Automatic shelving allocation pattern creation method

Abstract

To prepare a shelf allocation pattern which makes the best use of the experience of a person in charge of shelf allocation and does not give an unnatural feeling by using a general information processor.SOLUTION: In this shelf allocation pattern automatic preparation method, the shelf information and merchandise information of an existing shelf allocation pattern and a shelf allocation pattern to be prepared (called preparation pattern) are inputted to an arithmetic means, the automatic preparation processing of the shelf allocation pattern is performed, and the preparation pattern is outputted. The input data to the shelf allocation pattern automatic generation processing includes information on a commodity group (referred to as a minimum granularity continuous commodity group) included in the same attribute of the lowest layer when commodities are classified into a hierarchical structure by attributes in the existing shelf allocation pattern and continuous in one stage of the shelf, and the attribute arrangement processing for arranging the commodities according to the hierarchical structure of the attributes of the commodities and the commodity selection and arrangement processing for selecting and arranging the commodities in the lowest layer attribute are performed. In the commodity selection and arrangement processing, commodities included in the minimum granularity consecutive commodity group are continuously arranged in one stage of the shelf.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an automatic shelf allocation pattern creation method and system. [Background technology]

[0002] Retail stores such as drugstores and supermarkets regularly carry out "planograms" to classify and organize the wide variety of products on display shelves to make them easier for shoppers to see and choose. The overall arrangement of products on display when the product display positions are determined by planograms is also called a planogram pattern, and diagrams of this pattern are often used.

[0003] Planograms are determined for each store to meet various business requirements regarding display position, product lineup, and number of faces (the number of identical products lined up horizontally), taking into account the number of shelves in each store, shelf size (number of shelves, shelf width, shelf height, etc.), consumer purchasing trends, etc. Business requirements include those that reflect the store's strategy, those based on the experience of planogrammers (for example, ensuring that a predetermined layout trend is met for each product function or brand), and rebate conditions between the retailer and product suppliers (conditions based on mutual agreements regarding planograms, for example, filling all shelves with products of a specific brand).

[0004] The success or failure of a shelf planogram has a large impact on sales. However, as mentioned above, shelf planograms must be created for each store. Therefore, creating shelf planograms places a heavy burden on the shelf planogram manager.

[0005] In response to this, efforts have been made to automate shelf allocation. For example, Japanese Patent Application Laid-Open No. 2016-164754 (Patent Document 1) describes that three shelf allocation patterns, namely, a maximum shelf allocation pattern for a maximum sales floor size, a minimum shelf allocation pattern for a minimum sales floor size, and a standard shelf allocation pattern for a medium-sized sales floor size, are used as reference shelf allocation patterns, and a new shelf allocation pattern is created based on the reference shelf allocation patterns, and that the width of a section (hereinafter also referred to as a zone) for displaying products based on product attributes is used as the display space width, and the display space width of the new shelf allocation pattern is calculated using a deviation rate.

[0006] Here, the deviation rate is a numerical value derived from the ratio of the display space width of the minimum shelf allocation pattern to the display space width of the maximum shelf allocation pattern, and the ratio of the display space width of the temporary shelf allocation pattern to the display space width of the maximum shelf allocation pattern, and the closer the display space width is to the maximum shelf allocation pattern, the smaller the value is.

[0007] In addition, Patent Publication No. 2021-121904 (Patent Document 2) describes automatically creating shelf layout map data using a learning model.

[0008] The data acquired by this learning model includes shelf information such as shelf size and number of shelves, number of product faces, product name, product size, sales quantity, sales record, etc.

[0009] In the learning model, in order to group and recognize adjacently placed products, the degree of association between adjacently placed products is represented by a group coefficient. [Prior art documents] [Patent documents]

[0010] [Patent Document 1] Japanese Patent Application Laid-Open No. 2016-164754 [Patent Document 2] Patent Publication No. 2021-121904 Summary of the Invention [Problem to be solved by the invention]

[0011] The method for creating a shelf planogram described in Patent Document 1 calculates the display space width for each zone of a new shelf planogram from the existing maximum and minimum shelf planograms. However, this method does not take into account the specific layout of the zones or the order in which products are arranged within the zones. As a result, if the number of shelves in the existing shelf planogram and the new shelf planogram differ, the layout of the zones may appear unnatural, or products that should be placed together may end up separated.

[0012] Furthermore, the method of creating shelf layout patterns described in Patent Document 2 uses shelf layout data as training data to have a learning model learn product placement trends, but if shelf layouts from many past periods are learned, or if shelf layouts from multiple retail stores are learned, it becomes difficult to reflect the most recent placement strategy unique to that retail store, and products that should be placed consecutively on one shelf level end up being separated from each other.

[0013] As such, conventional automatic shelf planogram pattern creation methods do not fully utilize the experience of shelf planogram staff, and the shelf planogram staff may find the product placement in the created shelf planogram pattern unnatural, which requires manual correction before applying it to a physical store.Furthermore, shelf planogram patterns to be applied to physical stores need to be created at a speed that is applicable to business operations, but preparing high-speed calculation means for this purpose is a cost burden.

[0014] Therefore, an object of the present invention is to make it possible to create a shelf allocation pattern that makes use of the experience of a shelf allocation clerk and that does not seem unnatural to the shelf allocation clerk, using conventional calculation means. [Means for solving the problem]

[0015] The inventors have discovered that in planogram patterns created by experienced planogrammers to satisfy various business requirements, product arrangement is consistent with the hierarchical structure of product attributes, and that for products in the lowest attribute of this hierarchical structure, there is always a group of products that are arranged consecutively on one level of a shelf in any planogram pattern created by an experienced planogrammer. Therefore, they have conceived the idea that such product arrangement can be realized in planogram patterns created automatically using a calculation means, and have completed the present invention.

[0016] That is, the present invention is a method for automatically creating a shelf allocation pattern, which inputs existing shelf allocation patterns and shelf information and product information of a shelf allocation pattern to be created (hereinafter also referred to as a created pattern) into a calculation means, performs automatic generation processing of the shelf allocation pattern in the calculation means, and outputs the created pattern, When products are classified into a hierarchical structure by attributes in an existing shelf allocation pattern, information about a group of products that are included in the same attribute in the lowest layer and are continuous on one shelf level (hereinafter referred to as a group of continuous products within the minimum granularity) is included in the input data for the automatic generation process of the shelf allocation pattern, In the process of automatically generating shelf allocation patterns, an attribute allocation process is performed to allocate product attributes so as to match the hierarchical structure of product attributes, and a product selection and allocation process is performed to select and allocate products within the attributes of the lowest layer. In the product selection and placement process, a method for automatically creating a shelf planogram pattern is provided that places products included in a group of consecutive products within the smallest granularity consecutively on one shelf.

[0017] The present invention also provides an automatic shelf allocation pattern creation system that includes a calculation means for inputting existing shelf allocation patterns and shelf information and product information for a created pattern, performing automatic generation processing of shelf allocation patterns, and outputting a created pattern. The automatic generation process of the shelf allocation pattern includes an attribute allocation process in which the layout of a group of consecutive products within the minimum granularity is used as input data, and the attributes of the products are arranged so as to match the hierarchical structure when the products are classified into a hierarchical structure by attributes; A product selection and placement process is performed to select and place products within the attributes of the lowest layer. To provide an automatic shelf allocation pattern creation system in which a product selection and placement process places products included in a group of continuous products within the smallest granularity continuously on one shelf.

[0018] In addition, the present invention provides a shelf allocation pattern generation program in which a calculation means receives as input data a plurality of existing shelf allocation patterns with different shelf sizes, their shelf information, and product information, as well as shelf information and product information of a creation pattern, which is a shelf allocation pattern to be created, and outputs the shelf allocation pattern of the creation pattern, and the program performs an attribute placement process that places product attributes so as to match the hierarchical structure when products are classified into a hierarchical structure by attribute, and a product selection and placement process that selects and places products within the attributes of the lowest layer, and in the product selection and placement process, places products included in a group of consecutive products within the smallest granularity consecutively on one level of a shelf.

[0019] Furthermore, the present invention provides an automatic extraction program for a group of consecutive products within a minimum granularity, which automatically extracts a group of consecutive products within a minimum granularity from a plurality of existing planogram patterns that have been subjected to attribute allocation processing for different shelf sizes, in a calculation means, comprising: a step of acquiring, as reference planogram data from the plurality of existing planogram patterns, a maximum planogram pattern having a maximum shelf size, a minimum planogram pattern having a minimum shelf size, and a planogram pattern having a shelf size similar to that of the pattern to be created; In each of the acquired shelf allocation patterns, the arrangement order of the products with the lowest attribute is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the lowest attribute is detected as a one-end result, and the product at the other end is detected as an other-end result. a process of marking the one-end results and the other-end results detected in each of the planogram patterns on products in the product arrangement of the lowest attribute of the largest planogram pattern, detecting sets of products in which a product having an other-end result mark and a product having a one-end result mark are lined up in the arrangement direction, and automatically extracting contiguous product groups within the minimum granularity by dividing the sets of products into segments of contiguous product groups within the minimum granularity; The present invention provides an automatic extraction program for a continuous product group within the smallest granularity. [Effects of the Invention]

[0020] The shelf planogram pattern created using the calculation means by the automatic shelf planogram pattern creation method of the present invention is similar to a shelf planogram created by an experienced shelf planner to satisfy various business requirements, in that the product arrangement conforms to the hierarchical structure of product attributes and a group of consecutive products within the minimum granularity is arranged consecutively on one shelf level. Therefore, according to the present invention, it is possible to create a shelf planogram using the calculation means that does not seem unnatural to the shelf planner (i.e., a shelf planogram that reflects the product arrangement created by an experienced shelf planner, does not feel strange to the experienced shelf planner, and is deemed suitable for actual use), significantly reducing the burden on the shelf planner and significantly improving work efficiency.

[0021] Furthermore, when the shelf allocation pattern generating program of the present invention is incorporated into a calculation means, the calculation means can be used in the system of the present invention. The system of the present invention can then realize the shelf allocation pattern automatic generating method of the present invention.

[0022] Furthermore, by incorporating the program for automatically extracting consecutive products within the minimum granularity of the present invention into the calculation means, consecutive products within the minimum granularity can be automatically extracted from an existing planogram pattern. Therefore, by incorporating this automatic extraction program into the calculation means together with the planogram pattern generation program of the present invention, it becomes possible to implement the automatic planogram pattern creation method of the present invention even if information regarding consecutive products within the minimum granularity has not been input into the calculation means in advance. [Brief explanation of the drawings]

[0023] [Figure 1] FIG. 1 is a schematic diagram of an automatic shelf planogram creation system according to an embodiment of the present invention. [Figure 2] FIG. 2 shows the overall flow of processing performed by the shelving planogram generation processing server of the automatic shelving planogram generation system of the embodiment. [Figure 3] FIG. 3 is an explanatory diagram of attribute placement. [Figure 4A] FIG. 4A is a flowchart of the program for automatically extracting continuous products within the minimum granularity. [Figure 4B]FIG. 4B is an explanatory diagram of a method for extracting a group of continuous products within the minimum granularity. [Figure 5] FIG. 5 is an explanatory diagram of the weighted average regarding the layout width of the attribute. [Figure 6] FIG. 6 is an explanatory diagram of a weighted average regarding the ratio of new products to existing products for an attribute of a production pattern when the attribute is composed of new products and existing products. [Figure 7A] FIG. 7A is a flowchart for determining the priorities of existing products and new products on a product-by-product basis. [Figure 7B] FIG. 7B is an explanatory diagram of a method for determining the priority of each product and a method for arranging essential selection products. [Figure 7C] FIG. 7C is a flowchart of a method for determining priority for each face. [Figure 8] FIG. 8 is an explanatory diagram of a method for calculating the coordinate position of an attribute. [Figure 9] FIG. 9 is an explanatory diagram of attribute correction. DETAILED DESCRIPTION OF THE INVENTION

[0024] The present invention will now be described in detail with reference to the drawings, in which the same reference numerals represent the same or equivalent components.

[0025] 1. Overview of the method for automatically creating shelf allocation patterns in the embodiment FIG. 1 is a schematic diagram of an automatic shelving planogram creation system 1 that implements an automatic shelving planogram creation method according to one embodiment of the present invention, and FIG. 2 shows the overall flow performed by a shelving planogram creation processing server that constitutes the calculation means 2 of this system.

[0026] In this specification, the term "computing means" refers to a computer device that processes information according to a rewritable program, regardless of its architecture. Specific forms include commonly available computer devices such as portable devices like handheld computers and laptop computers, and stationary devices like desktop computers and server devices.

[0027] In this automatic shelving planogram pattern creation method, roughly speaking, an existing shelving planogram pattern, as well as shelf information and product information for the pattern to be created, are input to the calculation means 2, the automatic shelving planogram pattern creation process is performed, and the created pattern is output. In the automatic shelving planogram pattern creation process, a shelving planogram creation program is used.

[0028] In this embodiment, the calculation means 2 performs the following preprocessing for the automatic generation of shelf allocation patterns: (i) Extraction process of continuous products within the minimum granularity (ii) Calculating the ideal placement width for each attribute when products are classified into a hierarchical structure by attribute (iii) Determining the priority order to be used in the product selection and placement process for placing products in the placement width of the lowest-level attribute (iv) Calculation of the coordinate position of the center of gravity of the attribute Do the following.

[0029] Here, the "attributes" of a product refer to properties related to placement that are assigned based on the product placement tendency and product features. The hierarchy of product attributes can be managed using codes (e.g., 1, 2, 3, ...). In this specification, the smaller the number, the larger the attribute granularity, and the attribute granularity listed in Table 1 is attribute 1 > attribute 2 > attribute 3 > attribute 4.

[0030] "Layout width of attribute" refers to the sum of the width occupied by all products belonging to a certain attribute when viewed from the front of the shelf when products are arranged on the shelf, for all levels of all shelves included in the shelf planogram.

[0031] "Product layout width" refers to the sum of the horizontal length occupied by a product when viewed from the front of the shelf when the product is arranged on the shelf, for all levels of all shelves included in the shelf planogram.

[0032] If it is discovered that the shelves that make up the shelf planogram contain shelves with different numbers of shelves, the calculations are carried out separately for the shelves with different numbers of shelves and the other shelves. This process is called problem division (Figure 2).

[0033] The automatic generation process of a planogram pattern using a planogram pattern generation program involves (v) attribute placement processing and (vi) product selection and placement processing. The attribute placement processing is a process of placing product attributes so that they match the hierarchical structure of product attributes, and the product selection and placement processing is a process of selecting and placing products within the lowest-level attributes. In the product selection and placement processing, for selected products, products included in the group of consecutive products within the smallest granularity are placed consecutively on one shelf level. The above-mentioned processes (i) to (vi) will be described in detail in "2. Details of the method for automatically creating shelf allocation patterns and the system thereof according to the embodiment" below.

[0034] The process of automatically generating shelf planograms can be performed by performing mathematical optimization processing in the calculation means 2, and an application (mathematical optimization solver) can be used for this purpose. Examples of mathematical optimization solvers include commonly available Gurobi Optimizer, IBM CPLEX, and Hexaly. The mathematical optimization solver may also be a web application.

[0035] The attribute allocation process and product selection and allocation process performed in the automatic shelf allocation pattern generation process can also be performed by implementing a rule-based algorithm in a general calculation means.

[0036] The present invention is primarily configured to use a computing means to perform attribute placement processing and to arrange products in a group of consecutive products within the smallest granularity for each attribute in the lowest layer consecutively on one shelf level in a product selection and placement processing. To this end, the input data for the automatic planogram generation processing includes information about the group of consecutive products within the smallest granularity. This information about the group of consecutive products within the smallest granularity can be extracted in preprocessing of the mathematical optimization processing, as described above. It is preferable that the information about the group of consecutive products within the smallest granularity at least includes a product code. Here, "product code" refers to a character string used to identify a product, and an example of this may be a JAN code. In other words, it is preferable that the information about the group of consecutive products within the smallest granularity includes at least a list of product codes included in each group of consecutive products within the smallest granularity. In addition to the product code, the information about the group of consecutive products within the smallest granularity may also include product names, etc., as needed.

[0037] By performing these processes, it becomes possible to automatically create, using a calculation means, a shelf allocation pattern that an experienced shelf allocation clerk will not find unnatural.

[0038] (Attribute placement processing) The attribute placement process is a process of placing product attributes in a manner consistent with the hierarchical structure, and as described above, product attributes refer to properties related to product placement that are assigned based on product placement trends and product features.

[0039] An example of a product placement trend is the tendency for "heavy duty liquid detergent" and "heavy duty powder detergent" to be located adjacent to each other at the granularity of attribute 2 shown in Table 1.

[0040] Similarly, "heavy duty liquid detergent" and "heavy duty powder detergent" in attribute 2 are included in "laundry detergent" in attribute 1. In this way, product attributes contain hierarchical information.

[0041] Product features include product functions and brand names.

[0042] Product functions and placement trends are closely related.

[0043] As an example of a hierarchical structure of product attributes, product classifications independently used by retailers or manufacturers can be used. For example, the hierarchical structure of laundry detergents shown in Table 1 can be cited.

[0044] When products are classified according to their functions, they generally take on a hierarchical structure, so depending on the type of product, the hierarchical structure of product attributes can be directly derived from the product classifications of retailers and manufacturers.

[0045] [Table 1]

[0046] The information on the hierarchical structure may be obtained manually, or may be automatically generated by the calculation means from referable shelf allocation data for the past or the current period by cluster analysis, etc. In this embodiment, the information on the hierarchical structure is input information to the calculation means 2 that performs automatic generation processing of the shelf allocation pattern, but it may also be automatically generated by the calculation means 2.

[0047] (Minimum particle size continuous product group) A group of consecutive products within the minimum granularity is a group of products (a collection of products) that are included in the same attribute in the lowest layer when products are classified into a hierarchical structure by attribute in an existing planogram pattern and that are arranged consecutively on the same shelf level. In other words, even if product A and product B are arranged consecutively on the same shelf level in an existing planogram pattern, and product A and product B are included in the same attribute in a layer higher than the lowest layer, if they are not included in the same attribute in the lowest layer, they do not constitute a group of consecutive products within the minimum granularity.

[0048] Contiguous product groups within the minimum granularity exist in all planogram patterns, and planogram patterns created by experienced planogrammers are created so that contiguous product groups within the minimum granularity are consecutive on the same shelf. However, it is not easy to manually find contiguous product groups within the minimum granularity from existing planogram patterns. However, contiguous product groups within the minimum granularity can be easily found by performing the extraction method described below using a computing means incorporating the automatic extraction program for contiguous product groups within the minimum granularity of the present invention.

[0049] Once the group of consecutive products within the minimum granularity has been extracted using the calculation means, if the group of consecutive products within the minimum granularity is recorded in the storage device of the calculation means for each attribute of the lowest layer, it will be possible to omit the process of extracting the group of consecutive products within the minimum granularity each time a shelf planogram is created.

[0050] Even if the number of layers is increased when classifying products into a hierarchical structure by attributes instead of extracting a group of consecutive products within the minimum granularity, it is not guaranteed that products in the lowest layer will be arranged consecutively on one shelf level after the number of layers is increased. Therefore, the group of consecutive products within the minimum granularity in the lowest layer of the hierarchical structure of product attributes will differ from the group of products in the lowest layer when the number of layers in the hierarchical structure is increased. Note that, depending on the product, products with the lowest layer attribute may always be arranged consecutively on the same shelf level, and products with that lowest layer attribute will be a group of consecutive products within the minimum granularity.

[0051] In the planogram pattern, the product included in the group of consecutive products within the minimum granularity may be a single product, as shown by product C in Figure 4B (described later). In the planogram pattern, when the group of consecutive products within the minimum granularity consists of one product and multiple units of that product are placed on a shelf, the multiple products are placed consecutively in one layer.

[0052] In the hierarchical structure shown in Table 1, the lowest-level attribute 4 is brand, and the group of products within a brand is listed as the group of continuous products within the smallest granularity. However, in the present invention, the lowest-level attribute to which the group of continuous products within the smallest granularity belongs is not limited to brand.

[0053] For example, if attribute 3 is specified as the brand name "Cucute" and attribute 4 is specified as the function names of the brand "Cucute" such as "liquid detergent" or "foam spray," when "foam spray-1" and "foam spray-2" exist within attribute 4 "foam spray" due to differences in content volume, price range, etc., then "foam spray-1" and "foam spray-2" can become a group of continuous products within the smallest granularity within attribute 4.

[0054] Furthermore, in the case of products such as cosmetics where brands are at the top level, for example, if the existing shelf planogram contains brands such as Aube, Coffret D'or, KATE, and Sofina, attribute 1, attribute 2, and attribute 3 can be determined as follows based on the placement trends of these brands, and attribute 4 at the bottom level can be determined by function such as "lip" or "cheek." In this case, lip products with similar colors can be cited as a product group that could become a continuous product group within the smallest granularity.

[0055] Attribute 1 (Aube & Coffret D'or & KATE & Sofina) > Attribute 2 (Aube & Coffret D'or) > Attribute 3 (Aube) > Attribute 4 (Aube lip, Aube cheek, ...)

[0056] Also, even if the brand is the same, if the functions are significantly different, they may be placed in different locations on the shelves. For example, even if they are the same Haiter brand, "Laundry Drum Haiter" is placed in the washing machine drum cleaner section, and "Haiter" is placed in the laundry bleach section. In this case, too, the Haiter brand is not the lowest attribute.

[0057] On the other hand, although washing machine mold killer and washing machine bleach are different brands, they are conveniently grouped within the same lowest attribute and can be a continuous product group within the smallest granularity.In this way, in addition to classification by brand name, various classifications can exist at the lowest attribute level based on function and other factors.

[0058] (Example of attribute placement processing) As shown in Figure 3, the attribute placement process in the automatic generation of shelf planograms adheres to the conditions described by the parameters in the mathematical optimization process, placing attributes in order of decreasing granularity so that the placement of lower-level attributes is included in the placement of higher-level attributes. For example, at the granularity of attribute 1, there are attributes (2) and (3), and at the granularity of attribute 2, there are attributes (4), (5), (6), (7), (8), and (9), and of these, attributes (5), (6), (7), (8), and (9) are included in attribute (3) of the granularity of attribute 1. Note that parameters are values ​​used in the objective functions and constraints in the mathematical optimization process.

[0059] (Effects of attribute placement and product selection placement treatments) In the planogram pattern of the present invention, the attribute placement process aligns the attribute placement of products with the hierarchical structure of product attributes, resulting in a placement of products grouped together according to product function or brand. This prevents separation of product function or brand placement, and makes it easier to adjust the placement of products by function or brand to a rectangle when the shelf is viewed from the front.

[0060] Furthermore, the product selection and placement process ensures that contiguous product groups within the minimum granularity are lined up in a single row on a single shelf. This means that contiguous product groups within the minimum granularity are not folded over and placed across multiple shelves, and the product order in the planogram created by an experienced planogram manager can be maintained.

[0061] Therefore, the calculation means presents a product layout that does not look unnatural to the skilled planner and makes it easy for consumers to select products, just like a layout created by a skilled planner that satisfies various business requirements.

[0062] Furthermore, if there are products that tend to be placed on specific shelves in terms of business practices, product weight, size, etc., the present invention performs attribute placement processing, making it possible to realize placement that reflects these tendencies. In the product selection and placement processing, by specifying the priority of product placement, it is possible to specify products that have low sales but should be included in the planogram pattern.

[0063] In the mathematical optimization process, rebate conditions can also be reflected in the shelf allocation pattern as constraints.

[0064] As described above, according to the present invention, it is possible to create a shelving planogram that satisfies various business requirements using a calculation means. Moreover, since the shelving planogram is created by performing a mathematical optimization process using the calculation means, it is possible to create the shelving planogram at a speed that is applicable to business operations using calculation means that are conventionally used for ordinary people (for example, laptop computers, etc.). In other words, the time required to create a shelving planogram can be significantly reduced to approximately 1 / 7 to 1 / 2 compared to manually creating a shelving planogram of a different size from an existing shelving planogram.

[0065] 2. Details of the method and system for automatically creating shelf allocation patterns according to the embodiment The process and system of the embodiment will be described in detail below. (Overall system configuration) 1 includes a shelving planogram pattern generation processing server 2 as a calculation means for automatically generating shelving planogram patterns. The shelving planogram pattern automatic generation system 1 also includes a staff terminal 3, a data storage server 4, and a request-response server 5.

[0066] In this embodiment, the case where the shelf allocation pattern generation processing server 2, the data storage server 4, and the request-response server 5 are separate servers will be described, but the functions of the shelf allocation pattern generation processing server 2, the data storage server 4, and the request-response server 5 may be realized by the same server, or the functions of the shelf allocation pattern generation processing server 2, the data storage server 4, and the request-response server 5 may be realized by two servers.

[0067] The shelf planogram generation processing server 2 uses the web-based Gurobi Optimizer as an application for performing mathematical optimization. The web-based application is located on a specific web site and can be used by accessing it via the web.

[0068] The shelf allocation pattern generation processing server 2 performs the mathematical optimization processing and its pre-processing (i) to (iv) in accordance with the flow shown in FIG.

[0069] The application that performs the mathematical optimization process is not limited to a web application, but may be installed in advance on the shelving planogram pattern generation processing server 2. Furthermore, the preprocessing may be performed by a calculation means separate from the shelving planogram pattern generation processing server 2 that performs the mathematical optimization process.

[0070] The input data (input information) shown below is entered at the staff member terminal 3, and the entered input data is read by the planogram pattern generation processing server 2 via the request-response server 5. Note that the input data does not have to be entered from the staff member terminal 3; instead, data stored on another server such as the data storage server 4 may be selected at the staff member terminal 3, and the selected data may be entered into the planogram pattern generation processing server 2. The staff member terminal 3 is also capable of immediately correcting the planogram pattern displayed on the display. This allows the shelf allocation staff member to check and correct the planogram pattern generated by the planogram pattern generation processing server 2 on the staff member terminal 3. Note that it is preferable that the planogram pattern can be corrected on-site at a store or the like using a correction function on the staff member terminal 3, but this correction function on the staff member terminal 3 is not essential to the present invention.

[0071] The data storage server 4 outputs the stored data to generate a shelf allocation pattern in response to instructions and commands from the request-response server 5. The data storage server 4 also stores past shelf allocation patterns generated by the shelf allocation pattern generation processing server 2 and shelf allocation patterns modified by the person in charge terminal 3, and can output these in response to instructions and commands from the request-response server 5.

[0072] The request-response server 5 transmits the data stored in the data storage server 4 to the shelf allocation pattern generation processing server 2 in response to an instruction from the person in charge terminal 3. The request-response server 5 also receives the shelf allocation pattern generated by the shelf allocation pattern generation processing server 2 and transmits it to the data storage server 4.

[0073] (input data) The automatic planogram pattern creation method implemented by the automatic planogram pattern creation system 1 shown in Figure 1 is a method that reflects the know-how of the planogram manager as much as possible. Therefore, it is assumed that there is an existing planogram pattern. If the existing planogram pattern includes a pattern from the previous period, it is preferable to use the pattern from the previous period as input data, and if there are planogram patterns for the previous and current periods, it is even more preferable to use both of them. Therefore, as mentioned above, in this system, it is possible to input an existing planogram pattern on the manager terminal 3, or to select an existing planogram pattern saved in the data storage server 4 on the manager terminal 3 and input it to the planogram pattern generation processing server 2.

[0074] The existing planogram patterns are of different sizes. These are used to create new planogram patterns of different sizes from the existing planogram patterns. Here, size refers to the total width of the shelves on which products can be placed.

[0075] The input data preferably includes shelf information and product information of the retailer's existing shelf allocation pattern related to the creation pattern, and also includes shelf information and product information of the creation pattern.

[0076] If the pattern to be created is the first planogram pattern for the current period for the retailer, then a planogram pattern for the previous period exists, but no existing planogram pattern for the current period exists. In that case, the shelf information and product information for only the planogram pattern for the previous period are used as input data. Conversely, if the retailer is creating new planograms this period, then an existing planogram pattern for the current period exists, but no existing planogram pattern for the previous period exists. In that case, the shelf information and product information for only the planogram pattern for the current period are used as input data. If both an existing planogram pattern for the previous period and an existing planogram pattern for the current period exist, it is preferable to use both of them.

[0077] Here, the current period refers to the period in which the planogram pattern currently being created is adopted in the planogram pattern update cycle, and the previous period refers to a predetermined period prior to the period in which the planogram pattern currently being created is adopted in the update cycle, in which another planogram pattern was adopted. For planograms that display many seasonal products, it is preferable to use the period of the same season one year ago as the previous period, and for planograms that display non-seasonal products, it is preferable to use the period immediately preceding the current period as the previous period.

[0078] The current period pattern is a shelf allocation pattern that will be adopted in the same period as the shelf allocation pattern that is currently being created, and the previous period pattern is a shelf allocation pattern that was adopted in a specified period prior to the period in which the shelf allocation pattern that is currently being created will be adopted.

[0079] Typically, in retail businesses that operate multiple stores under the same name, such as chain stores or franchise stores, the shelf allocation patterns for each store are managed by the headquarters of the retail business to which the store belongs. Therefore, the shelf allocation patterns for the current and previous periods that serve as input data can be used across the multiple stores that make up the retail business.

[0080] Examples of the previous period's planogram pattern include a previous period's maximum pattern, which is a planogram pattern with the largest shelf size (i.e., the total width of the shelves on which products can be placed), a previous period's minimum pattern, which is a planogram pattern with the smallest shelf size, and a previous period's planogram pattern of a size similar to the shelf size to be created this time, and preferably also a previous period's basic pattern, which is a planogram pattern of a size intermediate between the previous period's maximum pattern and the previous period's minimum pattern. Note that the previous period's maximum pattern, previous period's minimum pattern, a previous period's planogram pattern of a size similar to the shelf size to be created this time, and a planogram pattern of an intermediate size are preferably from the same period.

[0081] The existing planogram patterns for the current period may include the existing maximum and minimum patterns for the current period. In addition, the planogram patterns to be input preferably also include a basic pattern for the current period that is an intermediate size between the maximum pattern for the current period and the minimum pattern for the current period.

[0082] It is preferable to input the shelf planogram using PTS (Planogram Transfer Specifications), a CSV format that allows data sharing between shelf planogram software.

[0083] The shelf allocation pattern input in the PTS includes, for example, the unit information shown in Table 2, the shelf level information shown in Table 3, and the product placement information shown in Table 4.

[0084] [Table 2]

[0085] [Table 3]

[0086] [Table 4] (In the table, "Shelf position" indicates product placement information. Shelf position "1" means that the product is placed first from the left on that shelf. "Null" indicates that no information is available.)

[0087] On the other hand, in this embodiment, the input data includes information other than the PTS referenced in the planogram pattern generation process as additional information for each type of information. The additional information may be input in a spreadsheet format such as Excel from the person in charge terminal 3, or may be input from the data storage server 4.

[0088] The additional information includes the shelf allocation name of the created pattern, reference information of the shelf allocation pattern, product information, rebate conditions, and the like.

[0089] An example of the shelf layout name of the created pattern is, for example, Store A, 4 bottles of laundry detergent, 5 tiers.

[0090] The reference information for the shelf allocation pattern may include, for example, information on the shelf allocation pattern (such as the shelf allocation pattern for the same shelf in the same store in the previous period) that is referenced in the attribute allocation.

[0091] The product information is information about the products to be displayed on the shelves of the created pattern. The product information includes the product code, product name, compression rate, attributes, and new product priority order shown in Table 5B, which are shown in the product master in Table 5A.

[0092] [Table 5A]

[0093] [Table 5B]

[0094] Products in forms such as standing pouches can be placed in a state where they are compressed in the width direction (narrower than the original product width), and putting products in this state is also called compression. In Table 5A, the compression rate is a numerical value that indicates the percentage by which the width that a product occupies in the actual shelf planogram can be compressed from the original product width, and the actual product width in the shelf planogram can be calculated by multiplying the product width by (1 - compression rate).

[0095] Although the information on the group of consecutive products within the minimum granularity may be included in the additional information, in this embodiment it is not included in the additional information and is extracted in preprocessing by the planogram pattern generation processing server 2.

[0096] In Table 5B, the priority order of new products may be the same for all new products, for example, number 1. In this case, any new product that belongs to a certain attribute can be placed on the shelf with the same probability. For example, if there are three candidate products that fit within the width of two products, it is possible to randomly determine which product will be placed on the shelf.

[0097] The new product priority order is referred to in the product selection and placement process of the mathematical optimization process.

[0098] The rebate conditions include the brand monopoly designation shown in Table 6A, the manufacturer ratio designation shown in Table 6B, the required selected product designation shown in Table 6C, and the required non-selected product designation shown in Table 6D. The rebate conditions are constraints in the mathematical optimization process, and are defined in the preprocessing "Definition settings for mathematical optimization" in the flowchart shown in Figure 2.

[0099] [Table 6A]

[0100] In Table 6A, "Layer Number_Minimum" and "Layer Number_Maximum" indicate which row from the bottom the item should be placed in the number of items placed (unit number), the number of items placed indicates which unit from the left the item should be placed in, and the number of items occupied indicates how many rows it should occupy. Therefore, according to Table 6A, it is entered that Brand A's product will occupy two of the second to fourth rows from the bottom on the second shelf from the left, and Brand B's product will occupy one of the first to third rows from the bottom on the fourth shelf from the left.

[0101] [Table 6B]

[0102] In Table 6B, the occupancy composition ratio indicates the ratio of the placement width of the designated manufacturer to the placement width of the attribute in question.

[0103] [Table 6C]

[0104] According to Table 6C, it is input that product code 10000001 and product name AAAA are required to be selected with one face, and product code 10000002 and product name BBBB are required to be selected with two faces.

[0105] [Table 6D]

[0106] According to Table 6D, product code 10000005, product name EEEE, and product code 10000006, product name FFFF are entered with a face count of 0, meaning they will not be selected. All products that exist in the current maximum pattern are candidates for products to be placed in the creation pattern, but when there are products that exist in the current maximum pattern that you do not want to place in the creation pattern, specifying a required non-selected product is useful.

[0107] The input data entered on the staff terminal 3 is passed through the data storage server 4 and the request / response server 5 and is then read as input information by the shelf allocation pattern generation processing server 2.

[0108] As described above, the shelf allocation pattern generation processing server 2 performs the following preprocessing for the automatic generation processing of shelf allocation patterns: (i) Extraction process of continuous products within the minimum granularity (ii) Calculating the ideal placement width for each attribute when products are classified into a hierarchical structure by attribute (iii) Determining the priority order to be used in the product selection and placement process for placing products in the placement width of the lowest-level attribute (iv) Calculation of the coordinate position of the center of gravity of each attribute These are described in detail below.

[0109] (Extraction process of continuous products within the minimum granularity) The group of consecutive products within the minimum granularity is extracted from an existing planogram pattern. Preferably, a plurality of existing planogram patterns for shelves of different sizes are used as reference planogram data, and more preferably, a maximum planogram pattern with the largest shelf size, a minimum planogram pattern with the smallest shelf size, and a planogram pattern with a shelf size similar to that of the created pattern are used. A basic planogram pattern with a shelf size intermediate between the maximum and minimum planogram patterns can also be used. Even more preferably, the maximum planogram pattern of the current period, the minimum planogram pattern of the current period, and a previous planogram pattern with a shelf size similar to that of the created pattern are used. Here, shelf size refers to the sum of the widths of shelves on which products can be placed, i.e., the value obtained by multiplying the width of the shelf by the number of levels.

[0110] As a shelf allocation pattern of the same shelf size as the above-mentioned creation pattern, it is possible to use an existing shelf allocation pattern whose shelf size is closest to the creation pattern, preferably an existing shelf allocation pattern with the same number of shelves and a difference in overall width of 5% or less is used, and more preferably an early shelf allocation pattern with the same shelf size as the creation pattern is used.

[0111] Below, with reference to Figures 4A and 4B, we will explain the operation of the "automatic extraction program for consecutive product groups within the minimum granularity" of the calculation means (hereinafter also referred to as the extraction process) for extracting consecutive product groups within the minimum granularity, which finds boundaries between consecutive product groups within the minimum granularity.

[0112] This process of extracting continuous product groups within the minimum granularity involves obtaining, from a plurality of existing planogram patterns with different shelf sizes, the maximum planogram pattern with the largest shelf size for the current or previous period, the minimum planogram pattern with the smallest shelf size, and a planogram pattern with the same shelf size as the created pattern as reference planogram data (S11), For each of the acquired shelf allocation patterns, a product array is created in which the products with the lowest attribute are arranged in the order of arrangement of the products with the lowest attribute in the largest shelf allocation pattern (S12). In each of the acquired shelf allocation patterns, the arrangement order of the products with the attribute of the lowest layer is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the attribute of the lowest layer is detected as a one-end result, and the product at the other end is detected as an other-end result (S13). The one-end and other-end results detected in each shelf allocation pattern are recorded in the product arrangement of the attribute at the lowest level of the largest shelf allocation pattern, and a combination of products in which a product with an other-end result and a product with a one-end result are arranged in the arrangement direction is detected, and the space between the combinations of products is used as a boundary between consecutive product groups within the minimum granularity (S14).

[0113] More specifically, first, as shown in Fig. 4A, the calculation means of the request-response server 5 acquires reference planogram data, i.e., planogram data to be used in the extraction process (reference planogram data acquisition step: S11). As the reference planogram data, preferably, a plurality of the above-mentioned existing planogram patterns of different sizes are acquired from the data storage server 4.

[0114] In the reference planogram data acquired in step S11, the products with the lowest layer attribute are arranged in a row, preferably in the arrangement order of the products with the lowest layer attribute in the existing maximum planogram pattern for this term (product arrangement creation step: S12). For example, as shown in Figure 4B, if the maximum planogram pattern for this term consists of products A, B, C / D, E, and F (where " / " indicates that the rows where the products are arranged are adjacent to each other vertically), the products are arranged in the order A, B, C, D, E, and F from left to right.

[0115] Next, for each acquired reference planogram data, the arrangement order of products with the lowest attribute is assumed to be from one end of the product array to the other end, and products at one end of the product array with the lowest attribute are detected as one-end results, and products at the other end are detected as other-end results. For example, as shown in Figure 4B, when the arrangement order of products with the lowest attribute in the current season's largest planogram pattern is from A on the left end to F on the right end, the arrangement direction is from left to right, and in a product array of A, B, C / D, E, and F, A and D have left-end results as one-end results, and C and F have right-end results as other-end results, and these are detected and marked (right-end / left-end marking step: S13).

[0116] The above steps S12 and S13 are performed for all reference planogram data (repeated between L11 and L11E).

[0117] Next, the one-end and other-end results detected in each planogram pattern are marked on the products in the product arrangement of the lowest attribute of the largest planogram pattern, and a pair of products in which a product with an other-end result mark and a product with a one-end result mark are arranged in this order is detected. In the example shown in Figure 4B, the arrangement direction of the products in the lowest attribute of the largest planogram pattern is from A on the left end to F on the right end. Therefore, in this product arrangement of A to F, a pair of products in which a product with an other-end result (right-end result) and a product with a one-end result (left-end result) are arranged in this order is found, and the space between the products included in that pair is considered to be a boundary of the continuous product group within the minimum granularity (boundary detection step: S14). In Figure 4(b), a broken line surrounds a pair of products in which both the right-end and left-end result marks are arranged in this order. Therefore, the space between the products surrounded by the broken line is considered to be a boundary of the continuous product group within the minimum granularity.

[0118] In the right-end / left-end marking step (S13), if the product arrangement direction is from right end to left end, opposite to the illustrated embodiment, the extraction of the separators is also reversed. That is, the left-end and right-end results are marked on the products in the product arrangement of the lowest attribute of the maximum planogram pattern, and when the marks are viewed in the arrangement direction from right end to left end, the space between the pair of products where the left-end and right-end results are lined up in this order is taken as the separator of the group of consecutive products within the minimum granularity.

[0119] Alternatively, the above-mentioned product arrangement creation step (S12) may be omitted, and one end result (left end result) and the other end result (right end result) may be detected in each reference shelf allocation data acquired in step S11 (S13), the detected one end result and the other end result may be marked on the product in the product arrangement of the lowest attribute of the maximum shelf allocation pattern, a combination of products in which a product with an other end result and a product with a one end result are lined up in this order in the arrangement direction may be detected, and the space between the detected products may be used as a separator for the group of consecutive products within the smallest granularity (S14).

[0120] The process for extracting consecutive product groups within the minimum granularity shown in Figures 4A and 4B makes it possible to find a set of products between divisions, i.e., a set of consecutive products that have no history of turning back within the product arrangement of the lowest attribute, and therefore makes it possible to automatically extract consecutive product groups within the minimum granularity.

[0121] In the above process, the products are arranged in the order of the largest existing planogram pattern for the current period for each attribute of the lowest level of each planogram pattern, based on the business practice that the order of products in the largest pattern should also be followed in planograms smaller than the largest pattern.

[0122] (Calculating the ideal placement width) In order to allocate attributes by mathematical optimization in the automatic generation of shelf planograms, it is necessary to determine the target value of the allocation width of each attribute in advance as a parameter.

[0123] As shown in Figure 5, the placement width W of the attribute generally expands or contracts according to the total shelf width L of the entire shelf, so the share ratio S, which is the proportion of the placement width W of the attribute to the total shelf width L, often changes depending on the shelf size.

[0124] Therefore, in this embodiment, the arrangement width W self The weighted average share ratio S is used to reflect whether the current period's maximum or minimum planogram pattern is closer to the current period's maximum or minimum planogram pattern. self Calculate the total shelf width L self Multiplying by the weighted average width W self Ask for.

[0125]

number

[0126]

number

[0127]

number

[0128] Furthermore, for the layout width of the corresponding attribute in the previous period's shelf planogram, the ratio of the actual layout width to the weighted average width is calculated as the rate of change, and this rate of change is used as the weighted average width W of the created pattern. self The ideal placement width is calculated by multiplying by W and used as a parameter in the mathematical optimization process. self The multiplication of the value and the rate of change is carried out in order to reflect the know-how of the shelf allocation staff as much as possible when determining the placement width using the calculation means, and this makes it possible to increase the validity of the placement width of each attribute calculated by the mathematical optimization process.

[0129] If necessary, the ideal layout width when compressed may be obtained by multiplying the ideal layout width by the product compression rate, and the ideal layout width when compressed may be used as a parameter in the process of automatically generating a shelf allocation pattern.

[0130] As shown in Figure 6, when the attribute of the creation pattern is composed of new products and existing products, first, the share ratio and layout width of the attribute for the new products and existing products combined are calculated using the method described above. That is, among the existing shelf planogram patterns for this period, the weighted average of the share ratio of the corresponding attribute in a shelf planogram pattern larger than the creation pattern and the share ratio of the corresponding attribute in a shelf planogram pattern smaller than the creation pattern is calculated, and the layout width of the attribute is calculated by multiplying the calculated weighted average share ratio by the total shelf width of the creation pattern, and the calculated share ratio and layout width of the attribute are used as parameters in the automatic generation process of the shelf planogram pattern. In addition, the share ratio S of the attribute is calculated. self The ratio of new products to existing products (S 新規品self ,S 既存品self ), first, calculate the share ratio S of existing products in the attribute range for this period. 既存品self The share ratio S of existing products in the corresponding attribute in the shelf planogram pattern that is larger than the created pattern 既存品large and the share ratio S of existing products in the corresponding attribute in the shelf allocation pattern smaller than the created pattern. 既存品small The share ratio S of new products in the attribute range of the creation pattern is calculated as a weighted average. 新規品self is the share ratio S of existing products within the attribute from 1 既存品self The share ratio S of only existing products within the attribute calculated in this way is 既存品self and the share ratio of new products only S 新規品self The ratio of the number of stocks to the number of stocks is used as a parameter in the automatic generation process of the shelf allocation pattern.

[0131] The calculated ideal layout width for each attribute can be changed by attribute correction, which will be described later.

[0132] (Determining the priority used in product selection and placement processing) The priorities of the products used in the product selection and placement process in the mathematical optimization process are generally determined as follows. (1) In order to reflect the know-how of the shelf allocation staff, first, the calculation means determines the priority of each product unit based on (a) Criterion 1 and (b) Criterion 2, which will be explained below (Fig. 7A, Fig. 7B). Note that the priority of each product unit may be determined and input by the shelf allocation staff. (2) In order to maintain a balance in the number of faces of each product, the priority of each face is determined by a calculation means (FIG. 7C).

[0133] That is, as shown in FIG. 7C, (i) Find the maximum number of faces for each product (calculation of the maximum number of faces: S21), (ii) Divide the number of faces of the product into multiple groups and assign priorities (called face priorities) to the multiple groups (dividing the number of faces: S22); (iii) Products are allocated in order of highest face priority according to the above-mentioned product-level priority criteria ((1) (a) Criterion 1, (b) Criterion 2) (product allocation: S23). (iv) The priorities of the products are re-assigned (sorting in order of priority: S24). Here, the maximum number of faces for each product is calculated based on the number of faces adopted in the maximum pattern for this season.

[0134] The finally determined product priorities are input to the shelf planogram generation processing server 2 and used as parameters of constraint conditions in the mathematical optimization processing.

[0135] Next, the above-mentioned (1) determination of priority for each product unit and (2) determination of priority for each face unit will be described in detail.

[0136] (1) Determining the priority of each product (a) Product-level priority criteria 1 In order to prevent the absence of products for a certain purpose, i.e., a specific category, products included in a planogram pattern of a size smaller than the size to be created among the planogram patterns for this period are made mandatory, and products separately specified by the person in charge of planogramming are also made mandatory.

[0137] (b) Product-level priority criteria 2 The priority is determined by checking whether the candidate product to be placed with that attribute was present in a similar planogram pattern in the previous period, how many times it has been used in all planogram patterns, and the amount of sales within the specified period. For example, as shown in the left diagram of Figure 7B, the priority can be determined automatically by setting the level of the condition priority.

[0138] (2) Setting priority for each face After the priority of each product is determined in (1), the maximum number of faces for each product is divided into multiple groups at a certain number of divisions to maintain a balance in the number of faces for each product, and a priority order (priority per face) is assigned to each group. The priorities are reassigned in order from the highest priority per face according to the priority per product determined in (1).

[0139] An example of this flowchart will be described with reference to FIG. 7C. First, the maximum number of faces for each product is calculated (calculation of the maximum number of faces: S21). The maximum number of faces for each product is preferably the number of faces for that product in the maximum pattern for this season. As an example, in FIG. 7C, the maximum number of faces is set to 6 for product A, 2 for product B, and 2 for product C.

[0140] Next, the number of faces is divided into multiple groups with different priorities. For example, the number of faces is divided into three groups, with the first group being the highest priority, followed by the second group and the third group. This priority is called the priority per face. In this case, the number of divisions is determined appropriately based on the maximum number of faces, etc. (Division of the number of faces: S22).

[0141] Next, products are allocated to each divided group (product allocation: S23). In this case, all products A, B, and C are allocated to group 1, which has the highest priority. The allocation quantity of each product A, B, and C can be determined by dividing the maximum number of faces of each product by the number of groups (3 in this embodiment) and rounding up any fractions. In this example, the allocation quantity of product A to group 1 is 6 / 3, which is 2 products, and for products B and C, it is 2 / 3, which is rounded up, which is 1 product. Next, products A, B, and C are similarly allocated to group 2 based on their respective maximum number of faces. After allocating products to group 1 and group 2, product B and product C have each reached their maximum number of faces, so only product A is allocated to group 3.

[0142] Then, the groups are sorted from highest to lowest priority for each face, and the products contained in each group are sorted in accordance with the priority of each product (sorting by priority: S24).

[0143] Since the adjustment of the attribute placement width and the actual product selection and placement process are performed in the subsequent mathematical optimization process, the final product selection and placement cannot be determined at the S24 stage, and the products are only sorted as described above.

[0144] In the product selection and placement process in the mathematical optimization process, the order sorted by face priority and product priority is used as a parameter. Note that this parameter may also be the product type and number of faces sorted in order of product priority.

[0145] Furthermore, in the product selection and placement process, products sorted by priority sorting (S24) are placed until the final placement width for the attribute is reached. Products that exceed the final placement width for the attribute are not placed. Therefore, for example, if the placement width for the attribute is reached by product C in the second group, the third group of product A will not be placed, and four product A, two product B, and two product C will be placed as shown in Figure 7C.

[0146] (Calculating the coordinate position of the center of gravity of the attribute) The position of the center of gravity of each attribute of an existing shelf allocation pattern is input to the calculation means.

[0147] On the other hand, the calculation means calculates the position of the center of gravity of each attribute in the created pattern in the preprocessing of the automatic generation process of the shelf allocation pattern. For example, as shown in FIG. 8, the x, y coordinates (x right , y top ) and the x,y coordinates of the bottom left side (x left , y bottom ) and the centroid (x center , y center The calculation means preferably refers to a shelf allocation pattern of the same size as the created pattern for the previous period or an existing shelf allocation pattern for the current period that is of the same size as the created pattern, sets the center of gravity as the ideal position of the center of gravity, and generates a shelf allocation pattern so that the difference between the ideal position of the center of gravity and the calculated position of the center of gravity is minimized, and preferably restricts the placement position of the attributes using the position of the center of gravity.

[0148] When the coordinates of the start and end points of an attribute are constrained for its placement position, the shape of the attribute in front view generated by the mathematical optimization process may easily deviate from a rectangle. In contrast, when the center of gravity of the attribute is constrained for its placement position, the shape of the attribute in front view generated by the mathematical optimization process can be easily adjusted to a rectangle.

[0149] When the hierarchical structure of attributes in the planogram pattern is four levels, it is preferable to calculate the coordinate position of the center of gravity for attributes of all granularities and arrange them so that the difference from the ideal center of gravity position is minimized. Here, the ideal center of gravity position is calculated from the ideal arrangement width.

[0150] In addition, the shelf allocation pattern that references the center of gravity of the attribute can be included in the supplementary information. This allows you to generate shelf allocation patterns that meet the user's needs.

[0151] (Attribute placement processing using mathematical optimization) The shelf allocation pattern generation processing server 2 uses the above-mentioned ideal placement width and the coordinate position of the center of gravity of the attributes as parameters, and performs attribute placement processing using mathematical optimization processing under the constraint that consecutive product groups within the minimum granularity must be placed consecutively in one layer (Figure 3).

[0152] For this reason, in the mathematical optimization process, an objective function is set for the layout width of the attribute so as to minimize the error between the ideal layout width and the layout width in the planogram to be generated. Also, an objective function is set for the attribute so as to minimize the error between the referenced center of gravity position and the center of gravity position in the generated pattern.

[0153] In the attribute allocation process, in order to create a planogram pattern that reflects the know-how of the planogram manager, it is preferable to set constraints so that the layout at each attribute granularity is as rectangular as possible. This prevents the attribute shape in the planogram pattern from becoming convex or concave when it is finally output.

[0154] It is also preferable to impose constraints so that the placement positions of attributes in the same layer are not separated, and therefore the positions of the centers of gravity of attributes in the same layer are determined so that they are as close to each other as possible.

[0155] In the attribute allocation process, if a rebate condition exists, it is preferable to set this as a constraint. One example of this constraint is to ensure that the allocation width is greater than the allocation width of the manufacturer's ratio specified in the rebate condition. This makes it possible to comply with the manufacturer's ratio specification.

[0156] It is also preferable to arrange the products so that all of the designated brands are located on the shelf positions designated in the rebate conditions, thereby ensuring that the brand exclusivity is maintained.

[0157] It is preferable to adopt the specified number of faces for products specified in the rebate conditions. This allows the mandatory selection product designation and mandatory non-selection product designation to be observed.

[0158] It is preferable that the rebate conditions are input in advance from a terminal of a person in charge and stored in a data storage server.

[0159] In order to avoid slowing down the calculation speed of the mathematical optimization process, it is preferable to set the number of attributes for attribute 1 to 5 or less, the number of attributes for attribute 2 to 10 or less, the number of attributes for attribute 3 to 30 or less, and the number of attributes for attribute 4 to 80 or less.

[0160] (Attribute correction) The shelf allocation pattern generated by the shelf allocation pattern generation processing server 2 can be subjected to attribute correction. Here, attribute correction refers to aligning the boundary of the attribute of the lowest layer with the position of the shelf divider by adjusting the width of the attribute of the lowest layer in mathematical optimization processing. In attribute correction, as shown in Figure 9, the layout width of each attribute is changed and when aligning the attribute boundary with the shelf boundary, it is preferable to maintain the rectangle of the attribute layout when the shelf is viewed from the front.

[0161] (Product selection and placement processing) In the product selection and placement process for selecting and placing products in the lowest layer attribute, the products are placed in accordance with the above-mentioned product priority order.

[0162] In the product selection and placement process, various conditions such as rebate conditions are also taken into consideration.

[0163] (Check before outputting the shelf allocation pattern) The shelf allocation pattern generation processing server 2 checks for problems where the creation of shelf allocation patterns has not been completed, which may occur when processing problem division for units with different numbers of shelves, and if there is a problem where the creation of shelf allocation patterns has not been completed, it returns to the step before the above (calculation of the coordinate position of each attribute) and repeats the series of processes.

[0164] The shelf allocation pattern generated by the shelf allocation pattern generation processing server 2 is output to the person in charge terminal 3. In this output, it is preferable that the shelf allocation pattern on the display is color-coded according to the product attribute so that the hierarchy of the product attributes can be seen.

[0165] By outputting the shelf allocation pattern, the shelf allocation person can check the shelf allocation pattern output on the display provided on the person in charge terminal 3, and can correct the shelf allocation pattern on the person in charge terminal 3 as necessary.

[0166] (Shelf planogram output) If there are no problems in the check of the shelf allocation pattern before output and if there are no problems in the check by the shelf allocation manager, the shelf allocation pattern is output. Note that the steps of checking and correcting before output may be omitted.

[0167] According to the method for automatically creating shelf allocation patterns using the system of this embodiment, it is possible to use a calculation means to create shelf allocation patterns that reflect the know-how of experienced shelf allocation managers at a speed that is applicable to business operations and that do not appear unnatural.

[0168] (Aspects of the present invention) The present invention includes the following aspects. (1) A method for automatically creating a shelf allocation pattern, which includes inputting shelf information and product information of an existing shelf allocation pattern and a creation pattern, which is a shelf allocation pattern to be created, into a calculation means, performing an automatic generation process of the shelf allocation pattern in the calculation means, and outputting the creation pattern, When products are classified into a hierarchical structure by attributes in an existing shelf planogram, information on a group of products that are included in the same attribute in the lowest layer and are continuous on one shelf (hereinafter referred to as a group of continuous products within the minimum granularity) is included in the input data for the automatic generation process of the shelf planogram, In the process of automatically generating shelf allocation patterns, an attribute allocation process is performed to allocate product attributes so as to match the hierarchical structure of product attributes, and a product selection and allocation process is performed to select and allocate products within the attributes of the lowest layer. In the product selection and placement process, the method automatically creates a shelf planogram pattern in which products included in a group of consecutive products within the smallest granularity are placed consecutively on one shelf level. (2) In the above (1), the calculation means performs an extraction process of a group of consecutive products within the minimum granularity from a plurality of existing shelf allocation patterns for shelves of different shelf sizes before the automatic generation process of the shelf allocation pattern. (3) In the above (1) or (2), the method for automatically creating a shelf allocation pattern is a shelf allocation pattern for a plurality of stores of a retailer that operates multiple stores under the same store name. (4) In any one of the above (1) to (3), In the process of extracting continuous product groups within the minimum granularity, the existing maximum and minimum shelf planogram patterns for the current period, as well as the shelf planogram patterns for the previous period with the same shelf size as the created pattern, are obtained. The arrangement direction of the products with the attribute of the lowest layer of each of the acquired shelf allocation patterns is determined to be from one end of the shelf to the other end in the order of arrangement, and in each of the acquired shelf allocation patterns, one end of the shelf to the other end of the arrangement of the products with the attribute of the lowest layer is detected as one-end actual results, and the other end is detected as the other-end actual results; Marking the detected one-end and other-end results on the products in the product arrangement of the attribute of the lowest layer of the acquired maximum planogram pattern, and detecting pairs of products in which a product having an other-end result mark and a product having a one-end result mark are arranged in the arrangement direction, A method for automatically creating shelf allocation patterns that automatically extracts groups of consecutive products within the minimum granularity by dividing the detected sets of products into groups of consecutive products within the minimum granularity. (5) In any one of the above (1) to (4), A method for automatically creating a shelf allocation pattern in which, before a product selection and placement process, a calculation means divides each candidate product to be placed in the product attribute of the lowest layer into a plurality of groups by a certain number of divisions and assigns a priority to each group to give a priority to each face, sorts the plurality of groups from a group with a high priority to a group with a low priority to each face, and sorts the products in each group according to the priority of each product, and uses the product placement ranking thus obtained as a parameter in the product selection and placement process. (6) In any one of (1) to (5) above, the method for automatically generating a shelf allocation pattern is performed by a mathematical optimization process. (7) In any one of (1) to (6) above, the method for automatically creating a shelf allocation pattern is to perform the attribute allocation process by mathematical optimization processing. (8) In the above (7), a method for automatically creating a shelf allocation pattern, which determines a target value for the layout width of each attribute of the created pattern as a parameter used when performing attribute layout processing by mathematical optimization. (9) In the above (8), a method for automatically creating a shelf allocation pattern in which the target value of the layout width for each attribute is calculated as follows: The share ratio S, which is the percentage of the attribute in the total shelf width self The share ratio of the attribute in the maximum shelf layout pattern for this period is S large and the share ratio S of the attribute in the minimum planogram for this period small The weighted average share ratio S self Total shelf width L self Multiplying by the weighted average width W self Calculate the weighted average width W self is the target value of the placement width of the attribute. (10) In the above (9), when the attribute of the created pattern is composed of existing products and new products, the parameters used when performing attribute allocation processing by mathematical optimization processing are calculated as follows: Share ratio S of the attribute self The ratio of new products to existing products (S 新規品self ,S 既存品self ) and On the other hand, the share ratio of the existing products in the current period in the attribute width is calculated by the share ratio S of the existing products in the corresponding attribute in the shelf allocation pattern that is larger than the created pattern. 既存品large and the share ratio S of existing products in the corresponding attribute in the shelf allocation pattern smaller than the created pattern. 既存品small The weighted average share ratio S of existing products in the attribute range is calculated as a weighted average of 既存品self Obtained, Next, the share ratio S of new products in the attribute range of the creation pattern 新規品self The share ratio of existing products in the attribute range from 1 to S 既存品self It is calculated by subtracting The share ratio S of only existing products in the attribute range calculated in this way 既存品self and the share ratio of new products only S 新規品self The ratio of and is calculated and used as a parameter. (11) In any one of (1) to (10) above, a method for automatically creating a shelf planogram, in which a product selection and placement process is performed by a mathematical optimization process, the priority of products to be used in the product selection and placement process is determined according to the following steps (i) to (iv), and the determined product priority is used as a parameter of a constraint condition: (i) Find the maximum number of faces for each product; (ii) Divide the number of products into multiple groups and assign priorities (face priorities) to the multiple groups (for example, group 1, group 2, group 3 in descending order of face priorities), (iii) Allocating products according to a per-product priority criterion, starting with the highest face priority; (iv) Reprioritize products. (12) In any one of the above (1) to (11), In the process of automatically generating a shelf allocation pattern, the position of the center of gravity of each attribute in a shelf allocation pattern for the previous or current period that is the same size as the pattern to be created is referenced, and the position of the center of gravity of each attribute in the pattern to be created is calculated as a parameter, and a process is performed to minimize the difference between the referenced position of the center of gravity and the position of the center of gravity in the shelf allocation pattern to be generated. (13) In any one of the above (1) to (12), A method for automatically generating shelf allocation patterns using rebate conditions as constraints in attribute allocation processing. (14) In any one of the above (1) to (13), A method for automatically creating shelf planograms by correcting attributes to align the attribute boundaries with the shelf separator positions by adjusting the width of the attribute of the lowest layer. (15) In any one of the above (1) to (14), A method for automatically generating shelf allocation patterns, which outputs shelf allocation patterns generated by an automatic shelf allocation pattern generation process to a display, and enables correction of product arrangement on the display. (16) In any one of the above (1) to (15), A method for automatically generating shelf allocation patterns, in which shelf allocation patterns generated by an automatic shelf allocation pattern generation process are output to a display, and the shelf allocation patterns on the display are color-coded according to product attributes. (17) An automatic shelf allocation pattern creation system including a calculation means for inputting shelf information and product information of an existing shelf allocation pattern and a creation pattern which is a shelf allocation pattern to be created, performing automatic generation processing of the shelf allocation pattern, and outputting the creation pattern. The automatic generation process of the shelf planogram pattern includes an attribute allocation process in which the layout of a group of consecutive products within the minimum granularity is used as input data, and the attributes of the products are arranged so as to match the hierarchical structure when the products are classified into a hierarchical structure by attributes; A product selection and placement process is performed to select and place products within the attributes of the lowest layer. The product selection and placement process places products included in the minimum granularity contiguous product group contiguously on one shelf. Automatic shelf layout pattern creation system. (18) In (17) above, A system for automatically creating shelf allocation patterns in which a calculation means performs an extraction process for a group of consecutive products within a minimum granularity, automatically extracting a group of consecutive products within a minimum granularity. (19) In (17) or (18) above, The maximum number of faces of the product to be subjected to the product selection and placement process and the priority of product placement are input to the calculation means; The calculation means divides candidate products into multiple groups with different priorities (called face-level priorities) before the product selection and placement process, sorts the multiple groups from highest to lowest face-level priorities, and sorts the products within each group according to their product-level priorities, and uses the product placement rankings obtained in this way as parameters in the product selection and placement process. This is an automatic shelf allocation pattern creation system. (20) In any one of the above (17) to (19), An automatic shelf allocation pattern creation system that uses mathematical optimization processing to automatically generate shelf allocation patterns. (21) In any one of the above (17) to (20), An automatic shelf allocation pattern creation system that uses mathematical optimization processing for attribute allocation processing. (twenty two) A shelf allocation pattern generation program in which a calculation means receives as input data a plurality of existing shelf allocation patterns with different shelf sizes, their shelf information and product information, as well as shelf information and product information of a creation pattern, which is a shelf allocation pattern to be created, and outputs the shelf allocation pattern of the creation pattern, and which performs attribute placement processing and product selection and placement processing, and in the product selection and placement processing, places products included in a group of consecutive products within the minimum granularity consecutively on one shelf level. (23) A shelf allocation pattern generation program in (22) above, in which attribute placement processing and product selection and placement processing are performed by mathematical optimization processing, and in the product selection and placement processing, a constraint is set such that products included in a group of consecutive products within the minimum granularity are placed consecutively on one shelf. (24) In paragraphs 22 or 23 above, an automatic extraction step of a group of consecutive products within a minimum granularity for automatically extracting a group of consecutive products within a minimum granularity from a plurality of existing shelf planograms with different shelf sizes; The automatic extraction step of the continuous product group within the minimum particle size, a reference planogram data acquisition step for acquiring, from the plurality of existing planogram patterns, a maximum planogram pattern having the largest shelf size in the current term or the previous term, a minimum planogram pattern having the smallest shelf size, and a planogram pattern having the same shelf size as the created pattern; In each of the acquired shelf allocation patterns, the arrangement order of the products with the lowest attribute is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the lowest attribute is detected as a one-end result, and the product at the other end is detected as an other-end result. A shelf allocation pattern generation program that marks the one-end and other-end results detected in each shelf allocation pattern on products in the product arrangement of the lowest attribute of the largest shelf allocation pattern, detects sets of products in which products with the other-end result mark and products with the one-end result mark are lined up in the arrangement direction, and automatically extracts groups of consecutive products within the minimum granularity by dividing the detected sets of products into groups of consecutive products within the minimum granularity. (25) In any one of the above (22) to (24), In the process of obtaining reference shelf planogram data, A shelf allocation pattern generation program that obtains the maximum shelf allocation pattern and the minimum shelf allocation pattern for the current period, as well as shelf allocation patterns of the same size as the patterns created in the previous period. (26) An automatic extraction program for a group of consecutive products within a minimum granularity, which automatically extracts a group of consecutive products within a minimum granularity from a plurality of existing planogram patterns that have been subjected to attribute arrangement processing for different shelf sizes, in a calculation means, a step of acquiring, as reference planogram data from the plurality of existing planogram patterns, a maximum planogram pattern having a maximum shelf size, a minimum planogram pattern having a minimum shelf size, and a planogram pattern having a shelf size similar to that of the pattern to be created; In each of the acquired shelf allocation patterns, the arrangement order of the products with the lowest attribute is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the lowest attribute is detected as a one-end result, and the product at the other end is detected as an other-end result. a process of marking the one-end results and the other-end results detected in each of the planogram patterns on products in the product arrangement of the lowest attribute of the largest planogram pattern, detecting sets of products in which a product having an other-end result mark and a product having a one-end result mark are lined up in the arrangement direction, and automatically extracting contiguous product groups within the minimum granularity by dividing the sets of products into segments of contiguous product groups within the minimum granularity; An automatic extraction program for continuous product groups within the smallest granularity. [Explanation of symbols]

[0169] 1. Automatic shelf layout pattern creation system 2. Calculation means, shelf allocation pattern generation processing server 3. Personnel terminal 4 Data storage server 5. Request-Response Server

Claims

1. An automatic shelving planogram pattern creation method in which existing shelving planogram patterns and shelf information and product information of a creation pattern which is a shelving planogram pattern to be created are input to a calculation means, and the calculation means performs automatic generation processing of the shelving planogram pattern and outputs the creation pattern, When products are classified into a hierarchical structure by attributes in an existing shelf planogram, information on a group of products that are included in the same attribute in the lowest layer and are continuous on one shelf (hereinafter referred to as a group of continuous products within the minimum granularity) is included in the input data for the automatic generation process of the shelf planogram, In the process of automatically generating shelf allocation patterns, an attribute allocation process is performed to allocate product attributes so as to match the hierarchical structure of product attributes, and a product selection and allocation process is performed to select and allocate products within the attributes of the lowest layer. In the product selection and placement process, the method automatically creates a shelf planogram pattern in which products included in a group of consecutive products within the smallest granularity are placed consecutively on one shelf level.

2. 2. The method for automatically creating shelf allocation patterns according to claim 1, wherein the calculation means performs an extraction process for a group of consecutive products within a minimum granularity from a plurality of existing shelf allocation patterns for shelves of different shelf sizes before the automatic generation process of the shelf allocation pattern.

3. In the process of extracting continuous product groups within the minimum granularity, the existing maximum and minimum shelf planogram patterns for the current period, as well as the shelf planogram patterns for the previous period with the same shelf size as the created pattern, are obtained. The arrangement direction of the products with the attribute of the lowest layer of each of the acquired shelf allocation patterns is determined from one end to the other end in order of arrangement, and in each of the acquired shelf allocation patterns, one end of the left and right of the arrangement of the products with the attribute of the lowest layer is detected as one-end actual results, and the other end is detected as the other-end actual results; Marking the detected one-end and other-end results on the products in the product arrangement of the attribute of the lowest layer of the acquired maximum planogram pattern, and detecting pairs of products in which a product having an other-end result mark and a product having a one-end result mark are arranged in the arrangement direction, 3. The method for automatically creating a shelf planogram pattern according to claim 2, wherein the detected sets of products are used as separators for groups of consecutive products within the minimum granularity, thereby automatically extracting groups of consecutive products within the minimum granularity.

4. A method for automatically creating shelf allocation patterns according to any one of claims 1 to 3, wherein, before the product selection and placement process, the calculation means divides each candidate product to be placed in the product attribute at the lowest level into a plurality of groups by a certain number of divisions and assigns priorities to each group to determine priorities on a face-by-face basis, sorts the plurality of groups from a group with high to a group with low face-by-face priority, and sorts the products within each group according to the priority on a product-by-product basis, and uses the product placement rankings obtained thereby as parameters in the product selection and placement process.

5. 4. The method for automatically creating a shelf allocation pattern according to claim 1, wherein in the process of automatically creating a shelf allocation pattern, the position of the center of gravity of each attribute in a shelf allocation pattern for the previous or current period that is the same size as the pattern to be created is referenced, and the position of the center of gravity of each attribute in the pattern to be created is calculated as a parameter, and a process is performed to minimize the difference between the referenced position of the center of gravity and the position of the center of gravity in the shelf allocation pattern to be created.

6. The method for automatically creating a shelf allocation pattern according to any one of claims 1 to 3, wherein in the process of automatically generating a shelf allocation pattern, attribute correction is performed to align the attribute boundary with the position of a shelf partition by adjusting the width of the attribute of the lowest layer.

7. An automatic shelf allocation pattern creation system including a calculation means for inputting shelf information and product information of an existing shelf allocation pattern and a creation pattern which is a shelf allocation pattern to be created, performing automatic generation processing of the shelf allocation pattern, and outputting the creation pattern. The automatic generation process of the shelf planogram pattern includes an attribute allocation process in which the layout of a group of consecutive products within the minimum granularity is used as input data, and the attributes of the products are arranged so as to match the hierarchical structure when the products are classified into a hierarchical structure by attributes; A product selection and placement process is performed to select and place products within the attributes of the lowest layer. A system for automatically creating shelf allocation patterns in which the product selection and placement process places products included in a group of consecutive products within the smallest granularity consecutively on one shelf level.

8. 8. The system for automatically creating a shelf planogram according to claim 7, wherein the calculation means performs an extraction process for a group of consecutive products within a minimum granularity, automatically extracting a group of consecutive products within a minimum granularity.

9. The maximum number of faces of the product to be subjected to the product selection and placement process and the priority of product placement are input to the calculation means; 9. The system for automatically creating shelf allocation patterns according to claim 7 or 8, wherein the calculation means divides candidate products into a plurality of groups with different priorities (these priorities are called priorities per face) before the product selection and placement process, sorts the plurality of groups from a group with high to a group with low priorities per face, and sorts the products within each group according to the priorities per product, and uses the product placement rankings obtained thereby as parameters in the product selection and placement process.

10. A shelf allocation pattern generation program in which a calculation means receives as input data a plurality of existing shelf allocation patterns with different shelf sizes, their shelf information, and product information, as well as shelf information and product information of a creation pattern, which is a shelf allocation pattern to be created, and outputs the shelf allocation pattern of the creation pattern, and performs an attribute placement process that places product attributes so as to match the hierarchical structure when products are classified into a hierarchical structure by attribute, and a product selection and placement process that selects and places products within the attributes of the lowest layer, and in the product selection and placement process, places products included in a group of consecutive products within the smallest granularity consecutively on one level of the shelf.

11. an automatic extraction step of a group of consecutive products within a minimum granularity for automatically extracting a group of consecutive products within a minimum granularity from a plurality of existing shelf planograms with different shelf sizes; The automatic extraction step of the continuous product group within the minimum particle size, a reference planogram data acquisition step for acquiring, from the plurality of existing planogram patterns, a maximum planogram pattern having the largest shelf size in the current term or the previous term, a minimum planogram pattern having the smallest shelf size, and a planogram pattern having the same shelf size as the created pattern; In each of the acquired shelf allocation patterns, the arrangement order of the products with the lowest attribute is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the lowest attribute is detected as a one-end result, and the product at the other end is detected as an other-end result. A shelf allocation pattern generation program as described in claim 10, which marks the one-end and other-end results detected in each shelf allocation pattern on products in the product arrangement of the lowest attribute of the largest shelf allocation pattern, detects pairs of products in which products with the other-end result mark and products with the one-end result mark are lined up in the arrangement direction, and automatically extracts consecutive product groups within the minimum granularity by dividing the detected pairs of products into consecutive product groups within the minimum granularity.

12. An automatic extraction program for a group of consecutive products within a minimum granularity, which automatically extracts a group of consecutive products within a minimum granularity from a plurality of existing planogram patterns that have been subjected to attribute arrangement processing for different shelf sizes, in a calculation means, a step of acquiring, as reference planogram data from the plurality of existing planogram patterns, a maximum planogram pattern having a maximum shelf size, a minimum planogram pattern having a minimum shelf size, and a planogram pattern having a shelf size similar to that of the pattern to be created; In each of the acquired shelf allocation patterns, the arrangement order of the products with the lowest attribute is assumed to be from one end of the product arrangement to the other end, and the product at one end of the product arrangement with the lowest attribute is detected as a one-end result, and the product at the other end is detected as an other-end result. a process of marking the one-end results and the other-end results detected in each of the planogram patterns on products in the product arrangement of the lowest attribute of the largest planogram pattern, detecting sets of products in which a product having an other-end result mark and a product having a one-end result mark are lined up in the arrangement direction, and automatically extracting contiguous product groups within the minimum granularity by dividing the sets of products into segments of contiguous product groups within the minimum granularity; An automatic extraction program for continuous product groups within the smallest granularity.

Images