Unit cost allocation device, unit cost allocation system, unit cost allocation program, and unit cost allocation method

Abstract

This technology provides the ability to assign appropriate unit costs to target items. [Solution] A unit cost allocation device comprising: a question receiving unit that receives the name of a target item subject to unit cost allocation; a unit cost candidate extraction unit that extracts candidate item names corresponding to the target item name from the unit cost item names in the emission coefficient master; a unit cost candidate allocation unit that transmits the candidates and causes a predetermined AI service to assign one unit cost item name from the candidates for the target item name; and an allocation result output unit that outputs the unit cost item name assigned by the unit cost candidate allocation unit.

Description

[Technical Field]

[0001] The present invention relates to a basic unit allocation device, a basic unit allocation system, a basic unit allocation program, and a basic unit allocation method. [Background Art]

[0002] Technologies related to the calculation of greenhouse gas emissions are undergoing remarkable development.

[0003] For example, the greenhouse gas emission calculation support method disclosed in Patent Document 1 is a technology for calculating the GHG (greenhouse gas) emission amount of metal products with high accuracy. In this technology, the types of energy (electricity, coke, LNG, etc.) used in the manufacture of metal products are analyzed in detail using production information, and the GHG emission basic unit is calculated. Specifically, Patent Document 1 describes: a first acquisition means for acquiring production information related to a target metal product item; a first calculation means for calculating a GHG emission basic unit derived from production-linked energy consumed in conjunction with the production of the target item based on the acquired production information; a second calculation means for calculating a GHG emission basic unit derived from fixed energy, which is part of the common fixed energy that is not directly linked to the production of the target item but is incidentally consumed in the production of metal products, and corresponds to the production of the target item; and a third calculation means for calculating the total GHG emission basic unit of the target item using the GHG emission basic unit derived from production-linked energy and the GHG emission basic unit derived from fixed energy. A greenhouse gas emission calculation support device comprising the above is disclosed. [Prior Art Documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Unexamined Patent Publication No. 2024-108427 [Summary of the Invention] [Problems to be Solved by the Invention]

[0005] When calculating LCA (Life Cycle Assessment) / CFP (Carbon Footprint of Products), which quantitatively evaluates the environmental impact of a product, it is necessary to assign an emission intensity to each raw material that makes up the product. Since there are thousands of emission intensity units, even just those defined in an inventory database, manually assigning an emission intensity unit to each raw material one by one is extremely inefficient. By assigning an emission intensity unit, it becomes possible to identify a predetermined emission coefficient for that unit, and by multiplying this by the activity level, it becomes possible to calculate the LCA.

[0006] However, the names of raw materials and the names of emission factors in the inventory database often do not match, making it difficult to resolve the issue using only conventional string matching methods. Therefore, it is conceivable to use generative AI to assign the appropriate emission factor to each raw material, taking into account the meaning of the item. However, since generative AI generates answers probabilistically using language models, it does not always pick the correct answer due to various factors. In addition, there is a phenomenon called hallucination, which generates low-accuracy answers, so the answers cannot be completely trusted. Detecting incorrectly assigned emission factors requires a very large amount of effort.

[0007] The object of the present invention has been made in view of the above points, and aims to provide a technology for assigning appropriate unit costs to target items. [Means for solving the problem]

[0008] The present invention includes several means for solving at least part of the above problems, and an example thereof is as follows: A unit cost allocation device according to one aspect of the present invention is characterized by comprising: a question receiving unit that receives the name of a target item subject to unit cost allocation; a unit cost candidate extraction unit that extracts candidate item names corresponding to the target item name from the unit cost item names of the emission coefficient master; a unit cost candidate allocation unit that transmits the candidates and causes a predetermined AI service to assign one of the unit cost item names from the candidates for the target item name; and an allocation result output unit that outputs the unit cost item name assigned by the unit cost candidate allocation unit.

[0009] Furthermore, in the above-described unit cost allocation device, the unit cost candidate extraction unit may determine similarity by unifying predetermined full-width characters contained in the target item name and the unit cost item name in the emission coefficient master to half-width characters.

[0010] Furthermore, in the above-described unit cost allocation device, the unit cost candidate extraction unit may determine similarity by converting the target item name and the unit cost item name in the emission coefficient master to phonetic readings and removing spaces.

[0011] Furthermore, in the above-described unit cost allocation device, the unit cost candidate extraction unit may unify predetermined full-width characters contained in the target item name and the unit cost item name in the emission coefficient master to half-width characters, convert them to phonetic readings, and remove spaces to determine similarity.

[0012] Furthermore, in the above-described unit cost allocation device, the unit cost candidate allocation unit may provide example answers to the predetermined AI service and prompt it to respond.

[0013] Furthermore, in the above-described unit cost allocation device, the unit cost candidate allocation unit may provide the predetermined AI service with the reason for estimation and allow it to infer step by step.

[0014] Furthermore, in the above-described unit cost allocation device, the unit cost candidate allocation unit may search for attribute information related to the target item name and the candidate (for example, "item classification," "item material," "item HS code," or information containing any of these, but not limited to these), and provide the predetermined AI service with the attribute information related to the target item name and the candidate, respectively, for allocation.

[0015] Furthermore, another aspect of the present invention provides a unit cost allocation system comprising: a question receiving unit that receives the name of a target item subject to unit cost allocation; a unit cost candidate extraction unit that extracts candidate item names corresponding to the target item name from the unit cost item names in the emission coefficient master; a unit cost candidate allocation unit that transmits the candidates and causes a predetermined AI service to assign one of the unit cost item names from the candidates for the target item name; and an allocation result output unit that outputs the unit cost item name assigned by the unit cost candidate allocation unit.

[0016] Furthermore, a unit cost allocation program according to another aspect of the present invention is characterized in that it causes an information processing device to perform a question reception step of receiving the name of a target item subject to unit cost allocation; a unit cost candidate extraction step of extracting candidate item names corresponding to the target item name from the unit cost item names in the emission coefficient master; a unit cost candidate allocation step of transmitting the candidates and causing a predetermined AI service to assign one of the unit cost item names from the candidates for the target item name; and an allocation result output step of outputting the unit cost item name assigned by the unit cost candidate allocation step.

[0017] Further, a basic unit allocation method according to another aspect of the present invention is a basic unit allocation method using an information processing apparatus, characterized by causing the information processing apparatus to execute: a question receiving step of receiving a target item name to which a basic unit is to be allocated; a basic unit candidate extraction step of extracting, for the target item name, corresponding item name candidates from basic unit item names in an emission factor master; a basic unit candidate allocation step of transmitting the candidates and causing a predetermined AI service to allocate one of the basic unit item names from the candidates for the target item name; and an allocation result output step of outputting the basic unit item name allocated in the basic unit candidate allocation step. [Effects of the Invention]

[0018] According to the present invention, an appropriate basic unit can be allocated to a target item.

[0019] Problems, configurations, and effects other than those described above will be clarified by the following description of embodiments. [Brief Description of Drawings]

[0020] [Figure 1] It is an example of a block diagram of a basic unit allocation system according to an embodiment. [Figure 2] It is a diagram showing an example of the data structure of an item master. [Figure 3] It is a diagram showing an example of the data structure of an emission factor master. [Figure 4] It is a diagram showing a hardware configuration example of a basic unit allocation apparatus. [Figure 5] It is a diagram showing a hardware configuration example of an operation terminal. [Figure 6] It is a diagram showing an example of a flow of basic unit candidate extraction processing. [Figure 7] It is a diagram showing an example of a flow of basic unit candidate allocation processing (by few-shot prompting). [Figure 8] It is a diagram showing an example of a flow of basic unit candidate allocation processing (by CoT). [Figure 9]This is an example block diagram of an emission intensity allocation system according to a second embodiment. [Figure 10] This figure shows an example flow chart of candidate emission intensity allocation processing (based on RAG). Mode for Carrying Out the Invention

[0021] Hereinafter, an emission intensity allocation system 1 to which an embodiment according to one aspect of the present invention is applied will be described with reference to the drawings.

[0022] Generally, descriptions of raw materials constituting a product include notation fluctuations. For example, fluctuations occur in katakana notation when English raw materials are written in Japanese, fluctuations in Japanese notation even for languages other than English, differences between full-width and half-width characters, differences between uppercase and lowercase letters, and fluctuations in hiragana, katakana, romaji, kanji numerals, and the like. Furthermore, regarding emission intensities, if different packages are used, the same material may have different notations. If generative AI is caused to perform allocation without absorbing these fluctuations, there is a possibility that incorrect information that becomes noise information may be allocated in some cases.

[0023] Using the emission intensity allocation system 1 according to the present embodiment makes it possible to allocate an appropriate emission intensity to a target item. As a result, it is possible to improve response speed and suppress the occurrence of hallucinations by reducing noise information, which is beneficial.

[0024] Figure 1 is an example block diagram of an emission intensity allocation system according to the present embodiment. The emission intensity allocation system 1 includes an emission intensity allocation device 100, a generative AI service 200, and an operation terminal 300.

[0025] The operation terminal 300 is a terminal owned and used by a user of the emission intensity allocation system 1, for example, a contact person in a company. The operation terminal 300 is communicably connected to the emission intensity allocation device 100 via a network 50.

[0026] Network 50 is a communication network such as a LAN (Local Area Network), WAN (Wide Area Network), the Internet, or a mobile phone network. Network 50 may also be a VPN (Virtual Private Network) on a wireless communication network such as a mobile phone network.

[0027] The unit cost allocation device 100 is a so-called server device. However, it is not limited to this, and the unit cost allocation device 100 may be any kind of information processing device such as a personal computer, smartphone, workstation, PDA (Personal Data Assistant), or tablet device. Furthermore, the unit cost allocation device 100 may be the same device as the operation terminal 300.

[0028] The unit cost allocation device 100 receives allocation requests from the operation terminal 300 via the network 50, extracts candidate unit costs for each target item in response to the allocation request, and then passes the extracted candidate unit costs to the generation AI service 200 and instructs it to allocate them, thereby realizing the unit cost allocation function. The unit cost allocation device 100 also receives a response from the generation AI service 200 and transmits the response to the operation terminal 300 via the network 50.

[0029] When the operating terminal 300 receives a predetermined response from the unit cost allocation device 100 via the network 50, it performs output processing such as display, printing, and data storage. The operating terminal 300 also transmits input information received from the input device, etc., provided in its own device (described later), via the network 50 to the unit cost allocation device 100 as information (such as a file path) indicating the storage location of the item master 121 for which allocation is requested.

[0030] The operating terminal 300 comprises a processing unit 310, an input receiving unit 320, a display unit 330, and a communication unit 340. The processing unit 310 includes an assignment request unit 311.

[0031] The allocation request unit 311 receives the file path of the item master 121 via the input reception unit 320 and passes the file path to the unit cost allocation device 100. When the allocation request unit 311 receives a predetermined response from the unit cost allocation device 100, it outputs it via the display unit 330.

[0032] The input reception unit 320 receives input instructions from the operator of the operation terminal 300, for example, a counter staff member, and transmits them to the unit cost allocation device 100. The display unit 330 receives screen information in a predetermined format instructed to be displayed by the unit cost allocation device 100 from the communication unit 340 and displays it on the display device. The communication unit 340 communicates with the unit cost allocation device 100 via the network 50.

[0033] The Generative AI Service 200 is a service that provides the functions of so-called generative AI, such as GPT and Gemini, via an API (Application Programming Interface). The Generative AI Service 200 provides commands (prompts) in natural language to the Generative AI and causes it to generate the desired result. The Generative AI Service 200 comprises an AI processing unit 210 and a communication unit 220.

[0034] In this embodiment, when the generation AI service 200 receives instructions, for example via an API, it causes the generation AI to generate response information and sends the result to the source of the instructions as a return value of the API. At that time, the generation AI can receive relevant information necessary for the response as RAG information, read the RAG information as part of the information to be added and considered, and reflect it in the generation of the response.

[0035] For example, the generation AI service 200 determines the unit cost to be allocated to raw materials by referring to information about the work process for a certain product. Alternatively, the generation AI service 200 may obtain similar information at runtime by searching for it from other devices via the internet or the like.

[0036] The unit cost allocation device 100 includes a processing unit 110, a storage unit 120, and a communication unit 130. The processing unit 110 includes a question receiving unit 111, a unit cost candidate extraction unit 112, a unit cost candidate allocation unit 113, a response request unit 114, and an allocation result output unit 115. The storage unit 120 includes an item master 121 and an emission factor master 122.

[0037] Figure 2 shows an example of the data structure of an item master. The item master 121 contains information including a list of one or more items to which unit consumption is to be assigned. The item master 121 stores the item name 121a, the item group 121b, and the HS code 121c in association with each other.

[0038] Item name 121a is information that displays the raw material and its unit quantity in language. For example, item name 121a includes items that are indicated as raw materials, such as "Dark Chocolate Drops 50% (1kg)". Item group 121b is information that identifies a predetermined group to which the item identified by item name 121a belongs. For example, item group 121b includes groups of raw materials, such as "Cake ingredients". HS code 121c is information that identifies the HS code of the item identified by item name 121a. For example, HS code 121c includes the HS code if the HS code assigned to the raw material is clear, such as "HS180632".

[0039] Figure 3 shows an example of the data structure of the emission factor master. The emission factor master 122 contains information that identifies the emission factor for each item name that is the unit of production. The emission factor master 122 stores the inventory database 122a, item name 122b, location 122c, emission factor 122d, and validity period 122e in association with each other.

[0040] Inventory database 122a is information that identifies the inventory database to which emission factor information is defined. For example, inventory database 122a includes information that identifies the inventory database that defines the emission intensity, such as "3EID", "IDEA", and "ecoinvent". Item name 122b is information that displays the intensity and its unit quantity in language. For example, item name 122b includes items that are indicated as intensity, such as "chocolate". Location 122c is information that identifies the country or region to which the emission intensity and emission factor identified by item name 122b apply. For example, location 122c includes country names such as "JP".

[0041] The emission factor 122d is information about the emission factor applied to the emission intensity identified by the item name 122b in the combination of inventory database 122a and location 122c. For example, emission factor 122d includes emission factors such as "1.9 kgCO2e / kilogram". The validity period 122e is information that identifies the period during which emission factor 122d is valid. For example, validity period 122e includes the start and end dates of a period expressed in years, such as "yyyy / mm / dd~yyyy / mm / dd".

[0042] Returning to the explanation of Figure 1, the question reception unit 111 receives the name of the target item to which unit consumption will be assigned. Specifically, the question reception unit 111 receives the name of the target item to which unit consumption will be assigned, or the file path of the item master 121 containing the name of the target item. If a file path is received, the question reception unit 111 obtains multiple item names 121a from the item master 121 identified by the file path, and extracts unit consumption candidates for each. If a target item name is received, the question reception unit 111 extracts unit consumption candidates limited to that target item.

[0043] The unit cost candidate extraction unit 112 corrects variations in the notation of the target item name and extracts corresponding item name candidates from the item name 122b of the emission coefficient master 122. For example, in correcting variations in notation, the unit cost candidate extraction unit 112 unifies predetermined full-width characters in the target item name and the item name 122b of the unit cost in the emission coefficient master 122 to half-width characters, converts them to phonetic readings and removes spaces to determine similarity. Note that the unit cost candidate extraction unit 112 does not necessarily have to correct variations in notation, and may perform only some of the following: unification to half-width characters, conversion to phonetic readings and removal of spaces, or a combination of any of them.

[0044] The unit cost candidate assignment unit 113 instructs the response request unit 114 to send unit cost candidates to the generation AI service 200. The unit cost candidate assignment unit 113 then instructs the generation AI service 200 to assign one unit cost item name from the unit cost candidates for the target item. The unit cost candidate assignment unit 113 then passes the assigned unit cost item name to the assignment result output unit 115 to output the result.

[0045] The unit cost candidate assignment unit 113 may also configure prompts to the generation AI service 200 to teach it a specific task by presenting a small number of concrete answer examples (future shot prompting), and instruct the answer request unit 114 accordingly.

[0046] Furthermore, the unit cost candidate assignment unit 113 may configure prompts to the generation AI service 200 to provide reasons for estimation and allow it to reason step by step (chain of sort prompting), and instruct the response request unit 114 accordingly.

[0047] Alternatively, the unit cost candidate assignment unit 113 may prompt the generation AI service 200 to search for relevant attribute information for the target item name and unit cost candidate, and to provide the generation AI service 200 with relevant attribute information (RAG: Retrieval Augmented Generation) for the target item name and unit cost candidate, thereby instructing the response request unit 114 to assign a unit cost candidate.

[0048] The response request unit 114, upon receiving instructions from the unit cost candidate assignment unit 113, transmits the unit cost candidate extracted by the unit cost candidate extraction unit 112 to the generation AI service 200. The response request unit 114 also instructs the generation AI service 200, using prompts or the like, to assign one unit cost item name from the unit cost candidate list to the target item.

[0049] The assignment result output unit 115 outputs the name of the unit cost item assigned by the unit cost candidate assignment unit 113.

[0050] The communication unit 130 communicates with other devices, namely the generation AI service 200 and the operation terminal 300, via the network 50.

[0051] Figure 4 shows an example of the hardware configuration of a unit cost allocation device. The unit cost allocation device 100 includes a processor 101, memory 102, storage 103, communication device 104, and a bus 105 connecting the devices. In addition, the unit cost allocation device 100 may also include an input device.

[0052] The processor 101 is a computing device such as a CPU (Central Processing Unit) or GPU (Graphics Processing Unit), and it performs processing according to a program recorded in memory 102 or storage 103. In the unit cost allocation device 100, processing is performed by the processor 101, which operates according to a program read from memory 102 or storage 103. The processing unit 110, the question receiving unit 111, the unit cost candidate extraction unit 112, the unit cost candidate allocation unit 113, the response request unit 114, and the allocation result output unit 115 each realize their respective functions by the processor 101 executing a program.

[0053] Memory 102 is a storage device such as RAM (Random Access Memory) or flash memory, and functions as a storage area where programs and data are temporarily read. Storage 103 is a writable and readable storage device. The storage unit 120's functions are realized by either memory 102 or storage 103. Alternatively, the storage unit 120 may be realized by a storage device connected via communication device 104.

[0054] The communication device 104 is an interface for connecting the unit allocation device 100 to an external device. For example, the communication device 104 uses an antenna that can utilize predetermined radio waves (e.g., 5GHz band, 2.4GHz band, etc.) to establish a connection with the generated AI service 200 and the operation terminal 300 according to the Wi-Fi standard to perform wireless communication.

[0055] Furthermore, the processing of each component of the unit cost allocation device 100 may be performed on one piece of hardware or on multiple pieces of hardware. Also, the processing of each component of the unit cost allocation device 100 may be implemented in one program or in multiple programs.

[0056] The communication unit 130 of the unit cost allocation device 100 described above is implemented by the communication device 104. The above is an example of the hardware configuration of the unit cost allocation device 100.

[0057] Each component of the unit allocation device 100 can be further classified into many more components depending on the processing content. Alternatively, each component can be classified to perform even more processing tasks.

[0058] Furthermore, each processing unit (processing unit 110, question receiving unit 111, unit cost candidate extraction unit 112, unit cost candidate assignment unit 113, response request unit 114, assignment result output unit 115) may be constructed using dedicated hardware (ASIC, GPU, etc.) to implement its respective function. Also, the processing of each processing unit may be executed on a single piece of hardware or on multiple pieces of hardware.

[0059] Figure 4 shows an example of the hardware configuration of an operating terminal. The operating terminal 300 includes an input device 301, a processor 302, storage 303, memory 304, a display 305, a communication device 306, and a bus 307 connecting the devices.

[0060] The input device 301 is a variety of input devices such as a keyboard, mouse, or touch panel. The input reception unit 320 of the operation terminal 300 is realized by the input device 301 and the processor 302. The processor 302 is a computing device such as a CPU or GPU, and it executes processing according to a program recorded in the memory 304 or storage 303. In the operation terminal 300, processing is performed by the processor 302, which operates according to a program read from the memory 304 or storage 303. The processing unit 310 and the allocation request unit 311 realize their respective functions by having the processor 302 execute a program.

[0061] Memory 304 is a storage device such as RAM (Random Access Memory) or flash memory, and functions as a storage area where programs and data are temporarily read. Storage 303 is a writable and readable storage device.

[0062] The display 305 is a display device such as a liquid crystal display or an organic EL display. The display unit 330 is realized by the display 305 and the processor 302.

[0063] The communication device 306 is an interface for connecting the operating terminal 300 to an external device. For example, the communication device 306 uses an antenna that can utilize predetermined radio waves (e.g., 5GHz band, 2.4GHz band, etc.) to establish a connection with the unit allocation device 100 according to the Wi-Fi standard, and performs wireless communication. The communication unit 340 of the operating terminal 300 is realized by the communication device 306 and the processor 302.

[0064] Furthermore, the processing of each component of the operating terminal 300 may be performed on one piece of hardware or on multiple pieces of hardware. Also, the processing of each component of the operating terminal 300 may be implemented by one program or by multiple programs.

[0065] The input device 301, processor 302, storage 303, memory 304, display 305, and communication device 306 are connected to each other by connecting wires such as a bus 307. The above is an example of the hardware configuration of the operating terminal 300.

[0066] Each component of the operating terminal 300 can be further classified into many more components depending on the processing content. Alternatively, each component can be classified to perform even more processing.

[0067] Furthermore, each processing unit (processing unit 310, assignment request unit 311, input reception unit 320, display unit 330, communication unit 340) may be constructed using dedicated hardware (ASIC, GPU, etc.) to realize its respective function. Also, the processing of each processing unit may be executed on a single piece of hardware or on multiple pieces of hardware.

[0068] The Generative AI Service 200 is a service provided outside of the Unit Cost Allocation System 1, which uses Generative AI. However, it is not limited to this, and it may also use Generative AI maintained by the Unit Cost Allocation System 1. In that case, the Generative AI includes Generative AI that uses Large Language Models (LLMs). The Generative AI is a pre-trained model, such as GPT-4, which has been trained in advance using language data to create a large language model using a neural network (NN) through machine learning or deep learning. The Generative AI is not limited to neural networks (NNs), and other known methods may also be used.

[0069] Next, the operation of the unit cost allocation system 1 in this embodiment will be described.

[0070] First, in this embodiment, a user of the unit cost allocation system 1, such as a company's customer service representative, acts as the operator and uses the operation terminal 300 to input unit cost allocation requests from customers or collaborators as text. This becomes the inquiry text (query), and the operation terminal 300 requests a response from the unit cost allocation device 100. The request is completed when the generation AI service 200 or the unit cost allocation device 100 responds to the operation terminal 300 with a response.

[0071] Figure 6 shows an example of the flow of the unit cost candidate extraction process. The unit cost candidate extraction process starts when the operation terminal 300 receives an instruction from the operator to assign unit costs.

[0072] First, the allocation request unit 311 of the operation terminal 300 transmits an allocation request to the unit allocation device 100 (step S001). The allocation request includes the name of the item master 121, i.e., the file path of the item master 121, which is received via the input reception unit 320, and the file path is passed on to the unit allocation device 100.

[0073] Then, the question receiving unit 111 of the unit allocation device 100 receives the file path of the item master 121. The question receiving unit 111 then retrieves the target item name from the item master (step S002). If the item master contains multiple target item names, the question receiving unit 111 repeatedly performs steps S003 to S006, described later, for each of the multiple target item names.

[0074] Then, the unit consumption candidate extraction unit 112 unifies predetermined full-width characters in the target item name to half-width characters (step S003). For example, the unit consumption candidate extraction unit 112 unifies predetermined full-width characters included in the target item name to half-width characters.

[0075] Then, the unit consumption candidate extraction unit 112 converts the target item name to its phonetic reading and removes spaces (step S004). For example, the unit consumption candidate extraction unit 112 converts the kanji characters in the target item name to their phonetic readings and removes spaces.

[0076] Then, the unit cost candidate extraction unit 112 unifies the predetermined full-width characters in the item name 112b of the unit cost in the emission coefficient master 122 to half-width characters, and performs conversion to phonetic readings and removal of spaces (step S005). For example, the unit cost candidate extraction unit 112 unifies the predetermined full-width characters in the item name 112b of the unit cost in the emission coefficient master 122 to half-width characters, and performs conversion to phonetic readings of kanji and removal of spaces.

[0077] Then, the unit cost candidate extraction unit 112 extracts multiple item names containing strings similar to the target item name from the emission coefficient master 122 as unit cost candidates (step S006). Specifically, the unit cost candidate extraction unit 112 determines the similarity between the phonetic reading of the target item name converted in step S004 and the phonetic reading of the item name of the unit cost in the emission coefficient master 122 converted in step S005, and extracts multiple item names containing strings that are more or less similar to the target item name from the emission coefficient master 122 as unit cost candidates. In determining this similarity, the unit cost candidate extraction unit 112 applies existing methods such as string matching determination and string distance measurement.

[0078] The above is an example of the unit cost candidate extraction process flow. According to this example of the unit cost candidate extraction process flow, multiple unit cost candidates can be extracted with high accuracy, thereby improving the matching accuracy and response speed of unit cost candidates related to the target item.

[0079] Figure 7 shows an example of the flow of the unit cost candidate assignment process (using Fusion Shot prompting). This example of the unit cost candidate assignment process (using Fusion Shot prompting) starts immediately after the unit cost candidate extraction process.

[0080] First, the unit cost candidate allocation unit 113 queries the generation AI service 200 using few-shot prompting. Specifically, the unit cost candidate allocation unit 113 instructs the response request unit 114 to process a request to the generation AI service 200 to allocate unit costs for each of the one or more target items that are target items in the unit cost candidate extraction process. The response request unit 114 passes the one or more target items that are target items in the unit cost candidate extraction process, along with a plurality of unit cost candidates extracted as candidates from the emission coefficient master 122, to the generation AI service 200, performs few-shot prompting with a predetermined prompt, and instructs the AI ​​service 200 to allocate from the unit cost candidates for each of the one or more target items that are target items in the unit cost candidate extraction process (step S101). Furthermore, the instructions (prompts) for the generation AI service 200 can include "instructions to assign unit consumption not only based on the name of the target item, but also by including the meaning of the target item (for example, the meaning based on the materials and uses included in the name of the target item) as a criterion." This allows for a more appropriate assignment of unit consumption that takes CO2 emissions into consideration, rather than just superficially understanding the target item.

[0081] Then, the AI ​​processing unit 210 of the generation AI service 200 assigns units from the delivered unit cost candidates to each of the one or more target items that are the target items in the unit cost candidate extraction process (step S102). Specifically, the AI ​​processing unit 210 selects one from the delivered unit cost candidates for each target item and assigns the unit cost to it, in accordance with the task queried by Fusion Shot Prompting. Then, the AI ​​processing unit 210 transmits the assigned unit cost to the unit cost assignment device 100 for each target item.

[0082] Then, the assignment result output unit 115 outputs the assigned unit cost candidate as a response to the operation terminal 300 (step S103).

[0083] The above is an example of the flow for assigning candidate unit consumption (using Fusion Shot prompting). According to this example of the flow for assigning candidate unit consumption (using Fusion Shot prompting), the AI ​​can be made to select a unit consumption from among candidate items with similar names to the target item, thus enabling accurate and fast assignment of unit consumption.

[0084] In the above-described unit cost candidate assignment process (using fusion shot prompting), a query is made to the generating AI using fusion shot prompting in step S101, but this is not the only method. For example, the generation AI service 200 may be given reasons for its estimation and allowed to infer step by step (chain of sort prompting) to improve the accuracy of unit cost assignment.

[0085] Figure 8 shows an example of the flow of the unit cost candidate assignment process (by COT). The example of the flow of the unit cost candidate assignment process (by COT) is basically the same as the unit cost candidate assignment process (by Fusion Shot Prompting) described above, but differs in that step S201, which will be described later, is performed instead of step S101. Step S201 will be explained below.

[0086] First, the unit cost candidate allocation unit 113 queries the generation AI service 200 using Chain of Thought. Specifically, the unit cost candidate allocation unit 113 instructs the response request unit 114 to process a request to the generation AI service 200 to allocate unit costs for each of the one or more target items that are target items in the unit cost candidate extraction process. The response request unit 114 passes the one or more target items that are target items in the unit cost candidate extraction process, along with a plurality of unit cost candidates extracted as candidates from the emission coefficient master 122, to the generation AI service 200, performs COT prompting with a predetermined prompt, and instructs the AI ​​service 200 to allocate from the unit cost candidates for each of the one or more target items that are target items in the unit cost candidate extraction process (step S201).

[0087] For example, in COT prompting, the unit cost candidate assignment unit 113 instructs the generation AI service 200 to provide an intermediate response and then instructs the response request unit 114 to request the assignment of unit costs for multiple unit cost candidates extracted as candidates from the emission coefficient master.

[0088] The above is an example of the unit cost candidate assignment process (by COT). According to this example of the unit cost candidate assignment process (by COT), the AI ​​can select a unit cost from among candidate items with similar names to the target item, thus enabling accurate and fast unit cost assignment. Furthermore, by performing COT prompting, it is possible to guide the reasoning behind the judgment criteria of the generating AI to some extent, thereby reducing the possibility of significantly incorrect unit cost assignments.

[0089] The unit cost allocation system 1, to which an embodiment of the present invention is applied, has been described above. According to this embodiment, an appropriate unit cost can be allocated to the target item.

[0090] The present invention is not limited to the embodiments described above. The embodiments described above can be modified in various ways within the scope of the technical idea of ​​the present invention. For example, in the embodiments described above, the allocation accuracy is improved by querying the generating AI using fusion shot prompting or COT, but the invention is not limited to this, and high-accuracy allocation may be performed using RAG data. Such embodiments will be described with reference to Figures 9 and 10. The unit allocation system according to the second embodiment is basically the same as the unit allocation system 1 described above, but differs in that it uses the generating AI with RAG data. The differences will be described below.

[0091] Figure 9 is an example of a block diagram of a unit cost allocation system according to the second embodiment. In the unit cost allocation system 1' according to the second embodiment, a RAG database 400 is included that can communicate with the unit cost allocation device 100 via a network 50, and the processing unit 110 of the unit cost allocation device 100 includes a RAG data acquisition unit 116.

[0092] The RAG database 400 contains various types of information, including the names of target items, item groups, and HS codes related to unit consumption. The RAG data acquisition unit 116 retrieves the results of searching the RAG database 400 using the target item name, item group, and HS code for each target item name as related information.

[0093] Figure 10 shows an example of the flow of the unit cost candidate assignment process (using RAG). The example of the flow of the unit cost candidate assignment process (using RAG) is basically the same as the unit cost candidate assignment process (using Fusion Shot Prompting) described above, but differs in that steps S301 to S303 described later are performed instead of step S101, and RAG data is used in step S102 (step S102'). The following explanation will focus on steps S301 to S303 and step S102'.

[0094] First, the RAG data acquisition unit 116 queries the RAG database 400 for attribute information about the target item (step S301). Specifically, the RAG data acquisition unit 116 queries the RAG database 400 for attribute information about the name of the item to which the unit cost will be assigned.

[0095] The RAG database 400 searches for relevant attribute information for the target item (step S302). For example, the RAG database 400 searches for the target item name, item group, and HS code for each target item name. The RAG database 400 then transmits the searched RAG data to the unit cost allocation device 100.

[0096] The unit cost candidate allocation unit 113 then adds attribute information to the multiple unit cost candidates extracted from the emission coefficient master 122 and requests the generation AI service 200 to allocate the units. Specifically, the unit cost candidate allocation unit 113 instructs the response request unit 114 to process a request to the generation AI service 200 to allocate units for each of the one or more target items that are target items in the unit cost candidate extraction process. The response request unit 114 passes the one or more target items that are target items in the unit cost candidate extraction process, the multiple unit cost candidates extracted from the emission coefficient master 122, and the attribute information added to the unit cost candidates to the generation AI service 200 and instructs it to allocate from the unit cost candidates for each of the one or more target items that are target items in the unit cost candidate extraction process using a predetermined prompt (step S303).

[0097] Then, the AI ​​processing unit 210 of the generation AI service 200 assigns units from the received unit cost candidates using RAG data for each of the one or more target items that are the target items in the unit cost candidate extraction process (step S102'). Specifically, for each target item, the AI ​​processing unit 210 selects one similar unit from the received unit cost candidates, taking into account the RAG data, and assigns the unit cost. This similarity generation criterion is a unique similarity criterion of the generation AI service 200, but generally, it prioritizes the range described in the RAG data and uses information within that range. For example, if HS codes are provided as RAG data, the AI ​​processing unit 210 identifies HS codes similar to the target item from the RAG data and determines the similarity. Alternatively, the response request unit 114 may instruct the generation AI service 200 in advance on the response criteria. Then, the AI ​​processing unit 210 transmits the assigned unit costs to the unit cost assignment device 100 for each target item.

[0098] The above is an example of the flow for assigning candidate unit costs (using RAG). According to this example of the flow for assigning candidate unit costs (using RAG), by using RAG data to guide similarity judgments, the possibility of the generated AI assigning unit costs being significantly incorrect can be reduced.

[0099] The above is an example of a unit cost candidate system according to the second embodiment. According to the unit cost candidate system according to the second embodiment, it is possible to assign appropriate unit costs to the target item.

[0100] Furthermore, the technical elements of the above-described embodiments may be applied individually, or they may be divided into multiple parts, such as program components and hardware components, and applied accordingly. Note that the processing performed by the response request unit 114 described above may also be performed by the unit cost candidate assignment unit 113.

[0101] The present invention has been described above, focusing on its embodiments. [Explanation of symbols]

[0102] 1...Unit allocation system, 50...Network, 100...Unit allocation device, 110...Processing unit, 111...Question reception unit, 112...Unit candidate extraction unit, 113...Unit candidate allocation unit, 114...Answer request unit, 115...Allocation result output unit, 120...Storage unit, 121...Item master, 122...Emission coefficient master, 130...Communication unit, 200...Generation AI service, 210...AI processing unit, 220...Communication unit, 300...Operation terminal, 310...Processing unit, 311...Allocation request unit, 320...Input reception unit, 330...Display unit, 340...Communication unit.

Claims

1. A question reception department that accepts the names of items subject to unit cost allocation, A unit consumption candidate extraction unit extracts candidate item names corresponding to the target item names from the unit consumption item names in the emission coefficient master, A unit consumption candidate assignment unit transmits the aforementioned candidates and causes a predetermined AI service to assign one of the aforementioned unit consumption item names from the aforementioned candidates to the aforementioned target item name, A unit allocation result output unit that outputs the unit item name assigned by the unit candidate allocation unit, A unit cost allocation device characterized by comprising the following features.

2. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate extraction unit is: The similarity is determined by unifying the predetermined full-width characters in the target item name and the unit cost item name in the emission coefficient master to half-width characters. A unit cost allocation device characterized by the following features.

3. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate extraction unit is: Similarity is determined by converting the aforementioned target item name and the aforementioned unit cost item name in the emission coefficient master to phonetic readings and removing spaces. A unit cost allocation device characterized by the following features.

4. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate extraction unit is: The following steps are performed to determine similarity: The full-width characters in the target item name and the unit cost item name in the emission coefficient master are standardized to half-width characters, and then converted to phonetic readings and spaces are removed. A unit cost allocation device characterized by the following features.

5. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate allocation unit is: The aforementioned predetermined AI service is presented with example answers and prompted to provide them. A unit cost allocation device characterized by the following features.

6. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate allocation unit is: The predetermined AI service is given reasons for estimation and is made to infer step by step. A unit cost allocation device characterized by the following features.

7. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate allocation unit is: The system searches for attribute information related to the target item name and the candidate, and provides the predetermined AI service with the attribute information related to the target item name and the candidate, and assigns them accordingly. A unit cost allocation device characterized by the following features.

8. A unit cost allocation device according to claim 1, The aforementioned unit cost candidate allocation unit is: When assigning the unit consumption item name to the target item name to the predetermined AI service, the meaning based on the material and use of the target item name is included in the judgment criteria, and the unit consumption item name is assigned accordingly. A unit cost allocation device characterized by the following features.

9. A question reception department that accepts the names of items subject to unit cost allocation, A unit consumption candidate extraction unit extracts candidate item names corresponding to the target item names from the unit consumption item names in the emission coefficient master, A unit consumption candidate assignment unit transmits the aforementioned candidates and causes a predetermined AI service to assign one of the aforementioned unit consumption item names from the aforementioned candidates to the aforementioned target item name, A unit allocation result output unit that outputs the unit item name assigned by the unit candidate allocation unit, A unit cost allocation system characterized by comprising the following features.

10. In an information processing device, A question submission step to receive the names of the items subject to the allocation of unit costs, A unit cost candidate extraction step is performed to extract candidate item names corresponding to the target item names from the unit cost item names in the emission coefficient master, A unit cost candidate assignment step involves transmitting the aforementioned candidates and having a predetermined AI service assign one of the aforementioned unit cost item names from the aforementioned candidates to the aforementioned target item name, A distribution result output step which outputs the name of the unit cost item assigned by the unit cost candidate assignment step, A unit cost allocation program characterized by implementing the following:

11. A method for allocating unit costs using an information processing device, The aforementioned information processing device, A question submission step to receive the names of the items subject to the allocation of unit costs, A unit cost candidate extraction step is performed to extract candidate item names corresponding to the target item names from the unit cost item names in the emission coefficient master, A unit cost candidate assignment step involves transmitting the aforementioned candidates and having a predetermined AI service assign one of the aforementioned unit cost item names from the aforementioned candidates to the aforementioned target item name, A distribution result output step which outputs the name of the unit cost item assigned by the unit cost candidate assignment step, A method for allocating unit costs, characterized by implementing such an arrangement.

Images