Recommended data generating device, recommended data generating method, and program

Abstract

A recommendation data generation device (2000) acquires a plurality of pieces of material specification information (10) indicating material specifications and acquires, for each piece of the material specification information (10), physical property information (20) indicating physical quantities for each of a plurality of physical properties of a product (70) that is possibly generated on the bases of the material specifications indicated by the material specification information (10). The recommendation data generation device (2000) uses the physical property information (20) to generate a self-organizing map (30) in which a material property vector indicating values related to the physical quantity of each of the plurality of types of physical properties of the product (70) is assigned to each node in a map space. The recommendation data generation device (2000) selects one or more target nodes from among the nodes of the self-organizing map (30) on the basis of an arrangement of nodes corresponding to respective physical property information (20) in the map space. The recommendation data generation device (2000) generates recommendation data (80) indicating material specifications corresponding to the target node(s).

Description

Recommendation data generating device, recommendation data generating method, and non-transitory computer-readable medium

[0001] The present disclosure relates to techniques for providing information related to product development.

[0002] In product development, it is useful to understand the relationship between materials and deliverables. Therefore, systems that support understanding of the relationship between materials and deliverables have been developed. For example, Patent Literature 1 discloses a system that uses a self-organizing map to support understanding of the causal relationship between design values ​​and physical property values ​​of a tire.

[0003] JP 2016-148988 A

[0004] In Patent Document 1, a self-organizing map is used to identify which of multiple design variables of a tire is an important factor. Therefore, the use of the self-organizing map for other purposes is not anticipated. The present disclosure has been made in consideration of such problems, and an object of the present disclosure is to provide a new technology that provides information useful for product development.

[0005] The recommended data generation device of the present disclosure includes an acquisition means for acquiring a plurality of pieces of material parameter information indicating material parameters, and for each piece of material parameter information, acquiring physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated using the material parameters indicated in the material parameter information; a first generation means for using the physical property information to generate a self-organizing map in which physical property vectors indicating values ​​for the physical property quantities of each of a plurality of types of physical properties of the product are assigned to each node in a map space; a selection means for selecting one or more target nodes from the nodes of the self-organizing map based on the arrangement in the map space of the nodes corresponding to each piece of physical property information; and a second generation means for generating recommended data indicating the material parameters corresponding to the target nodes.

[0006] A recommendation data generation method according to the present disclosure is executed by a computer and includes the steps of: acquiring a plurality of pieces of material specification information indicating material specifications and, for each of the pieces of material specification information, acquiring physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated using the material specifications indicated in the material specification information; a first generation step of using the physical property information to generate a self-organizing map in which physical property vectors indicating values ​​for the physical property quantities of each of a plurality of types of physical properties of the product are assigned to each node in a map space; a selection step of selecting one or more target nodes from the nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each piece of physical property information; and a second generation step of generating recommendation data indicating the material specifications corresponding to the target nodes.

[0007] The non-transitory computer-readable medium of the present disclosure stores a program that causes a computer to execute the recommendation data generating method of the present disclosure.

[0008] According to the present disclosure, a new technology is provided that provides information useful for product development.

[0009] 1 is a diagram illustrating an example of an outline of the operation of a recommended data generating device of embodiment 1. FIG. 2 is a block diagram illustrating an example of the functional configuration of a recommended data generating device of embodiment 1. FIG. 3 is a block diagram illustrating an example of the hardware configuration of a computer that realizes the recommended data generating device of embodiment 1. FIG. 4 is a flowchart illustrating an example of a processing flow executed by the recommended data generating device of embodiment 1. FIG. 5 is a diagram illustrating an example of material specification information in table format. FIG. 6 is a diagram illustrating an example of physical property information in table format. FIG. 7 is a diagram illustrating an example of the configuration of a self-organizing map in table format. FIG. 8 is a flowchart illustrating an example of a processing flow for selecting a target node. FIG. 9 is a flowchart illustrating an example of a processing flow for selecting a target node. FIG. 10 is a diagram illustrating an example of a map space divided into a plurality of partial regions. FIG. 11 is a diagram illustrating an example of a configuration of a self-organizing map in which a specification vector is assigned to each node in table format. FIG. 12 is a diagram illustrating an example of a map image. FIG. 13 is a diagram illustrating an example of an outline of the operation of a recommended data generating device of embodiment 2. FIG. 14 is a block diagram illustrating an example of the functional configuration of a recommended data generating device of embodiment 2. FIG. 15 is a flowchart illustrating an example of a processing flow executed by the recommended data generating device of embodiment 2.

[0010] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In each drawing, identical or corresponding elements are designated by the same reference numerals, and duplicate explanations will be omitted as necessary for clarity. Furthermore, unless otherwise specified, predetermined information such as predetermined values ​​and thresholds is stored in advance in a storage device accessible from a device that uses the information. Furthermore, unless otherwise specified, the storage unit is composed of one or any number of storage devices.

[0011] [First Embodiment] <Overview> Fig. 1 is a diagram illustrating an example of an outline of the operation of a recommendation data generation device 2000 according to the first embodiment. Fig. 1 is a diagram for facilitating understanding of the outline of the recommendation data generation device 2000, and the operation of the recommendation data generation device 2000 is not limited to that shown in Fig. 1.

[0012] The recommendation data generating device 2000 generates a self-organizing map 30 representing the distribution of physical properties of various deliverables 70 that can be generated in a specific process (hereinafter, target process) of product development. The deliverables 70 are those that are predicted to be generated by processing a material 60 in the production process of the target process, or those that have actually been generated. The material 60 is a material used to generate the deliverables 70. Various patterns of material 60 can be used in the target process. The physical properties of the deliverables 70 can vary depending on the material 60 used.

[0013] One pattern of the material 60 is specified by the material specifications. In other words, materials 60 having different material specifications are treated as materials 60 having different patterns. On the other hand, materials 60 having the same material specifications are treated as materials 60 having the same pattern.

[0014] The material specifications are expressed, for example, by the type of material, the types of substances constituting the material, the blending ratios of each substance, and the type of processing performed to create the material. Examples of material types include carbon fiber reinforced plastic and stainless steel. For example, assume that material 60 is carbon fiber reinforced plastic. In this case, the material specifications of material 60 include the type of each of one or more carbon fibers constituting material 60 (e.g., polyacrylonitrile fiber or carbonized cellulose fiber), the type of each of one or more resins constituting material 60 (e.g., epoxy or polyether terephthalate), and the blending ratios of these substances. The material specifications may also include the type of fiber-directional polymerization method, the type of crimping method, and the resin composition.

[0015] The target process may be a single process or a combination of multiple consecutive processes. In the latter case, the deliverable 70 is a deliverable that can be generated by processing the material 60 in these multiple consecutive processes. For example, assume that the target process is a combination of processes P1 and P2. In this case, the deliverable 70 is obtained by processing the material 60 in the generation process of process P1, and then processing the deliverable of process P1 in the generation process of process P2.

[0016] The self-organizing map 30 has a plurality of nodes arranged in an m-dimensional map space, where m is set to 2 or 3 so that the map space can be visually represented (e.g., as an image). In the visually represented map space, each node is represented, for example, by a square on a grid or a lattice on a grid.

[0017] Multidimensional data (hereinafter referred to as a physical property vector) representing the magnitude of each of multiple types of physical properties is assigned to each node of the self-organizing map 30. For example, assume that four types of physical properties, namely flame retardancy, heat resistance, elastic modulus, and toughness, are used. In this case, the physical property vector is four-dimensional data representing the magnitude of each of the physical property quantities of these four types of physical properties. Hereinafter, the number of dimensions of the physical property vector is defined as n, where n>m. That is, in the self-organizing map 30, the space of the physical property vectors is a high-dimensional space, and the map space is a low-dimensional space.

[0018] To generate such a self-organizing map 30, the recommendation data generating device 2000 acquires, for each of the multiple patterns of materials 60 (in other words, for each material 60 identified by each of the multiple patterns of material specifications), material specification information 10 representing the material specifications of the material 60 and physical property information 20 corresponding to the material specification information 10. That is, the recommendation data generating device 2000 acquires multiple pairs of material specification information 10 and physical property information 20. The physical property information 20 corresponding to the material specification information 10 indicates the quantity of each of multiple types of physical properties for a deliverable 70 that can be produced in a target process using the material 60 whose material specifications are represented by the material specification information 10. The types of physical properties include, for example, the aforementioned flame retardancy, heat resistance, elastic modulus, and toughness. The number of types of physical properties indicated by the physical property information 20 is equal to or greater than the number of dimensions n of the physical property vector.

[0019] The recommendation data generating device 2000 identifies a node (hereinafter, a corresponding node) in the self-organizing map 30 that corresponds to each piece of physical property information 20. Here, the corresponding node of the physical property information 20 is a node in the self-organizing map 30 that is associated with a physical property vector that is most similar to the physical property vector obtained from the physical property information 20. In other words, the corresponding node of the physical property information 20 is a node in the self-organizing map 30 that represents a physical property that is closest to the physical property represented by the physical property information 20.

[0020] The recommendation data generating device 2000 selects, as target nodes, one or more nodes that do not correspond to any of the physical property information 20 (in other words, nodes that are not corresponding nodes) based on the arrangement of the corresponding nodes of each physical property information 20 in the map space. For example, a node located far away from each corresponding node in the map space is selected as the target node.

[0021] The recommendation data generating device 2000 estimates material specifications corresponding to the target node using the self-organizing map 30. Then, the recommendation data generating device 2000 generates recommendation data 80 indicating the estimated material specifications.

[0022] <Example of Effects> The pair of material specification information 10 and physical property information 20 acquired by the recommendation data generation device 2000 is generated based on the results of simulations and experimental generation of the product 70 that have already been performed. Therefore, the pair of material specification information 10 and physical property information 20 can be said to be knowledge about the correspondence between material specifications and physical properties that has been obtained by simulations and the like that have been performed up to now.

[0023] When further performing simulations or the like to increase knowledge, it is useful to efficiently increase knowledge by appropriately selecting material specifications to be used in the simulations or the like. Here, one method for efficiently increasing knowledge is to perform simulations or the like so as to obtain a product 70 whose physical properties are as dissimilar as possible to the product 70 obtained so far. In this way, knowledge can be obtained about the correspondence between various physical properties that are not highly similar to each other and material specifications. In other words, it is possible to avoid obtaining knowledge only about similar physical properties.

[0024] In this regard, according to the recommendation data generating device 2000, nodes (corresponding nodes) of the self-organizing map 30 corresponding to each piece of physical property information 20 are identified, and one or more target nodes are selected based on the arrangement of the corresponding nodes in the map space of the self-organizing map 30. Here, the arrangement of the corresponding nodes in the map space represents the distribution of physical properties in the map space for which knowledge has already been obtained. Therefore, by selecting target nodes based on the arrangement of the corresponding nodes in the map space, it is possible to select, as target nodes, nodes that represent physical properties that are not highly similar to physical properties for which knowledge has already been obtained.

[0025] The recommendation data generation device 2000 then estimates material specifications corresponding to the target node and generates recommendation data 80 indicating the material specifications. Here, the material specifications corresponding to the target node are material specifications that are estimated to result in a product 70 having physical properties corresponding to the target node. Therefore, by performing a simulation or the like using these material specifications, there is a high probability that a product 70 having physical properties that are not highly similar to properties already known will be obtained through the simulation or the like. Therefore, the recommendation data generation device 2000 can easily obtain material specifications that allow efficient simulation of the generation of the product 70.

[0026] The recommendation data generating device 2000 of this embodiment will be described in more detail below.

[0027] <Example of Functional Configuration> FIG. 2 is a block diagram illustrating an example of the functional configuration of the recommendation data generation device 2000 according to the first embodiment. The recommendation data generation device 2000 includes an acquisition unit 2020, a first generation unit 2040, a selection unit 2060, and a second generation unit 2080. The acquisition unit 2020 acquires material specification information 10 and physical property information 20 for each of multiple patterns of materials 60. The first generation unit 2040 generates a self-organizing map 30 using physical property vectors obtained from each piece of physical property information 20. The selection unit 2060 identifies corresponding nodes for each piece of physical property information 20. The selection unit 2060 selects one or more non-corresponding nodes as target nodes based on the arrangement of each corresponding node in the map space. The second generation unit 2080 generates recommendation data 80 indicating material specifications corresponding to the target node.

[0028] <Example of Hardware Configuration> Each functional component of the recommendation data generation device 2000 may be realized by hardware that realizes the respective functional component (e.g., a hardwired electronic circuit, etc.), or may be realized by a combination of hardware and software (e.g., a combination of an electronic circuit and a program that controls it, etc.). Below, a case where each functional component of the recommendation data generation device 2000 is realized by a combination of hardware and software will be further described.

[0029] 3 is a block diagram illustrating an example of the hardware configuration of a computer 1000 that realizes the recommendation data generation device 2000 of embodiment 1. The computer 1000 is any computer. For example, the computer 1000 is a stationary computer such as a PC (Personal Computer) or a server machine. Alternatively, the computer 1000 may be a portable computer such as a smartphone or a tablet terminal. The computer 1000 may be a dedicated computer designed to realize the recommendation data generation device 2000, or may be a general-purpose computer.

[0030] For example, by installing a predetermined application on the computer 1000, each function of the recommendation data generation device 2000 is realized on the computer 1000. The application is configured as a program for realizing each functional component of the recommendation data generation device 2000. Note that any method for acquiring the program is possible. For example, the program can be acquired from a storage medium (such as a DVD disc or USB memory) on which the program is stored. Alternatively, the program can be acquired by downloading the program from a server device that manages the storage device on which the program is stored.

[0031] The computer 1000 has a bus 1020, a processor 1040, a memory 1060, a storage device 1080, an input / output interface 1100, and a network interface 1120. The bus 1020 is a data transmission path for the processor 1040, the memory 1060, the storage device 1080, the input / output interface 1100, and the network interface 1120 to transmit and receive data to and from each other. However, the method of connecting the processor 1040 and the like to each other is not limited to bus connection.

[0032] The processor 1040 is a processor such as a central processing unit (CPU), a graphics processing unit (GPU), or a field-programmable gate array (FPGA). The memory 1060 is a main storage device realized using a random access memory (RAM) or the like. The storage device 1080 is an auxiliary storage device realized using a hard disk, a solid state drive (SSD), a memory card, a read only memory (ROM), or the like.

[0033] The input / output interface 1100 is an interface for connecting the computer 1000 to an input / output device. For example, the input / output interface 1100 is connected to an input device such as a keyboard and an output device such as a display device.

[0034] The network interface 1120 is an interface for connecting the computer 1000 to a network. This network may be a LAN (Local Area Network) or a WAN (Wide Area Network).

[0035] The storage device 1080 stores a program (a program that realizes the above-mentioned application) that realizes each functional component of the recommendation data generating device 2000. The processor 1040 reads this program into the memory 1060 and executes it to realize each functional component of the recommendation data generating device 2000.

[0036] The recommendation data generating device 2000 may be realized by one computer 1000 or by multiple computers 1000. In the latter case, the configurations of the computers 1000 do not need to be the same, and can be different from each other.

[0037] <Processing Flow> FIG. 4 is a flowchart illustrating the processing flow executed by the recommendation data generating device 2000 according to the first embodiment. The acquiring unit 2020 acquires material specification information 10 and physical property information 20 for each of multiple patterns of materials 60 (S102). The first generating unit 2040 generates a self-organizing map 30 using physical property vectors obtained from each piece of physical property information 20 (S104). The selecting unit 2060 identifies corresponding nodes for each piece of physical property information 20 (S106). The selecting unit 2060 selects a target node based on the arrangement of the corresponding nodes in the map space (S108). The second generating unit 2080 generates recommendation data 80 indicating material specifications corresponding to the target node (S110).

[0038] <Acquisition of Material Specifications Information 10 and Physical Properties Information 20: S102> The acquisition unit 2020 acquires, for each of a plurality of patterns of materials 60, material specifications information 10 indicating the material specifications of the material 60 and physical properties information 20 regarding the product 70 that can be generated using the material 60 (S102). FIG. 5 is a diagram illustrating an example of the material specifications information 10 in table format. The table 100 in FIG. 5 has columns for material identification information 102 and material specifications 104. The material identification information 102 indicates identification information assigned to the material 60. The material specifications 104 indicate the specifications of the material 60.

[0039] 5, the material specification information 10 is represented by one record in a table 100. That is, the material specification information 10 associates the identification information of a material 60 with the material specifications of the material 60 having that identification information.

[0040] FIG. 6 is a diagram illustrating an example of the physical property information 20 in table format. The table 200 in FIG. 6 has columns of deliverable identification information 202 and physical properties 204. The deliverable identification information 202 indicates identification information of the deliverable 70. The physical properties 204 indicate the physical properties of the deliverable 70. In the table 200, the physical properties of the deliverable 70 are represented by showing, for each physical property, a correspondence between a "label indicating the type of physical property: the physical property amount of that physical property."

[0041] 6, the physical property information 20 is represented by one record in a table 200. That is, the physical property information 20 associates the identification information of the deliverable 70 with the physical property of the deliverable 70 having that identification information.

[0042] The acquisition unit 2020 acquires a plurality of pairs of material specification information 10 and physical property information 20. There are various methods for the acquisition unit 2020 to acquire pairs of material specification information 10 and physical property information 20. For example, pairs of material specification information 10 and physical property information 20 are stored in advance in any storage unit accessible from the recommendation data generation device 2000. The acquisition unit 2020 acquires pairs of material specification information 10 and physical property information 20 by accessing this storage unit. Alternatively, for example, the acquisition unit 2020 may acquire pairs of material specification information 10 and physical property information 20 by accepting user input of pairs of material specification information 10 and physical property information 20. Alternatively, for example, the acquisition unit 2020 may acquire pairs of material specification information 10 and physical property information 20 by receiving pairs of material specification information 10 and physical property information 20 transmitted from another device.

[0043] Here, there are various methods for generating a pair of material specification information 10 and physical property information 20. For example, the pair of material specification information 10 and physical property information 20 is generated by simulating the generation of a product 70. Specifically, by providing specific material specifications as input and running a simulation, physical property information 20 indicating predicted values ​​of the physical properties of each physical property of the product 70 is generated. Then, a pair of the generated physical property information 20 and material specification information 10 indicating the material specifications provided as input is obtained. Here, existing technology can be used to realize a simulation that acquires material specifications as input and, for a material specified by the material specifications, outputs predicted data on the physical properties of a product generated in a specific process using that material.

[0044] Alternatively, for example, a pair of material specification information 10 and physical property information 20 may be generated by actually generating a product 70. Specifically, the product 70 is experimentally generated by using a material 60 represented by specific material specifications in a target process. Furthermore, the physical property information 20 is generated by measuring the physical quantities of each physical property of the generated product 70. As a result, a pair of the generated physical property information 20 and material specification information 10 representing the used material 60 is obtained.

[0045] The physical property information 20 acquired by the acquisition unit 2020 may include information that expresses data in different ways. For example, different labels may be used for essentially identical physical properties. Furthermore, physical quantities of the same physical property may be expressed in different units. In such cases, it is preferable for the acquisition unit 2020 to unify the data expression methods by unifying the labels or converting units. Such a situation in which the data expression methods differ between the physical property information 20 may occur, for example, when acquiring both physical property information 20 generated using a simulation and physical property information 20 generated by actually generating a deliverable 70. It is preferable that such unification of the data expression methods is also performed for the material specification information 10.

[0046] <Generation of Self-Organizing Map 30: S104> The first generator 2040 generates a self-organizing map 30 using each piece of physical property information 20 (S104). The self-organizing map 30 has multiple nodes arranged in an m-dimensional map space (m=2 or m=3). Whether the number of dimensions of the map space is two or three may be determined in advance or may be specified by the user. An n-dimensional physical property vector is assigned to each node of the self-organizing map 30.

[0047] 7 is a diagram illustrating an example of the configuration of a self-organizing map 30 in table format. Table 300 has two columns: nodes 302 and physical property vectors 304. Each record in table 300 indicates that a physical property vector indicated in the physical property vector 304 of that record is assigned to a node identified in the node 302 of that record. Note that in FIG. 7, node 302 indicates the coordinates of the node in map space.

[0048] The allocation of physical property vectors to each node is performed by training the self-organizing map 30. The self-organizing map 30 can be trained by inputting n-dimensional training data to be used for training into the self-organizing map 30. Here, an existing method can be used as a specific method for training the self-organizing map using the training data.

[0049] For example, the first generation unit 2040 initializes the self-organizing map 30 by an arbitrary method. For example, the initialization method may involve initializing the physical property vector of each node to a random value. The first generation unit 2040 obtains a plurality of physical property vectors by obtaining the physical property vector from each of the plurality of pieces of physical property information 20. The first generation unit 2040 generates the self-organizing map 30 by training the self-organizing map 30 by treating each of the plurality of physical property vectors as training data. As a result, the physical property vector corresponding to each node of the self-organizing map 30 becomes n-dimensional data indicating values ​​related to the physical property quantities of n types of physical properties.

[0050] <Identifying Corresponding Nodes: S106> The selection unit 2060 identifies (S106) the corresponding nodes of each piece of physical property information 20. The corresponding node of the physical property information 20 is the node in the self-organizing map 30 that has a physical property vector that is most similar to the physical property vector obtained from the physical property information 20.

[0051] The degree of similarity between physical property vectors can be determined based on, for example, the distance between the physical property vectors. Therefore, for example, the selection unit 2060 performs the following process for each piece of physical property information 20. First, the selection unit 2060 calculates the distance between the physical property vector obtained from the physical property information 20 and the physical property vector of each node in the self-organizing map 30. Then, the selection unit 2060 identifies the node with the smallest calculated distance as the node having the physical property vector most similar to the physical property vector obtained from that physical property information 20. Therefore, the node identified here is identified as the corresponding node of that physical property information 20.

[0052] <Selection of Target Node: S108> The selection unit 2060 selects a target node from among nodes other than the corresponding node based on the arrangement of the corresponding node in the map space (S108). Conceptually, the selection unit 2060 selects a node that is located far away from the corresponding node in the map space as the target node. A specific example of a method for selecting a target node will be described below.

[0053] 8A and 8B are flowcharts illustrating an example of the flow of a process for selecting a target node. Hereinafter, FIGS. 8A and 8B will be collectively referred to as FIG. 8.

[0054] The selection unit 2060 divides the map space into a plurality of partial regions (S202). Each partial region is configured to include a predetermined number of nodes. FIG. 9 is a diagram illustrating a map space 32 divided into a plurality of partial regions. In the map space 32 of FIG. 9, each partial region 36 is a rectangular region surrounded by a thick frame and has four nodes 34 therein. The nodes 34 marked with a circle are corresponding nodes.

[0055] Steps S204 to S212 constitute a loop process L1 that is executed for each of a plurality of partial regions. In step S204, the selection unit 2060 determines whether or not the loop process L1 has been executed for all partial regions. If the loop process L1 has already been executed for all partial regions, the process in FIG. 8 proceeds to step S214. On the other hand, if there are partial regions that have not yet been subjected to the loop process L1, the selection unit 2060 selects one of them. The partial region selected here is referred to as "partial region R." The process in FIG. 8 then proceeds to step S206.

[0056] The selection unit 2060 calculates an evaluation score for the representative point of the partial region R based on the positional relationship between the representative point and each corresponding node (S206). The representative point of the partial region R is, for example, one of the vertices of the partial region R or the center of the partial region R.

[0057] The evaluation score of the representative point is defined, for example, by the following formula (1). Here, S[i][j] represents the evaluation score of the representative point P(i,j) of the partial region R that is located at the i-th position from the top and the j-th position from the left. d_k[i][j] represents the distance between the representative point P(i,j) and the k-th corresponding node. Note that the underscore represents a subscript. n represents the total number of corresponding nodes. In the example of Figure 9, there are 15 corresponding nodes, so n = 15.

[0058] According to formula (1), the representative point located at a location with lower density in the distribution of corresponding nodes in the map space has a smaller evaluation score.

[0059] Here, if the representative point is located at the center of a node, the coordinates of the representative point are expressed by the coordinates of that node. On the other hand, if the representative point is not located at the center of the node, for example, the coordinates of the representative point are determined based on the coordinates of each node adjacent to the representative point. For example, in the example of Figure 9, the representative point of a partial area is assumed to be one vertex of that partial area. In this case, the coordinates of the representative point are calculated as the coordinates of the centers of the four nodes that have the representative point as a vertex (the nodes at the top left, top right, bottom right, and bottom left of the representative point).

[0060] The distance between the representative point and the corresponding node may be expressed as a distance in the map space, or may be expressed as a distance in the high-dimensional space (space of physical property vectors) of the self-organizing map 30. In the latter case, the distance between the representative point and the corresponding node is expressed as the distance between the physical property vectors corresponding to these two points.

[0061] The definition of the evaluation score of the representative point is not limited to the definition of formula 1. For example, the evaluation score of the representative point may be defined by the following formula (2).

[0062] According to equation (2), the farther a representative point is from the nearest corresponding node, the smaller the evaluation score.

[0063] The selection unit 2060 determines whether the evaluation score of the representative point of the partial region R is lower than the evaluation scores of any of the other representative points adjacent to that representative point (S208). If the evaluation score of the representative point of the partial region R is lower than the evaluation scores of any of the other representative points adjacent to that representative point (S208: YES), the selection unit 2060 selects the representative point of the partial region R as a reference point to be used in subsequent evaluations and stores it in the list of reference points Lp (S210).

[0064] On the other hand, if there is a representative point adjacent to the representative point of partial region R whose evaluation score is equal to or lower than the evaluation score of the representative point of partial region R (S208: NO), the processing in Fig. 8 proceeds to S212. In this case, the representative point of partial region R is not selected as the reference point.

[0065] 9, for example, each representative point has eight adjacent representative points, and the representative point having an evaluation score smaller than any of the eight adjacent representative points is selected as the reference point.

[0066] S212 is the end of the loop process L1, so the process in FIG. 8 proceeds to S204.

[0067] When loop process L1 ends, S214 is executed. S214 to S226 constitute loop process L2, which is repeatedly executed until a predetermined termination condition is met. For example, the predetermined termination condition is that loop process L2 is executed a predetermined number of times. Another example of the predetermined termination condition is that the size of the evaluation area, which will be described later, is equal to or smaller than a predetermined threshold. The predetermined number of times and the predetermined threshold that define the termination condition are set in advance, for example, by the user of the recommendation data generation device 2000.

[0068] Steps S216 to S224 constitute a loop process L3 that is executed for each reference point included in the list Lp. In step S216, the selection unit 2060 determines whether or not the loop process L3 has been executed for all reference points included in Lp. If the loop process L3 has already been executed for all reference points, the process in FIG. 8 proceeds to step S226. Since step S226 is the end of the loop process L2, the process in FIG. 8 proceeds to step S214.

[0069] On the other hand, if there are any reference points that have not yet been subjected to the loop process L3, the selection unit 2060 selects one of them. The reference point selected here is referred to as reference point q. Then, the process of FIG. 8 proceeds to S218.

[0070] The selection unit 2060 determines an evaluation area, which is an area in the map space and is centered around the reference point q, and divides the determined evaluation area into multiple partial areas (S218). If the current step S218 is included in the first iteration of the loop process L2, the size of the evaluation area is calculated, for example, by multiplying the size of the map space by a predetermined ratio α that is less than 1. On the other hand, if the current step S218 is included in the second or subsequent iteration of the loop process L2, the size of the evaluation area is calculated, for example, by multiplying the size of the previous evaluation area by a ratio α that is less than 1. In this way, the size of the evaluation area becomes smaller each time the loop process L2 is executed.

[0071] The ratio α may be determined by any method. For example, the ratio α may be set as a fixed value by an administrator of the recommendation data generation device 2000. Alternatively, the ratio α may be specified by a user of the recommendation data generation device 2000. Here, in the repeated execution of the loop process L2, the value of the ratio α may be the same each time, or a different value may be used for each execution. In the latter case, for example, the value of the ratio α is set separately for the first repeated execution of the loop process L2 and for subsequent executions.

[0072] The selection unit 2060 calculates the aforementioned evaluation score for the representative point of each partial area obtained from the evaluation area (S220). The selection unit 2060 replaces the reference point q stored in the list Lp with the representative point with the smallest evaluation score calculated in S220 (S222).

[0073] According to the process of S222, the reference point q included in Lp is replaced with a point with a smaller evaluation score that is included in the evaluation area surrounding the reference point q. As mentioned above, the size of the evaluation area becomes smaller each time the loop process L2 is executed. Therefore, by repeatedly executing the loop process L2, the position of each reference point is repeatedly corrected while gradually narrowing the correction range.

[0074] S224 is the end of the loop process L3, so the process in FIG. 8 proceeds to S216.

[0075] After completing the loop process L3, the selection unit 2060 selects a node corresponding to each reference point in the list Lp as a target node. For example, when the map space is divided into multiple nodes as shown in Figure 9, the node corresponding to the reference point is a node that includes the reference point in that area. Alternatively, for example, the node corresponding to the reference point is a node having coordinates closest to the coordinates of the reference point.

[0076] According to the method shown in Fig. 8, a node with a small evaluation score is searched for in the map space by a so-called grid search. When the evaluation score defined by Equation (1) is used, a node located in a low density area in the distribution of corresponding nodes in the map space is selected as the target node. On the other hand, when the evaluation score defined by Equation (2) is used, a node located far from the nearest corresponding node is selected as the target node.

[0077] Here, the method of selecting the target node is not limited to the method shown in Fig. 8. For example, the selection unit 2060 may select, as the target node, a node corresponding to each reference point included in the list Lp obtained in the loop process L1, without performing the loop processes L2 and L3 in Fig. 8B. Alternatively, for example, the selection unit 2060 may calculate an evaluation score for each node other than the corresponding node, and select, as the target node, a node with the smallest evaluation score or a predetermined number of nodes at the top in ascending order of the evaluation score.

[0078] <Generation of Recommended Data 80: S110> The second generation unit 2080 estimates material specifications corresponding to the target node and generates recommended data 80 indicating the estimated material specifications (S110). To this end, the second generation unit 2080 uses the material specification information 10 to assign multidimensional data (hereinafter, "specification vector") representing values ​​for each of multiple types of parameters (hereinafter, "specification parameters") of the material specifications to each node of the self-organizing map 30. For example, the specification vector is data in which the information indicated in the material specifications 104 in FIG. 5 is expressed as a vector. The second generation unit 2080 then estimates the specification vector assigned to the target node as the material specifications corresponding to the target node.

[0079] A method for assigning a specification vector to each node will now be described.

[0080] First, the second generator 2080 assigns the specification vector obtained from each piece of material specification information 10 to a node corresponding to that material specification information 10. Here, the node corresponding to the material specification information 10 is the corresponding node of the physical property information 20 corresponding to that material specification information 10.

[0081] The specification vector obtained from the material specification information 10 may indicate values ​​for all specification parameters indicated by the material specification information 10, or may indicate values ​​for some of the specification parameters. In other words, if the number of dimensions of the specification vector is k, the value of k may be the same as the number of specification parameters indicated by the material specification information 10, or may be smaller than the number of specification parameters indicated by the material specification information 10.

[0082] For example, suppose that the material specification information 10 indicates both parameters that take continuous values ​​(e.g., the compounding ratio of materials) and parameters that do not take continuous values ​​(e.g., the type of processing). In this case, for example, the specification vector is generated by the parameters that take continuous values.

[0083] It should be noted that which parameters among those indicated by the material specification information 10 are used to generate the specification vector may be predetermined or may be designated by the user. Furthermore, the specification vector may directly indicate the values ​​of the parameters indicated by the material specification information 10, or may indicate values ​​obtained by converting the values ​​of each parameter using a predetermined method (for example, by normalizing or standardizing).

[0084] Furthermore, the second generating unit 2080 estimates the specification vectors to be assigned to nodes that are not associated with material specification information 10. Specifically, the second generating unit 2080 estimates the distribution of specification vectors in the map space based on the specification vectors of each node that is associated with material specification information 10 and the arrangement of those nodes in the map space. Then, the second generating unit 2080 assigns specification vectors to nodes that are not associated with material specification information 10, using the estimated distribution.

[0085] There are various methods for estimating the distribution of the specification vectors in the map space. For example, the second generation unit 2080 estimates the distribution of the specification vectors in the map space from the specification vectors of each node associated with the material specification information 10 and the arrangement of those nodes in the map space, using any interpolation process such as linear interpolation or spline interpolation. Alternatively, for example, the second generation unit 2080 may estimate the distribution of the specification vectors in the map space using sparse estimation. Note that when estimating the distribution of the specification vectors, Bayesian estimation may be further applied to improve the estimation accuracy.

[0086] 10 is a diagram illustrating, in table form, the configuration of a self-organizing map 30 in which a property vector is assigned to each node. The table 300 in FIG. 10 has four columns: a node 302, a physical property vector 304, material identification information 306, and a property vector 308. The table 300 has one record per node.

[0087] A node 302 indicates the coordinates of the node in map space. A physical property vector 304 indicates a physical property vector assigned to the node. For a node to which material property information 10 is assigned, material identification information 306 indicates the identification information of the material 60 indicated by the material property information 10 assigned to that node. For records of nodes to which material property information 10 is not assigned, the material identification information 306 indicates "-". A property vector 308 indicates a property vector assigned to the node.

[0088] <Output of Results> The recommendation data generation device 2000 outputs information representing the processing results (hereinafter, output information) in any format. Hereinafter, the functional component of the recommendation data generation device 2000 that outputs the output information will be referred to as the output unit. For example, the output information indicates, for each of one or more target nodes, recommendation data 80 representing the material specifications corresponding to the target node, in association with the coordinates of the target node.

[0089] For example, the output unit stores the output information in an arbitrary storage unit. Alternatively, for example, the output unit outputs the output information to a display device to display the output information on the display device. Alternatively, for example, the output unit transmits the output information to an arbitrary other device (for example, the simulator described above).

[0090] The recommendation data generating device 2000 may also generate a map image that visually represents the map space of the self-organizing map 30 and include the map image in the output information. The map image is an image that represents the correspondence between the distribution of material specifications and the distribution of physical properties.

[0091] 11 is a diagram illustrating an example of a map image. In FIG. 11, a map image 40 visually represents the map space of the self-organizing map 30. A material specification display 42 representing part or all of the material specification information 10 is superimposed on a node to which material specification information 10 is assigned.

[0092] The nodes in the map image 40 are divided into clusters based on the physical property vectors. The thick frames in the map image 40 represent the boundaries of the clusters. To divide the nodes into clusters, the recommendation data generating device 2000 performs clustering on the physical property vectors corresponding to each node in the self-organizing map 30. By dividing the physical property vectors into clusters in this way, the nodes corresponding to the physical property vectors can also be divided into clusters. For example, the recommendation data generating device 2000 performs clustering of the physical property vectors by using various clustering algorithms such as the k-means method.

[0093] Each node of the map image 40 may be colored based on the physical property vector. Various existing methods can be used to color each node of the self-organizing map according to the data associated with that node.

[0094] <Example of How the Recommended Data 80 is Used> For example, the recommended data 80 is used for simulating the generation of the deliverable 70 or for experimentally generating the deliverable 70. Here, the recommended data 80 may be referenced by the worker performing the simulation or experimental generation, or may be referenced by a simulator performing the simulation. In the former case, the worker performs a simulation by inputting the recommended data 80 into a simulator, or experimentally generates the deliverable 70 by preparing a material 60 identified by the material specifications indicated in the recommended data 80. The latter case will be described below as embodiment 2.

[0095] [Embodiment 2] Fig. 12 is a diagram illustrating an example of an outline of the operation of a recommendation data generating device 2000 according to embodiment 2. Fig. 12 is a diagram for facilitating understanding of the outline of the recommendation data generating device 2000, and the operation of the recommendation data generating device 2000 is not limited to that shown in Fig. 12.

[0096] The recommendation data generation device 2000 inputs the recommendation data 80 into the simulator 400, causing the simulator 400 to execute a simulation. The simulator 400 simulates the production process of the target process. Specifically, the simulator 400 acquires input data representing material specifications and generates prediction data 410 representing predicted physical properties of the deliverable 70 to be produced using a material 60 identified by the input data. Here, the simulator 400 may be any existing simulator that acquires input data representing material specifications, performs a simulation using the input data, and outputs prediction data for the physical properties of the deliverable. The simulator 400 may be implemented by the computer that implements the recommendation data generation device 2000, or by another computer.

[0097] The recommendation data generating device 2000 acquires the prediction data 410 generated by the simulator 400, and outputs a pair of the recommendation data 80 and the prediction data 410. As a result, a pair of the recommendation data 80 and the prediction data 410 obtained by inputting the recommendation data 80 to the simulator 400 is output.

[0098] It can be said that the above pair corresponds to the pair of material specification information 10 and physical property information 20 described above. Therefore, by using the recommendation data generation device 2000, it is possible to improve the efficiency of subsequent simulations based on the results of simulations and experiments that have already been conducted to generate the product 70.

[0099] For example, for the first number of simulations or experimental generation runs, the material specifications to be used for simulation or experimental generation are determined based on past knowledge or randomly. These results are then input into the recommended data generation device 2000, which estimates material specifications that are expected to produce a product with physical properties that are less similar to the physical properties of each product 70 obtained in previous simulations, and outputs the estimated material specifications as recommended data 80. The material specifications indicated in the recommended data 80 are then input into the simulator 400, whereby new pairs of material specifications and physical properties are obtained.

[0100] By using the recommendation data generating device 2000 in this way, it is possible to reduce the degree of similarity between the physical properties of the deliverables 70 obtained in a new simulation and the physical properties of the deliverables 70 obtained in previous simulations, etc. This makes it possible to avoid a situation in which only deliverables 70 with similar physical properties are obtained through simulations, etc., and to increase the variety of physical properties of the deliverables 70 with fewer simulations, etc.

[0101] The pair of recommendation data 80 and prediction data 410 may be output in any manner. For example, the recommendation data generating device 2000 stores the pair in any storage unit accessible from the recommendation data generating device 2000. Alternatively, for example, the recommendation data generating device 2000 displays the pair on any display device controllable by the recommendation data generating device 2000. Alternatively, for example, the recommendation data generating device 2000 transmits the pair to any device communicably connected to the recommendation data generating device 2000.

[0102] The recommendation data generating device 2000 may also use the predicted data 410 as new training data to further train the self-organizing map 30, thereby updating the self-organizing map 30. Furthermore, the recommendation data generating device 2000 may generate a map image 40 for the updated self-organizing map 30 and output it in any format.

[0103] In addition, the recommended data generation device 2000 may cause the simulator 400 to perform a simulation using the recommended data 80 each time it generates the recommended data 80, or may generate multiple pieces of recommended data 80 and then cause the simulator 400 to perform a simulation for each of the multiple pieces of recommended data 80 in sequence.

[0104] <Example of Functional Configuration> Fig. 13 is a block diagram illustrating an example of the functional configuration of a recommendation data generation device 2000 according to the second embodiment. The recommendation data generation device 2000 according to the second embodiment further includes a simulator control unit 2120 and a simulation result output unit 2140. The simulator control unit 2120 inputs the recommendation data 80 to the simulator 400, thereby causing the simulator 400 to execute a simulation of generating a deliverable 70 using a material 60 specified by the material specifications represented by the recommendation data 80. The simulation result output unit 2140 acquires the prediction data 410 output from the simulator 400, and outputs a pair of the recommendation data 80 and the prediction data 410.

[0105] <Example of Hardware Configuration> The hardware configuration of the recommendation data generation device 2000 of the second embodiment is shown in Fig. 3, for example, similar to the recommendation data generation device 2000 of the first embodiment. However, the storage device 1080 of the second embodiment stores a program that realizes each functional component of the recommendation data generation device 2000 of the second embodiment.

[0106] <Processing Flow> FIG. 14 is a flowchart illustrating the processing flow executed by the recommendation data generating device 2000 of the second embodiment. Since steps S102 to S110 are the same as those shown in FIG. 4 , they are not illustrated in FIG. 14 . In S112, the simulator control unit 2120 inputs the recommendation data 80 to the simulator 400 and causes the simulator 400 to execute a simulation. The simulation result output unit 2140 acquires the predicted data 410 from the simulator 400 (S114). The simulation result output unit 2140 outputs a pair of the recommendation data 80 and the predicted data 410 (S116). The first generation unit 2040 updates the self-organizing map 30 through training using the predicted data 410 as training data (S118).

[0107] Although the present invention has been described above with reference to the embodiments, the present invention is not limited to the above embodiments. Various modifications that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

[0108] In the above examples, the program includes instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, computer-readable media or tangible storage media include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technologies, CD-ROM, digital versatile disc (DVD), Blu-ray (registered trademark) disc or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may also be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, transitory computer-readable media or communication media include electrical, optical, acoustic, or other forms of propagated signals.

[0109] Some or all of the above embodiments may be described as, but are not limited to, the following supplementary notes: (Supplementary Note 1) A recommendation data generation device comprising: an acquisition means for acquiring a plurality of pieces of material specification information indicating material specifications, and for each piece of material specification information, acquiring physical property information indicating physical property quantities for each of a plurality of physical properties of a deliverable that can be generated using the material specifications indicated in the material specification information; a first generation means for using the physical property information to generate a self-organizing map in which physical property vectors indicating values ​​for the physical property quantities for each of a plurality of types of physical properties of the deliverable are assigned to each node in a map space; a selection means for selecting one or more target nodes from nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each piece of physical property information; and a second generation means for generating recommendation data indicating the material specifications corresponding to the target nodes. (Supplementary Note 2) The recommendation data generation device according to Supplementary Note 1, wherein the selection means calculates, for each of a plurality of positions on the map space, an evaluation score based on the distance between the position and the node corresponding to each of the physical property information, selects one or more of the plurality of positions based on the evaluation score, and selects the node corresponding to the selected position as the target node. (Supplementary Note 3) The recommendation data generation device according to Supplementary Note 2, wherein the evaluation score is the sum of reciprocals of the distances between the position and the nodes corresponding to each of the physical property information, or the reciprocal of the minimum value of the distances between the position and the nodes corresponding to each of the physical property information. (Supplementary Note 4) The recommendation data generation device according to any one of Supplements 1 to 3, wherein the second generation means assigns, to each of the nodes, a specification vector indicating a value related to a material specification, using each of the material specification information, and generates the recommendation data indicating the specification vector assigned to the target node.(Supplementary Note 5) A recommended data generation device as described in any one of Supplementary Notes 1 to 3, comprising: simulator control means for causing a simulator that generates predicted data of physical properties of the product that can be generated using a material specified by input material specifications to execute a simulation using the material specifications represented by the recommended data as input; and simulation result output means for acquiring the predicted data generated by the simulator and outputting the predicted data and the recommended data. (Supplementary Note 6) A recommended data generation method executed by a computer, comprising: an acquisition step of acquiring a plurality of pieces of material parameter information indicating material parameters and, for each piece of material parameter information, acquiring physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated using the material parameters indicated in the material parameter information; a first generation step of using the physical property information to generate a self-organizing map in which physical property vectors indicating values ​​for the physical property quantities for each of a plurality of types of physical properties of the product are assigned to each node in a map space; a selection step of selecting one or more target nodes from nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each piece of physical property information; and a second generation step of generating recommended data indicating the material parameters corresponding to the target nodes. (Supplementary Note 7) The recommended data generation method according to Supplementary Note 6, wherein in the selection step, for each of a plurality of positions in the map space, an evaluation score is calculated based on the distance between the position and the node corresponding to each piece of physical property information, one or more of the plurality of positions is selected based on the evaluation score, and the node corresponding to the selected position is selected as the target node. (Appendix 8) The recommended data generation method described in Appendix 7, wherein the evaluation score is the sum of the reciprocals of the distances between the position and the nodes corresponding to each of the physical property information, or the reciprocal of the minimum value of the distances between the position and the nodes corresponding to each of the physical property information.(Supplementary Note 9) The recommended data generation method according to any one of Supplements 6 to 8, wherein in the second generation step, a parameter vector indicating a value related to a material parameter is assigned to each of the nodes using each of the material parameter information, and the recommended data indicating the parameter vector assigned to the target node is generated. (Supplementary Note 10) The recommended data generation method according to any one of Supplements 6 to 8, comprising: a simulator control step of causing a simulator to generate prediction data of physical properties of the deliverable that can be generated using a material specified by the input material parameter, and executing a simulation using the material parameter represented by the recommendation data as input; and a simulation result output step of acquiring the prediction data generated by the simulator and outputting the prediction data and the recommendation data. (Supplementary Note 11) A non-transitory computer-readable medium storing a program that causes a computer to execute the following steps: an acquisition step of acquiring a plurality of pieces of material property information indicating material properties, and acquiring, for each of the pieces of material property information, physical property information indicating physical property quantities for each of a plurality of physical properties of a deliverable that can be generated with the material properties indicated in the material property information; a first generation step of using the physical property information to generate a self-organizing map in which physical property vectors indicating values ​​for the physical property quantities for each of a plurality of types of physical properties of the deliverable are assigned to each node in a map space; a selection step of selecting one or more target nodes from nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each of the physical property information; and a second generation step of generating recommendation data indicating the material properties corresponding to the target nodes. (Supplementary Note 12) A computer-readable medium according to Supplementary Note 11, wherein in the selection step, for each of a plurality of positions on the map space, an evaluation score is calculated based on the distance between the position and the node corresponding to each of the physical property information, one or more of the plurality of positions is selected based on the evaluation score, and the node corresponding to the selected position is selected as the target node.(Supplementary Note 13) The computer-readable medium according to Supplementary Note 12, wherein the evaluation score is a sum of reciprocals of distances between the position and the nodes corresponding to each of the physical property information, or a reciprocal of a minimum value of distances between the position and the nodes corresponding to each of the physical property information. (Supplementary Note 14) The computer-readable medium according to any one of Supplements 11 to 13, wherein in the second generation step, a property vector indicating a value related to a material property is assigned to each of the nodes using each of the material property information, and the recommendation data indicating the property vector assigned to the target node is generated. (Supplementary Note 15) The computer-readable medium according to any one of Supplements 11 to 13, comprising: a simulator control step of causing a simulator to generate prediction data of physical properties of the deliverable that can be generated using a material specified by the input material property, to execute a simulation using the material property represented by the recommendation data as input; and a simulation result output step of acquiring the prediction data generated by the simulator and outputting the prediction data and the recommendation data.

[0110] REFERENCE SIGNS LIST 10 Material specification information 20 Physical property information 30 Self-organizing map 32 Map space 34 Node 36 Partial region 40 Map image 42 Material specification display 60 Material 70 Product 80 Recommended data 100 Table 102 Material identification information 104 Material specifications 200 Table 202 Product identification information 204 Physical properties 300 Table 302 Node 304 Physical property vector 306 Material identification information 308 Specifications vector 400 Simulator 410 Prediction data 1000 Computer 1020 Bus 1040 Processor 1060 Memory 1080 Storage device 1100 Input / output interface 1120 Network interface 2000 Recommended data generation device 2020 Acquisition unit 2040 First generation unit 2060 Selection unit 2080 Second generation unit 2120 Simulator control unit 2140 Simulation result output unit

Claims

1. an acquisition means for acquiring a plurality of pieces of material specification information indicating material specifications, and acquiring, for each of the pieces of material specification information, physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated from the material specifications indicated in the material specification information; a first generating means for generating a self-organizing map in which a physical property vector indicating a value related to each of a plurality of types of physical properties of the deliverable is assigned to each node in a map space, using the physical property information; a selection means for selecting one or more target nodes from the nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each of the physical property information; and a second generating means for generating recommendation data indicating the material specifications corresponding to the target node.

2. The recommendation data generation device of claim 1, wherein the selection means calculates an evaluation score for each of a plurality of positions on the map space based on the distance between the position and the node corresponding to each of the physical property information, selects one or more of the plurality of positions based on the evaluation score, and selects the node corresponding to the selected position as the target node.

3. The recommendation data generation device of claim 2, wherein the evaluation score is the sum of the inverses of the distances between the position and the nodes corresponding to each of the physical property information, or the inverse of the minimum value of the distances between the position and the nodes corresponding to each of the physical property information.

4. The second generating means is Using each of the material property information, a property vector indicating a value related to the material property is assigned to each of the nodes; The recommendation data generating device according to claim 1 , further comprising: a device configured to generate the recommendation data indicating the specification vector assigned to the target node.

5. a simulator control means for causing a simulator that generates prediction data of physical properties of the product that can be generated using a material specified by the input material specifications to execute a simulation using the material specifications represented by the recommended data as input; 4. The recommended data generating device according to claim 1, further comprising: a simulation result output unit that acquires the predicted data generated by the simulator and outputs the predicted data and the recommended data.

6. an acquisition step of acquiring a plurality of pieces of material specification information indicating material specifications, and acquiring, for each of the pieces of material specification information, physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated from the material specifications indicated in the material specification information; a first generation step of generating a self-organizing map in which a physical property vector indicating a value related to each of a plurality of types of physical properties of the deliverable is assigned to each node in a map space using the physical property information; a selection step of selecting one or more target nodes from the nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each of the physical property information; and a second generating step of generating recommendation data indicative of the material specifications corresponding to the target node.

7. The recommendation data generation method of claim 6, wherein in the selection step, an evaluation score is calculated for each of a plurality of positions on the map space based on the distance between the position and the node corresponding to each of the physical property information, one or more of the plurality of positions is selected based on the evaluation score, and the node corresponding to the selected position is selected as the target node.

8. The recommended data generation method described in claim 7, wherein the evaluation score is the sum of the inverses of the distances between the position and the nodes corresponding to each of the physical property information, or the inverse of the minimum value of the distances between the position and the nodes corresponding to each of the physical property information.

9. an acquisition step of acquiring a plurality of pieces of material specification information indicating material specifications, and acquiring, for each of the pieces of material specification information, physical property information indicating physical property quantities for each of a plurality of physical properties of a product that can be generated from the material specifications indicated in the material specification information; a first generation step of generating a self-organizing map in which a physical property vector indicating a value related to each of a plurality of types of physical properties of the deliverable is assigned to each node in a map space using the physical property information; a selection step of selecting one or more target nodes from the nodes of the self-organizing map based on an arrangement in the map space of the nodes corresponding to each of the physical property information; A second generation step of generating recommendation data indicating the material specifications corresponding to the target node.

10. The program according to claim 9, wherein in the selection step, an evaluation score is calculated for each of a plurality of positions on the map space based on the distance between the position and the node corresponding to each of the physical property information, one or more of the plurality of positions is selected based on the evaluation score, and the node corresponding to the selected position is selected as the target node.