Information processing device and information processing method

Abstract

It accurately calculates greenhouse gas emission factors and emissions without requiring any effort from the user. [Solution] The information processing device comprises: a classification unit that classifies each of multiple accounting data into classification items that include at least one of major classification, minor classification, or summary; a data conversion unit that converts the classified classification items into input data for a machine learning model; an inference unit that inputs the input data into a trained machine learning model and obtains greenhouse gas emission-related information inferred by the machine learning model; a master selection unit that selects master data corresponding to the emission-related information obtained by the inference unit from among multiple master data, each containing greenhouse gas emission factors in a global supply chain; a search unit that searches for emission factors corresponding to the emission-related information obtained by the inference unit from the master data selected by the master selection unit; and a calculation unit that calculates greenhouse gas emissions based on the searched emission factors.

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus and an information processing method.

Background Art

[0002] As efforts to reduce greenhouse gases (GHGs) such as carbon dioxide accelerate, many businesses are required to conduct environmentally conscious management. Calculating the amount of greenhouse gas emissions associated with activities not only in products but also in a business's supply chain is important for managing the entire business's activities. To calculate the amount of greenhouse gas emissions, it is necessary to know the emission factor for each individual activity volume. The emission factor is also called the emission unit.

[0003] Techniques have been disclosed for deriving the amount of greenhouse gas emissions for each of a plurality of scopes and categories based on the activity content, activity information indicating the activity volume of the activity content, and ratio specifying information for specifying the ratio of a plurality of scopes and categories that are the scope for calculating the amount of greenhouse gas emissions (see Patent Document 1).

[0004] Also, techniques have been disclosed for selecting an optimal emission unit format for new activity volume data in consideration of the user's industry type and the application type of the source of the activity volume data from existing emission unit formats (see Patent Document 2).

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0006] In the case of Patent Document 1, the process of determining the proportion of activity content and activity volume to be allocated to multiple scopes and categories is time-consuming, and there is no guarantee that the determined proportions are correct, which may prevent accurate prediction of greenhouse gas emissions. Furthermore, in the case of Patent Document 2, users are required to create the existing emission intensity format themselves, which is time-consuming for the user. In addition, there is a risk that an inappropriate emission intensity format may be created due to user input errors, and there is no guarantee that the emission coefficient (emission intensity) can be calculated correctly.

[0007] Therefore, this disclosure provides an information processing device and an information processing method that can accurately calculate greenhouse gas emission factors and emissions without burdening the user. [Means for solving the problem]

[0008] To solve the above problems, according to this disclosure, a classification unit classifies each of multiple accounting data into classification items that include at least one of a major category, a minor category, or a summary, A data conversion unit converts the classification items classified by the classification unit into input data for a machine learning model, An inference unit that inputs the aforementioned input data into the trained machine learning model and obtains greenhouse gas emission-related information inferred by the machine learning model, A master selection unit selects a master data from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, that corresponds to the emission-related information acquired by the inference unit. A search unit searches for emission coefficients corresponding to the emission-related information obtained by the inference unit from the master data selected by the master selection unit, An information processing device is provided, comprising: a calculation unit that calculates the amount of greenhouse gas emissions based on the emission coefficients retrieved by the search unit. [Brief explanation of the drawing]

[0009] [Figure 1]Block diagram showing the schematic configuration of an information processing apparatus according to an embodiment. [Figure 2] Diagram for explaining greenhouse gases in the supply chain. [Figure 3] Diagram showing the data configuration of the first master data. [Figure 4] Diagram showing the data configuration of the second master data. [Figure 5] Diagram showing the data configuration of the third master data. [Figure 6] Diagram showing the data configuration of the fourth master data. [Figure 7] Diagram showing the data configuration of the fifth master data. [Figure 8] Diagram showing the data configuration of the sixth master data. [Figure 9] Diagram showing the data configuration of the seventh master data. [Figure 10] Flowchart showing the processing operation of an information processing apparatus according to an embodiment. [Figure 11] Diagram showing an example of listing. [Figure 12] Diagram showing an example of part-of-speech analysis. [Figure 13] Diagram showing an example of stop words. [Figure 14] Flowchart showing the details of the processing operations in steps S6 and S7 of FIG. 10. [Figure 15] Diagram showing an example of the emission factor addition data generated in step S11 of FIG. 10.

Mode for Carrying Out the Invention

[0010] Hereinafter, embodiments of an information processing apparatus and an information processing method will be described with reference to the drawings. Hereinafter, the description will focus on the main components of the information processing apparatus, but there may be components and functions that are not shown or described in the information processing apparatus. The following description does not exclude components and functions that are not shown or described.

[0011] FIG. 1 is a block diagram showing a schematic configuration of an information processing apparatus 1 according to an embodiment. The processing of the information processing apparatus 1 according to an embodiment is executed, for example, by a computer. In this case, the block diagram shown in FIG. 1 is a functional block diagram of the processing executed by the computer. Alternatively, the information processing apparatus 1 according to an embodiment can also be configured by a semiconductor chip or discrete circuits. In this case, each block shown in FIG. 1 is composed of a hardware circuit incorporated in a semiconductor chip or discrete circuits.

[0012] As shown in FIG. 1, the information processing apparatus 1 according to an embodiment includes a classification unit 2, a data conversion unit 3, an inference unit 4, a master selection unit 5, a search unit 6, and a calculation unit 7.

[0013] The classification unit 2 classifies each of a plurality of accounting data into classification items including at least one of a major classification, a middle classification, or an abstract. The classification items are items representing the characteristics of the accounting data. The accounting data is data related to the flow of money due to the activities of a business operator. The accounting data includes information such as, for example, year, date, business operator name, aggregation unit, activity content, activity volume, unit, etc. The activity volume includes, for example, an amount of money or a quantity of goods. The unit is the unit of the activity volume.

[0014] The process of the classification unit 2 classifying the accounting data into classification items may be a process manually performed by an operator or an automated process using a computer or the like. The abstract included in the classification items is an accounting item that more specifically represents the middle classification accounting item. The middle classification is an accounting item that more specifically represents the major classification accounting item. That is, the abstract accounting item is included in the middle classification accounting item, and the middle classification accounting item is included in the major classification accounting item.

[0015] There may be a case where the accounting data is classified into classification items including the three of major classification, middle classification, and abstract, and a case where it is classified into classification items not including any one or more of major classification, middle classification, or abstract. Also, depending on the accounting data, there may be a case where it is classified into classification items other than major classification, middle classification, and abstract.

[0016] The data transformation unit 3 converts the classification items classified by the classification unit 2 into input data for a machine learning model. The machine learning model is, for example, a deep learning model derived from the BERT (Bidirectional Encoder Representations from Transformers) model, and is a model specialized for natural language processing of Japanese. The detailed processing operation of the data transformation unit 3 will be described later.

[0017] The inference unit 4 inputs the input data transformed by the data transformation unit 3 into a trained machine learning model, performs inference using the machine learning model, and obtains greenhouse gas emission-related information, which is the inference result output from the machine learning model. In this specification, emission-related information may be referred to as linked content. As will be described later, the linked content is linked to emission coefficients in the master data.

[0018] The master selection unit 5 selects a master data from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, that corresponds to the emission-related information (linking content) obtained by the inference unit 4. Each master data set contains data that includes the correspondence between emission-related information (linking content) and emission factors. Multiple master data sets are generated based on various databases containing emission factors, such as those provided by the National Institute for Environmental Studies, the Ministry of the Environment, the Tokyo Metropolitan Government Environmental Bureau, or the National Institute of Advanced Industrial Science and Technology. The specific contents of the master data will be described later.

[0019] The search unit 6 searches for emission coefficients corresponding to the emission-related information (linking details) obtained by the inference unit 4 from the master data selected by the master selection unit 5.

[0020] The calculation unit 7 calculates greenhouse gas emissions based on the emission factors retrieved by the search unit 6. More specifically, for each of the multiple accounting data sets, the calculation unit 7 calculates the emissions by multiplying the emission factor retrieved by the search unit 6 by the corresponding activity level.

[0021] An information processing device 1 according to one embodiment may include an additional information generation unit 8, as shown in Figure 1. The additional information generation unit 8 generates emission factor additional data for each of the multiple accounting data, which includes a classification item, information from the master data selected by the master selection unit 5, and the emission factor retrieved by the search unit 6. Specific examples of emission factor additional data will be described later.

[0022] Figure 2 illustrates greenhouse gas emissions in a supply chain. A supply chain refers to a series of activities carried out by businesses, such as raw material procurement, manufacturing, logistics, sales, and disposal. In this specification, greenhouse gas emissions generated in a supply chain are referred to as supply chain emissions.

[0023] Supply chain emissions are classified into Scope 1, Scope 2, and Scope 3. Greenhouse gas emissions upstream and downstream of the supply chain are Scope 3, while greenhouse gas emissions from the business itself are Scope 1 and Scope 2.

[0024] Scope 1 refers to direct greenhouse gas emissions by the business itself. Scope 2 refers to indirect greenhouse gas emissions resulting from the use of electricity, heat, and steam supplied by other companies. Scope 3 refers to indirect greenhouse gas emissions other than those in Scope 1 and Scope 2, which are greenhouse gas emissions by other businesses related to the business's activities.

[0025] Scope 3 is classified into 15 categories and any number of other categories. The other categories are unique to the basic guidelines of the Ministry of the Environment and the Ministry of Economy, Trade and Industry. The 15 categories are consistent with the Scope 3 criteria of the GHG Protocol. The greenhouse gases covered are carbon dioxide (CO2), methane (CH4), nitrous oxide (N2O), hydrofluorocarbons (HFCs), perfluorocarbons (PFCs), sulfur hexafluoride (SF6), and nitrogen trifluoride (NF3). In this specification, the CO2 equivalent amount of these greenhouse gases is referred to as "CO2," and "CO2 equivalent emissions" are referred to as "emissions."

[0026] Upstream Scope 3 in the supply chain includes categories 1 through 8. Category 1 is emissions associated with activities from the manufacturing of raw materials, parts, containers, and packaging. Category 2 is emissions associated with the construction or manufacture of capital goods by the business operator (the company). Category 3 is emissions associated with the upstream processes of procured fuel (extraction or refining, etc.) and upstream processes of procured electricity, etc. (extraction or refining of fuel used for power generation, etc.). Category 4 is (1) emissions associated with logistics (transportation, handling, storage) from suppliers of products and services purchased in the reporting year to the business operator (the company), and (2) emissions associated with logistics services other than (1) purchased in the reporting year (emissions associated with logistics at the expense of the business operator). Category 5 is emissions associated with the transportation and processing of waste generated by the business operator (the company). Category 6 is emissions associated with employee business trips. Category 7 is emissions associated with employee travel when commuting to and from work. Category 8 includes emissions resulting from the operation of leased assets that the business operator (the company itself) leases (excluding cases where calculations are made using Scope 1 and Scope 2).

[0027] Scope 3 downstream in the supply chain includes categories 9 through 15. Category 9 is emissions associated with the logistics (transportation, handling, storage, and sale) of products sold by the business to the end consumer (limited to those not borne by the business). Category 10 is emissions associated with the processing of intermediate products by the business. Category 11 is emissions associated with the use of products by users (consumers, businesses). Category 12 is emissions associated with the disposal of products by users (consumers, businesses). Category 13 is emissions associated with the application of leased assets owned by the business as a lessor and leased to others. Category 14 is emissions from franchisees. Category 15 is emissions associated with the operation of investments.

[0028] Supply chain emissions classified as Scope 1, Scope 2, and Scope 3 are registered in multiple master data sets. These multiple master data sets are generated based on databases of emission intensity (emission factors) published by countries or organizations. The multiple master data sets include, for example, seven master data sets. In this specification, these seven master data sets are referred to as Master Data Sets 1 through 7. Figures 3 through 9 show the data structure of Master Data Sets 1 through 7.

[0029] The first master data shown in Figure 3 is an emission intensity database estimated using the GLIO (Global Link Input-Output) model provided by the National Institute for Environmental Studies. Figure 3 is an excerpt of the first master data. The first master data has the following items: row ID, row code, department name, unit, and GLIO1 to GLIO5. The department name refers to, for example, the specific product or service name purchased by the business operator. The department name corresponds to emission-related information (linked content). GLIO1 is the price of the purchased product or service including shipping costs, and is the selling price of the producer without wholesale or retail. GLIO2 is the price including shipping costs, and is the purchase price of the buyer with wholesale or retail. GLIO3 is the selling price of the producer without shipping costs and without wholesale. GLIO4 is the purchase price of the buyer with wholesale or retail, without shipping costs, in accordance with the Ministry of the Environment's DB[5]. GLIO5 is the selling price of the producer in accordance with the Ministry of the Environment's DB[5]. In this embodiment, the emission coefficients are determined in the order of priority GLIO2 → GLIO4 → GLIO5.

[0030] The second master data shown in Figure 4 is data extracted from the emission intensity database provided by the Ministry of the Environment, containing emission factors for Scope 3 categories 4, 6, and 7. Figure 4 is an excerpt of the second master data. As shown in Figure 4, the second master data has the fields for linking content, emission factor_unit, and emission factor. The second master data is generated by extracting parts of several other master data, as indicated in the source column.

[0031] The third master data shown in Figure 5 is data provided by the Ministry of the Environment, including emission factors for each electric utility company. As shown in Figure 5, the third master data includes items such as fiscal year, emission factor, unit, and source.

[0032] The fourth master data shown in Figure 6 is data including emission factors for the calculation, reporting, and publication system (SHK system) provided by the Ministry of the Environment. Figure 6 is an excerpt of the fourth master data. As shown in Figure 6, the fourth master data has the following items: fiscal year, database version, sheet, linkage content 2, linkage content, emission factor, unit, and source.

[0033] The fifth master data shown in Figure 7 is a list of coefficients provided by the Tokyo Metropolitan Government Bureau of Environment under the Global Warming Countermeasures Reporting System. As shown in Figure 7, the fifth master data includes the following items: fiscal year, linked content, emission coefficient, unit, and source.

[0034] The sixth master data shown in Figure 8 is data issued by the Agency for Natural Resources and Energy based on coefficients from the IDEAv3.1 inventory database provided by the National Institute of Advanced Industrial Science and Technology, and includes the unit calorific value of each energy type and the emission coefficient per unit price calculated based on the unit price of each energy type. The sixth master data is fuel emission coefficient data. As shown in Figure 8, the sixth master data has the following items: year, fuel name, IDEA product code, combustion energy_emission coefficient, combustion energy_unit, unit calorific value_high, unit calorific value_unit, fuel_average unit price, fuel_unit, usage_emission coefficient, usage_unit, monetary emission coefficient_t-co2eq / million yen, and data source_emission coefficient. Of these, the fuel name corresponds to the linked content.

[0035] The seventh master data shown in Figure 9 is an excerpt from the inventory database IDEAv3.1 mentioned above. As shown in Figure 9, the seventh master data includes emission factors for various items related to water supply, cement, and general lumber. Water supply, cement, and general lumber are the linked items.

[0036] As described above, the first to seventh master data all include emission-related information (linking details) and emission coefficients. It should be noted that the multiple master data in this embodiment do not necessarily need to include all of the first to seventh master data described above; data other than the first to seventh master data may be used as master data, but they must include emission-related information (linking details) and emission coefficients.

[0037] The master selection unit 5 selects master data containing emission-related information (linked content) acquired by the inference unit 4. For example, if the linked content is "advertising," "office supplies," "goods rental business (excluding rental vehicles)," "mobile telecommunications," "bolts, nuts, and springs," "petroleum products," "piping work accessories, powders, metal products, and tools," or "metal products for construction," the master selection unit 5 selects the first master data (GLIO). Also, if the linked content is, for example, "Category 4_Road freight transport (excluding private transport)," "Category 5_Waste disposal (public)★★," "Category 5_Sewerage★★," "Category 6_Passenger railway," or "Category 7_Passenger railway," the master selection unit 5 selects the second master data (Categories 4-6, etc.). Also, if the linked content is, for example, "Scope2_Electricity," the master selection unit 5 selects the third master data (by electricity provider). Also, if the linked content is, for example, "Scope2_Chilled water," the master selection unit 5 selects the fourth master data (SHK system). Furthermore, the master selection unit 5 selects the sixth master data (fuel emission coefficient) if, for example, the linked content is "Scope1_LPG". Also, the master selection unit 5 selects the seventh master data (IDEAv3.1) if, for example, the linked content is "water supply (unit m3)", "cement, 4 digits", or "general sawmilling, 4 digits".

[0038] Figure 10 is a flowchart showing the processing operation of the information processing device 1 according to one embodiment. First, the business operator's accounting data is entered (step S1). A file containing multiple accounting data may be entered, or each accounting data may be entered individually.

[0039] Next, in the classification unit 2, the entered accounting data is classified into classification items that include at least one of the major category, medium category, or description (step S2). For example, major categories include utilities, miscellaneous expenses, commuting expenses, business trip expenses, office supplies, miscellaneous expenses, communication and transportation expenses, XX factory, etc. Medium categories include, for example, utilities, commuting expenses, business trip expenses, XX department XX section, etc. Descriptions include, for example, "XX office gas bill," "XX office electricity bill," "XX office water bill," etc.

[0040] Next, the data conversion unit 3 lists the classification items classified in the classification unit 2 (step S3). Figure 11 shows an example of listing. In listing, the text information of the major category, minor category, and summary is concatenated to generate a single text item. In the example in Figure 11, the major category is "Water and electricity expenses", the minor category is "Energy expenses", and the summary is "Gas bill for XX business". In this case, listing generates a single text item: "Water and electricity expenses, Energy expenses, Gas bill for XX business". As shown in Figure 11, in listing, each text item is separated by a comma, and a single text item is generated by concatenating multiple text items.

[0041] Next, the data conversion unit 3 performs part-of-speech analysis on one of the listed text pieces (step S4 in Figure 10). Figure 12 shows an example of part-of-speech analysis. In part-of-speech analysis, morphological analysis of the text piece is performed, followed by part-of-speech decomposition. In the example in Figure 12, morphological analysis is performed on the text piece "Water, electricity, utilities, XX business gas bill," and the text piece "Water, electricity, expenses, XX, business, gas, bill" is generated. The text piece divided by morphological analysis is called a subword.

[0042] Next, a part-of-speech analysis is performed, breaking it down into text information containing parts of speech such as "water / noun, utilities / noun, expense / noun, XX / noun, business / noun, gas / noun, substitute / noun". In the example in Figure 12, all parts of speech are nouns, but depending on the text information, it may be broken down into parts of speech such as "verb," ​​"particle," "auxiliary verb," ​​or "symbol."

[0043] Next, the data conversion unit 3 removes stop words (step S5 in Figure 10). Figure 13 shows an example of stop words. Stop words are characters or strings separated by commas. Stop words are characters or strings that are not necessary for natural language processing. Stop words included in the text information analyzed in step S4 are removed.

[0044] Next, the data conversion unit 3 performs tokenization of the text information (step S6 in Figure 10). In step S6, the words that have undergone morphological analysis are divided into tokens and assigned token IDs. The token ID is identification information that identifies each token.

[0045] Next, the data transformation unit 3 performs vectorization (step S7 in Figure 10). In step S7, a vector is generated with the ID of each token as its element. Vectorization generates the input data that will be fed into the deep learning model.

[0046] Next, in the inference unit 4, the input data generated in step S7 is input to the trained deep learning model to perform inference, and the association content, which is output related to output from the deep learning model, is obtained (step S8 in Figure 10).

[0047] Next, the master selection unit 5 selects the master data from the first to seventh master data that corresponds to the linking content obtained in step S8 (step S9 in Figure 10).

[0048] Next, the search unit 6 searches for the emission coefficient corresponding to the linked content from the master data selected in the master selection unit 5 (step S10 in Figure 10).

[0049] Next, the additional information generation unit 8 generates emission coefficient additional data that includes classification items, information from the master data selected by the master selection unit 5, and emission coefficients retrieved by the search unit 6 (step S11 in Figure 10).

[0050] Next, the calculation unit 7 converts the units (step S12 in Figure 10). Here, for example, the units are standardized to tCO2 / million yen.

[0051] Next, the CO2 emissions are calculated by multiplying the input values ​​included in the emission factor data by the emission factor (step S13 in Figure 10). The input values ​​represent the amount or quantity of activity corresponding to the classification item.

[0052] Figure 14 is a flowchart detailing the processing operations of steps S6 and S7 in Figure 10. First, the input data, which has been divided into words by morphological analysis, is divided into subwords, which are smaller units than words (step S21). For the processing in step S21, WordPiece is used, for example. Each subword is called a token.

[0053] Next, special tokens are inserted at the beginning and end of the token (step S22). The special tokens inserted at the beginning and end of the token are called CLS and SEP, respectively.

[0054] Next, each of the multiple tokens is converted into an ID (hereinafter referred to as token ID), and padding is generated to equalize the length of the batch processing for processing multiple tokens together (step S23). A vector is generated from the multiple token IDs and padding.

[0055] Next, an attention mask is generated to distinguish between multiple token IDs and padding contained in the vector (step S24). The vectorized input data, including the multiple token IDs and attention mask, is input into the machine learning model (step S25).

[0056] Figure 15 shows an example of emission factor data generated in step S11 of Figure 10. The emission factor data in Figure 15 includes the following items: year, company name, aggregation unit, major category, minor category, description, input value, unit, linking content, emission factor, and emission factor unit. The year, company name, aggregation unit, input value, and unit included in the emission factor data in Figure 15 are accounting data entered in step S1 of Figure 10. The major category, minor category, and description in Figure 15 are classification items used to classify the accounting data in step S2 of Figure 10.

[0057] Each row in the emission factor-added data represents different accounting data. By multiplying the input value of each row by the emission factor, the amount of CO2 emissions can be calculated.

[0058] In this embodiment, accounting data is classified into classification items consisting of major categories, minor categories, and summaries. Master data matching the classification items is selected, and the corresponding linking content and emission coefficients are selected from the selected master data. The greenhouse gas emissions are then calculated from the selected emission coefficients. Furthermore, emission coefficient-added data is generated, which includes the accounting data, classification items, and the selected linking content and emission coefficients. This allows for the simple and accurate calculation of greenhouse gas emissions corresponding to the accounting data.

[0059] At least a part of the information processing device described in the above-described embodiment may be configured as hardware or as software. In the case of software configuration, a program that realizes at least a part of the functions of the information processing device may be stored on a recording medium such as a flexible disk or optical disk, and loaded into a computer for execution. The recording medium is not limited to removable ones such as magnetic disks or optical disks, but may also be a fixed recording medium such as a hard disk drive or memory.

[0060] Furthermore, a program that implements at least some of the functions of the information processing device may be distributed via communication lines such as the Internet (including wireless communication). In addition, the program may be encrypted, modulated, or compressed and distributed via wired or wireless lines such as the Internet, or stored on a recording medium.

[0061] The embodiments of the present invention are not limited to the individual embodiments described above, but also include various modifications that a person skilled in the art could conceive, and the effects of the present invention are not limited to those described above. In other words, various additions, modifications, and partial deletions are possible as long as they do not depart from the conceptual idea and spirit of the present invention derived from the claims and their equivalents.

[0062] [Note] [Item 1] A classification unit that classifies each of multiple accounting data into classification items that include at least one of the following: major category, minor category, or description, A data conversion unit converts the classification items classified by the classification unit into input data for a machine learning model, An inference unit that inputs the aforementioned input data into the trained machine learning model and obtains greenhouse gas emission-related information inferred by the machine learning model, A master selection unit selects a master data from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, that corresponds to the emission-related information acquired by the inference unit. A search unit searches for emission coefficients corresponding to the emission-related information obtained by the inference unit from the master data selected by the master selection unit, The system includes a calculation unit that calculates the greenhouse gas emissions based on the emission coefficients retrieved by the search unit, Information processing device. [Item 2] The aforementioned multiple master data are generated based on a database containing emission factors provided by at least one of the following: the National Institute for Environmental Studies, the Ministry of the Environment, the Tokyo Metropolitan Government Bureau of Environment, and the National Institute of Advanced Industrial Science and Technology. The information processing device described in item 1. [Item 3] For each of the aforementioned plurality of accounting data, the system includes an additional information generation unit that generates emission coefficient additional data including the classification item, the information of the master data selected by the master selection unit, and the emission coefficient retrieved by the search unit. An information processing device as described in item 1 or 2. [Item 4] The summary included in the aforementioned classification item is an accounting item that more specifically represents the accounting item of the aforementioned sub-classification, The subcategories included in the aforementioned classification items are accounting items that more specifically represent the aforementioned major classifications. An information processing device as described in any one of items 1 to 3. [Item 5] The data conversion unit sequentially lists the text information contained in each classification item classified by the classification unit and performs part-of-speech analysis, converts the text information into multiple tokens, and generates the input data based on a vector whose elements are the identification information of each of the multiple tokens. An information processing device as described in any one of items 1 through 4. [Item 6] When performing part-of-speech analysis, the data conversion unit first performs morphological analysis on the listed text information, and then performs part-of-speech parsing. The information processing device described in item 5. [Item 7] The data conversion unit, after performing the part-of-speech analysis, removes words from the text information that are unnecessary for conversion into tokens. Information processing device as described in item 5 or 6. [Item 8] Each of the aforementioned multiple accounting data includes an activity level representing the amount or quantity of the corresponding classification item, The calculation unit calculates the emission amount for each of the plurality of accounting data by multiplying the emission coefficient retrieved by the search unit by the corresponding activity amount. An information processing device as described in any one of items 1 through 7. [Item 9] The aforementioned machine learning model is a deep learning model derived from the BERT (Bidirectional Encoder Representations from Transformers) model. An information processing device as described in any one of items 1 through 8. [Item 10] The aforementioned multiple master data include emission factors in Scope 1, Scope 2, and Scope 3 in the supply chain. An information processing device as described in any one of items 1 through 9. [Item 11] Computers The steps include classifying each of the multiple accounting data into a classification item that includes at least one of the following: major category, minor category, or description, The steps include converting the classified classification items into input data for a machine learning model, The steps include inputting the aforementioned input data into the trained machine learning model to obtain greenhouse gas emission-related information inferred by the machine learning model, The steps include selecting a master data set corresponding to the acquired emission-related information from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, The steps include: searching for emission coefficients corresponding to the acquired emission-related information from the selected master data; The process includes the step of calculating the greenhouse gas emissions based on the emission factors that have been searched, Information processing methods.

[0063] The aspects of this disclosure are not limited to the individual embodiments described above, but include various modifications that a person skilled in the art could conceive, and the effects of this disclosure are not limited to those described above. In other words, various additions, modifications, and partial deletions are possible, as long as they do not depart from the conceptual idea and spirit of this disclosure derived from the claims and their equivalents. [Explanation of symbols]

[0064] 1. Information processing unit, 2. Classification unit, 3. Data conversion unit, 4. Inference unit, 5. Master selection unit, 6. Search unit, 7. Calculation unit, 8. Additional information generation unit

Claims

1. A classification unit that classifies each of multiple accounting data into classification items that include at least one of the following: major category, minor category, or summary. A data conversion unit converts the classification items classified by the classification unit into input data for a machine learning model, An inference unit that inputs the aforementioned input data into the trained machine learning model and obtains greenhouse gas emission-related information inferred by the machine learning model, A master selection unit selects a master data from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, that corresponds to the emission-related information acquired by the inference unit. A search unit searches for emission coefficients corresponding to the emission-related information obtained by the inference unit from the master data selected by the master selection unit, The system includes a calculation unit that calculates the greenhouse gas emissions based on the emission coefficients retrieved by the search unit, Information processing device.

2. The aforementioned multiple master data are generated based on a database containing emission factors provided by at least one of the following: the National Institute for Environmental Studies, the Ministry of the Environment, the Tokyo Metropolitan Government Bureau of Environment, and the National Institute of Advanced Industrial Science and Technology. The information processing apparatus according to claim 1.

3. For each of the aforementioned plurality of accounting data, the system includes an additional information generation unit that generates emission coefficient additional data including the classification item, the information of the master data selected by the master selection unit, and the emission coefficient retrieved by the search unit. The information processing apparatus according to claim 1.

4. The summary included in the aforementioned classification item is an accounting item that more specifically represents the accounting item of the aforementioned sub-classification, The subcategories included in the aforementioned classification items are accounting items that more specifically represent the aforementioned major classifications. The information processing apparatus according to claim 1.

5. The data conversion unit sequentially lists the text information contained in each classification item classified by the classification unit and performs part-of-speech analysis, converts the text information into multiple tokens, and generates the input data based on a vector whose elements are the identification information of each of the multiple tokens. The information processing apparatus according to claim 1.

6. When performing part-of-speech analysis, the data conversion unit first performs morphological analysis on the listed text information, and then performs part-of-speech parsing. The information processing apparatus according to claim 5.

7. The data conversion unit, after performing the part-of-speech analysis, removes words from the text information that are unnecessary for conversion into tokens. The information processing apparatus according to claim 5.

8. Each of the aforementioned multiple accounting data includes an activity level representing the amount or quantity of the corresponding classification item, The calculation unit calculates the emission amount for each of the plurality of accounting data by multiplying the emission coefficient retrieved by the search unit by the corresponding activity amount. The information processing apparatus according to claim 1.

9. The aforementioned machine learning model is a deep learning model derived from the BERT (Bidirectional Encoder Representations from Transformers) model. The information processing apparatus according to claim 1.

10. The aforementioned multiple master data include emission factors in Scope 1, Scope 2, and Scope 3 in the supply chain. The information processing apparatus according to any one of claims 1 to 9.

11. Computers The steps include classifying each of the multiple accounting data into a classification item that includes at least one of the following: major category, minor category, or description, The steps include converting the classified classification items into input data for a machine learning model, The steps include inputting the aforementioned input data into the trained machine learning model to obtain greenhouse gas emission-related information inferred by the machine learning model, The steps include selecting a master data set corresponding to the acquired emission-related information from among multiple master data sets, each containing greenhouse gas emission factors in the global supply chain, The steps include: searching for emission coefficients corresponding to the acquired emission-related information from the selected master data; The process includes the step of calculating the greenhouse gas emissions based on the emission factors that have been searched, Information processing methods.

Images