JP2023036000A - Combination determination support system, combination determination support method, and combination determination support program - Google Patents

Abstract

To aggregate combinations composed of multiple options on a user-by-user basis, and determine an optimum combination by reflecting user's preferences including predetermined restrictions on the combination.SOLUTION: A server device S includes: an acquisition unit configured to acquire combination information for each user modeled based on a combination; a storage unit configured to store the combination information; a setting unit configured to set, based on the stored combination information, constraint information that contributes to combination constraint satisfaction and an objective function that contributes to optimization; and a calculation unit configured to calculate desired combination information based on the stored combination information and also the constraint information and the objective function having been set. The combination information includes variable information corresponding to multiple options and/or mutation information associated with the variable information. Each piece of variable information, each piece of mutation information, and at least one of the variable information and the mutation information are accompanied by dependency constraint information indicating their mutual relationship. The variable information and/or mutation information includes attribute value information associated with each.SELECTED DRAWING: Figure 1

Description

There is an application for exception to loss of novelty

The present invention provides, for example, a combination decision support system, a combination decision support method, and a combination decision support that support the determination of a desired combination from a plurality of options that constitute the specifications of products designed, developed, produced, and manufactured in the manufacturing industry. Regarding the program.

Conventionally, in the field of manufacturing, for example, computers are used to design, develop, produce, and manufacture products, as typified by tools (software) such as CAD (Computer Aided Design) and CAE (Computer Aided Engineering). (hereinafter also referred to as “manufacturing, etc.”). As a technology related to such a tool, for example, Japanese Patent Laid-Open No. 2002-200001 discloses a technique for planning manufacturing of a product or the like.

Specifically, a plurality of plan candidates are drafted based on specification information on a plurality of elements that specify the specifications of the plan and constraint information that defines constraints for each of the plurality of elements, A planning device that generates plan data from the contents of each of a plurality of proposed plan candidates, a plan evaluation device that evaluates each plan data and generates a plurality of evaluation data indicating evaluation results for each plan data, and planning An evaluation input device that transmits each plan data generated by the device to a user terminal and receives user data indicating the user's evaluation result for each plan data from the user terminal, and at least each user by receiving the evaluation input device and an evaluation learning device that learns data as learning data and constructs an evaluation learner for each plan candidate from the learning result.

Japanese Patent Application Laid-Open No. 2020-201611

However, in the technique of Patent Document 1, constraints for each user (for example, delivery date, quantity, continuous amount) is not fully reflected in planning. Therefore, even if the prediction value and prediction reliability are highly evaluated, it cannot be said that it is an optimal combination of multiple elements if the plan does not consider the constraints depending on the user.

For products manufactured in a general manufacturing industry, the planning department selects the design, color, function, etc. to determine the product to be developed, and the design department selects materials, parts, application technology, etc. In order to optimize the combination of options for specifying product specifications, each department needs to Constraints that can only be known (per user) should be reflected.

In addition to products in the manufacturing industry, there are a wide variety of constraints and optimization problems for combinations consisting of multiple options in the world, and for users who encounter such problems, constraints and combinations regarding options and user convenience It is not easy to accurately reflect the objective function for the optimization of . Therefore, the inventor has arrived at the need for improvement of usability in tasks such as generating combinations, adding constraints for each user, and setting objective functions after a lot of ingenuity.

Therefore, the object of the present invention is to provide combination determination support that aggregates combinations composed of a plurality of options on a user-by-user basis, and can determine the optimum combination by reflecting the user's wishes, including predetermined restrictions, on the above combinations. An object of the present invention is to provide a system, a combination decision support method, and a combination decision support program.

That is, the present invention is a computer system that supports the determination of a combination composed of a plurality of options, an acquisition unit that acquires combination information for each user modeled based on the combination, and the acquired combination a storage unit for storing information; a setting unit for setting constraint information contributing to constraint satisfaction of the combination and an objective function contributing to optimization based on the stored combination information; the stored combination information and the set a calculation unit that calculates desired combination information based on the constraint information and the objective function, wherein the combination information includes variable information corresponding to a plurality of options and/or mutation information related to the variable information, Each of the variable information, each of the mutation information, and at least one of the variable information and the mutation information are accompanied by dependency constraint information indicating mutual relationships, and the variable information and/or the mutation information are each A combination decision support system characterized by including associated attribute value information.

According to such a configuration, all of the combination information in which the dependency constraints between the multiple options are taken into account by modeling are aggregated for each user, and the constraint information and the objective function of the predetermined user among all the combinations are further set. As a result, it is possible to calculate the desired combination information for the user, so that the efficiency of narrowing down, extracting, and determining combinations can be expected without relying on individual experience, intuition, and trial and error. In other words, since the modeled combination information is accompanied by dependency constraint information that is added by the user, it is not necessary to coordinate or agree on dependency constraints between users or individuals (persons in charge, etc.) who are subordinate to each user. Therefore, it is possible to expect the effect of improving the calculation accuracy of the optimum combination information. Therefore, especially for product development such as the manufacturing industry, where there are a huge number of choices, combination restrictions, and users from upstream processes (planning, design, etc.) to downstream processes (production, distribution, etc.), the present invention If the above effect can be obtained, productivity can be improved.

In the following, in order to supplement the content of the present invention, the meanings and examples of words expressing this are listed below.

A "computer system" is composed of an information processing terminal such as a personal computer managed for each user communicating with each other via a network line such as the Internet, and a server device managed at a predetermined location. It may be a physical server, a cloud server, or a compound server that integrates these, and instead of the server device, it may be composed of information processing terminals that function as server devices, and the server device operated via the information processing terminal ( computer) may be composed of a single unit, or may be composed of a quantum computer. The information processing terminal and the server device are not limited in terms of specifications as long as they perform well-known techniques such as arithmetic processing, storage, and transmission/reception of electronic information.

"Multiple options" means, for example, products manufactured in the manufacturing industry, parts that make up the products, specifications of products and parts (hereinafter also referred to as "parts, etc.") (for example, shape, design, color, structure, dimensions, etc.) ingredients, precision, function, performance, grade, applied technology, line manufacturing method, distribution method, test method), but elements that realize a predetermined combination (for example, items related to clothing, food and drink, housing, race, Things related to people such as compatibility and nationality, and things related to concepts such as money, time, and value) are not limited.

In addition, "multiple options" refers to common identification information for parts, etc. given for management of parts, etc., non-common identification information for parts, etc. given for management of parts, etc., flow until parts etc. are distributed to the market All or part of (hereinafter also referred to as "flow"), flow or flow given for management of items adopted in flow (for example, parts, software packages, drawings, raw materials, materials, consumables, jigs) Common identification information, flow non-common identification information that is combined with flow common identification information and given for management of flows or items adopted in these, non-identification information such as parts and / or in addition to flow non-identification information Order information indicating a predetermined order of non-common identification additional information, non-identification information such as parts, and/or non-identification information for flows (for example, information regarding instructions for predetermined execution, change, etc.) and/or Non-common identification identification information identified by time-series information indicating year, month, day, hour, minute, and second, common identification information such as specifications and parts, classification of parts categorized by common identification information of flows, and classification of flows may be used. The parts common identification information may correspond to the classification of the parts and the like, and the flow common identification information may correspond to the flow classification.

Parts, etc. may be hardware, software, or electronic data composed of predetermined character strings. In the case of hardware, materials obtained from predetermined raw materials, members obtained by forming the above materials, and processing the above members. It may be a finished product obtained by assembling or assembling.

A flow is also to change the state of a given item to another state. Criteria) (hereinafter also referred to as "process"), procedures necessary for the establishment of the manufacturing process (hereinafter also referred to as "steps"), actions necessary for the establishment of the above procedures (hereinafter also referred to as "tasks". ).

Common identification information for parts, etc. is, for example, a function number, a standard number, a measurement unit, and a regulation associated with the function, and may be a name or specification of a part, or a number or symbol obtained by converting these into character strings.

The non-common identification information for parts, etc. is, for example, part numbers and drawing numbers, and may be combined with the common identification information for parts, etc. to enable identification of products and parts. can also be a number or symbol that is converted to a string.

The flow common identification information is, for example, the standard process number that identifies the standard process (hereinafter also referred to as "standard process"), the standard number, measurement unit, and regulation associated with the process, and the names of the processes, steps, and tasks. , specifications, or numbers and symbols obtained by stringifying these.

The flow non-common identification information is, for example, an applied process number that identifies a process adjusted to be applied to the standard process (hereinafter also referred to as "applied process"), and is combined with the flow non-common identification information to It may be possible to identify all or part of the flow until it is distributed to the market, or the items adopted in these, the name and specification of the process, step, task, or the number that converts these into strings It can be a symbol.

In other words, the parts common identification information and the flow common identification information are information that does not change regardless of predetermined conditions (external environment to be identified, for example, conditions related to design and manufacturing such as temperature and location). The information and flow non-common identification information is information that changes according to the above conditions.

The non-common identification additional information is, for example, information (hereinafter also referred to as "route information") related to the setting of routes (hereinafter also referred to as "routes") by areas, groups, etc. that target parts and flows. It may be the name or specification of or a number or symbol that converts these into character strings, and information (hereinafter also referred to as "choice information") regarding selection (hereinafter referred to as "choice") for parts, etc. and flows It may be the name of the choice or a number or symbol obtained by converting this into a character string.

Non-common identification specific information means information (hereinafter also referred to as "instruction information") related to instructions such as execution and change for parts, etc., flows, the above paths, and the above selections (hereinafter also referred to as "instructions"), For example, it may be a design change number, an approval number, or an instruction number, or it may be the name of an instruction or a number or symbol obtained by converting this into a character string.

The route information/choice information/instruction information may correspond to the flow, or may correspond to the classification of the flow information. An item may be employed in any one or more of a manufacturing process, procedure, action, instruction, route, choice.

A product or a part may be selected as a combination of common identification information for parts or common identification information for parts and non-common identification information for parts. Flows may be selected as flow common identities or combinations of flow common identities and flow non-common identities.

The "plurality of options" may be converted into the variable information and/or the mutation information, and have a predetermined relationship with the variable information and/or the mutation information by dependency constraint information indicating mutual configurations. good too. That is, the combination information includes the above variable information, the above mutation information, common identification information such as parts included in a plurality of options, non-common identification information such as parts, common flow identification information, non-common flow identification information, non-common additional identification information, It may be composed of one or more of the non-common identification information, according to which, the restrictions caused by the combination information and the variation of the search range can be increased, and the non-common identification information can be combined with the non-common identification information. Since it is possible to add information and perform modeling, an effect of obtaining desired combination information with higher accuracy (accuracy) can be expected.

"Combination" means any relationship between at least two or more of a plurality of options, for example, products and products, parts and parts, products and parts , between things in a subordinate relationship (or selection relationship) such as parts and specifications of parts, between things in a relationship between a predetermined superordinate concept and a subordinate concept, relationships in which one is essential to the other or many other relationships (hereinafter "essential relationship"), requirements (hereinafter also referred to as "requirements"), mutual exclusion (hereinafter also referred to as "exclusion"), one or two or more (hereinafter also referred to as "affiliation relationship"), one party owns one other (hereinafter also referred to as "ownership relationship"), one party shares one or more of the other relationship (hereinafter also referred to as "shared relationship") and relationship in which one is derived from the other (hereinafter also referred to as "derivative relationship").

“Modeling” refers to creating a diagram of a combination of multiple options using prescribed figures, lines, symbols, etc., and listing the types (items), names, attribute values, etc. of options. and may be performed using a predetermined tool that can be operated by the user's information processing terminal.

“User” means a corporation, individual, or other predetermined organization that considers, creates, selects, or decides on one or more combinations of multiple options, such as a finished product manufacturer in the manufacturing industry, Parts makers, OEM (Original Equipment Manufacturing) makers, suppliers of parts, materials, and software (so-called suppliers, vendors), software houses, distribution companies, etc., and other related companies, consisting of two or more of these companies. It may be a group, and may be added/changed/deleted according to the configuration of selective vision elements and combinations. "Each user" refers to each user of two or more companies, including groups, which may or may not be in a relationship between the ordering side and the receiving side, and may or may not be in a capital relationship. and joint development agreements.

“Constraint satisfaction of combinations” means narrowing down the combinations to be extracted in order for the user to obtain a desired combination on the premise of the dependency constraints of the combinations. “Combination optimization” means narrowing down the combinations to be extracted so that the user can obtain a desired combination on the premise that the constraints on the combinations are satisfied. The number of combinations that satisfy the constraint or the number of optimal combinations may be one, two or more, or any number desired by the user.

A "dependency constraint" indicates at least one of a selection relationship, an essential relationship, a requirement relationship, an exclusion relationship, an belonging relationship, an ownership relationship, a sharing relationship, and a derivation relationship that realizes a predetermined combination.

The "attribute value" is a numerical value related to each item such as price, evaluation, size, weight, temperature, speed, and area.

"Combination information", "constraint information", "objective function", "variable information", "mutation information", "dependent constraint information", and "attribute value information" can be sent, received, stored, browsed, added, It is electronic information for which various processes such as editing and calculation are performed by an information processing terminal device or a server device, and the electronic information may include characters, moving images, still images, and sounds.

Preferred inventions included in the present invention are listed below.

In the present invention, the constraint information is set using the attribute value information or the calculated value of the attribute value information as a threshold, and the objective function is set as the maximum, minimum, or extreme value of the calculated value of the attribute value information. may

According to such a configuration, since the number of constraint information and objective functions to be set is finite, it is expected that the accuracy of the combinations extracted by the user to obtain a desired combination is increased and that the combinations can be easily narrowed down.

Specifically, the constraint information must be within a predetermined numerical range using the attribute value information (individual value) or the calculated value of the attribute value information as a threshold, must be greater than or equal to a predetermined numerical value, or must be less than or equal to the predetermined numerical value. , and the objective function may be set as any of all three ways that the calculated value of the attribute value information is the maximum, minimum, or extreme value. One or two or more of the constraint information and the objective function may be set according to each attribute value item, and if there are two or more objective functions, there may be an order of priority. A "calculated value" is, for example, the sum of a plurality of pieces of attribute value information or the total value of any two or more pieces of attribute value information.

The present invention may further include a determination unit that determines whether or not the user can access the stored combination information. Further, in the present invention, the storage unit stores a first combination information that accepts access only from a single or specified user, and a second combination information that accepts access from all users. The setting unit may encrypt and set the constraint information and the objective function, and decrypt the desired combination information calculated by the calculation unit.

According to such a configuration, according to the configuration of the computer system, the access authority of other users to the combination information for each user is logically set by the determination unit, or is set by the first storage unit and the second storage unit. Since it can be set physically, it is possible to generate combination information according to the disclosure or non-disclosure of dependency constraint information for each user, so it is expected that the user can easily manage the reason and process of generating combination information as a history. can. In other words, the combination information with the dependency constraint information that can be disclosed and the combination information with the dependency constraint information that cannot be disclosed can be separated and managed in association with each other, so that the addition, change, and deletion of the combination information can be easily performed. You can expect results.

The present invention uses a recognition unit that recognizes the calculated desired combination information as an image, and a trained first model that is machine-learned using the combination image information recognized as the image as teacher data, and uses the combined image information. and an analysis unit that analyzes the feature amount from the above, and the constraint information may be set in consideration of the analyzed feature amount.

According to such a configuration, it is possible to expect an effect of avoiding redundancy in the calculated number of pieces of desired combination information due to the existence of individual options with high relevance. In other words, by adding constraint information that bundles options with high relevance, the number of desired combination information to be calculated is minimized, facilitating optimization.

In the present invention, the calculation unit performs machine learning using the combination information stored in the past, the set constraint information and the objective function, and the desired combination information calculated in the past as teacher data. A second model may be used to calculate new desired combination information estimated from the newly stored combination information.

According to such a configuration, the more the learned second model is machine-learned using the teacher data, the faster and more accurate the calculation of the combination information desired by the user can be expected.

“Teacher data” is learning data in the first trained model or the second trained model, and corresponds to supervised learning using a data set combining predetermined input data and predetermined output data. However, it may be one corresponding to unsupervised learning using only the above input data, and may be character information or image information.

The “learned first model” and the “learned second model” are defined as, when predetermined input data is input, predetermined output data (only the feature amount may be used) from predetermined feature amounts included in the input data. ) and is a model for which machine learning using teacher data has been completed, but the accuracy of the program may be improved by learning continuously or as needed using the teacher data. The analysis method employed in the "learned first model" and/or the "learned second model" may be known non-hierarchical clustering or hierarchical clustering.

The “learned first model” and the “learned second model” are linear models such as SVM (Support Vector Machine), CNN (Convolutional Neural Network) and BNN (Bayesian Neural Network). It is generated by a learning method such as deep learning using a neural network or multi-layer neural network, and a predetermined device or part (model generation unit) that is the subject of execution of the above learning method is included in the same computer system. It doesn't matter if it is or not. If the model generator is not included in the computer system, the trained first model and the trained second model may be provided to the computer system via a predetermined network line.

Moreover, since the following are inventions relating to methods and programs that differ from the present invention only in categories, the meanings and examples of words expressing the effects and inventions of the invention will be omitted.

A combination decision support method executed by a computer system corresponding to the present invention acquires combination information for each user modeled based on a combination composed of a plurality of options, stores the acquired combination information, setting constraint information that contributes to constraint satisfaction of the combination and an objective function that contributes to optimization, and calculating desired combination information based on the stored combination information, the set constraint information, and the objective function; The combination information includes variable information corresponding to a plurality of options and/or mutation information related to the variable information, each of the variable information, each of the mutation information, and at least one of the variable information and the mutation information The variable information and/or the mutation information include attribute value information associated with each of the variable information and/or the mutation information.

A computer-readable program executed as a computer system corresponding to the present invention obtains combination information for each user modeled based on a combination consisting of a plurality of options, and stores the obtained combination information. set constraint information that contributes to constraint satisfaction of the combination and an objective function that contributes to optimization, and calculate desired combination information based on the stored combination information, the set constraint information, and the objective function. , the combination information includes variable information corresponding to a plurality of options and/or mutation information related to the variable information, each of the variable information, each of the mutation information, and at least the variable information and the mutation information Either one is accompanied by dependency constraint information indicating a mutual relationship, and the variable information and/or the mutation information include attribute value information related to each.

According to the present invention, it is possible to expect an effect of being able to aggregate combinations composed of a plurality of options on a user-by-user basis, and to determine the optimum combination by reflecting the user's wishes, including predetermined restrictions, on the above combinations.

1 is a configuration diagram of a computer system in one embodiment of the present invention; FIG. It is a figure which shows an example of the software screen performed by the said system. It is a figure which shows an example of the software screen performed by the said system. It is a figure which shows an example of the software screen performed by the said system. It is a figure which shows another example about a part of said software screen. It is a figure which shows an example of the software screen performed by the said system. It is a figure which shows an example of the modeled combination information handled by the said system. It is a figure which shows an example of the modeled combination information handled by the said system. It is a flow chart which shows an example of execution processing by the above-mentioned system. It is a flow chart which shows an example of execution processing by the above-mentioned system. FIG. 4 is a configuration diagram of another computer system in one embodiment of the present invention; FIG. 4 is a configuration diagram of yet another computer system in one embodiment of the present invention; It is a figure which shows the outline|summary of the execution process by the said system. It is a figure which shows an example of the software screen performed by the said system. It is a figure which shows an example of the software screen performed by the said system.

An example of a combination determination support system according to an embodiment of the present invention will be described below with reference to FIGS. 1 to 9. FIG.

<Outline of combination decision support system>
The combination decision support system supports users involved in product development in determining the optimum combination among the combinations consisting of multiple options necessary for product development in the manufacturing industry. A plurality of options in product development are, for example, parts and specifications of parts. (any of the specifications) must be required or required, or mutually exclusive (either the part or the specification of the part). It may have one or more relationships depending. The users are, for example, manufacturers of finished products and OEMs who are on the ordering side, and suppliers (so-called suppliers and vendors) of parts, materials, and software, software houses, and distribution companies on the receiving side.

<Basic hardware and software constituting the combination decision support system>
The combination decision support system is a computer system composed of a computer or two or more computers, and electronic information {for example, OS (Operating System), middleware, firmware, software such as applications, programs for executing these, other characters, A microprocessor including a CPU (Central Processing Unit) that performs arithmetic processing of static images, moving images, and sounds and commands to each hardware, a ROM (Read Only Memory) such as a hard disk or SSD that stores electronic information, and a CPU RAM (random access memory) that temporarily stores the above electronic information under the control of , communication devices such as RF chips, baseband chips, and other communication modules, keyboards, mice, touch panels, microphones, cameras, voice recorders, sensors, etc. Hardware such as input devices, output devices such as displays, printers, speakers, and headphones, power devices such as batteries, and operating devices such as motors may be combined as appropriate. They may be interconnected via an output interface.

The computer is, for example, a server device or an information processing terminal such as a personal computer, a smart phone, or a tablet. The type, number, and size of hardware and software installed in the computer may be appropriately determined according to the application and specifications, and may be configured as a quantum computer. Each of the two or more computers may be connected to each other via a communication network from an input/output interface or communication device, or may be connected in other manners. The communication network is, for example, the Internet, an intranet, an extranet, a LAN, a CATV communication network, a VPN, a telephone line, a mobile communication network, or a satellite communication network. The transmission media that make up the communication network are wired such as IEEE1394, power line carriers, telephone lines, etc., and wireless such as IrDA, Bluetooth (registered trademark), IEEE802.11 (wifi, etc.), mobile phone networks, satellite circuits, and terrestrial digital networks. It's okay. Each of the computers may send and receive electronic information to and from each other via the communication network from the communication device.

<Relationship between combination decision support system and hardware>
As shown in FIG. 1, the combination determination support system includes a server device S and an information processing terminal C. The server device S and the information processing terminal C communicate with each other via a communication network. The server device S includes a control unit 1 equivalent to a microprocessor, a storage unit 2 equivalent to a ROM, and an attachment equivalent to a communication device for transmitting electronic information to and receiving electronic information from the information processing terminal C. However, it may be provided with an output unit corresponding to an output device and an input unit corresponding to an input device. The information processing terminal C, although not shown, corresponds to a control unit corresponding to a microprocessor, a storage unit corresponding to a ROM, and a communication device for transmitting electronic information to and receiving electronic information from the server device S. It has a communication section, an output section corresponding to an output device, and an input section corresponding to an input device. Note that the server device S and the information processing terminal C may have all the parts and functions of a known computer that handles electronic information.

<Combination modeling>
The combination decision support system is modeled by an application (hereinafter also referred to as an "OVM modeler") that describes a combination composed of a plurality of options using a known OVM (Orthogonal Variability Model). "OVM" is a method for modeling variability information, and the model is composed of variable points that indicate the target of mutation (hereinafter also referred to as "variation point" or "VP" (abbreviation of variation point). )}, a variant (hereinafter also referred to as “variant” or “V” (abbreviation of Variant)) indicating the content of the variation (a specific instance of the variable point), and their dependencies.

That is, the combination decision support system uses OVM to represent multiple options necessary for product development by variation points and variants, and to represent selection relationships, essential relationships, requirement relationships, and exclusive relationships between multiple options as dependency relationships. to model a combination of options. Specifically, in the combination decision support system, when the server device S is equipped with an OVM modeler, the information terminal device C uses the OVM modeler as cloud software and sends electronic information to the server device according to the operation input from the user. Send to S. On the other hand, when the information terminal device C is equipped with the OVM modeler, the information terminal device C with the OVM modeler installed transmits electronic information to the server device S according to the operation input from the user.

The combination information modeled by the OVM modeler includes variable information corresponding to the variation point and mutation information corresponding to the variant as multiple options, and each of the two or more variable information, two Each of the above mutation information and at least one of the set of the variable information and the mutation information is accompanied by dependency constraint information indicating a mutual relationship corresponding to a dependency relationship, and the variable information and/or the mutation information are each Contains relevant attribute value information.

2 and 3 show an example of the OVM modeler screen. As shown in FIG. 2, by the user's drag operation, lines and arrows indicating the relationship between "VP", "V", "VP" and "V" are drawn, and the corresponding "VP" and "V", " VP' and 'VP' are connected by the above-mentioned line or the above-mentioned arrow, and given the predetermined dependency constraint information, the user's combination information is modeled based on the combination of a plurality of options. Further, by the user's click operation, a "table view" listing the combination information is expanded, the combination information is set to be "disclosed to other users", or "saved". Then, when the "table view" is expanded, as shown in Fig. 3, the attribute values of "V" (for example, "price" and "evaluation") are set by the user's input operations such as key touches and voice. Thus, predetermined attribute value information is added to the combination information.

That is, the OVM modeler, for example, the supplier who accepts an order from an OEM manufacturer can select the types of parts M (variable information) as parts M1, M2, M3, M4, and M5 (mutation information), and parts N (variable information). Parts N1, N2, N3 (mutation information) and their dependencies (dependency constraint information) as well as the price/evaluation (attribute value information) associated with each of the mutation information are given to model the combination information . Similarly, another supplier who accepts an order from an OEM manufacturer also gives variable information, mutation information, and dependency constraint information, as well as attribute value information related to mutation information, to model combination information. Thereby, the combination information for each user is generated.

The combination information is stored (saved) in the storage unit 2 of the server device S in real time, but may or may not be saved in a file format. That is, the combination information is saved via the OVM modeler, and the saved combination information is called via the OVM modeler. Note that the combination information may be output in a predetermined file format by a known export function.

<Details of server device S>
The control unit 1 included in the server device S includes an acquisition unit 11 that acquires combination information for each user modeled by the OVM modeler, and contributes to satisfying combination constraints based on the combination information stored in the storage unit 2. A setting unit 12 for setting constraint information and an objective function contributing to optimization, and optimal combination information based on the combination information stored in the storage unit 2 and the constraint information and the objective function set in the setting unit 12. A calculation unit 13 for calculating and a determination unit 14 for determining whether or not a user can access the combination information stored in the storage unit 2 are provided.

<Acquisition unit>
The acquisition unit 11 performs processing of acquiring the combination information for each user transmitted from the information terminal device C via the communication unit, but may perform processing of providing the combination information to the storage unit 2 . The timing of each process by the acquisition unit 11 is after a predetermined time (for example, 1 to 3 seconds) has elapsed before the information terminal device C recognizes the operation input from the user via the OVM modeler, but when the predetermined operation input from the user is received. It may be immediately after recognition.

<Setting section>
The setting unit 12 stores data as a function of an interface (hereinafter also referred to as an "ASP solver interface") for a program (hereinafter also referred to as an "ASP solver") that executes a known ASP (Answer Set Programming). A process of setting constraint information and an objective function for the combination information stored in the unit 2 is performed. A program encoding the combination information, the set constraint information and the objective function is operated as a function of the ASP solver. processing to input to the calculation unit 13, processing to provide the set constraint information and the objective function to the storage unit 2, and the calculated desired combination information as character information or image information (or as character information or image information It is also possible to perform a process of converting and providing the information, or a process of setting a class (classification) or whether or not it is required for each of two or more sets of constraint information. "ASP" is a concept that fuses the concepts of logic programming and constraint programming to encode a program containing variables and to compute the solution set of the encoded program.

FIG. 4 shows an example of an ASP solver interface screen, and the process of setting constraint information and objective functions is executed by operating this screen. As shown in FIG. 4, first, attribute items (“price”, “evaluation”) in the attribute value information set via the OVM modeler are changed to variables (for example, “W1”, “W2 ”).

Next, as setting of constraint information, "type" (for example, "calculated value is value 1 or more") is set by selection from the pull-down menu, and "value 1" is set as the threshold value via the OVM modeler. It is set by inputting a calculated value (for example, "300") based on the attribute value information (price) of information (parts M1 to M5 and parts N1 to N3), "value 2" is set as a blank, and "formula" is a variable ("W1"). Further, as the setting of the objective function, "type" (for example, "maximize") is set by selection from a pull-down menu, and "formula" is set as a variable (for example, "W2").

That is, according to the setting unit 12, based on the combination information for each supplier, for example, the OEM manufacturer sets "calculated values (attribute value information) of prices (attribute value information) of parts M1 to M5 and parts N1 to N3 (mutation information) as constraint information. (variable) is 300 (value 1) or less", and as an objective function, "the calculated value (variable) of the evaluation (attribute value information) of parts A1 to A5 and parts B1 to B3 (mutation information) is the maximum ”. If multiple combinations of information (solutions) that satisfy the objective function are calculated as a result, instead of all the calculated solutions, a solution that "minimizes variant options" or "maximizes variant options" A setting may be made to extract either of the solutions as follows.

Here, the "type" set in the constraint information includes "calculated value is 1 or less", "calculated value is 1 or more", and "calculated value is between value 1 and value 2". , "individual value is less than or equal to 1", "individual value is greater than or equal to 1", and "individual value is between value 1 and value 2". In addition, the "type" set in the objective function can be "maximization" that seeks the solution (combination) that maximizes the calculated value of the "formula", "minimization" that seeks the minimum solution (combination), It also includes “extremization” to find a solution (combination) that becomes an extreme value, and maximization or minimization and extremization may be equivalent in result. In other words, the constraint information has six types of attribute value information (individual value) or the calculated value of the attribute value information as a threshold, and the objective function has three types of the maximum, minimum, or extreme value of the calculated value of the attribute value information. , it is easy for the user to set a desired constraint for a combination of multiple options.

FIG. 5 shows another example of the setting portion of the constraint information shown in FIG. The difference between FIG. 5 and FIG. 4 is that the setting unit 12 performs processing for setting a class (classification) or whether or not it is required for each of two or more pieces of constraint information. Therefore, only the above differences will be described below, and descriptions of other points will be omitted.

As shown in FIG. 5A, for example, there are four types of "formula", and "classes" are set to "1", "2", "2", and "3" from the top for each type. The actual constraint information pattern includes all classes below each number, specifically, for class 1 only "W1", for class 2 "W1" and "W2", for class 3 "W1", "W2" and "W3" are all three patterns. The same number means the same class. Specifically, class 2 is an AND condition of 'W2' and 'W3'. Note that a pattern including "formulas" with conflicting relationships is not adopted as constraint information.

As shown in FIG. 5B, for example, there are three types of "formula", and the "essential (check box)" of "W1" is checked. The patterns of the actual constraint information are four patterns in total: "W1" only, "W1" and "W2", "W1" and "W3", and "W1" and "W2" and "W3". Note that a pattern including "formulas" with conflicting relationships is not adopted as constraint information.

According to these configurations, constraint information that must always be satisfied and constraint information that is not (lower priority) are specified in advance, patterns of the constraint information are set, and desired combinations are obtained for each of the patterns. Get information. That is, by setting two or more pieces of constraint information in a two-stage manner, the accuracy of obtaining desired combination information is improved.

<Calculation unit>
As one function included in the ASP solver, the calculation unit 13 performs processing for calculating optimum combination information based on the program encoded by the setting unit 12. The calculated optimum combination information is sent to the ASP solver interface. The providing process and the process of providing the calculated optimal combination information to the storage unit 2 may be performed.

FIG. 6 shows an example of an ASP solver interface screen, through which optimal combination information is confirmed by the user. As shown in FIG. 6, "VP" corresponding to variable information, "V" corresponding to mutation information, "expression" corresponding to constraint information and objective function, and "solution N" corresponding to combination pattern " 〇” or “×” is listed.

That is, according to the calculation unit 12, for example, the OEM manufacturer sets the variable “W1” corresponding to the “expression” of the constraint information set by the setting unit 12 to “280” and the variable “W1” corresponding to the “expression” of the objective function “ W2” satisfies “27” as a combination pattern “solution 1”, and “parts M3/M5 and part N1・N3" is obtained.

<Determination part>
The determination unit 14 performs processing for determining whether or not the user can access the combination information for each user stored in the storage unit 2 . As shown in FIG. 2, for example, when the combination information modeled by the user's click operation is set to be "disclosed to other users", the determination unit 14 selects the combination information for disclosure from other users. Allow access. Note that "access" means browsing, but may also include writing.

7 and 8 show an example of combination information for each user modeled by the OVM modeler. As shown in FIG. 7, for example, as the contents of an order from an OEM manufacturer, "part p1" and "part p2" as "part P", "part q1" and "part q2" as "part Q", and "part R" as Public combination information of "parts r1" and "parts r2" and "parts s1" and "parts s2" as "parts S" is generated. As an order condition for the supplier 1, "part p1", "part p2", "part q1", "part q2" and public combination information indicating an exclusive relationship between "part p1" and "part q1", The order conditions for supplier 2 are "part q1", "part q2", "part r1", and "part r2", and combination information for disclosure indicating an exclusive relationship between "part q2" and "part r1", supplier 3's Public combination information is generated that indicates the exclusive relationship between "part r1", "part r2", "part s1", and "part s2" and "part r2" and "part s2" as order conditions.

On the other hand, as shown in FIG. 8, non-disclosure combination information related to public combination information is generated as an order condition for supplier 1, for example. The combination information for non-disclosure is, among the options of "part p1", "part p2", "part q1", and "part q2" of supplier 1, "part p1" is related to "part X", "Part q1" is related to "Part Y", "Part X" is accompanied by combination information including "Part x1" and "Part x2", and "Part Y" is associated with "Part y1" and "Part y2". Together with the combination information, it finally indicates an exclusive relationship between "component xxx" related to "component X" and "component yyy" related to "component Y".

That is, the supplier 1 generates and stores in the storage unit 2 not only combination information to be disclosed to the OEM manufacturer and suppliers 2 and 3, but also non-disclosure combination information resulting from the combination information to be disclosed. This makes it easier to manage confidential know-how. On the other hand, the OEM manufacturer is provided with only the necessary combination information regarding the order from the supplier 1, so it is easy to extract the desired combination information.

Hereinafter, the flow of execution processing of the server device S constituting the combination determination support system will be described with reference to the above-described contents.

<Flow of basic execution process>
As shown in FIG. 9, first, based on a combination consisting of multiple options such as parts and part specifications, each user including OEM manufacturers on the ordering side and several suppliers on the receiving side modeled with the OVM modeler The acquisition unit 11 acquires the combination information (Step 1). Next, the storage unit 2 stores the combination information acquired by the acquisition unit 11 (Step 2). Then, based on the combination information stored in the storage unit 2, the setting unit 13 sets constraint information that contributes to, for example, satisfying the constraints of the combination of OEM manufacturers and an objective function that contributes to optimization (Step 3). Then, the calculation unit 13 calculates combination information desired by the OEM based on the combination information stored in the memory 2 and the constraint information and the objective function set in the setting 12 (Step 4).

According to this flow, all combination information that takes into account the dependency constraints between multiple options can be aggregated by user by modeling, and the constraint information and purpose of a predetermined user (for example, an OEM manufacturer) among all combinations By further setting the function, it is possible to calculate the desired combination information for the user. Therefore, the efficiency of narrowing down, extracting, and determining the combination is realized without relying on the experience, intuition, and trial and error of one individual. In other words, since the modeled combination information is accompanied by dependency constraint information that is added by the user, it is not necessary to coordinate or agree on dependency constraints between users or individuals (persons in charge, etc.) who are subordinate to each user. Therefore, it is possible to improve the calculation accuracy of the optimum combination information. Therefore, especially for product development such as the manufacturing industry where there are a huge number of choices, restrictions on combinations, and users from upstream processes (planning, design, etc.) to downstream processes (production, distribution, etc.), combination decision support system resulting in overall productivity gains

<Flow of execution processing related to user access>
As shown in FIG. 10, when the combination information stored in the storage unit 2 is accessed by a predetermined user via an information terminal device, a single The judging unit 14 judges whether the user is a user or a specified user who has been previously permitted to access the combination information (Step 2-1). Then, the user who is approved by the determination unit 14 is permitted to access all the combination information including the private one, and the user who is not permitted is permitted to access only the public combination information.

According to this flow, the access authority of other users to the combination information for each user can be logically set by the determination unit 14. Therefore, the combination information is determined according to whether the dependency constraint information is disclosed or not for each user. Since it can be generated, it becomes easier for the user to manage the reason and process of generating the combination information as a history. In other words, it is relatively easy to add, change, or delete the combination information because the combination information with the dependency constraint information that can be disclosed and the combination information with the dependency constraint information that cannot be disclosed can be separated and managed in association with each other. I will be able to do it.

Next, with reference to FIGS. 11 to 15, another combination determination support system according to an embodiment of the present invention will be described with respect to portions different from those described above, and description of equivalent portions will be omitted. Parts or portions equivalent to those shown in FIGS. 1 to 9 are numbered uniformly by adding 100 in FIGS. 1 to 9 for ease of reference.

<Decentralization of combination decision support system>
As shown in FIG. 11, the server device 100S is distributed to a first server device 100Sa and a second server device 100Sb via a communication network. The first server device 100Sa may be operated from the information processing terminal 100C via the network, but the first server device 100Sa may also serve as the information processing terminal. The first server device 100Sa may be a local physical server for each user, and the second server device 100Sb may be a cloud server.

The first server device 100Sa includes a control unit 101a, a first storage unit 102a, and an unnumbered communication unit. The control unit 101 a includes an acquisition unit 111 and a setting unit 112 . The second server device 100Sb includes a control unit 101b, a calculation unit 113, a second storage unit 102b, and an unnumbered communication unit.

The first storage unit 102a stores first combination information that includes private combination information and that accepts access only from a single or specified user. The second storage unit 102b stores second combination information that is open to the public and receives access from all users. With such a configuration, management of private combination information and public combination information is physically separated, making it easier to manage private combination information for each user.

The acquisition unit 111 performs processing of providing first combination information to the first storage unit 102a and second combination information to the second storage unit 102b among the combination information for each user acquired via the OVM modeler. conduct.

The setting unit 112 sets the constraint information and the objective function for the first combination information stored in the first storage unit 102a and the second combination information stored in the second storage unit 102b; A process of encrypting a program in which the set constraint information and the objective function are encoded and input to the calculation unit 113 and a process of decrypting and providing the calculated desired combination information to the calculation unit 113 are performed. The encryption and decryption may be performed by a known cryptosystem, and the decryption may be performed by querying the first storage unit 102 . Such a configuration makes it easier for each user to obtain desired combination information without disclosing private combination information to other users.

<Analysis of combination information by machine learning>
As shown in FIG. 12, the server device 200S includes an acquisition unit 211, a setting unit 212, a calculation unit 213, a determination unit 214, a recognition unit 215, and an analysis unit 216. The recognition unit 215 recognizes the desired combination information calculated by the calculation unit 213 and provided from the setting unit 212 as an image. The analysis unit 216 analyzes the feature amount from the combination image information using the first trained model machine-learned using the combination image information recognized by the recognition unit 215 as teacher data.

FIG. 13 shows part of the desired combination information in which the calculation results of the calculator 213 are output in a matrix. As shown in FIGS. 12 and 13, the recognition unit 215 may recognize the desired combination information as combination image information converted into image information. The analysis unit 216 performs cluster analysis using a learned first model obtained by machine learning a plurality of the combined image information as teacher data, analyzes the feature amount of the desired combination information output in a matrix, and performs the feature amount The combined image information is classified for each. The learned first model may be one function of the analysis unit 216, but may function as another part.

In FIG. 13, for example, the desired combination information includes cables (parts) as options, and a plurality of matrix-like combination image information having a plurality of solutions (desired combination information) have been learned as teacher data. When cluster analysis was performed using the first model, the portions surrounded by squares on the diagram were analyzed as feature amounts. That is, from this analysis result, it was found that the relevance of each of cables 1, 2, and 7 and the relevance of each of cables 5, 6, and 9 are high by machine learning using a plurality of combined image information. The cluster analysis employs a known technique such as the kmeans method.

The constraint information is set in the setting unit 212 in consideration of the feature amount analyzed by the analysis unit 216 . That is, the setting unit 212 sets the constraint information with the cables 1, 2, and 7 as one set and the cables 5, 6, and 9 as one set. As a result, the desired combination information (solution ) to avoid redundancy and optimization is achieved.

<Expansion of options in combination decision support system>
A plurality of options processed by the combination decision support system are, for example, parts, etc., specifications of parts, common identification information for parts, non-common identification information for parts, common flow identification information, non-common flow identification information, route information, choice Information and instruction information, which may be treated in the same way as variable information and mutation information.

<Diversification of combinatorial information models>
The combination decision support system may be modeled by an application that describes a combination composed of a plurality of options as a BOM (Bill Of Material), or may be modeled by also describing the BOM with an OVM modeler. A model described as a BOM may be composed of "VP" and "V", but may also be composed of, for example, icons for flowcharts. Accompanied by

In other words, the combination decision support system uses an OVM modeler described as a BOM to represent multiple options necessary for product development with variation points, variants, and predetermined icons, and select relationships, essential relationships, and requirements between multiple options. Representing relationships, exclusion relationships, affiliation relationships, ownership relationships, shared relationships, and derived relationships as dependency relationships can be expected to diversify combinatorial information models composed of multiple options.

<Example of dependency>
Variation point (VP), variant (V), parts common identification information (hereinafter also referred to as "MasterItem"), parts etc. non-common identification information (hereinafter also referred to as "AffectedItem") described by the OVM modeler as a plurality of options. ), flow common identification information (hereinafter also referred to as “MasterLink”), flow non-common identification information (hereinafter also referred to as “AffectedLink”), route information (hereinafter also referred to as “CodeLine”), choice information (hereinafter referred to as “Choice” ) and instruction information (hereinafter also referred to as “EngineeringChange” or “WorkItem”).

In Table 1, the vertical axis (From) indicates the side of the subject, and the horizontal axis (To) indicates the side having a predetermined relationship with the subject. It may not have a predetermined relationship, but may have other relationships not listed in the table, or have a predetermined relationship other than the "x" mark, and is not limited.

A 'VP' has a required or exclusive relationship to another 'VP' and a 'V' has a required, exclusive, mandatory, or optional relationship. In other words, a 'VP' claims or excludes another 'VP' and claims, excludes, requires, or selects a 'V'.

A 'VP' and another 'V' to a 'V' have a required or exclusive relationship. In other words, a 'V' demands or excludes a 'VP' and another 'V'.

'MasterItem', 'MasterLink' and 'Choice' have a materialization relationship to 'VP'. In other words, "VP" is an abstract description, while "MasterItem", "MasterLink", and "Choice" are descriptions with concreteness.

'MasterItem', 'MasterLink', 'AffectedLink', 'CodeLine', 'Choice', and 'EngineeringChange' are related to 'V'. In other words, "V" is an abstract description, while "MasterItem", "MasterLink", "AffectedLink", "CodeLine", "Choice", and "EngineeringChange" are concrete descriptions.

Another "MasterItem", "AffectedLink" and "Choice" have an belonging relation to "MasterItem". In other words, a "MasterItem" may belong to another "MasterItem", "AffectedLink" or "Choice".

A "MasterItem" or a "MasterLink" has an ownership relationship or a sharing relationship with an "AffectedItem", and another "AffectedItem" has a derived relationship. In other words, an "AffectedItem" owns or shares a "MasterItem" or "MasterLink" and derives another "AffectedItem".

Other 'MasterLink', 'AffectedLink' and 'CodeLine' belong to 'MasterLink'. In other words, "MasterLink" has other "MasterLink", "AffectedLink" and "CodeLine" belong to it.

"MasterItem" and "MasterLink" have an ownership relationship or a shared relationship with respect to "AffectedLink", and another "AffectedLink" has a derived relationship. In other words, "AffectedLink" owns "MasterItem" and "MasterLink" and derives another "AffectedLink".

"AffectedItem" and "AffectedLink" have a relationship of belonging to "CodeLine". In other words, "CodeLine" makes "AffectedItem" and "AffectedLink" belong.

"AffectedItem" and "AffectedLink" are in a sharing relationship with respect to "Choice". In other words, "Choice" shares "AffectedItem" and "AffectedLink".

"AffectedItem", "AffectedLink", "CodeLine" and "Choice" are in an ownership relationship with "EngineeringChange", and another "EngineeringChange" is in a sharing relationship. In other words, an 'EngineeringChange' owns 'AffectedItem', 'AffectedLink', 'CodeLine' and 'Choice' and shares another 'EngineeringChange'.

Next, with reference to FIGS. 14 and 15, an example of a combination information model created based on the above-described multiple options and the inter-option relationships and their solutions will be described.

As shown in FIG. 14, "VP" is part A, and "V" having a selection relationship with "VP" is parts A-1 and A-2. As a result, it can be described that the types of component A include component A-1 and component A-2. On the other hand, since "VP" and "V" alone cannot model the combination information about part A and parts A-1 and A-2, the combination information model as BOM is described by combining with other options. easier.

First, the "VP" part A is materialized as "MasterItem", and the "V" parts A-1 and A-2 are materialized as "AffectedItem". Then, part A-1, which is 'Affected Item', belongs to part A which is 'Master Item' through the first factory which is 'CodeLine', and part A-1 which is 'Affected Item' through the second factory which is 'CodeLine'. Assign A-2. As a result, for example, part A defined as a higher level concept, part A-1 defined as a lower level concept, can be obtained through manufacturing at the first factory, and part A-2 can be obtained at the second factory. Can describe what is obtained through manufacturing, etc.

Next owns a change (for example, from January 1, 2022) where the manufacturing order is "Choice" which is "EngineeringChange" and shares part A-1 and part A-2 where the above change is "AffectedItem" do. As a result, it can be described that the manufacturing instruction will be changed from January 1, 2022 to manufacture part A-1 or part A-2.

Also, the part A-1 which is "AffectedItem" derives to another part A-1a which is "AffectedItem". As a result, for example, a component A-1a that is substantially the same as the component A-1 but differs only in part in specifications can be described.

Also, the part A-1 which is "AffectedItem" owns the assembly process of part A-1 which is "MasterLink", and the assembly process of part A-1 is the assembly procedure of part A-1 which is "AffectedLink". 1 and assembly procedure 2 belong. From this, it can be described that assembly procedure 1 and assembly procedure 2 are included in the assembly process for part A-1.

Assembly procedure 1 and assembly procedure 2 of part A-1, which is "AffectedLink", share part B, which is "MasterItem", and part B belongs to part B-1 and part B-2, which are "AffectedItem". Let This makes it possible to describe that the parts B-1 and/or the parts B-2 are used in the assembly procedure 1 and the assembly procedure 2 of the part A-1.

In addition, there are two patterns of assembly instructions for part A-1, which is "EngineeringChange". In the second pattern, "EngineeringChange" shares the design change instruction 2 of the assembly process of part A-1 and the design change instruction of the specification of part B-2. According to this, it can be described that either the assembly procedure 1 or the assembly procedure 2 of the part A-1 follows the assembly instruction of the part A-1. That is, when the assembly process of part A-1 includes two or more patterns, and part B-1 and part B-2 classified as part B are used in each pattern, the assembly instruction for part A-1 is Design change instruction 1 for the assembly process of part A-1 and design change instruction for the specification of part B-1, or design change instruction 2 for the assembly process of part A-1 and design change for the specification of part B-2 Instructions can be written respectively.

As shown in FIG. 15, patterns of combinations of options that constitute the combination information model of FIG. 14 are calculated as solutions via the ASP solver interface. In FIG. 15, "VP" is a variable point and "V" is this option (displacement point), "MasterItem", "Choice" and "CodeLine" are variable points and "AffectedItem" is these options (displacement point), If "MasterLink" is a variable point and "AffectedLink" is this option (displacement point), "EngineeringChange" is a variable point and this content is a choice (displacement point), and if the combination of each variable point and choice is satisfied The calculated solutions 1 to 5 are listed as "O", "X" if they are not satisfied, and "△" if they are satisfied but some of the other related combinations are not satisfied. Constraint information and objective functions are not set for the calculation of solutions 1 to 5, but the number of solutions may be reduced if these are set.

It should be noted that the present embodiment is not limited to the contents described above, and includes all system configurations, methods, software, hardware, functions, and interrelationships thereof as long as equivalent effects can be obtained. Specifically, even a solver that solves a combinatorial problem by a program other than ASP has the same configuration as the present embodiment, and the same effect can be obtained by the same operation. A solver executed by a predetermined program such as SAT (Satisfiability Problem) or a solver executed by a quantum computer may be employed.

S, 100S, 200S Server device C, 100C Information processing terminal 1, 101, 201 Control unit 11, 111, 211 Acquisition unit 12, 112, 212 Setting unit 13, 113, 213 Calculation unit 14, 214 Determination unit
215 recognition unit
216 Analysis Department

Claims (8)

A computer system that supports the determination of a combination consisting of multiple options,
an acquisition unit that acquires combination information for each user modeled based on the combination;
a storage unit that stores the acquired combination information;
a setting unit that sets constraint information that contributes to constraint satisfaction of the combination and an objective function that contributes to optimization based on the stored combination information;
a calculation unit that calculates desired combination information based on the stored combination information, the set constraint information, and the objective function;
The combination information includes variable information corresponding to multiple options and/or mutation information related to the variable information,
each of the variable information, each of the mutation information, and at least one of the variable information and the mutation information is accompanied by dependency constraint information indicating a mutual relationship;
A combination decision support system, wherein the variable information and/or the mutation information includes attribute value information associated with each. The constraint information is set using the attribute value information or a calculated value of the attribute value information as a threshold,
The combination decision support system according to claim 1, wherein the objective function is set as a maximum, minimum, or extreme value of the calculated values of the attribute value information. 3. The combination determination support system according to claim 1, further comprising a determination unit that determines whether or not a user can access the stored combination information. The storage unit includes a first storage unit that stores first combination information that accepts access only from a single or a specified user, and a second storage unit that stores second combination information that accepts access from all users. has
4. The setting unit according to claim 1, wherein the setting unit encrypts and sets the constraint information and the objective function, and decrypts the desired combination information calculated by the calculation unit. Combination decision support system. a recognition unit that recognizes the calculated desired combination information as an image;
an analysis unit that analyzes a feature amount from the combined image information using a trained first model machine-learned using the combined image information recognized as the image as teacher data;
5. The combination decision support system according to any one of claims 1 to 4, wherein the constraint information is set in consideration of the analyzed feature amount. The calculation unit generates a trained second model obtained by machine-learning the previously stored combination information, the set constraint information and the objective function, and the previously calculated desired combination information as teaching data. 6. The combination decision support system according to any one of claims 1 to 5, wherein new desired combination information estimated from the newly stored combination information is calculated using the new combination information. A combination decision support method executed by a computer system that supports the decision of a combination consisting of a plurality of options,
Acquiring combination information for each user modeled based on the combination,
storing the acquired combination information;
Setting constraint information that contributes to constraint satisfaction of the combination and an objective function that contributes to optimization,
calculating desired combination information based on the stored combination information and the set constraint information and objective function;
The combination information includes variable information corresponding to multiple options and/or mutation information related to the variable information,
each of the variable information, each of the mutation information, and at least one of the variable information and the mutation information is accompanied by dependency constraint information indicating a mutual relationship;
A combination decision support method, wherein the variable information and/or the mutation information includes attribute value information associated with each. A computer that supports the decision of a combination consisting of multiple options,
Obtaining combination information for each user modeled based on the combination,
storing the acquired combination information;
setting constraint information that contributes to constraint satisfaction of the stored combination and an objective function that contributes to optimization;
calculating desired combination information based on the stored combination information and the set constraint information and objective function;
The combination information includes variable information corresponding to multiple options and/or mutation information related to the variable information,
each of the variable information, each of the mutation information, and at least one of the variable information and the mutation information is accompanied by dependency constraint information indicating a mutual relationship;
A computer-readable combination decision support program, wherein the variable information and/or the mutation information includes attribute value information associated with each.

Images