Determination support system, determination support method, and determination support program

Abstract

To make a proper judgment on the activity related to the growth of a cultured marine product.SOLUTION: A determination support system 10 is a system that supports determination on an activity related to growth of a cultivated fishery product, and includes an inquiry acquisition unit 11 that acquires subjective information on experience of an expert of the activity and objective information on the activity, an answer acquisition unit 12 that makes an inquiry about the activity to a large scale language model 20 that is a language model generated by machine learning using information acquired by the inquiry acquisition unit 11 as reference information and acquires an answer to the inquiry from the large scale language model 20, and an output unit 13 that outputs information based on the answer acquired by the answer acquisition unit 12.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a decision support system, a decision support method, and a decision support program for supporting decisions regarding activities related to the growth of cultivated marine products. [Background technology]

[0002] Patent Documents 1 and 2 disclose systems for feeding farmed fish. These systems are said to feed fish in accordance with information obtained by cameras, microphones, or sonar. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2018 / 042651 [Patent Document 2] Japanese Patent Publication No. 2022-49753 Summary of the Invention [Problem to be solved by the invention]

[0004] The systems disclosed in Patent Documents 1 and 2 enable automatic feeding. However, simply using information obtained by cameras, microphones, or sonar may not necessarily result in appropriate feeding. The same applies to decisions about activities related to the growth of farmed aquatic products other than feeding.

[0005] The present invention has been made in consideration of the above, and aims to provide a decision support system, a decision support method, and a decision support program that enable appropriate decisions to be made regarding activities related to the growth of cultivated marine products. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, the decision support system of the present invention is a decision support system that supports decisions about activities related to the growth of farmed marine products, and is equipped with a query acquisition means that acquires subjective information about the experience of experts in the activity and objective information about the activity, an answer acquisition means that uses the information acquired by the query acquisition means as reference information to make an inquiry about the activity to a language model generated by machine learning and acquires an answer to the inquiry from the language model, and an output means that outputs information based on the answer acquired by the answer acquisition means.

[0007] The decision support system according to the present invention obtains an answer from a language model in response to an inquiry about activities related to the growth of cultivated aquatic products, using subjective and objective information as reference information, and outputs information based on the answer. Therefore, the output information can be used to make appropriate decisions about activities related to the growth of cultivated aquatic products. Therefore, the decision support system according to the present invention allows appropriate decisions to be made about activities related to the growth of cultivated aquatic products.

[0008] The subjective information may include information linking the subjective information with the objective information. With this configuration, the output information can be used to make more appropriate decisions regarding activities related to the growth of cultivated aquatic products. As a result, more appropriate decisions can be made regarding activities related to the growth of cultivated aquatic products.

[0009] The objective information may include at least one of information indicating past activities, information related to the marine products themselves, and information indicating the environment the marine products are in. With this configuration, appropriate decisions can be made about activities related to the growth of the cultivated marine products based on at least one of the information indicating past activities, information related to the marine products themselves, and information indicating the environment the marine products are in.

[0010] The subjective information may include information indicating the expert's answers to prepared questions. This configuration allows the expert to make appropriate decisions about activities related to the growth of cultivated aquatic products based on the expert's answers to the prepared questions.

[0011] Incidentally, the present invention can be described not only as an invention of a decision support system as described above, but also as an invention of a decision support method and a decision support program as described below. These are essentially the same inventions, just in different categories, and have similar functions and effects.

[0012] In other words, the decision support method of the present invention is a decision support method that is an operating method of a decision support system that supports decisions about activities related to the growth of farmed aquatic products, and includes an inquiry acquisition step that acquires subjective information about the experience of an expert in the activity and objective information about the activity, an answer acquisition step that uses the information acquired in the inquiry acquisition step as reference information to make an inquiry about the activity to a language model generated by machine learning and acquires an answer to the inquiry from the language model, and an output step that outputs information based on the answer acquired in the answer acquisition step.

[0013] In addition, the decision support program of the present invention is a decision support program that causes a computer to operate as a decision support system that supports decisions regarding activities related to the growth of cultivated marine products, and causes the computer to operate as: a query acquisition means that acquires subjective information regarding the experience of an expert in the activity and objective information regarding the activity; an answer acquisition means that uses the information acquired by the query acquisition means as reference information to make an inquiry about the activity to a language model generated by machine learning and acquires an answer to the inquiry from the language model; and an output means that outputs information based on the answer acquired by the answer acquisition means. [Effects of the Invention]

[0014] According to the present invention, it is possible to make appropriate decisions regarding activities related to the growth of cultivated marine products. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a diagram illustrating a configuration of a decision support system according to an embodiment of the present invention. [Figure 2] 10 is an example of a display by the decision support system. [Figure 3] 10 is another example of a display by a decision support system. [Figure 4] This is an example of subjective information used in a decision support system. [Figure 5] 1 is an example of an image that may be included in subjective information. [Figure 6] This is an example of objective information used in a decision support system. [Figure 7] 1 is a flowchart illustrating a decision support method, which is a process executed in a decision support system according to an embodiment of the present invention. [Figure 8] FIG. 2 is a diagram showing the configuration of a decision support program according to an embodiment of the present invention, together with a recording medium. DETAILED DESCRIPTION OF THE INVENTION

[0016] A decision support system, a decision support method, and a decision support program according to the present invention will be described in detail below with reference to the drawings. In the description of the drawings, the same elements are given the same reference numerals, and duplicated explanations will be omitted.

[0017] 1 shows the configuration of a decision support system 10 according to this embodiment. The decision support system 10 is a system (device) that supports decisions regarding activities related to the growth of farmed marine products. The decision support system 10 may also support decisions regarding activities related to the growth of specific marine products (e.g., activities related to yellowtail farming).

[0018] The decision support system 10 is used by those who cultivate aquatic products such as fish (e.g., aquaculture farmers). The decision support system 10 supports decisions regarding the activities by providing answers to inquiries about the activities. The activities are, for example, aquaculture operations. Specific examples of the activities include feeding the aquatic products, improving the environment in which the aquatic products are grown, and responding when abnormalities occur in the aquatic products (e.g., when disease or parasites occur in the aquatic products). The activities may also be other than those mentioned above. By referring to the answers from the decision support system 10, appropriate activities can be taken.

[0019] The decision support system 10 is realized as a device by a computer such as a PC (personal computer) or a server device that includes hardware such as a CPU (central processing unit) and memory. Each function of the decision support system 10, which will be described later, is realized by these components operating through programs or the like. The decision support system 10 may be realized by a single computer, or may be realized by a computer system configured by connecting multiple computers to each other via a network. The decision support system 10 may also be realized by cloud computing.

[0020] Responses to inquiries related to the decision support system 10 are generated by a large-scale language model (LLM) 20, which is a language model generated by machine learning. That is, the decision support system 10 uses AI (artificial intelligence) to support decisions regarding the above activities.

[0021] The large-scale language model 20 receives a query as input and generates and outputs a response to the query. The query input to the large-scale language model 20 and the response output from the large-scale language model 20 can both be text. The large-scale language model 20 can also receive reference information for generating a response to a query and generate a response based on the reference information. The large-scale language model 20 may also be a multimodal AI. The large-scale language model 20 may be similar to conventional models, as long as it is realized in the same way as conventional models.

[0022] The large-scale language model 20 is realized by a computer such as a PC or a server device. The decision support system 10 and the computer on which the large-scale language model 20 is realized can transmit and receive information to and from each other. The large-scale language model 20 does not have to be configured separately from the decision support system 10, but may be included in the decision support system 10.

[0023] The decision support system 10, for example, receives an inquiry from a user terminal 30 used by a user (e.g., a person who cultivates marine products such as the above-mentioned fish) as described below, and outputs a response to the inquiry to the user terminal 30. The user terminal 30 is a conventional computer such as a PC or a smartphone. The decision support system 10 and the user terminal 30 can send and receive information to each other. In the user terminal 30, the function for using the decision support system 10 may be realized by a specific application (e.g., a smartphone app) installed on the user terminal 30. Information exchange between the decision support system 10 and the user terminal 30 may also be performed via a website.

[0024] 2 and 3 show examples of display images relating to decision support by the decision support system 10. FIG. 2 shows an example in which a user, an aquaculture farmer, inquires about the health condition of farmed fish XX when the water temperature is 25 degrees. In this case, the answer provided by the decision support system 10 is that a water temperature of 25 degrees is above the comfortable temperature for farmed fish. FIG. 3 shows an example in which a user inquires about feeding decisions based on an image. In this case, the answer provided by the decision support system 10 is that it would be better to increase the amount of feed.

[0025] As shown in Figures 2 and 3, in the decision support system 10, a user makes an inquiry 40 about activities related to the growth of marine products, and a response 50 to the inquiry 40 is provided. Also, similar to the use of conventional large-scale language models, the decision support system 10 may be configured to interact with the user in a conversational format, as shown in Figures 2 and 3. However, the interaction with the user does not necessarily have to be in a conversational format.

[0026] Next, a description will be given of the functions of the decision support system 10 according to this embodiment. As shown in Fig. 1, the decision support system 10 includes an inquiry acquisition unit 11, an answer acquisition unit 12, and an output unit 13.

[0027] The inquiry acquisition unit 11 is an inquiry acquisition means for acquiring subjective information regarding the experience of an expert in an activity and objective information regarding the activity. The subjective information may include information linking the subjective information with objective information. The objective information may include at least one of information indicating the content of past activities, information regarding the seafood itself, and information indicating the environment in which the seafood exists. The subjective information may include information indicating the expert's answers to questions prepared in advance.

[0028] The subjective information and objective information acquired by the query acquisition unit 11 is information that serves as reference information when the large-scale language model 20 generates an answer to a query. The more numerous and diverse the subjective information and objective information acquired by the query acquisition unit 11, the more appropriate the answer the large-scale language model 20 can generate by referring to it.

[0029] FIG. 4 shows an example of subjective information. The subjective information is written information that describes the experience of an expert in an activity. An expert in an activity is, for example, someone who is currently performing or has previously performed an activity related to the growth of marine products. As shown in FIG. 4, for example, the written experience includes written questions prepared in advance about the activity and written answers from the expert to the questions. The questions may relate, for example, to how to perform the activity, the timing of the activity, points to note about the activity, and how to make decisions related to the activity. The answers from the expert reflect the expert's experience and intuition. The subjective information may also include written annotations for each answer. The annotations provide concrete details to ambiguous answers. The subjective information may include information related to multiple questions.

[0030] Furthermore, the subjective information may include information of a type other than text. For example, the subjective information may include images such as photographs. FIG. 5 shows examples of images included in the subjective information. The images shown in FIG. 5 are images obtained by photographing farmed fish underwater. FIG. 5(a) is an image during feeding. FIG. 5(b) is an image under normal conditions. For example, the image may be associated with a question or answer sentence. For example, the question or answer sentence may refer to the image. The image may be obtained by capturing an image with a camera or a fish finder. Furthermore, the subjective information may include numerical data, audio, video, etc.

[0031] The subjective information may be other than the above information as long as it relates to the experience of an expert in the activity and can be used for the functions of the decision support system 10, which will be described later.

[0032] An example of objective information is shown in Figure 6. Objective information is information obtained objectively through measurements or the like regarding past activities. For example, the objective information may be the numerical data shown in Figure 6. In this case, the objective information is various data associated with the past time (date) when the activity was carried out. The data in question is data on the amount of bait (kg), average weight (kg), water temperature (°C), number of dead individuals, number of sick individuals, and red tide occurrence.

[0033] The amount of feed (kg) is the amount of feed given to the aquatic products being cultivated. The average body weight (kg) is the average body weight of the aquatic products being cultivated. Note that the body weight of the aquatic products used as objective information may be something other than the average body weight (for example, a representative value of the body weights of multiple aquatic products other than the average weight). The water temperature (°C) is the water temperature of the environment in which the aquatic products are cultivated. The water temperature (°C) may be the water temperature at a specific, pre-set time on that day, or the average water temperature for that day. The number of dead individuals and the number of sick individuals are the number of dead and sick aquatic products among the aquatic products being cultivated on that day. The occurrence of red tide indicates whether or not a red tide has occurred in the environment in which the aquatic products are being cultivated.

[0034] As described above, the objective information may include information indicating past activity details, information related to the marine product itself, and information indicating the environment in which the marine product exists. For example, in the example of FIG. 6, the information indicating past activity details is the amount of feed (kg). For example, in the example of FIG. 6, the information related to the marine product itself is the average body weight (kg), the number of dead individuals, and the number of sick individuals. As the information related to the marine product itself, information indicating the type of marine product (e.g., fish species) to be cultivated may be used. Furthermore, the information related to the marine product itself may indicate the growth status (condition) of the marine product. For example, the body weight (kg) indicates the growth status of the marine product. Furthermore, information indicating the body length of the marine product may be used as information indicating the growth status of the marine product. In the example of FIG. 6, the information indicating the environment in which the marine product exists is the water temperature (°C) and the occurrence of red tide.

[0035] The objective information does not need to include all of the information indicating past activity details, information related to the seafood itself, and information indicating the environment in which the seafood exists, but may include only one type of information. Furthermore, these pieces of information may be other than the above examples. In the above examples, the objective information was associated with time, but the objective information does not have to be associated with time. Note that the objective information may be other than the above as long as it is objectively obtained regarding the activity and can be used for the functions of the decision support system 10, which will be described later. For example, the objective information may include text, images, audio, video, etc.

[0036] An annotation included in the subjective information may include information linking the subjective information with objective information. That is, the annotation may be an intermediate representation between the answer in the subjective information and the objective information. For example, the annotation may be a sentence including information that may be included in the objective information (e.g., numerical data). By including information linking the subjective information with the objective information, the large-scale language model 20 can take the link between the subjective information and the objective information into consideration when generating an answer to a query.

[0037] The subjective information and objective information are prepared in advance by an administrator of the decision support system 10 and stored in a database accessible to the decision support system 10. The subjective information is prepared, for example, by asking questions to aquaculture experts and obtaining their answers. The objective information is prepared, for example, by performing measurements on past activities. The inquiry acquisition unit 11 acquires the subjective information and objective information by reading them from the database. The inquiry acquisition unit 11 may also acquire the subjective information and objective information by any method other than those described above. The inquiry acquisition unit 11 outputs the acquired subjective information and objective information to the response acquisition unit 12.

[0038] The answer acquisition unit 12 is an answer acquisition means that uses the information acquired by the inquiry acquisition unit 11 as reference information to make an inquiry about activities to the large-scale language model 20, which is a language model generated by machine learning, and acquires an answer to the inquiry from the large-scale language model 20.

[0039] The answer acquisition unit 12 makes an inquiry to the large-scale language model 20 and acquires an answer, for example, as follows: The answer acquisition unit 12 inputs subjective information and objective information from the inquiry acquisition unit 11. The answer acquisition unit 12 causes the large-scale language model 20 to use the subjective information and objective information input from the inquiry acquisition unit 11 as reference information when generating an answer to a future inquiry.

[0040] For example, the answer acquisition unit 12 generates a prompt including information (e.g., a sentence to that effect) indicating that the subjective information and the objective information will be used as reference information when generating answers to future inquiries, and the subjective information and the objective information. The answer acquisition unit 12 transmits the generated prompt to the large-scale language model 20 (or a computer on which the large-scale language model 20 is implemented) and inputs it. The large-scale language model 20 accepts the input of the prompt and, in accordance with the prompt, uses the subjective information and the objective information included in the prompt as reference information when generating answers to future inquiries.

[0041] The answer acquiring unit 12 may use the subjective information and objective information as reference information in the large-scale language model 20 by a method other than the above. Furthermore, the answer acquiring unit 12 may input the subjective information and objective information to the large-scale language model 20 before making an inquiry to the large-scale language model 20.

[0042] The answer acquisition unit 12 acquires information for making an inquiry about an activity to the large-scale language model 20. The information for the inquiry is, for example, a query sentence. The information for the inquiry may also include images such as photographs. The information for the inquiry may also include other information such as numerical data, audio, and video. The answer acquisition unit 12, for example, receives and acquires the information for the inquiry transmitted from the user terminal 30.

[0043] Specifically, as shown in Figures 2 and 3, the answer acquisition unit 12 presents an interface (in this example, display content) to the user terminal 30, through which information for an inquiry can be input, and receives and acquires information input by the user using the user terminal 30 via the interface. The example shown in Figure 2 is an example in which the information for an inquiry consists only of the inquiry text. The example shown in Figure 3 is an example in which the information for an inquiry includes an image in addition to the inquiry text. Note that, as shown in Figures 2 and 3, the interface may also be used to present an answer to the inquiry to the user.

[0044] The answer acquisition unit 12 may also provide an interface presented to the user with a function that allows the user to easily input inquiries. For example, the interface may be provided with buttons for inputting standard inquiries, so that when a user presses the button via the user terminal 30, information for the inquiry corresponding to the button can be acquired. For example, the buttons may be "expert opinion," "feeding judgment from images," and "feelings of farmed fish." The answer acquisition unit 12 may also acquire information for an inquiry by any method other than those described above.

[0045] The answer acquisition unit 12 makes an inquiry to the large-scale language model 20 based on the acquired information for the inquiry. For example, the answer acquisition unit 12 transmits the acquired information for the inquiry as a prompt to the large-scale language model 20 (or the computer on which the large-scale language model 20 is implemented) and inputs it. The large-scale language model 20 accepts the input of the prompt and generates an answer to the inquiry in accordance with the prompt. As described above, when generating an answer, subjective information and objective information are used as reference information. The generated answer is, for example, a sentence answer. As shown in FIG. 2, the answer may include a question and answer that are the subjective information used as reference. The answer generated by the large-scale language model 20 may be something other than a sentence. A query to the large-scale language model 20 may be made by a method other than the above.

[0046] As described above, the large-scale language model 20 may be the same as a conventional model. Therefore, the generation of an answer to a query by the large-scale language model 20 using subjective information and objective information as reference information may be performed in the same manner as the generation of an answer by a conventional large-scale language model. The large-scale language model 20 transmits the generated answer to the decision support system 10. The answer acquisition unit 12 receives and acquires the answer transmitted from the large-scale language model 20. The answer acquisition unit 12 outputs the answer acquired from the large-scale language model 20 to the output unit 13.

[0047] The output unit 13 is an output means that outputs information based on the answer acquired by the answer acquisition unit 12. The output unit 13 outputs information based on the answer, for example, as follows: The output unit 13 inputs the answer of the large-scale language model 20 from the answer acquisition unit 12. The output unit 13 transmits the input answer to the user terminal 30. For example, the output unit 13 transmits the answer to the user terminal 30 so that the answer is displayed on the interface shown in FIGS. 2 and 3 on the user terminal 30.

[0048] The output unit 13 may output information based on the answer of the large-scale language model 20 instead of outputting it as is. The output unit 13 may also output information in a manner other than the above to an output destination other than the above. The above are the functions of the decision support system 10 according to this embodiment.

[0049] Next, a decision support method, which is a process executed by the decision support system 10 according to this embodiment (a method of operation performed by the decision support system 10), will be described using the flowchart in Fig. 7. In this process, first, the query acquisition unit 11 acquires subjective information and objective information (S01, acquisition step). Next, the answer acquisition unit 12 inputs the subjective information and objective information into the large-scale language model 20 (S02, answer acquisition step). This input is intended to allow the large-scale language model 20 to use the subjective information and objective information as reference information when generating answers to future inquiries.

[0050] Next, the answer acquisition unit 12 acquires information for an inquiry related to the growth of the cultivated marine products (S03, answer acquisition step). Next, the answer acquisition unit 12 makes an inquiry to the large-scale language model 20 based on the information for the inquiry (S04, answer acquisition step). The large-scale language model 20 accepts the inquiry and generates an answer to the inquiry. The generated answer is transmitted from the large-scale language model 20 to the decision support system 10.

[0051] In the decision support system 10, the answer acquisition unit 12 receives and acquires the answer transmitted from the large-scale language model 20 (S05, answer acquisition step). Subsequently, the output unit 13 outputs information based on the answer (S06, output step). The above is the decision support method, which is the processing executed by the decision support system 10 according to this embodiment.

[0052] In this embodiment, a response is obtained from the large-scale language model 20 in response to an inquiry about activities related to the growth of cultivated aquatic products, using subjective information and objective information as reference information, and information based on the response is output. Therefore, the output information is information that can be used to make appropriate decisions about activities related to the growth of cultivated aquatic products. Specifically, by using subjective information, the output response can be based on the experience and intuition of an expert. Furthermore, by using objective information, the output response can be based on data from past aquaculture, etc.

[0053] For example, as described above, when carrying out activities related to the growth of cultivated aquatic products, the activities can be carried out while virtually listening to the knowledge of an expert. Therefore, according to this embodiment, even people with little or no experience in aquaculture can make appropriate decisions regarding activities related to the growth of cultivated aquatic products based on the answers output from the decision support system 10. As a result, even people with little or no experience in aquaculture can carry out appropriate activities, such as appropriate feeding, appropriate measures when aquatic products become ill, and the creation of an appropriate environment for aquaculture.

[0054] As in this embodiment, the subjective information may include annotations, which are information linking the subjective information with the objective information. This configuration allows the output information to be used to make more appropriate decisions regarding activities related to the growth of farmed aquatic products. For example, including annotations in the subjective information allows for more specific responses. For example, when an inquiry is made about the amount of feeding, including annotations in the subjective information allows for a response with specific numerical values ​​regarding the timing to stop feeding or the amount of feeding.

[0055] As a result, more appropriate decisions can be made regarding activities related to the growth of cultivated aquatic products. However, the information linking the subjective information and the objective information may be something other than annotations. Furthermore, the information linking the subjective information and the objective information does not necessarily have to be used.

[0056] The objective information may include at least one of information indicating past activities, information related to the marine products themselves, and information indicating the environment in which the marine products exist. With this configuration, appropriate decisions can be made regarding activities related to the growth of farmed marine products based on at least one of information indicating past activities, information related to the marine products themselves, and information indicating the environment in which the marine products exist. However, the objective information does not have to be the above, and may be any objective information related to activities.

[0057] The subjective information may include information indicating the expert's answers to questions prepared in advance. This configuration allows an appropriate decision to be made regarding the activities related to the growth of cultivated aquatic products based on the expert's answers to the questions prepared in advance. Furthermore, by providing the subjective information in this manner, the subjective information can be easily and appropriately prepared. However, the subjective information does not have to be the above-described information, and may be information regarding the expert's experience in the activity.

[0058] Next, a description will be given of a decision support program for executing the above-mentioned series of processes by the decision support system 10. As shown in Fig. 8, the decision support program 100 is inserted into a computer and accessed, or is stored in a program storage area 111 formed on a computer-readable recording medium 110 provided in the computer. The recording medium 110 may be a non-transitory recording medium.

[0059] The decision support program 100 is configured to include an inquiry acquisition module 101, an answer acquisition module 102, and an output module 103. The functions realized by executing the inquiry acquisition module 101, the answer acquisition module 102, and the output module 103 are similar to the functions of the inquiry acquisition unit 11, the answer acquisition unit 12, and the output unit 13 of the decision support system 10 described above, respectively.

[0060] The decision support program 100 may be configured such that a part or all of it is transmitted via a transmission medium such as a communication line, and is received and recorded (including installed) by another device. Furthermore, each module of the decision support program 100 may be installed on multiple computers, rather than on a single computer. In this case, the above-described series of processes are performed by a computer system consisting of the multiple computers.

[0061] The decision support system, decision support method, and decision support program of the present disclosure have the following configuration. [1] A decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, comprising: an acquisition means for acquiring subjective information about the experience of a skilled person in the activity and objective information about the activity; an answer acquisition means for making an inquiry about the activity to a language model generated by machine learning using the information acquired by the inquiry acquisition means as reference information, and acquiring an answer to the inquiry from the language model; an output means for outputting information based on the answer acquired by the answer acquisition means; A decision support system comprising: [2] The decision support system according to [1], wherein the subjective information includes information linking the subjective information with the objective information. [3] The decision support system described in [1] or [2], wherein the objective information includes at least one of information indicating past activity, information relating to the fishery product itself, and information indicating the environment in which the fishery product exists. [4] The decision support system according to any one of [1] to [3], wherein the subjective information includes information indicating the expert's answers to questions prepared in advance. [5] A decision support method that is a method for operating a decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, comprising: an acquisition step for acquiring subjective information about the experience of experts in the activity and objective information about the activity; an answer acquisition step of making an inquiry about the activity to a language model generated by machine learning using the information acquired in the inquiry acquisition step as reference information, and acquiring an answer to the inquiry from the language model; an output step of outputting information based on the answer acquired in the answer acquisition step; A decision support method comprising: [6] A decision support program that causes a computer to operate as a decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, The computer an acquisition means for acquiring subjective information about the experience of a skilled person in the activity and objective information about the activity; an answer acquisition means for making an inquiry about the activity to a language model generated by machine learning using the information acquired by the inquiry acquisition means as reference information, and acquiring an answer to the inquiry from the language model; an output means for outputting information based on the answer acquired by the answer acquisition means; A decision support program that operates as a [Explanation of symbols]

[0062] 10...decision support system, 11...inquiry acquisition unit, 12...answer acquisition unit, 13...output unit, 20...large-scale language model, 30...user terminal, 100...decision support program, 101...inquiry acquisition module, 102...answer acquisition module, 103...output module, 110...recording medium, 111...program storage area.

Claims

1. A decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, an acquisition means for acquiring subjective information about the experience of a skilled person in the activity and objective information about the activity; an answer acquisition means for making an inquiry about the activity to a language model generated by machine learning using the information acquired by the inquiry acquisition means as reference information, and acquiring an answer to the inquiry from the language model; an output means for outputting information based on the answer acquired by the answer acquisition means; A decision support system comprising:

2. 2. The decision support system according to claim 1, wherein the subjective information includes information linking the subjective information with the objective information.

3. 3. The decision support system according to claim 1, wherein the objective information includes at least one of information indicating past activity details, information relating to the marine product itself, and information indicating the environment in which the marine product exists.

4. 3. The decision support system according to claim 1, wherein the subjective information includes information indicating the expert's answers to questions prepared in advance.

5. A decision support method that is an operating method of a decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, comprising: an acquisition step for acquiring subjective information about the experience of experts in the activity and objective information about the activity; an answer acquisition step of making an inquiry about the activity to a language model generated by machine learning using the information acquired in the inquiry acquisition step as reference information, and acquiring an answer to the inquiry from the language model; an output step of outputting information based on the answer acquired in the answer acquisition step; A decision support method comprising:

6. A decision support program that causes a computer to operate as a decision support system that supports decisions regarding activities related to the growth of cultivated aquatic products, The computer an acquisition means for acquiring subjective information about the experience of a skilled person in the activity and objective information about the activity; an answer acquisition means for making an inquiry about the activity to a language model generated by machine learning using the information acquired by the inquiry acquisition means as reference information, and acquiring an answer to the inquiry from the language model; an output means for outputting information based on the answer acquired by the answer acquisition means; A decision support program that operates as a

Images