in the list is extracted as a new restaurant, and as shown in FIG. 6, a new restaurant table 410 in which new restaurants corresponding to each similar preference user are arranged in the order of similarity of similar preference users. Generate. The portion indicated by X in FIG. 6 is a restaurant included in the restaurant ranking list corresponding to the user A and is excluded from the new restaurant.

  Based on the new restaurant table 410, the recommended score is calculated based on the similarity of the corresponding similar taste user and the ranking set by the corresponding similar taste user for each new restaurant, but the restaurant 3 (411 413), the ranking set by the user B with the highest degree of similarity is second, and the ranking set by the user D with the next highest degree of similarity is first, so this is taken into account. If so, the recommended number is calculated high. In the case of the restaurant 6 (415, 417), the ranking set by the user B with the highest degree of similarity is third, and the ranking set by the user C with the lowest degree of similarity is first. A recommendation score lower than that of restaurant 3 (411) is calculated.

  As an example, the recommended score for a new product can be calculated by the following [Equation 2].

  Here, SCORE is the recommended score for the corresponding new product, M is the total number of similar preference users who set the ranking for the new product, SI is the similarity of the i-th similar preference user, and Rank is the i-th similarity The ranking set by the intended user for the corresponding new product, k is a weighting value set as necessary.

  As described above, the recommended score is calculated for each new product, and the recommended product is selected based on the calculated recommended score. Only one new product with the highest recommended score is calculated as the recommended product. It is also possible to select a large number of new products up to the set higher order among the recommended points, and it is also possible to select the recommended products.

  When the recommended product is selected, the product recommendation unit 150 of the recommendation service device 100 generates recommendation information including the recommended product and provides it to the user. For example, the recommended service method according to the present invention illustrated in FIG. As in the embodiment for the recommended product information, the recommended information 420 including the recommended restaurants in the sequential order along with the recommended order is provided to the user terminal. At this time, various kinds of additional information can be included in the recommendation information. For example, in the case of restaurant recommendation as shown in FIG. 7, information such as menu, price, location, and telephone number can be included. A user's evaluation sentence can be included, and further, a URL for a web page on which such various information is presented can be included.

(Improved example 1)
Here, the product recommendation process (S300) can be improved as follows. Hereinafter, the product recommendation process (S300) executed by the recommendation service method according to the improved example of the embodiment will be described with reference to FIGS. 3, 8, and 9. FIG.

  The product recommendation unit 150 of the recommendation service device 100 compares the product ranking list of the specific user with the product ranking list of the similar preference user (S320), and selects the specific user from the product ranking list corresponding to the similar preference user. A product not included in the corresponding product ranking list is extracted as a new product (S330).

  Hereinafter, the comparison (S320) and the new product extraction (S330) process will be described in more detail with reference to FIG.

  First, the product ranking list corresponding to the user B, the user D, the user E, the user C,... The user F selected as the user A and the similar preference user is extracted, and the product ranking list is extracted from each product ranking list. A new product table 510 in which products not included in the product run ranking list corresponding to the user A are extracted as new products, and new restaurants corresponding to the similar preference users in order of similarity of the similar preference users are arranged. Generate. The part indicated by * in FIG. 8 is a product that is included in the product run ranking list corresponding to the user A and is excluded from the new product.

  Based on such a new product table 510, the recommended score is calculated based on the similarity of the corresponding similar preference user and the ranking set by the corresponding similar preference user for each new product. It becomes as follows.

  A rank score is assigned to each rank in advance. In this example, a ranking score 100 is given to the first place, a ranking score 90 is given to the second place, a ranking score 80 is given to the third place, and a ranking score 70 is given to the fourth place. Then, the degree of similarity SIRawi of each comparison target person, that is, a user with a similar preference is converted into a ratio SIi such that the sum is 1. This conversion is based on the following [Equation 3].

Here, SI _i is a ratio related to the similarity, SIRaw _i is the similarity of each user having a similar preference, and M is the number of comparison subjects.

After calculating the ratio SI _i related to the similarity in this way, the recommended score SCORE of the product to be evaluated is calculated by the following [Formula 4] based on the ratio of each user and the rank score RankScore _i .

Here, SI _i is a ratio relating to the degree of similarity, RankScore _i is a ranking score of products in each user's product list, and M is the number of comparison subjects.

For example, in FIG. 8, the recommended score and the predicted ranking of the product P12 are determined as follows.
SCOREP12 = (SI ₁ * 80 + SI ₂ * 80 +... + SI _i * 80)
= 80 * ΣSI _i = 80
Therefore, in the user A's product ranking list, the recommended score is the same as that of the product P3, so the predicted ranking is third.

In FIG. 8, the recommended score and the predicted ranking of the product P10 are determined as follows.
SCOREP10
= (SI ₁ * 100 + SI ₂ * 90 + SI ₃ * 70... + SI _i * 90)
The recommended ranking SCOREP10 is compared with the ranking score of the product ranking list of the user A to determine the expected ranking.

  As described above, the recommended score is calculated for each new product, and the recommended product is selected based on the calculated recommended score. Only one new product with the highest recommended score is calculated as the recommended product. It is also possible to select a large number of new products up to the set higher order among the recommended points, and it is also possible to select the recommended products.

  When the recommended product is selected, the product recommendation unit 150 of the recommendation service device 100 generates recommendation information including the recommended product and provides it to the user. For example, as shown in FIG. Recommendation information 520 including the name is provided to the user terminal. At this time, the recommendation information can include various additional information. For example, if the product is a restaurant recommendation, it can include information such as menus, prices, locations, phone numbers, etc., it can also include various user evaluation statements, and such diverse information A URL for the presented web page can also be included.

(Improvement example 2)
Further, the user subjectively gives a relative preference to his / her product, generates a product ranking list, and judges similar preference users and recommended products based on the product ranking list. It may be. In this case, since it becomes possible to judge the similar preference user and the recommended product based on the subjective relative preference for the product of the user and the minimum information, the database capacity, calculation, It becomes possible to reduce processing load.

  In this case, the similar preference user determination process (S200) can be improved as follows. In the above-described method, a user whose product ranking list includes products common to the user A (for example, in order of the number of common products) is to be compared, and the products of the user A and other users who are compared with each other are compared. The ranking list was divided into groups, cross comparisons were made for each group, similarity was calculated, and similar preference users were judged.

  On the other hand, a user who includes many products belonging to the upper group of the user A in the upper group may be determined as a similarity-oriented user.

  Alternatively, the ranking of each product included in the product ranking list of user A is confirmed in the product ranking list of the user to be compared, and the similarity is calculated by comprehensively considering the difference in ranking, A similar preference user may be determined. This may be calculated, for example, by comparing the ranking of the product X in the product ranking list of the user A with the ranking of the product X in the product ranking list of another user.

(Improved example 3)
In FIG. 10, the structure of the recommended goods service apparatus which concerns on the improvement example 3 is shown and demonstrated.

  Since the basic configuration is the same as in FIG. 2, the same components are denoted by the same reference numerals and different portions will be mainly described. This improved example has one of the features that a specific user verifies and evaluates in advance whether or not it is close to his / her preference before using the product. In order to realize this, a product designation receiving unit 180 and a pre-verification / evaluation unit 190 are provided. The product designation receiving unit 180 accepts designation of products that the user has never experienced before. Then, the pre-verification / evaluation unit 190 makes a request for pre-verification and evaluation of the designated product. Detailed processing of each part will be described later.

  Hereinafter, with reference to the flowchart of FIG. 11, a processing procedure performed by the recommended service device according to Modification 3 will be described. Here, the same processing steps as those shown in FIG. The concept of FIGS. 4 to 10 is basically applicable in the same manner.

  A product ranking list is generated and edited for each user (S110 to S140). Details of this processing are as described above with reference to FIG.

  Subsequently, the specification of the product that the specific user is interested in is received (S410), and the specific user requests evaluation information for the specified product from the system side for prior verification and evaluation of the specified product (S420).

  A similar preference user is determined (S220 to S240). Then, only the designated product is extracted from the similarity (ratio) of the similar preference user and the product ranking list set by the similar preference user, and based on the rank information for the designated product, the recommended score of the designated product Is calculated (S310 to S340).

  That is, the product recommendation unit 150 of the recommendation service device 100 extracts the designated product from the product ranking list corresponding to the similar preference user (S330), calculates the recommended score of the designated product, and determines the ranking (S340).

  Note that in the extraction of the designated product, it is not necessary to extract all new products. Next, pre-verification / evaluation result information (recommended score, ranking) of the designated product is provided (S350). Since these processes are also as described above with reference to FIG. 3, details thereof will be omitted.

  This completes the series of processing up to provision of recommended product information.

  In S340, the calculation is performed based on the similarity ratio of the similar preference user and the ranking / ranking score for the designated product. In S350, the method for displaying the designated product on the product ranking list of the specific user, the method for displaying on the menu of the restaurant, the method for displaying on the map, and the like are given to the specific user. Provide evaluation results for designated products.

  As for existing methods, it was impossible to specify a product to be verified and evaluated by finding a new product. In addition, since new products as a whole are subject to evaluation, system resources are often consumed.

  On the other hand, the method of the present improved example 3 is suitable for checking in advance whether a specific product that has not yet been experienced is close to the user's taste, and evaluates only products that the user is interested in. Can be specified as a product. Another feature is that the consumption of system resources is low because only the products of interest are evaluated.

  The processing procedure related to actual pre-verification / evaluation will be described more specifically as follows.

  For example, user A designates a product that he / she is interested in from street signs, televisions, newspapers, magazines, blogs, SNSs, advertisements, job offers, the Internet, catalogs, restaurant menus, etc. The evaluation unit 190 requests prior verification and evaluation. Then, the recommended service device calculates the recommended score of the designated product based on the evaluation of the designated product (s) by the similar preference user (s), and the expected rank and recommendation on the product ranking of the user A Recommendation information such as availability is provided to user A.

  Here, Fig.12 (a) has shown the display screen displayed during the user A's product ranking. FIG. 12B shows a display screen displayed together with the restaurant menu. In FIG. 12B, the expected rank on the product ranking of the user A is displayed. FIG. 12C shows a display screen that displays mainly the recommendation information of the corresponding product. In the screen of FIG. 12C, the expected rank on the product ranking of the user A is displayed. In the screen of FIG. 13, the evaluation result is displayed on the map.

  More specifically, for example, in the case of verification and evaluation of a ramen shop found on the street, for example, when it is desired to check whether “AA ramen” “tsukemen” suits his / her mouth, first, the product designation receiving unit 180 However, the designation of “tsukemen” of “AA ramen” is accepted, and the pre-verification / evaluation unit 190 requests pre-verification and evaluation. Then, the predicted ranking on the product ranking and the recommendation possibility (option) are displayed on the screen.

  As another example, in the case where you want to select a menu that suits your mouth at the first restaurant “BB 庵”, for example, the product designation receiving unit 180 displays the entire menu, Accepts the designation of okonomiyaki with plenty of pork, mentaiko, salmon, cheese monja, monji made with Heidi, potato cheese okonomiyaki, and okonomiyaki to eat with soy sauce. Make a request. Then, the predicted order on the product ranking of each menu and the recommendation availability (option) are displayed on the screen (menu).

  Furthermore, in another example, the predicted ranking on the product ranking of “tsukemen” around Akasaka Station is displayed on the map with characters and colors. In other words, the product and menu rankings that interest you in a specific area are displayed on the map in letters and colors. In conjunction with the map, search for the menu “Tsukemen” within the area of the specific area, request verification and evaluation from the system with the information (ID number etc.) of the searched “Tsukemen”, and for each “Tsukemen” Display on the map the expected rank on your product ranking or the recommendation availability.

  The group classification method and comparison method in the similar user judgment process are as follows.

  In other words, the classification method of a group that is “classify multiple products into multiple groups in sequential order” is not limited to “sequential order”, and only product groups with a specific range of ranks. To calculate the similarity.

  For example, when calculating the similarity by cross-comparison for each group between a group composed of the top 1 to 10th products and a group composed of products below a specific rank, the following first and second The processing procedure is as follows.

  That is, in the first processing procedure, a plurality of products may be classified into a plurality of groups according to a sequential order, or a group may be classified without being in a sequential order.

  Furthermore, in the second processing procedure, a plurality of groups classified by the first processing procedure described above, that is, a method of classifying a plurality of products into a plurality of groups in a sequential order, is selectively cross-compared. To be applicable to. In the case of this second method, only the group composed of the top 1 to 10th products and the group composed of the 30th or lower products are included in the comparison process for calculating the similarity, and ranked 11th. It is recommended that the group composed of the 29th product is not included in the comparison process.

  As described above, the recommended service device according to the improvement example 3 sets the rankings for the products for each user based on the user's preference product selection based on the relative comparison with respect to the plurality of products according to the preferences of the plurality of users. A product ranking setting unit 110 to perform, a product designation receiving unit 180 that accepts designation of a product that the user has never experienced, and a pre-verification / evaluation unit 190 that makes a request for pre-verification and evaluation of the designated product. And a ranking for a product set by a specific user and a ranking for a product set by one or more other users to determine other users having a similar preference to that of the specific user To the rankings for products set by other users who have similar preferences to that of the specific user And a product recommendation unit 150 for recommending a product to the specific user's Zui. The product ranking setting unit 110 receives a plurality of product information input from the user, generates a product list, and performs a relative comparison according to the user's preference for the plurality of products included in the product list. Based on the user's preferred product selection, rankings for the plurality of products are set, and a product ranking list is generated. The similar user determination unit 130 performs cross-comparison between a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more other users among the plurality of users to obtain a similarity. Based on the similarity, one or more other users having an interest similar to that of the specific user among the one or more other users are determined. In the calculation of the similarity by the similar user determination unit 130, a plurality of products are sequentially added to each of the product ranking list corresponding to the specific user and the product ranking list corresponding to the other user. Classify into multiple groups according to a random order or non-sequential order, and cross-comparison the product ranking list group corresponding to the specific user and the product ranking list group corresponding to the other user by group The similarity of the other user to the specific user is calculated based on the number of the same product for each group.

  Accordingly, it is possible to accept designation of a product that the user is interested in, and to select an optimal product based on the prior verification and evaluation results of the designated product. Thus, it is suitable when confirming in advance whether a specific product that has not yet been experienced is close to his / her preference, and only products that the user is interested in can be designated as evaluation target products. Another feature is that the consumption of system resources is low because only the products of interest are evaluated.

  As detailed above, according to the present invention, by recommending a product based on an evaluation for a product set by another user having a taste similar to that of the user, a product that the user can satisfy is recommended. Recommendations can be made.

  In addition, since the product is recommended based on the evaluation of a large number of other users who have similar preferences to the user who requested the product recommendation, it is possible to eliminate advertisement information and the reliability of the recommendation. Can be increased.

  In addition, the user can select a product more effectively by overcoming the limitations of simple evaluation for products using the number of stars, etc., and performing comparative evaluation among a number of products and making recommendations.

  The above description is merely illustrative of the technical idea of the present invention, and a person who has ordinary knowledge in the technical field to which the present invention belongs does not depart from the essential characteristics of the present invention. Various modifications and variations are possible. Therefore, each embodiment described in the present invention is not intended to limit the technical idea of the present invention, but to explain it, and the technical idea of the present invention is not limited by such an embodiment. . The protection scope of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the right of the present invention.

  For example, the present invention is also realized as a program that causes a computer to function as the ranking setting unit 110, the similar user determination unit 130, the product recommendation unit 150, the product designation reception unit 180, and the pre-verification / evaluation unit 190.

10a, 10b, ... 10n User 30 Network 50 Search service server 100 Recommended service device 110 Ranking setting unit 130 Similar user determination unit 150 Product recommendation unit 170 User information DB
180 Product designation reception unit 190 Prior verification / evaluation unit

Claims (11)

A product ranking setting unit that sets a ranking for a product for each user based on a user's preferred product selection by a relative comparison with a plurality of products according to each user's preferences;
A product designation receiving unit that accepts designation of a product that the user has never experienced, and
A pre-verification / evaluation unit that requests pre-verification and evaluation of the designated product;
Similarity in which a ranking for a product set by a specific user and a ranking for a product set by one or more other users are cross-compared to determine other users who have similar preferences to that of the specific user A user judgment unit;
A product recommendation unit that recommends a product to the specific user based on a ranking for a product set by another user having a preference similar to the preference of the specific user,
The product ranking setting unit generates a product list upon receiving a plurality of product information input from the user, and the user through a relative comparison according to the user's preference for a plurality of products included in the product list. Set rankings for the plurality of products based on the preferred product selection, and generate a product ranking list,
The pre-verification / evaluation unit requests prior verification / evaluation as to whether or not the specified product related to the specific user is suitable for the specific user before experiencing the specified product,
The similar user determination unit calculates a similarity by cross-comparing a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more other users among the plurality of users. And determining one or more other users having a preference similar to that of the specific user among the one or more other users based on the similarity,
In the calculation of the similarity by the similar user determination unit, a plurality of products are sequentially arranged in order for each of the product ranking list corresponding to the specific user and the product ranking list corresponding to the other user. Or, it is classified into a plurality of groups by non-sequential ordering, and the group of the product ranking list corresponding to the specific user and the group of the product ranking list corresponding to the other user are cross-compared by group, Separately, the recommendation service apparatus calculates the similarity of the other user to the specific user based on the number of the same products. The product recommendation unit extracts a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more selected similar preference users, and a product ranking list corresponding to the similar preference users The designated product is extracted from the list, the designated product is evaluated based on a ranking for the designated product set by the similar preference user, and the evaluation result is provided to the specific user. Recommendation service device. Each ranking of the product ranking list is given a ranking score in advance,
The product recommendation unit converts the similarity of the similar preference user into a ratio such that the total is 1 when evaluating the designated product, and based on the ratio related to the similarity and the ranking score The recommendation service device according to claim 2, wherein the recommendation score of a new product or a designated product is calculated. The recommendation service apparatus according to claim 3, wherein the product recommendation unit determines a recommendation order based on the recommended score and displays the recommendation order in the product ranking of the specific user. The recommendation service device according to claim 3, wherein the product recommendation unit determines a recommendation order based on the recommended score, and displays a product having a high recommendation order on a map. A product recommendation service method by a recommendation service device comprising a product ranking setting unit, a similar user determination unit, and a product recommendation unit,
A step of setting a ranking for the product for each user based on the user's preferred product selection by a relative comparison with respect to the plurality of products according to the preferences of each of the plurality of users;
A step in which a product designation receiving unit accepts designation of a product that the user has never experienced, and
A step in which the prior verification / evaluation section requests prior verification and evaluation for evaluating in advance whether or not the specified user is satisfied with the preference before the specific user experiences the designated product relating to the specific user;
The similar user determination unit crosses and compares the ranking for the product set by the specific user with the ranking for the product set by one or more other users, and has similar taste to that of the specific user Determining the user of
A product recommendation unit recommending a product to the specific user based on a ranking for a product set by another user having a preference similar to the preference of the specific user;
Have
The product ranking setting unit receives a plurality of product information input from the user to generate a product list, and the user through a relative comparison according to the user's preference for a plurality of products included in the product list Set rankings for the plurality of products based on the preferred product selection, and generate a product ranking list,
The similar user determination unit calculates a similarity by cross-comparing a product ranking list corresponding to the specific user with a product ranking list corresponding to one or more other users among the plurality of users. And determining one or more other users having a preference similar to that of the specific user among the one or more other users based on the similarity,
In the calculation of the similarity by the similar user determination unit, a plurality of products are sequentially arranged in order for each of the product ranking list corresponding to the specific user and the product ranking list corresponding to the other user. Or, it is classified into a plurality of groups by non-sequential ordering, and the group of the product ranking list corresponding to the specific user and the group of the product ranking list corresponding to the other user are cross-compared by group, Separately, the product recommendation service method is characterized in that the similarity of the other user to the specific user is calculated based on the number of the same product. The product recommendation unit extracts a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more selected similar preference users, and the similarity of similar preference users and similar preference 7. The product recommendation service according to claim 6, wherein only the designated product is extracted from the product ranking list set by the user, and the recommended score of the designated product is calculated based on the ranking information for the designated product. Method. Each ranking of the product ranking list is given a ranking score in advance,
When the product recommendation unit evaluates the designated product, the similarity level of the similar preference user is converted into a ratio such that the total is 1, and based on the ratio related to the similarity level and the rank score The product recommendation service method according to claim 7, wherein the recommended number of designated products is calculated. The product recommendation service method according to claim 8, wherein the product recommendation unit determines a recommendation order based on the recommended score, and displays the recommendation order in the product ranking of the specific user. The product recommendation service method according to claim 8, wherein the product recommendation unit determines a recommendation order based on the recommended score, and displays a product having a high recommendation order on a map. Recommended service device
A product ranking setting unit that sets a ranking for a product for each user based on a user's preferred product selection by a relative comparison with a plurality of products according to each user's preferences;
A product designation receiving unit that accepts designation of a product that the user has never experienced, and
A pre-verification / evaluation unit that requests pre-verification and evaluation of the designated product;
Similarity in which a ranking for a product set by a specific user and a ranking for a product set by one or more other users are cross-compared to determine other users who have similar preferences to that of the specific user A user judgment unit;
Function as a product recommendation unit that recommends a product to the specific user based on a ranking for a product set by another user having a preference similar to that of the specific user,
The product ranking setting unit generates a product list upon receiving a plurality of product information input from the user, and the user through a relative comparison according to the user's preference for a plurality of products included in the product list. Set rankings for the plurality of products based on the preferred product selection, and generate a product ranking list,
The pre-verification / evaluation unit requests prior verification / evaluation as to whether or not the specified product related to the specific user is suitable for the specific user before experiencing the specified product,
The similar user determination unit calculates a similarity by cross-comparing a product ranking list corresponding to the specific user and a product ranking list corresponding to one or more other users among the plurality of users. And determining one or more other users having a preference similar to that of the specific user among the one or more other users based on the similarity,
In the calculation of the similarity by the similar user determination unit, a plurality of products are sequentially arranged in order for each of the product ranking list corresponding to the specific user and the product ranking list corresponding to the other user. Or, it is classified into a plurality of groups by non-sequential ordering, and the group of the product ranking list corresponding to the specific user and the group of the product ranking list corresponding to the other user are cross-compared by group, A program for calculating the similarity of the other user with respect to the specific user based on the number of the same product.