Information processing device, information processing method, promotion effect prediction method, and program

Abstract

To provide an information processing device that utilizes big data from purchases to provide information useful for understanding product positioning. [Solution] According to the embodiment, the information processing device comprises an input receiving unit, a cluster generation unit, and an overview diagram generation unit. The input receiving unit receives input of a population setting condition that sets the conditions for the product group constituting the population, and a product of interest designation condition that specifies a product of interest that deserves attention within the product group. Based on the population setting condition, the cluster generation unit analyzes the purchase data of the product group constituting the population and generates multiple clusters, including a product of interest cluster to which the product of interest designated by the product of interest designation condition belongs, and other clusters. The overview diagram generation unit generates an overview diagram that represents the product of interest in a diagram representing the product of interest cluster, showing its first relationship to the product of interest cluster and its second relationship to other products in the product of interest cluster, so that it can be identified from other products.

Description

[Technical Field]

[0001] Embodiments of the present invention relate to an information processing apparatus, an information processing method, a promotion effect prediction method, and a program that utilize purchasing big data to provide information useful for grasping the positioning of products from the standpoints of both retailers and producers. [Background Art]

[0002] Conventionally, in data analysis performed by manufacturers and retailers, product development, sales promotion and other activities are effectively carried out by analyzing the positioning of products and understanding the products. [Prior Art Documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Unexamined Patent Application Publication No. 2024-040549 [Non-Patent Documents]

[0004] [Non-Patent Document 1] https: / / www.microsoft.com / en-us / research / uploads / prod / 2006 / 01 / Bishop-Pattern-Recognition-and-Machine-Learning-2006.pdf [Summary of the Invention] [Problem to be Solved by the Invention]

[0005] However, although conventional techniques such as co-sales analysis can grasp local co-occurrence where one product and another product are purchased at the same time, they cannot clarify based on what preferences a huge number of individual products are purchased (purchasing axes), nor what position each product occupies among all products.

[0006] Therefore, retailers only know the sales figures for products sold in their own stores, and do not know which products are selling well and to what extent in other stores. Even if they did know which products are selling well and to what extent in other stores, they would not know the positioning of each product, and given the limited shelf space in their own store, they would not know which products to replace and which new products to stock to increase sales.

[0007] Therefore, retailers are seeking technology that can evaluate the appropriateness of their current product assortment and, based on the results, propose specific policies regarding their product lineup.

[0008] On the other hand, manufacturers and other producers often lack an understanding of their own positioning and the positioning of their products in the market, that is, their strengths and weaknesses. As a result, they are unable to make decisions about what products to develop, which products to discontinue, or what kind of product promotion is necessary to increase sales.

[0009] Therefore, producers are seeking technology that can analyze the sales of their current market products and, based on the results, propose future product development and marketing strategies.

[0010] The problem that this invention aims to solve is to provide an information processing device, an information processing method, a promotional effect prediction method, and a program that utilize big data on purchases to provide information useful for understanding product positioning from the perspectives of both retailers and producers. [Means for solving the problem]

[0011] The information processing device of this embodiment comprises an input receiving unit, a cluster generation unit, and an overview diagram generation unit. The input receiving unit receives input of population setting conditions, which set the conditions for the product group constituting the population, and attention product designation conditions, which specify attention products that deserve attention within the product group. Based on the population setting conditions, the cluster generation unit analyzes the purchase data of the product group constituting the population and generates multiple clusters, including attention clusters to which the attention products specified by the attention product designation conditions belong, and other clusters. The overview diagram generation unit generates an overview diagram that represents the attention clusters, showing the attention products in a way that allows them to be distinguished from other products, while also showing their first relationship to the attention cluster and their second relationship to other products within the attention cluster. [Brief explanation of the drawing]

[0012] [Figure 1A] Figure 1A is a block diagram showing an example of the functional configuration of an information processing device to which the information processing method of the first embodiment is applied. [Figure 1B] Figure 1B is an explanatory diagram showing an example of the hardware configuration of an information processing device. [Figure 2] Figure 2 is a conceptual diagram illustrating purchasing characteristics information. [Figure 3] Figure 3 shows an example of prompts for obtaining label names from the generating AI based on a list of all products belonging to a cluster. [Figure 4A] Figure 4A shows an example of prompts for obtaining label names from the generating AI based on a list of some products belonging to a cluster. [Figure 4B] Figure 4B is a data structure diagram showing an example of the relationship between product purchase characteristics and clusters. [Figure 5A] Figure 5A shows a list of all products belonging to a cluster and an example of prompts for obtaining label names from the generated AI based on product purchase characteristics. [Figure 5B] Figure 5B is a data structure diagram showing an example of product purchase characteristics information. [Figure 6]FIG. 6 is a diagram showing an example of a prompt for obtaining a label name from generative AI based on product purchase characteristics of all products belonging to a cluster. [Figure 7] FIG. 7 is a diagram showing an example of a bird's-eye view generated by an information processing apparatus. [Figure 8] FIG. 8 is a diagram showing an example of a bird's-eye view in which the contour color of a diagram representing a cluster is changed according to the sales share of a target product in the cluster. [Figure 9] FIG. 9 is a diagram showing an example of screen transition of the information processing apparatus when the user is a retailer. [Figure 10] FIG. 10 is a flowchart showing an operation example during positioning analysis by the information processing apparatus when the user is a retailer. [Figure 11] FIG. 11 is a data structure diagram showing an example of purchase data. [Figure 12] FIG. 12 is a data structure diagram showing an example of a product. [Figure 13] FIG. 13 is a data structure diagram showing an example of a purchaser. [Figure 14] FIG. 14 is a data structure diagram showing an example of a store. [Figure 15] FIG. 15 is a flowchart showing in detail processing for calculating product hidden state information. [Figure 16] FIG. 16 is a data structure diagram showing an example of a purchase matrix. [Figure 17] FIG. 17 is a data structure diagram showing an example of product purchase characteristic information. [Figure 18] FIG. 18 is a diagram for explaining a first variation of a proposal based on a bird's-eye view. [Figure 19] FIG. 19 is a diagram for explaining a second variation of a proposal based on a bird's-eye view. [Figure 20] FIG. 20 is a diagram explaining a proposal made to a user who is a retailer (in the case of a high-attention cluster). [Figure 21]Figure 21 illustrates a proposal made to a retailer user (in the case of a strong attention cluster). [Figure 22] Figure 22 illustrates a proposal made to a retailer user (in the case of a weak attention cluster). [Figure 23] Figure 23 illustrates a proposal made to a retailer user (in the case of a weak attention cluster). [Figure 24] Figure 24 illustrates a proposal made to a retailer user (in the case of a weak attention cluster). [Figure 25] Figure 25 shows an example of screen transitions in an information processing device when the user is a producer. [Figure 26A] Figure 26A shows an example of an overview view provided to the user, who is a producer. [Figure 26B] Figure 26B is an overview diagram illustrating the method for determining whether or not each cluster has a market share in the segment it represents. [Figure 27] Figure 27 is a bar graph created by the graph display unit, showing the sales share of the products. [Figure 28] Figure 28 illustrates a method for determining cannibalism. [Figure 29] Figure 29 shows an example of an overview diagram used to support new product development. [Figure 30] Figure 30 illustrates the cluster movement of products in a simulation of promotion and improvement of existing products. [Modes for carrying out the invention]

[0013] Embodiments of the present invention will be described below with reference to the drawings. The drawings are schematic or conceptual, and the relationship between the thickness and width of each part, the ratio of the sizes of the parts, etc., are not necessarily the same as those of reality. Furthermore, even when representing the same part, the dimensions and ratios may be represented differently in the drawings. In this specification and each drawing, elements similar to those described in previously shown drawings are denoted by the same reference numerals, and detailed explanations and redundant explanations are omitted as appropriate.

[0014] (Device configuration) Figure 1A is a block diagram showing an example of the functional configuration of an information processing device to which the information processing method of the first embodiment is applied.

[0015] The information processing device 10 processes purchase big data and provides information useful for understanding product positioning. Specifically, it performs product positioning analysis and outputs the results in the form of an overview diagram W, as exemplified in Figure 7 described later. Furthermore, by analyzing the overview diagram W from various perspectives, retailers can obtain information useful for shelf allocation and sales promotion, and producers can obtain information useful for understanding sales status and developing new products.

[0016] As shown in Figure 1A, the information processing device 10 includes an input receiving unit 11, an acquisition unit 12, a cluster generation unit 14, a label name generation unit 16, an overview diagram generation unit 18, an analysis unit 20, a proposal unit 22, a graph display unit 24, an output control unit 26, and a display device 28.

[0017] The input receiving unit 11 is the part that receives input from the user for setting conditions b necessary to create the overview diagram W. Specifically, setting conditions b include a population setting condition b1 that sets the conditions for the group of products that constitute the population considered in creating the overview diagram W, and a focus product designation condition b2 that specifies a focus product that deserves attention within the group of products.

[0018] A featured product is a product that is subject to positioning analysis by the information processing device 10. While not limited to this, for example, if the user is a retailer, the user can specify products sold in their own store as featured products in the featured product designation condition b2.

[0019] On the other hand, if the user is a producer involved in development, such as a manufacturer, they can designate products produced by a specific producer or group of producers as featured products in the featured product designation condition b2. A specific producer could be, for example, the company to which the user belongs (e.g., "Food Manufacturer YYYY"), and a specific group of producers could be, for example, the corporate group to which that company belongs (e.g., "YYYY Group"). A specific group of producers may include multiple different producers (e.g., "Food Manufacturer YYYY" and "Food Manufacturer ZZZZ").

[0020] Regardless of whether a user is a retailer or a producer, in the featured product designation condition b2, users can designate not just one product, but multiple products as featured products.

[0021] Furthermore, setting condition b is not limited to the population setting condition b1 and the product of interest designation condition b2. Additional conditions can be added to setting condition b and input from the input reception unit 11 in order to create a more customized overview diagram W.

[0022] The acquisition unit 12 acquires purchase data a for creation in the overhead view W. The acquisition unit 12 can acquire purchase data a from, for example, an external server.

[0023] Purchase data a includes purchase characteristics information a1 showing the purchase characteristics for each product, multiple buyer identification information (e.g., "buyer ID") a2 identifying buyers, multiple product identification information (e.g., "product ID") a3 identifying products, and performance information a4. Performance information a4 is information that includes, for example, the price of the product and at least one of the number of units purchased (purchase quantity).

[0024] Figure 2 is a conceptual diagram illustrating the purchasing characteristics information a1.

[0025] Purchase characteristic information a1, as shown in Figure 2, is information that expresses the purchases of various people using, for example, around 100 purchase axes. One example of such purchase characteristic information a1 is the hidden state information a11 of purchase data a. Purchase axes indicate, for example, what kind of preferences lead to the purchase of a product, such as a preference for rich flavors or a preference for dietary fiber. Purchase axes can be related to taste, texture, ingredients, health consciousness and other functional aspects, as well as package design, container shape, material, and capacity.

[0026] For example, these could include package designs using bright colors, packages made from environmentally friendly materials, products containing multiple items (such as a 6-pack of beer or a box of ice cream), or products with larger or smaller capacities than usual. The approximately 100 purchasing criteria can include a mix of various perspectives, such as purchasing criteria related to taste preferences or preferences related to package design.

[0027] A hidden state refers to a state that is not observed or is not visible. Therefore, the hidden state information a11 of purchase data a means information that cannot be directly obtained from purchase data a.

[0028] The acquisition unit 12 obtains a purchase matrix from the purchase data a, for example as disclosed in Patent Document 1, using multiple buyer IDs and multiple product IDs as row and column indices, respectively, and using non-negative values ​​calculated based on performance information a4 as element values ​​(for example, non-negative values ​​indicating whether a purchase was made, price, or number of purchases). By matrix decomposing the purchase matrix, the acquisition unit 12 can calculate hidden state information a11 consisting of buyer hidden state information a12 and product hidden state information a13.

[0029] Buyer Hidden State Information a12 shows the relationship between multiple buyer IDs and hidden states related to purchases. Product Hidden State Information a13 shows the relationship between hidden states and multiple product IDs.

[0030] The purchase data a acquired by the acquisition unit 12 can be acquired by any method, such as receiving it from an external device (not shown) or reading it from a storage medium on which the purchase data a is stored.

[0031] The cluster generation unit 14 analyzes the purchase data a of the product groups that constitute the population from the purchase data a acquired by the acquisition unit 12, based on the population setting condition b1. Then, it generates multiple clusters d, which include the cluster to which the product of interest specified by the product of interest designation condition b2 belongs, and the other clusters.

[0032] The cluster generation unit 14 further obtains purchase characteristic information a1 for each product from the purchase data a, and classifies each product into one of several clusters d such that products with similar purchase characteristic information a1 belong to the same cluster d.

[0033] The label name generation unit 16 generates the label name e for cluster d. Specifically, the label name generation unit 16 generates the label name e for cluster d based on one of the following: (1) first information, which is a list of all products belonging to cluster d; (2) second information, which is a list of some products belonging to cluster d; (3) third information, which is a list of all (or some) products belonging to cluster d and their product purchase characteristics; and (4) fourth information, which is product purchase characteristics common to all (or many) products belonging to cluster d.

[0034] The label name generation unit 16 can generate label e using a generation AI or a large language model (LLM). In this case, the label name generation unit 16 can input any of the first to fourth pieces of information described above into the generation AI, and the output from the generation AI can be the label name e. The prompts for inputting any of the first to fourth pieces of information into the generation AI are described below.

[0035] Figure 3 shows an example of prompts for obtaining label names from the generating AI based on (1) the first piece of information, which is a list of all products belonging to cluster d.

[0036] In this case, after the question, "There are five groups of products that are purchased in a similar way (share common preferences). Please provide the names of these product groups so that we can understand the common preferences of the buyers," prompt p1 will be set to prompt the user to enter product name, manufacturer, JAN code, etc., for the entire product list within cluster d.

[0037] When such a prompt p1 is input to the generating AI, the generating AI outputs an answer q1 such as "Fruit Flavor Healthy Carbonated Drink". The label name generation unit 16 can use this answer q1, "Fruit Flavor Healthy Carbonated Drink", as the label name e for this cluster d.

[0038] Figure 4A shows an example of prompts for obtaining label names from the generating AI based on second information, which is a list of some products selected from a list of all products belonging to cluster d.

[0039] Figure 4B illustrates how to select some products from the list of all products belonging to cluster d.

[0040] Figure 4B is a data structure diagram showing an example of the relationship between product purchase characteristics and clusters.

[0041] The label name generation unit 16 refers to the data structure diagram shown in Figure 4B and selects only the products I that are strongly associated with the product purchase characteristics H that have a strong influence on the formation of the cluster d. Specifically, it takes the average of the product purchase characteristics H of the products within the cluster and extracts the top predetermined number of product purchase characteristics H. It then extracts (representative) products whose product purchase characteristics H are equal to or greater than a threshold (e.g., 0.7).

[0042] In Figure 4B, products I00001 to I00003 are classified into cluster 1, and products I00004 to I00006 are classified into cluster 2. The average of the product purchase characteristics H of the products within each cluster is taken, and for example, the top two product purchase characteristics H are extracted.

[0043] The top two product purchase characteristics H in cluster 1 are product purchase characteristics H5 and H6. The top two product purchase characteristics H in cluster 2 are product purchase characteristics H1 and H2. We then extract representative products (products strongly linked to purchase characteristics) whose product purchase characteristic H is 0.7 or higher.

[0044] In cluster 1, product I00001 has a product purchase characteristic H5 of 0.7 or higher, and product I00002 has a product purchase characteristic H6 of 0.7 or higher. Therefore, these two products are designated as representative products of cluster 1.

[0045] On the other hand, in cluster 2, product I00004 is a representative product of product purchase characteristic H1, and products I00004 and I00006 are representative products of product purchase characteristic H0.7 or higher. Therefore, these three products are considered representative products (since product I00004 is a representative product of both product purchase characteristic H1 and product purchase characteristic H2, in reality, products I00004 and I00006 are considered representative products).

[0046] By selecting products that are strongly linked to the cluster's representative product purchasing characteristic H (i.e., products with high preference), it becomes possible to select a subset of products that better represent the cluster's characteristics, thereby improving the accuracy of labeling generated by the AI ​​based on the second set of information.

[0047] In Figure 4A, after the question, "There are two groups of products that are purchased in a similar way (share common preferences). Please provide the names of these product groups so that we can understand the common preferences of the buyers," prompt p2 is set, prompting the user to enter product names, manufacturers, JAN codes, etc., for a selection of products (in this case, two) within the chosen cluster d, as described above.

[0048] When such a prompt p2 is input to the generating AI, the generating AI outputs an answer q2 such as "Fruit Flavor Healthy Drink". The label name generation unit 16 can use this answer q2, "Fruit Flavor Healthy Carbonated Drink," as the label name e for this cluster d.

[0049] Figure 5A shows an example of prompts for obtaining label names from the generated AI based on (3) a list of all products belonging to cluster d and a third piece of information which is product purchase characteristics.

[0050] The label name generation unit 16 uses as third information one or more label names of product purchase characteristics H that are strongly associated with cluster d and whose value of product purchase characteristic H is above a predetermined threshold, and a list of all or some of the products belonging to the label names and cluster d.

[0051] For example, for product purchase characteristic H, a label name for product purchase characteristic H is created by a generation AI and used. The label name for product purchase characteristic H is generated by providing the generation AI with a list of representative products whose product purchase characteristic value is equal to or greater than a threshold (e.g., 0.6).

[0052] Figure 5B is a data structure diagram showing an example of product purchase characteristics information.

[0053] In the case of product purchase characteristic H1 as illustrated in Figure 5B, products I0001 (product purchase characteristic = 0.9) and I0003 (purchase characteristic = 0.8), whose product purchase characteristic H values ​​are equal to or greater than a threshold (e.g., 0.6), are selected as representative products. By selecting representative products (products with strong preferences) that are strongly linked to product purchase characteristic H in this way, and generating a label name for product purchase characteristic H, it is possible to create a label name that better captures the characteristics of product purchase characteristic H.

[0054] In Figure 5A, it is assumed that this cluster is strongly linked to three product purchase characteristics H, and that the label names generated for each are "fan of xxx manufacturer," "fruit-based beverage," and "health-conscious." In this case, after the question, "There are five product groups that consumers buy in a similar way. Analysis has shown that these products are strongly linked to three purchase characteristics: 'fan of xxx manufacturer,' 'fruit-based beverage,' and 'health-conscious.' Please provide the names of these product groups so that the common preferences (purchase characteristics) can be identified," prompt p3 is set to prompt the user to enter the product name, manufacturer, JAN code, etc., for the entire product list within cluster d.

[0055] When such a prompt p3 is input to the generating AI, the generating AI outputs an answer q3 such as "A fruit-flavored health drink favored by fans of xxx". The label name generation unit 16 can use this answer q3, "A fruit-flavored health drink favored by fans of xxx", as the label name e for this cluster d.

[0056] Figure 6 shows an example of a prompt for obtaining a label name from the generated AI based on the fourth piece of information, which is the product purchasing characteristic H common to all (or many) products belonging to cluster d.

[0057] In this case, prompt p4 would be a question like this: "There is a group of products that are strongly linked to (and serve as purchasing criteria for) three purchasing characteristics: 'fans of xxx manufacturer,' 'fruit-based beverages,' and 'health consciousness.' Please provide the name of this product group so that the purchasing criteria are clearer."

[0058] When such a prompt p4 is input to the generating AI, the generating AI outputs an answer q4 such as "A fruit-flavored health drink favored by fans of xxx". The label name generation unit 16 can use this answer q4, "A fruit-flavored health drink favored by fans of xxx", as the label name e for this cluster d.

[0059] The overview generation unit 18 uses the cluster d generated by the cluster generation unit 14 and the label name e generated by the label name generation unit 16 to generate an overview W, for example, as illustrated in Figure 7.

[0060] Figure 7 shows an example of an overhead view.

[0061] The overview diagram W is a diagram that represents cluster d, in which product c1 of interest is represented in a different color from the other products c2, c3, c4, ..., c6, while showing its relationship to cluster d (first relationship) and its relationship to the other products c2, c3, c4, ..., c6 in cluster d (second relationship).

[0062] The overview generation unit 18 places the label name e near the diagram representing cluster d.

[0063] The overview generation unit 18 also changes the size of the diagram representing cluster d according to the total sales of each product c1 to c6 belonging to cluster d. That is, cluster d with high total sales are displayed larger, and cluster d with low total sales are displayed smaller.

[0064] The overview generation unit 18 also changes the size of each product c1 to c6 figure represented in the figure representing cluster d according to their respective sales. That is, the size of product c with high sales is displayed larger, and the size of product c with low sales is displayed smaller. In this way, the overview W represents the first relationship described above.

[0065] Furthermore, the overview generation unit 18 arranges the figures of each product c1 to c6, which are represented in the figure representing cluster d, at a distance corresponding to the similarity between the corresponding products. When the similarity is high, the distance corresponding to the similarity is short and the products are placed close together, and when the similarity is low, the distance corresponding to the similarity is long and the products are placed far apart. In addition, the overview generation unit 18 displays the product of interest c1 and the other products c2, c3, c4, ..., c6 in different colors. In this way, the overview W represents the second relationship described above. For details on similarity, please refer to Patent Document 1.

[0066] The overview generation unit 18 makes the shape of the figure representing cluster d the same as the shape of each figure representing each product c1 to c6 within cluster d. In the example shown in Figure 7, both the shape of the figure representing cluster d and the shape of the figures representing products c1 to c6 are circles. Circles are preferred, but the unit is not limited to circles; ellipses, or polygons such as triangles and quadrilaterals can also be used. If polygons are used, equilateral polygons such as equilateral triangles and squares are preferred, but unequal polygons are also acceptable as long as the shape of the figure representing cluster d and the shape of each figure representing each product c1 to c6 are similar.

[0067] The overview generation unit 18 displays the outline color of the diagram representing cluster d, changing it according to the sales share of the featured product in cluster d.

[0068] Figure 8 shows an example of an overview diagram in which the outline color of the clusters is changed according to the sales share of the featured product within the cluster. Note that in Figure 8, the label name 'e' is omitted to avoid cluttering the diagram.

[0069] As shown in Figure 8, for example, if the average sales share of each product belonging to cluster d1 is 10%, and the sales share of product c11 belonging to cluster d1 is 20%, and is greater than or equal to the average of 10%, then the overview generation unit 18 will display the outline of the circle representing cluster d1 in thick red.

[0070] On the other hand, if the average sales share of each product belonging to cluster d2 is 30%, and the sales share of the featured product c22 belonging to cluster d2 is 15%, which is less than or equal to the average of 30%, then the overview generation unit 18 will display the outline of the circle representing cluster d2 in thick blue.

[0071] Furthermore, in the case of cluster d3, the average sales share of each product belonging to cluster d3 is 20%, and the sales share of the featured product c31 belonging to cluster d3 is also around 20%. In this way, when the sales share of the featured product c31 is about the same as the average sales share of cluster d3, the overview generation unit 18 displays the outline of the circle representing cluster d3 in a thin black line.

[0072] In the case of cluster d4, the average sales share of each product belonging to cluster d4 is 40%, and the sales share of the featured product c41, which belongs to cluster d4, is also around 40%. In this way, when the sales share of the featured product c41 is about the same as the average sales share of cluster d4, the overview generation unit 18 displays the outline of the circle representing cluster d4 in a thin black line.

[0073] In this way, by viewing the overview W, in which the outline of the circle representing cluster d is displayed in different colors depending on the comparison of the share of the featured product c with the average share, retail users can visually grasp the degree of influence (large, small, or average) of the featured product c within cluster d.

[0074] The analysis unit 20 analyzes the overview view W and outputs the analysis result f to the proposal unit 22.

[0075] The proposal unit 22 makes suggestions g related to the product of interest to the user based on the analysis result f. For example, it can output predetermined suggestions by pre-setting a correspondence table between the analysis result f meeting certain conditions and the suggested content. Although specific variations of suggestions will be described later, for example, if the analysis results show that other products are within a predetermined distance from a product of interest in a certain cluster, and the sales volume of those other products meets predetermined conditions, it can suggest placing those other products near the product being analyzed.

[0076] Alternatively, a generative AI method, such as a large-scale language model, may be used to make suggestions g to users related to the product of interest based on the analysis results f from the analysis unit 20.

[0077] The specific method of making a proposal is not limited, but for example, it could be done by outputting the proposal content as text or images from the display device 28, or by outputting it as audio.

[0078] If the user is a retailer, the proposal unit 22 makes a proposal g to the user, for example, as follows:

[0079] (g1) In the featured cluster, we propose selling products that have a high degree of similarity to the featured product and have a larger sales volume than the featured product.

[0080] (g2) In the case of (g1) above, each product belonging to the cluster of interest is further classified to belong to one of several subclusters according to a predetermined method such as k-means described in Non-Patent Document 1, and it is proposed to sell the products with the largest sales volume in each subcluster.

[0081] (g3) In a cluster of interest, if there are multiple interest products, and a certain number of these interest products have a high degree of similarity, we propose that the interest products with low sales figures not be sold, and that the interest products with low similarity to all of the other interest products and high sales figures be sold instead.

[0082] If the user is a producer, the proposal unit 22 makes the following proposal to the user:

[0083] (h1) In a cluster of interest, if the sales share of products produced by the producer is lower than a predetermined value, we propose developing a new product belonging to that cluster of interest.

[0084] (h2)The proposal for the development of new products described in (h1) above further proposes developing new products that have characteristics similar to those of products that have a sales share higher than a predetermined value in the target cluster.

[0085] (h3)The proposal for the development of the new product described in (h2) above further includes notification of the estimated sales volume of the new product, estimated from the overview based on its similarity to products that have a sales share higher than a predetermined value.

[0086] (h4) If a cluster of interest contains multiple products produced by a producer, and the total sales share of these multiple products in the cluster of interest is higher than a predetermined value, the proposal includes a notification that the supply of products in the cluster of interest is sufficient.

[0087] The graph display unit 24 displays a graph, such as a bar graph, for each of the multiple clusters generated by the cluster generation unit 14, which clearly shows the sales breakdown for each producer in each cluster.

[0088] The graph display unit 24, for each cluster displayed in the graph, indicates the breakdown of sales by products produced by producers in a first color if the sales share of products produced by producers is above a predetermined high threshold, and indicates it in a second color different from the first color if it is below a predetermined low threshold.

[0089] The user can refer to the overview diagram W generated by the overview diagram generation unit 18, the proposal output from the proposal unit 22 according to the analysis results f from the analysis unit 20, the graphs displayed by the graph display unit 24, etc., change the setting conditions b, input them through the input reception unit 11, and regenerate the overview diagram W with the changed setting conditions b, and perform positioning analysis based on this overview diagram W.

[0090] The output control unit 26 controls the output of various data output by the information processing device 10. For example, the output control unit 26 controls the output of the overview diagram W generated by the overview diagram generation unit 18, the proposals g and h output from the proposal unit 22, and the graphs generated by the graph display unit 24. The output method by the output control unit 26 can be any method, but examples include displaying on a display device 28 such as a liquid crystal display, transmitting data to an external device (server, other information processing device, etc.), and outputting to a recording medium using an image forming device such as a printer.

[0091] Next, the hardware configuration of the information processing device 10 will be explained using Figure 1B.

[0092] Figure 1B is an explanatory diagram showing an example of the hardware configuration of the information processing device 10.

[0093] The information processing device 10 includes a control device such as a CPU 51, a storage device such as a ROM (Read Only Memory) 52 and RAM 53, a communication interface 54 for connecting to a network and communicating, and a bus 61 for connecting the various parts.

[0094] The program to be executed by the information processing device 10 is provided pre-loaded into a ROM 52 or the like.

[0095] The program executed by the information processing device 10 may be configured to be provided as a computer program product by recording it in an installable or executable file format onto a computer-readable recording medium such as a CD-ROM (Compact Disk Read Only Memory), a flexible disk (FD), a CD-R (Compact Disk Recordable), or a DVD (Digital Versatile Disk).

[0096] Furthermore, the information processing device 10 may be configured to store the program executed on a computer connected to a network such as the Internet and provide it by allowing download via the network. Alternatively, the information processing device 10 may be configured to provide or distribute the program via a network such as the Internet.

[0097] The program executed by the information processing device 10 can enable the computer to function as one of the parts of the information processing device 10 described above. This computer can read the program from a computer-readable storage medium into its main memory and execute it using the CPU 51.

[0098] Next, we will describe an example of the operation of the information processing device 10 with the above configuration.

[0099] As examples of operation, we will explain Operation Example 1, which assumes the user is a retailer, and Operation Example 2, which assumes the user is a producer.

[0100] (Example 1: When the user is a retailer) Figure 9 shows an example of screen transitions in the information processing device 10 when the user is a retailer.

[0101] When the information processing device 10 is started, the login button 100 is displayed on the display device 28.

[0102] The user, a retailer, enables login button 100. This can be done using a mouse and keyboard (not shown). In this example, the user is assumed to be an analyst at "XXXX Store".

[0103] When the user activates the login button 100, screen 101 is displayed on the display device 28. Screen 101 is for entering the product selection conditions b2 into the input reception unit 11. Following screen 101, the user selects [Retailer] as the analyst's attribute. Next, in "For stores: Store ID", the user enters the store ID of XXXX store. There is also an item called "For producers: Producer ID", but if the analyst's attribute is retailer, there is no need to enter anything here, and even if something is entered, it will be ignored.

[0104] Once the input of the featured product specification condition b2 is complete, screen 102 is then displayed on the display device 28. Screen 102 is a screen for inputting the population setting condition b1 into the input reception unit 11. Following screen 102, the user inputs the population setting condition b1, for example, "Period: 2022 / 1 / 1-2022 / 12 / 31, User: All users, Store: Supermarkets nationwide".

[0105] Once the user has completed entering the population setting condition b1, they activate the Start Overview Map Generation button 104. This causes the information processing device 10 to start a series of processes consisting of cluster generation, label name generation, and overview map generation.

[0106] Figure 10 is a flowchart showing an example of the operation of the information processing device 10 during positioning analysis when the user is a retailer.

[0107] As shown in Figure 10, four steps, S1 to S4, are carried out.

[0108] In step S1, the acquisition unit 12 acquires the purchase data a to be analyzed.

[0109] Figure 11 is a data structure diagram showing an example of purchase data a.

[0110] The purchase data a shown in Figure 11 includes the following data items: time, buyer ID, product ID, quantity, price, and store ID.

[0111] Figure 12 is a data structure diagram showing an example of a product ID.

[0112] As illustrated in Figure 12, the product ID is information that links the product name, product category, and JAN code.

[0113] Figure 13 is a data structure diagram showing an example of a buyer ID.

[0114] As illustrated in Figure 13, the buyer ID is information that links the buyer's gender and age group.

[0115] Figure 14 is a data structure diagram showing an example of a store ID.

[0116] As illustrated in Figure 14, the store ID is information that links the store name, location, and store category.

[0117] In the cluster generation unit 14, based on the population setting condition b1, the purchase data a to be analyzed is narrowed down from the purchase data a acquired by the acquisition unit 12 (S1).

[0118] Next, in step S2, product hidden status information a13 is calculated from the narrowed-down purchase data a (S2).

[0119] Figure 15 is a flowchart that shows the process in step S2 in detail.

[0120] When acquiring hidden product status information a13, the acquisition unit 12 generates a purchase matrix from the purchase data a to be analyzed, representing the relationship between "purchasers" and "all products to be analyzed, including the group of products of interest and other products" (S21). Next, the purchase matrix is ​​decomposed into a matrix, and as hidden status information a11, purchaser purchase characteristics information (purchaser hidden status information) a12 and product purchase characteristics information (product hidden status information) a13 are calculated (S22).

[0121] Figure 16 is a data structure diagram showing an example of a purchase queue.

[0122] As illustrated in Figure 16, the purchase matrix is ​​a two-dimensional matrix consisting of buyer IDs and product IDs.

[0123] Figure 17 is a data structure diagram showing an example of product purchase characteristics information.

[0124] As illustrated in Figure 17, the product purchase characteristic information a13 is a matrix consisting of hidden states (H1, H2, ...), a product ID, and two dimensions.

[0125] Next, as shown in the flowchart of Figure 10, in step S3, the cluster generation unit 14 generates clusters. Specifically, the cluster generation unit 14 analyzes the purchase data a of the product groups that constitute the population, based on the population setting conditions b1, from the purchase data a acquired by the acquisition unit 12.

[0126] Next, multiple clusters d are generated, including the cluster to which the featured product specified in the featured product designation condition b2 belongs, and other clusters.

[0127] Furthermore, purchase characteristic information a1 for each product is obtained from purchase data a, and each product is classified into one of several clusters d such that products with similar purchase characteristic information a1 belong to the same cluster d.

[0128] Figure 17 is used to explain how clusters are generated. By classifying the product purchase characteristics information of each product by similarity (or distance), groups of products with similar purchasing patterns (purchase axes and preferences) are classified into multiple clusters.

[0129] For example, product purchase characteristic information for each product ID is represented by a vector with element values ​​equal to the number of product purchase characteristics H (100 in Figure 17). Product IDs with high similarity between vectors are grouped into the same cluster. Note that similarity can be calculated using cosine similarity, which represents the similarity between two vectors in a multidimensional space, but is not limited to this method.

[0130] Furthermore, for dimensionality reduction, which is generally necessary to map product purchase characteristics H for each product ID, represented by a multidimensional vector, into a 2D or 3D space, UMAP (Uniform Manifold Approximation and Projection) or t-SNE (t-distributed Stochastic Neighbor Embedding) may be used, but are not limited to these methods.

[0131] Next, in step S4, the label name generation unit 16 displays each cluster d so that the product groups of interest can be identified, and assigns a label name e to each cluster, so that the overview view W is displayed as shown in screen 106 of Figure 9. The label name e can also be generated by prompting the generation AI.

[0132] In each cluster d shown in the overview W illustrated on screen 106, the featured product c is a product located in the XXXX store, which has its store ID set on screen 101, while the other products are products located in supermarkets nationwide, which are the stores set on screen 102.

[0133] In the overview diagram W, the label name e is placed near the diagram representing cluster d.

[0134] To explain using Figure 7 again, in the overview W, within the diagram representing cluster d, product c1 of interest is represented in a way that distinguishes it from the other products c2, c3, c4, ..., c6 by a different color, while showing its relationship to cluster d (first relationship) and its relationship to the other products c2, c3, c4, ..., c6 within cluster d (second relationship).

[0135] The size of the diagram representing cluster d corresponds to the total sales of each product c1-c6 belonging to cluster d. Furthermore, the size of each diagram representing each product c1-c6 within the diagram representing cluster d corresponds to their respective sales. Thus, the overview diagram W can represent the first relationship described above.

[0136] Furthermore, each product c1 to c6, represented within the diagram of cluster d, is positioned at a distance corresponding to the similarity between the corresponding products. Products with high similarity are placed close together, while products with low similarity are placed far apart. In addition, the product of interest c1 is displayed in a different color from the other products c2, c3, c4, ..., c6. This allows the overview diagram W to represent the second relationship mentioned above.

[0137] The outline color of the diagram representing cluster d is displayed according to the sales share of the featured product c1 in cluster d. For example, as shown in Figure 8, if the average sales share of each product belonging to cluster d1 in each supermarket nationwide is 10%, and the sales share of the featured product c11 belonging to cluster d1 in that supermarket is 20% or more than the national average of 10%, then the outline of the circle representing cluster d will be displayed in thick red. On the other hand, if the average sales share of each product belonging to cluster d2 is 30%, and the sales share of the featured product c22 belonging to cluster d2 is 15% or less than the average of 30%, then the outline of the circle representing cluster d2 will be displayed in thick blue to distinguish them.

[0138] Furthermore, by displaying the outlines of the circles representing the other clusters d3 and d4 in black, retailers can visually grasp the degree of influence (large, small, or average) of the featured product c within each cluster d.

[0139] For example, a store with a sales share that is large compared to the national average for supermarkets can be considered a strong cluster, while a store with a small sales share can be considered a weak cluster.

[0140] Furthermore, retailers can significantly streamline their analysis by comparing the sales ratio composition of each cluster in other stores with that of their own store, allowing them to identify clusters that deserve attention for sales improvement from among a large number of clusters. However, this is not the only method for identifying clusters of interest.

[0141] Furthermore, as shown in Figure 9, screen 106 also displays a list of label names 108 assigned to each cluster d shown in the overview W. This allows users to easily understand the link between purchasing characteristics (preferences) and products.

[0142] On screen 106, below the list of label names 108, a "Return to Population Settings" button 110 and a "Start Analysis with Created Overview Map" button 112 are also displayed.

[0143] If the created overview diagram W is not what the user wants to use for position analysis, the user can enable the "Return to Population Settings button" 110 to return to screen 101 and change the featured product specification condition b2, or then proceed to screen 102 to change the population setting condition b1 and recreate the overview diagram W with the changed conditions.

[0144] On the other hand, if the user wants to perform a position analysis on the created overview map W, the user activates the "Start Analysis with Created Overview Map" button 112. This causes the analysis unit 20 to analyze the overview map W and display the analysis results as shown in the example on screen 114. In addition to the overview map W, screen 114 also displays the retail input field 116, the analysis menu 118, and the mode switching button 120.

[0145] The retail input field 116 displays the store name of the retailer, corresponding to the entered analyst's attributes (i.e., the set store ID).

[0146] The analysis menu 118 displays a list of various suggestions g made by the suggestion unit 22 to the user based on the analysis results f (for example, "Recommended products to include" 118a, "Recommended products to remove" 118b, etc.).

[0147] The mode switching button 120 is a button used by retailers to switch modes for analysis from a producer's perspective. When the mode switching button 120 is activated, a display switching request is sent to the output control unit 26, and in response to this request, the output control unit 26 switches the screen displayed on the display device 28 from screen 114 to screen 122. Screen 122 will be explained in Operation Example 2 below.

[0148] Next, we will explain the analysis of the overhead view W in Operation Example 1 and several variations of proposal g based on the analysis result f.

[0149] As the first variation, we will explain how to use the overview diagram W to understand the relationship between the "product group of interest" and the "total product group." This makes it possible to understand, for example, the relationship between "products available at the store under analysis" and "other product groups: products not available at our store but available at supermarkets nationwide."

[0150] Figure 18 is a diagram illustrating the first variation, showing (a) a comparative example with a single store, (b) a comparative example with multiple stores, and (c) a comparative example with only multiple stores of the same chain.

[0151] As shown in Figure 18(a), a comparison of single stores, it becomes possible to grasp the sales of a particular product sold at supermarket α in a broad or local context by representing the stores selling that product within the cluster of supermarkets nationwide, or by narrowing the scope to, for example, supermarkets in Tokyo, or even further narrowing the scope to, for example, supermarkets in City T.

[0152] As shown in Figure 18(b), a comparison across multiple stores, it becomes possible to grasp the sales performance of a particular product in a broad or local context by representing the multiple stores selling that product at Supermarket α within the cluster of supermarkets nationwide, or by narrowing the scope to, for example, supermarkets within Tokyo, or even further narrowing the scope to, for example, supermarkets within City T.

[0153] In this case, within a cluster, by distinguishing between products sold only at one store (e.g., XXXX Store xx branch), products sold only at another store (e.g., XXXX Store yy branch), and products sold at both stores using colors or other means, it becomes possible to understand which products are selling well in which regions.

[0154] Figure 18(c) shows a comparison of only multiple stores within the same chain. For example, by representing another store selling the same product as one store (e.g., XXXX Store xx) within a cluster of stores in the same chain, it becomes possible to understand the characteristics of each store in a chain store.

[0155] In the second variation, the cluster size in the overview view W is switched and displayed according to the conditions.

[0156] Figure 19 is a diagram illustrating the second variation.

[0157] The overview map Wa shown in Figure 19(a) was created using one year's worth of purchasing data from supermarkets nationwide. The size of the circle for each product within cluster d is initially set to be proportional to the total annual sales of supermarkets nationwide, which constitute the population. In this store, only product B is sold; products A and C are not. Therefore, in the overview map Wa, only product B is displayed in color, while products A and C are displayed in white, indicating that this store sells product B and not products A and C.

[0158] Figure 19(b) shows an overview view Wb obtained when the conditions are changed from the initial settings shown in Figure 19(a).

[0159] The specific conditions that were changed from the initial settings in Figure 19(a) are the area and the period.

[0160] The initial area was changed from a nationwide supermarket to a supermarket in City T. This was done to understand which products sell well in the surrounding area.

[0161] The initial setting for the timeframe was one year, but it was changed to the most recent month.

[0162] According to the overview diagram Wb obtained by changing the conditions in this way, for example, in City T, it can be seen that product C has been selling better than product A in the past month. Therefore, the user, a retailer, can suggest that product C should also be stocked in the store because it is closer to product B, which is currently being sold (corresponding to "Recommended Products to Stock" 118a in Figure 9).

[0163] Figure 19(c) is an overview view Wc in which the size of each cluster's circle is displayed in proportion to the total sales of the products within that cluster.

[0164] As described above, multiple overview views W obtained by changing the setting condition b are analyzed by the analysis unit 20, and based on the analysis results f, suggestions g regarding the featured product are made to the user, who is a retailer.

[0165] The specific analysis and proposals based on the analysis results f are explained below, using diagrams, as examples, divided into cases of strong focus clusters and weak focus clusters. A strong focus cluster is, for example, a cluster that is considered to be a strength of the store, where high sales are achieved when comparing the sales ratio composition of each cluster with that of other stores. However, there may be cases where too many similar products are placed in the same cluster, so it should be actively analyzed.

[0166] On the other hand, weak clusters are those that, for example, are not generating sufficient sales when compared to other stores in terms of sales ratio composition, and should be actively analyzed and strengthened.

[0167] Figures 20 and 21 illustrate the proposal made to a retailer user (in the case of a strong attention cluster).

[0168] Figure 20 shows an overview W containing a cluster d of interest labeled "rich-flavored instant noodles." Furthermore, within this cluster d, all products from A to F except product C are shown as filled in, indicating that they are available at the stores being analyzed. In other words, the stores being analyzed are focusing too much on this cluster d of interest.

[0169] This highly-watched cluster has too many cannibalizing products, but it is also a cluster that can be improved to more efficiently maintain or increase sales. In cases where a store has too many similar products belonging to cluster d, it is possible to reduce the number of similar products without causing a drop in sales and replace them with other products, thereby increasing sales.

[0170] Specifically, if there are a certain number of products with a certain level of similarity or higher, it is preferable to remove the products with low sales. In such cases, when the "Recommended Products to Remove" 118b in Figure 9 is enabled, the proposal unit 22 determines which products to recommend for removal based on the analysis results f from the analysis unit 20, and clearly communicates the recommendation by flashing the selected products on the overview W displayed on the screen 114, displaying them as text information, or notifying the user via voice.

[0171] The suggestion unit 22 can also suggest alternative products, such as products within the same cluster but located far apart and with high sales at other stores, or products from different clusters. When suggesting alternative products in this way, the suggested products can be displayed by flashing them on the overview diagram W shown on screen 114, displaying them as text information, or providing audio notifications.

[0172] For example, if the overhead view W shown in Figure 21 is displayed on screen 114, and the proposal is to remove product C and replace it with product D, the proposal can be clearly communicated by making product C in the overhead view W flash in a dark color and product D flash in a bright color.

[0173] On the other hand, Figures 22, 23, and 24 illustrate the proposal made to a retailer user (in the case of a weak attention cluster).

[0174] Figure 22 shows an overview W containing cluster d labeled "Specialty Beers and Non-Alcoholic Beverages." Overview W also shows that the store under analysis only stocks product B, which has a low sales volume, and that most of the products in this focus cluster d are not stocked at the store under analysis. In other words, the store under analysis not only lacks sufficient focus on this focus cluster d, but also fails to offer other products that would be suitable choices for buyers. There is room for improvement even in such weak focus clusters.

[0175] For such weakly focused clusters, for example, we propose stocking products that are already available in stores and have a certain level of sales, but that have a large sales volume in other stores and are likely to be purchased together with that product in those stores—in other words, products with high similarity and that are shown close together within cluster d.

[0176] For example, as shown in the overview W in Figure 23(a), if product B is available, products D and E, which are shown nearby within cluster d, are likely to be purchased together at other stores. Products with larger sales volumes can increase the average customer spending, so product D, which is within a certain distance within cluster d and has a large sales volume at other stores, is suggested as a candidate with the expectation that customers will also purchase it.

[0177] Furthermore, since product D is likely to be purchased together with product B, it can be suggested to place it on a shelf near product B. Based on this idea, the information processing device 10 can suggest product D as the first choice by highlighting product D when the mouse hovers over the portion of product B in the overview view W displayed on the display device 28, such as a display screen.

[0178] On the other hand, as shown in Figure 23(b), if product B is already available in a store but is not found within a certain distance, product A, which has the largest sales volume within the attention cluster d, is proposed as the first candidate.

[0179] On the other hand, as shown in Figure 24(a), if a product with a certain level of sales is not stocked in a store, as shown in Figure 24(b), the products within the target cluster d are re-clustered, for example, using k-means (see Non-Patent Literature 1), and it is proposed that products with high sales volume in other stores be stocked in each subcluster s1 and s2. In a case like Figure 24(b), product A in subcluster s1 and product D in subcluster s2 are proposed as such products.

[0180] Furthermore, in the case of such weakly attention-grabbing clusters, proposals will be clearly communicated in the same way as for strongly attention-grabbing clusters, by flashing on the overview diagram W, displaying them as text information, or by providing audio notifications.

[0181] As explained above, a retailer user can use the overview diagram W generated by the information processing device 10 to compare, for example, the positions of products that are stocked and those that are not stocked in the retail store being analyzed. This allows them to visualize the characteristics of the product assortment in each cluster, such as whether the product assortment is too large or too small.

[0182] Furthermore, based on these results, it becomes possible to identify clusters of interest for in-depth analysis and use the similarity between products within those clusters to diagnose the appropriateness of the product assortment and consider product replacements.

[0183] (Example 2: When the user is a producer) Figure 25 shows an example of screen transitions in the information processing device 10 when the user is a producer.

[0184] The screen transition diagram shown in Figure 25 is similar to the screen transition diagram shown in Figure 9. Therefore, we will briefly explain the similarities and focus on explaining the differences between Figure 25 and Figure 9.

[0185] The user, a retailer, enables login button 100. In this example, the user is assumed to be an analyst for "YYYY Beer".

[0186] When the user activates the login button 100, screen 101 is displayed on the display device 28. Screen 101 is for entering the product selection conditions b2 into the input reception unit 11. Following screen 101, the user selects [Producer] as the attribute of the analyst. Next, ignoring "Store ID (if store)", the user enters the producer ID of "YYYY Beer" in "Producer ID (if producer)".

[0187] Once the input of the featured product designation condition b2 is complete, screen 102 is then displayed on the display device 28. Screen 102 is a screen for inputting the population setting condition b1 into the input reception unit 11.

[0188] After completing the input of the population setting condition b1, the information processing device 10 starts a series of processes consisting of cluster generation, label name generation, and overview diagram generation by specifying the overview diagram generation start button 104.

[0189] An example of the operation of the information processing device 10 when the user is a producer is shown in the flowchart in Figure 10.

[0190] In each cluster d shown in the overview W illustrated on screen 106, the product c of interest is the YYYY beer product whose producer ID was set on screen 101.

[0191] On screen 106, a list of label names 108 is displayed below the overview W, and below the list of label names 108, a "Return to Population Settings" button 110 and a "Start Analysis with Created Overview" button 112 are displayed.

[0192] If the created overview diagram W is not what the user wants to use for position analysis, the user can activate the "Return to Population Settings button" 110 to return to screen 101 and change the featured product specification condition b2, or then move to screen 102 to change the population setting condition b1 and recreate the overview diagram W with the changed conditions.

[0193] On the other hand, if the user wants to perform a position analysis on the created overview map W, the user activates the "Start Analysis with Created Overview Map" button 112. This causes the analysis unit 20 to analyze the overview map W and display the analysis results as shown in the example on screen 122. In addition to the overview map W, screen 122 also displays the producer name input field 124, the analysis menu 126, and the mode switching button 128.

[0194] The producer name input field 124 displays the producer name (for example, "YYYY Beer") corresponding to the entered analyst's attributes (i.e., the set producer ID).

[0195] The analysis menu 126 displays a list of various proposals h made to the user by the proposal unit 22 based on the analysis results (for example, "product analysis" 126a, "cannibalization analysis" 128b, etc.).

[0196] The mode switching button 128 is a button that allows producers to switch modes to analyze from a retailer's perspective. When the mode switching button 128 is activated, a display switching request is sent to the output control unit 26, and in response to this request, the output control unit 26 switches the displayed screen from screen 122 to screen 114. Screen 114 has already been explained in operation example 1, so we will avoid repeating the explanation.

[0197] Next, we will describe the analysis of the overhead view W in operation example 2 and several variations of proposal h based on the results.

[0198] As the first variation, we will explain how to propose products to be developed based on the analysis of the overhead view W.

[0199] Figure 26A shows an example of an overview view provided to the user, who is a producer.

[0200] For example, with respect to an overview view W as shown in Figure 26A(a), the analysis unit 20 can analyze that in the segment indicated by cluster d (a segment refers to a division when the market is subdivided by the clustering method of this method. This division is not a division based on the conventional classification by type of product, but a data-driven division based on the similarity of how products are purchased), only product A has been introduced to the market and has not gained any market share, but there is a product B from another manufacturer that is often purchased together with product A and is selling well, located near product A.

[0201] Based on these analysis results f, proposal 22 proposes h that a new product should be developed that can take the position of product B using product A as a foothold. Proposal h may also include, for example, recommendations for promotions that sell the new product together with product A, and shelf allocation plans that place the new product in a similar position to product B in stores, given the potential to capture market share from product B.

[0202] The content of such proposals h can also be clearly communicated to the user by outputting them as text or images from the display device 28, or as audio.

[0203] It should be noted that product B from another manufacturer may be a private brand product. Private brand products are products developed or manufactured by retailers themselves. Private brand products can be displayed in a way that does not reveal which specific retailer developed them, and in that case, the specific name of product B cannot be identified by the user, but it is clear that there is a large market for product B.

[0204] Private label products may be displayed in a different color to distinguish them from products manufactured by so-called food manufacturers or other manufacturers (referred to as national brand products, etc.). In this case, the overview diagram W is a diagram that represents cluster d, in which product c1 of interest is shown in a different color from the other products c2, c3, c4, ..., c6 in cluster d, and the other products c2, c3, c4, ..., c6 are further distinguished by showing them in two different colors depending on whether they are private label products or other products.

[0205] Figure 26B will explain how to determine whether or not each cluster has secured a market share in the segment it represents.

[0206] Figure 26B is an overview diagram illustrating the method for determining whether or not each cluster has a market share in the segment it represents.

[0207] Whether or not a market share is secured in the segment represented by each cluster d is determined based on whether the market share of the featured product (the product group of the manufacturer being analyzed) in each cluster d is above or below a predetermined threshold.

[0208] In the overview diagram W illustrated in Figure 26B, in cluster d1, the sales of product c11 of the manufacturer under analysis exceed 40% of the total sales of all products within cluster d1. In other words, its market share exceeds 40%. If the threshold for determining whether or not a market share is achieved is set to, for example, 30%, then cluster d1 is judged to be a noteworthy cluster with a high market share.

[0209] On the other hand, sales of product c21 from the analyzed manufacturer within cluster d2 do not even reach 20% of the total sales of all products within cluster d2. In other words, the market share is below the threshold, so cluster d2 is determined to be a cluster of interest with a low market share.

[0210] The sales share of products for each cluster d is displayed graphically by the graph display unit 24.

[0211] Figure 27 is a bar graph created by the graph display unit 24, showing the sales share of the products.

[0212] As shown in Figure 27, the graph display unit 24 displays the sales share of products produced by each producer for each cluster on a bar graph. In this case, the graph display unit 24 can display the sales share of products produced by each producer in each cluster in a first color (e.g., red) if it is above a predetermined high threshold, and in a second color (e.g., blue) different from the first color if it is below a predetermined low threshold.

[0213] On the other hand, with respect to the overview view W shown in Figure 26A(b), the analysis unit 20 outputs analysis result f, which indicates that while products A and B introduced into this segment have secured a sufficient market share, products A and B are competing with each other (cannibalizing each other).

[0214] Figure 28 illustrates a method for determining cannibalism.

[0215] As shown in Figure 28(a), input two products (Product A and Product B) that you want to check for cannibalization. If these two products have different purchasing characteristics, they will appear in separate segments (the segment labeled #1 and the segment labeled #2), as shown in Figure 28(b), which generally indicates that they are not cannibalizing each other.

[0216] In such a case, for example, a manufacturer can use this as leverage to persuade a retailer to stock both product A and product B. In response, the retailer can increase sales in a different segment by stocking both product A and product B in their store.

[0217] For example, if you have two products whose contents (ingredients, etc.) are the same (or nearly the same), but whose only differences are the shape of the container, packaging design, and volume, then they would fall into the same category according to traditional classifications based on product type (e.g., classifications by genre such as beer, wine, and whiskey). However, this method is useful when you want to see if there are differences in classification based on actual purchases and data-driven classifications (differences in how they are purchased).

[0218] If two products belong to different clusters, it means they have different purchasing characteristics, indicating that they are being purchased based on different purchasing criteria. By examining the purchasing characteristics associated with each product, it's possible to identify which purchasing criteria are being used (and which criteria are different).

[0219] In other words, the analysis unit 20 determines whether two products that are classified the same by product type are competing products based on whether or not they belong to the same cluster. If they belong to the same cluster and are determined to be competing products, the analysis result f outputs the purchasing characteristics of each of the two products, and can show the differences in the purchasing axes of the two products.

[0220] Other proposals h output from the proposal unit 22 include support for new product development, product analysis, and product promotion.

[0221] These will be explained using Figures 29 and 30.

[0222] Figure 29 shows an example of an overview diagram used to support new product development.

[0223] Figure 29(a) shows an overview W of a weak cluster d in which no products have been introduced. This cluster contains product B from another manufacturer, which sells well (has a large market). In such a case, the proposal unit 22 conveys to the user, through text, images, or audio, a suggestion h that the user should develop a product that can take this position (compete). Furthermore, the analysis unit 20 analyzes the product purchase characteristics information 200 of product B, and the proposal unit 22 conveys a suggestion h based on the analysis results to the user, through text, images, or audio.

[0224] Figure 29(b) shows an example of product purchase characteristic information a13 for product B (product ID is I0003). The analysis unit 20 analyzes what kind of product product B is (what purchasing axes it is selling based on, or what preferences it is selling based on) by displaying the purchase characteristics associated with product B shown in Figure 29(b) that have large values ​​(H1 and H30, which indicate strong associations), and also displays the label names assigned by the generating AI (for example, H1: high protein, H30: low calorie).

[0225] The analysis unit 20 then analyzes that the first purchasing criterion for purchasing product B is high protein, and the second purchasing criterion is low calorie. Based on this analysis result f, the proposal unit 22 outputs a proposal h stating that producers should develop such products, and that it would be effective to display health-related words on the product packaging (this would effectively appeal to consumers who purchase product B based on those purchasing criteria, and increase the likelihood of capturing market share currently held by product B).

[0226] Figure 30 illustrates the movement of product clusters in a simulation of the promotion and improvement of existing products.

[0227] The overhead view W shown in Figure 30(a) illustrates the first cluster d1 labeled as #1 and the second cluster d2 labeled as #2.

[0228] As shown in Figure 30(a), if you want to promote or improve product A, which is currently classified in the first cluster d1 (a cluster with a small sales volume), so that it is classified in the second cluster d2 (a cluster with a large sales volume), the overview generation unit 18 will simulate this by changing the product purchase characteristic H (for example, changing H2 from 0.1 to 0.8 to appeal to the preference for dietary fiber, H2). Then, as shown in Figure 30(b), the overview generation unit 18 will display an overview W in which product A is placed in the second cluster d2. This is intended for use, for example, when a product possesses certain characteristics but its correlation with purchase characteristics is low.

[0229] Specifically, if product B is rich in dietary fiber but this fact is not effectively promoted, resulting in a low H2 value of 0.1 (weak connection between product B and the purchase axis of dietary fiber), and then, hypothetically, this purchase axis is promoted and consumers begin to buy the product based on that axis, resulting in a high H2 value of 0.8 (stronger connection between product B and the purchase axis of dietary fiber), this shows the predicted position of product A within which cluster it would be placed.

[0230] In other words, to predict the effectiveness of a promotion that appeals to a product's purchasing characteristics, if the value of that purchasing characteristic is changed, a cluster corresponding to the changed purchasing characteristic and its position within that cluster will be displayed. The cluster corresponding to the changed purchasing characteristic and its position within that cluster may also be the position of the product that has the purchasing characteristic closest to the changed purchasing characteristic.

[0231] Thus, the information processing device 10 can simulate which purchasing axes to promote and how changing the association between those purchasing axes and products will result in them being classified into the clusters and positions desired by producers, enabling more efficient promotions. The information processing device 10 can also realize this method of predicting the effectiveness of promotions.

[0232] Similarly, when a user sets purchasing characteristics information for a new product under development before market launch, the cluster in which the product is expected to be placed after launch, and its position within that cluster, are displayed. This makes it possible to estimate the potential sales volume of product A from the sales volume of the second cluster d2.

[0233] As explained above, the user, as a producer, can use the overview diagram W generated by the information processing device 10 to compare their own products with those of other companies, for example, and visualize the characteristics of their own products in each cluster (for example, whether they have a market share or not).

[0234] Furthermore, based on these results, it becomes possible to identify clusters of interest for in-depth analysis and use the similarity between products within those clusters to diagnose product lineups and plan future product development.

[0235] According to the information processing device 10 to which the information processing method of this embodiment is applied, as described above, by utilizing purchase big data to extract product purchase characteristics that represent characteristic purchase axes (or preferences) that constitute the entire purchase, and by clustering products with similar product purchase characteristics, an overview of the entire product is created based on the classification of the buyer's needs axis, and using this overview, analysts can intuitively explore the product space and deepen their understanding of the products.

[0236] Furthermore, it becomes possible to propose new products to retailers, and to offer producers product analysis results, product promotion proposals, and new product development plans. It can even predict the effectiveness of promotions.

[0237] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These novel embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims of the invention and its equivalents.

[0238] The following is a list of embodiments corresponding to the claims described in the claims of this application at the time of filing, as well as embodiments that were not described.

[0239] [Aspect 1] An input receiving unit that receives input for population setting conditions that set the conditions for the group of products that make up the population, and focus product designation conditions that specify a product of interest within the said group of products, A cluster generation unit analyzes the purchase data of the product groups constituting the population based on the population setting conditions and generates multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. An overview diagram generation unit generates an overview diagram in which the product of interest is represented in a diagram representing the cluster of interest, while showing a first relationship to the cluster of interest and a second relationship to other products in the cluster of interest, so as to be distinguishable from other products. Equipped with, information processing device.

[0240] [Aspect 2] (Subordinate to Aspect 1) The information processing apparatus according to embodiment 1, wherein the aforementioned product of interest is one or more.

[0241] [Aspect 3] (Subordinate to Aspect 1) The information processing device according to Embodiment 1, wherein the aforementioned product designation condition specifies that the product in question is a product sold at a specific store or group of stores.

[0242] [Aspect 4] (Subordinate to Aspect 1) The information processing apparatus according to Embodiment 1, wherein the aforementioned product designation condition specifies a product produced by a specific producer or group of producers as the product of interest.

[0243] [Aspect 5] (Dependent to Aspect 3 or 4) The information processing apparatus according to embodiment 3 or 4, wherein the overhead view generation unit represents the product of interest in a diagram representing the cluster of interest using a different color from the other products.

[0244] [Aspect 6] (Subordinate to Aspect 1) The aforementioned purchase data includes purchase characteristics information that shows the purchase characteristics for each product, The information processing apparatus according to Embodiment 1, wherein the cluster generation unit obtains purchase characteristic information for each product from the purchase data and performs a classification process to classify each product into one of the plurality of clusters such that products with similar purchase characteristic information belong to the same cluster.

[0245] [Aspect 7] (Subordinate to Aspect 6) The information processing apparatus according to embodiment 6, wherein the cluster generation unit represents the purchase characteristic information for each product as a vector having element values ​​equal to the number of product purchase characteristics, and performs the classification process by grouping the products with high similarity between the vectors so that they belong to the same cluster.

[0246] [Aspect 8] (Subordinate to Aspect 6) The information processing device according to embodiment 6, wherein the analysis unit determines whether two products that are classified the same by type belong to the same cluster or not, and if the two products belong to the same cluster and are determined to be competing products, it outputs purchasing characteristic information of the two products as the analysis result.

[0247] [Aspect 9] (Subordinate to Aspect 1) The information processing apparatus according to Embodiment 1, wherein the overhead view generation unit represents the first relationship by making the size of the diagram representing the cluster of interest correspond to the total sales of each product belonging to the cluster of interest, and by making the size of each diagram of each product represented in the diagram representing the cluster of interest correspond to the sales of the corresponding product.

[0248] [Aspect 10] (Subordinate to Aspect 9) The information processing apparatus according to embodiment 9, wherein the overhead view generation unit represents the second relationship by arranging the figures of each product represented in the figure representing the cluster of interest at a distance corresponding to the similarity between the corresponding products.

[0249] [Aspect 11] (Subordinate to Aspect 10) The information processing apparatus according to embodiment 10, wherein when the similarity is high, the distance corresponding to the similarity is short, and when the similarity is low, the distance corresponding to the similarity is long.

[0250] [Aspect 12] (Subordinate to Aspect 9) The information processing apparatus according to embodiment 9, wherein the shape of the figure representing the cluster of interest and the shape of the figure representing the product represented in the figure are identical.

[0251] [Aspect 13] (Subordinate to Aspect 12) The information processing apparatus according to embodiment 12, wherein the shape of the figure representing the cluster of interest and the shape of the figure representing the product represented in the figure are circles.

[0252] [Aspect 14] (Subordinate to Aspect 9) The information processing apparatus according to embodiment 9, wherein the overhead view generation unit changes the color of the outline of the diagram representing the cluster of interest according to the sales share of the product of interest in the cluster of interest.

[0253] [Aspect 15] (Subordinate to Aspect 1) The system further comprises a label name generation unit that generates the label name of the aforementioned cluster of interest, The information processing apparatus according to Embodiment 1, wherein the overhead view generation unit places the label name in the vicinity of the diagram representing the cluster of interest in the overhead view.

[0254] [Aspect 16] (Subordinate to Aspect 1) The information processing apparatus according to embodiment 1, wherein the label name generation unit generates the label name based on any of the following: first information, which is a list of all products belonging to the cluster of interest; second information, which is a list of some products belonging to the cluster of interest; third information, which is a list of all or some products belonging to the cluster of interest and product purchase characteristics; and fourth information, which is the product purchase characteristics of all products belonging to the cluster of interest.

[0255] [Aspect 17] (Subordinate to Aspect 16) The information processing apparatus according to embodiment 16, wherein the label name generation unit generates the second information by taking the average of the product purchase characteristics of products within the cluster of interest, extracting the top predetermined number of product purchase characteristics, and selecting products whose extracted product purchase characteristics are equal to or greater than a predetermined threshold.

[0256] [Aspect 18] (Subordinate to Aspect 16) The information processing apparatus according to embodiment 16, wherein the label name generation unit comprises the label names of one or more product purchase characteristics strongly associated with the attention cluster, whose product purchase characteristic values ​​are greater than or equal to a predetermined threshold, and a list of all or some products belonging to the attention cluster, as the third information.

[0257] [Aspect 19] (Subordinate to Aspect 16) The information processing apparatus according to embodiment 16, wherein the label name generation unit inputs any of the first to fourth pieces of information to the generation AI, and the output obtained from the generation AI in response to the input is used as the label name.

[0258] [Aspect 20] (Subordinate to Aspect 1) An analysis unit that analyzes the aforementioned overhead view, The information processing apparatus according to aspect 1, further comprising: a proposing unit configured to output a proposal to a user associated with the product of interest based on an analysis result obtained by the analyzing unit.

[0259] [Aspect 21] (Dependent on Aspect 20) further comprising an output control unit that controls display of the bird's-eye view generated by the bird's-eye view generation unit, The information processing apparatus according to aspect 20, wherein the output control unit switches between displaying the bird's-eye view from the perspective of a retailer and displaying the bird's-eye view from the perspective of a producer that produces the product of interest.

[0260] [Aspect 22] (Dependent on Aspect 20) The information processing apparatus according to aspect 20, wherein the input receiving unit receives an input of at least one of the population setting condition and the target product specifying condition that has been changed in accordance with the proposal.

[0261] [Aspect 23] (Dependent on Aspect 20) The information processing apparatus according to aspect 20, wherein the user is a retailer that sells the product of interest.

[0262] [Aspect 24] (Dependent on Aspect 23) The information processing apparatus according to aspect 23, wherein the proposal includes a proposal to sell a product that has a high similarity to the product of interest in the cluster of interest and has a larger sales scale than the product of interest.

[0263] [Aspect 25] (Dependent on Aspect 23) The information processing apparatus according to aspect 23, wherein the proposal includes classifying each product belonging to the cluster of interest into one of a plurality of sub-clusters according to a predetermined method, and proposing to sell a product having a large sales scale in each sub-cluster.

[0264] [Aspect 26] (Dependent on Aspect 25) The information processing apparatus according to aspect 25, wherein the predetermined method is k-means.

[0265] [Aspect 27] (Dependent on Aspect 23) The proposal is an information processing device according to embodiment 23, wherein, in the attention cluster, there are multiple attention products, and among the multiple attention products, there are a certain number of attention products with high similarity, the proposal is to not sell the attention products with low sales, and instead sell the attention products that have low similarity to all of the multiple attention products and have high sales.

[0266] [Aspect 28] (Subordinate to Aspect 20) The information processing apparatus according to embodiment 20, wherein the user is a producer that produces the product of interest.

[0267] [Aspect 29] (Subordinate to Aspect 28) The proposal is an information processing device according to embodiment 28, which includes a proposal to develop a new product belonging to the cluster of interest if the sales share of the product produced by the producer of the product of interest in the cluster of interest is lower than a predetermined value.

[0268] [Aspect 30] (Subordinate to Aspect 29) The information processing apparatus according to embodiment 29, wherein the proposal for the development of the new product includes a proposal for the development of a new product having characteristics similar to those of a product having a sales share higher than a predetermined value in the cluster of interest.

[0269] [Aspect 31] (Subordinate to Aspect 30) The information processing apparatus according to embodiment 30, wherein the proposal for the development of the new product includes estimating the sales scale of the new product, which is estimated from the overview diagram based on its similarity to a product that has a sales share higher than the predetermined value.

[0270] [Aspect 32] (Subordinate to Aspect 28) The information processing device according to embodiment 28, wherein if the cluster of interest contains multiple products produced by producers who produce the product of interest, and the total sales share of the multiple products in the cluster of interest is higher than a predetermined value, the proposal includes notification that the input of products in the cluster of interest is sufficient.

[0271] [Aspect 33] (Subordinate to Aspect 28) The information processing apparatus according to embodiment 28, further comprising a graph display unit that displays a graph showing the sales breakdown for each producer in each cluster for each of the multiple clusters generated by the cluster generation unit.

[0272] [Aspect 34] (Subordinate to Aspect 33) The information processing apparatus according to embodiment 33, wherein the graph display unit, for each cluster displayed in the graph, if the sales share of the goods produced by the producer is above a predetermined high threshold, the sales breakdown of the goods produced by the producer is indicated in a first color, and if it is below a predetermined low threshold, it is indicated in a second color different from the first color.

[0273] [Aspect 35] An information processing method performed by an information processing device, The processor of the aforementioned information processing device The system accepts input for population setting conditions, which define the conditions for the product group that constitutes the population, and focus product designation conditions, which specify the focus products within the said product group that deserve attention. Based on the aforementioned population setting conditions, the purchase data of the product groups constituting the population is analyzed to generate multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. An information processing method that generates an overview diagram in which the product of interest is represented in a diagram representing the cluster of interest, while expressing a first relationship to the cluster of interest and a second relationship to other products in the cluster of interest, so as to be distinguishable from other products.

[0274] [Aspect 36] A function that accepts input for population setting conditions, which define the conditions of the product group that constitutes the population, and focus product designation conditions, which specify the focus product that deserves attention within the said product group. Based on the aforementioned population setting conditions, a function is provided to analyze the purchase data of the product groups constituting the aforementioned population and generate multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. a function for generating an overhead view in which the target product is represented distinguishably from other products while expressing a first relationship of the target product with respect to the target cluster and a second relationship between the target product and the other products in the target cluster, within a diagram representing the target cluster A program for causing a processor to implement the function.

[0275] [Aspect 37] (dependent on Aspect 6) A promotion effect prediction method, which uses the information processing apparatus according to Aspect 6, changes a value of a purchasing characteristic of a target product, and predicts an effect when a promotion that appeals to the purchasing characteristic of the target product is performed based on a cluster displayed according to the purchasing characteristic after the change and a position of the target product within the cluster. [Description of Reference Numerals]

[0276] 10 Information processing apparatus 11 Input receiving unit 12 Acquisition unit 14 Cluster generation unit 16 Label name generation unit 18 Overhead view generation unit 20 Analysis unit 22 Proposal unit 24 Graph display unit 26 Output control unit 28 Display device 51 CPU 52 ROM 53 RAM 54 Communication interface 61 Bus 100 Login button 101 Screen 102 Screen 104 Overhead view generation start button 106 Screen 108 Label name list 114 Screen 116 Retail input field 118 Analysis menu 120 Mode switching button 122 Screen 124 Producer name input field 126 Analysis Menu 128 Mode switching button 200 Product purchasing characteristics information a. Purchase data A product B product C product D product E-product F product H Product purchasing characteristics I product W Overhead view α Super a. Purchase data a1 Purchasing characteristics information a2 Purchaser identification information (purchaser ID) a3 Product identification information (product ID) a4 Performance Information a11 Hidden Status Information a12 Buyer purchasing characteristics information (buyer hidden status information) a13 Product Purchase Characteristics Information (Product Hidden Status Information) b Setting conditions b1 Population setting conditions b2 Conditions for specifying featured products d cluster e. Label name f Analysis results g suggestion h suggestion p1~p4 prompts q1~q4 Answer s1, s2 subclusters

Claims

1. An input receiving unit that receives input for population setting conditions that set the conditions for the group of products that make up the population, and focus product designation conditions that specify a product of interest within the said group of products, A cluster generation unit analyzes the purchase data of the product groups constituting the population based on the population setting conditions and generates multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. An overview diagram generation unit generates an overview diagram in which the product of interest is represented in a diagram representing the cluster of interest, while showing a first relationship to the cluster of interest and a second relationship to other products in the cluster of interest, so as to be distinguishable from other products. Equipped with, information processing device.

2. The information processing apparatus according to claim 1, wherein the aforementioned product designation condition specifies a product sold at a specific store or group of stores as the product of interest.

3. The information processing apparatus according to claim 1, wherein the aforementioned product designation condition specifies a product produced by a specific producer or group of producers as the product of interest.

4. The information processing apparatus according to claim 2 or 3, wherein the overhead view generation unit represents the product of interest in a diagram representing the cluster of interest using a different color from the other products.

5. The aforementioned purchase data includes purchase characteristics information that shows the purchase characteristics for each product, The information processing apparatus according to claim 1, wherein the cluster generation unit obtains purchase characteristic information for each product from the purchase data and performs a classification process to classify each product into one of the plurality of clusters such that products with similar purchase characteristic information belong to the same cluster.

6. The information processing apparatus according to claim 5, wherein the cluster generation unit represents the purchase characteristic information for each product as a vector having element values ​​equal to the number of product purchase characteristics, and performs the classification process by grouping the products with high similarity between the vectors so that they belong to the same cluster.

7. The system further includes an analysis unit that analyzes the aforementioned overhead view, The information processing device according to claim 5, wherein the analysis unit determines whether two products that are classified the same by type are competing products based on whether they belong to the same cluster, and if the two products belong to the same cluster and are determined to be competing products, it outputs purchasing characteristic information of the two products as an analysis result.

8. The information processing apparatus according to claim 1, wherein the overhead view generation unit represents the first relationship by making the size of the diagram representing the cluster of interest correspond to the total sales of each product belonging to the cluster of interest, and by making the size of each diagram of each product represented in the diagram representing the cluster of interest correspond to the sales of the corresponding product.

9. The information processing apparatus according to claim 8, wherein the overhead view generation unit represents the second relationship by arranging the figures of each product represented in the figure representing the cluster of interest at a distance corresponding to the similarity between the corresponding products.

10. The information processing apparatus according to claim 8, wherein the overhead view generation unit changes the color of the outline of the diagram representing the cluster of interest according to the sales share of the product of interest in the cluster of interest.

11. The system further comprises a label name generation unit that generates the label name of the aforementioned cluster of interest, The information processing apparatus according to claim 1, wherein the label name generation unit generates the label name based on any of the following: first information, which is a list of all products belonging to the cluster of interest; second information, which is a list of some products belonging to the cluster of interest; third information, which is a list of all or some products belonging to the cluster of interest and product purchase characteristics; and fourth information, which is the product purchase characteristics of all products belonging to the cluster of interest.

12. The information processing apparatus according to claim 11, wherein the label name generation unit generates the second information by taking the average of the product purchase characteristics of the products in the attention cluster, extracting the top predetermined number of product purchase characteristics, and selecting products whose extracted product purchase characteristics are equal to or greater than a predetermined threshold.

13. The information processing apparatus according to claim 11, wherein the label name generation unit includes, as the third information, one or more label names of product purchase characteristics that are strongly associated with the attention cluster and whose value of the product purchase characteristics is greater than or equal to a predetermined threshold, and a list of all or some products belonging to the label names and the attention cluster.

14. The information processing apparatus according to claim 11, wherein the label name generation unit inputs any of the first to fourth pieces of information to the generation AI, and the output obtained from the generation AI in response to the input is used as the label name.

15. An analysis unit that analyzes the aforementioned overhead view, The information processing apparatus according to claim 1, further comprising a proposal unit that outputs suggestions to users related to the product of interest based on the analysis results from the analysis unit.

16. The system further includes an output control unit that controls the display of the overhead view generated by the overhead view generation unit, The information processing apparatus according to claim 11, wherein the output control unit switches between displaying an overview from the retailer's perspective and displaying an overview from the producer's perspective that produces the product of interest.

17. The information processing apparatus according to claim 15, wherein the proposal includes a proposal to sell products in the attention cluster that have a high degree of similarity to the attention product and have a larger sales volume than the attention product.

18. The information processing apparatus according to claim 15, wherein the proposal includes classifying each product belonging to the aforementioned cluster of interest to belong to one of a plurality of subclusters according to a predetermined method, and selling products with a large sales volume in each subcluster.

19. The information processing device according to claim 15, wherein, in the attention cluster, there are multiple attention products, and among the multiple attention products, there are a certain number of attention products with high similarity, the information processing device according to claim 15, wherein the attention products with low sales are not sold, and the attention products that have low similarity to all of the multiple attention products and have high sales are sold.

20. The information processing apparatus according to claim 15, wherein the proposal includes, when the sales share of products produced by producers of the product of interest in the product of interest is lower than a predetermined value in the product of interest cluster, a proposal to develop a new product belonging to the product of interest cluster.

21. The information processing apparatus according to claim 20, wherein the proposal for the development of the new product includes a proposal for the development of a new product having characteristics similar to those of a product having a sales share higher than a predetermined value in the cluster of interest.

22. The information processing apparatus according to claim 21, wherein the proposal for the development of the new product includes estimating the sales scale of the new product, which is estimated from the overview diagram based on the similarity to a product having a sales share higher than the predetermined value.

23. The information processing apparatus according to claim 15, wherein, if the cluster of interest contains multiple products produced by producers who produce the product of interest, and the total sales share of the multiple products in the cluster of interest is higher than a predetermined value, the proposal includes notification that the input of products in the cluster of interest is sufficient.

24. An information processing method performed by an information processing device, The processor of the aforementioned information processing device The system accepts input for population setting conditions, which define the conditions for the product group that constitutes the population, and focus product designation conditions, which specify the focus products within the said product group that deserve attention. Based on the aforementioned population setting conditions, the purchase data of the product groups constituting the population is analyzed to generate multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. An information processing method that generates an overview diagram in which the product of interest is represented in a diagram representing the cluster of interest, while expressing a first relationship to the cluster of interest and a second relationship to other products in the cluster of interest, so as to be distinguishable from other products.

25. A function that accepts input for population setting conditions, which define the conditions of the product group that constitutes the population, and focus product designation conditions, which specify the focus product that deserves attention within the said product group. Based on the aforementioned population setting conditions, a function is provided to analyze the purchase data of the product groups constituting the aforementioned population and generate multiple clusters, including the cluster to which the featured product specified in the featured product designation conditions belongs, and other clusters. The function generates an overview diagram within the diagram representing the cluster of interest that shows the product of interest in a way that allows it to be distinguished from other products, while representing its first relationship to the cluster of interest and its second relationship to other products in the cluster of interest. A program to be implemented by the processor.

26. A method for predicting the effectiveness of a promotion, which uses the information processing device described in claim 5 to change the purchase characteristics of a product of interest, and predicts the effectiveness of a promotion that appeals to the purchase characteristics of the product of interest based on a cluster displayed according to the changed purchase characteristics and the position of the product within that cluster.

Images