SDS generation system, SDS generation method, and SDS generation program

Abstract

Efficiently generate Safety Data Sheets (SDS) for composite raw materials. [Solution] The SDS generation device 10 comprises an SDS generation system 1 that generates a Safety Data Sheet (SDS) for composite raw materials consisting of multiple materials, a material information acquisition unit 11 that acquires material SDSs which are SDSs for multiple materials, a text conversion unit 12 that acquires a text-converted SDS consisting of text extracted from a non-text-converted material SDS when the material SDS is not composed of text data, and a composition table generation unit 13 that generates a composition table for composite raw materials based on the material SDS or text-converted SDS composed of text data for each of the multiple materials, wherein the composition table is used for generating the SDS for composite raw materials and is information that associates the attribute values ​​of multiple items associated with each material with the material and with each of the common items common to all materials.

Description

Technical Field

[0001] One aspect of the present disclosure relates to an SDS generation system, an SDS generation method, and an SDS generation program.

Background Art

[0002] There is a demand for efficiently creating a Safety Data Sheet (SDS) for a composite raw material (product) composed of multiple materials. An SDS is a document that describes the dangerous and harmful information of chemical substances when an operator uses chemical substances and products containing chemical substances in the working environment and transfers or provides them to other operators. The purpose of the SDS is to inform all parties in the supply chain of the foreseeable risks caused by chemical substances and products containing chemical substances, and to prevent disasters and accidents to human health and the environment. For example, Patent Document 1 describes a technique for creating an SDS with a minimum of manual labor.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Conventionally, the SDS of a composite raw material has been created by referring to the SDS of the materials contained in the composite raw material and extracting and transcribing the necessary information. However, creating such an SDS for a composite raw material has required a great deal of labor.

[0005] Therefore, the present invention has been made in view of the above problems, and an object thereof is to efficiently generate an SDS for a composite raw material.

Means for Solving the Problems

[0006] An SDS generation system relating to one aspect of this disclosure is an SDS generation system that generates a Safety Data Sheet (SDS) for a composite raw material consisting of multiple materials, comprising: a material information acquisition unit that acquires material SDSs which are SDSs for multiple materials; a text conversion unit that, when the material SDS is not composed of text data, acquires a text-converted SDS consisting of text extracted from the non-text data material SDS; and a composition table generation unit that generates a composition table for the composite raw material based on the material SDS or text-converted SDS composed of text data for each of the multiple materials, wherein the composition table is used for generating the SDS for the composite raw material and is information that associates the attribute values ​​of multiple items associated with each material with the material and with each of the common items common to all materials.

[0007] An SDS generation method relating to one aspect of the present disclosure is an SDS generation method performed by an SDS generation system comprising at least one processor and generating a safety data sheet for a composite raw material consisting of multiple materials, comprising: a material information acquisition step of acquiring a material SDS which is an SDS of multiple materials; a text conversion step of acquiring a text-based SDS consisting of text extracted from a non-text-based material SDS when the material SDS is not composed of text data; and a composition table generation step of generating a composition table for a composite raw material based on a material SDS or text-based SDS composed of text data for each of the multiple materials, wherein the composition table is used for generating an SDS for a composite raw material and is information that associates the attribute values ​​of multiple items associated with each material with the material and with each of the common items common to all materials.

[0008] An SDS generation program relating to one aspect of this disclosure is an SDS generation program for causing a computer to function as an SDS generation system for generating safety data sheets for composite raw materials consisting of multiple materials, and causes the computer to execute the following steps: a material information acquisition step for acquiring material SDS which are SDS for multiple materials; a text conversion step for acquiring a text-based SDS consisting of text extracted from a non-text-based material SDS when the material SDS is not composed of text data; and a composition table generation step for generating a composition table for composite raw materials based on the material SDS or text-based SDS composed of text data for each of the multiple materials, wherein the composition table is used for generating the SDS for composite raw materials and is information that associates the attribute values ​​of multiple items associated with each material with that material and with each of the common items common to all materials.

[0009] From this perspective, since the material SDSs of the multiple materials contained in the composite raw material are obtained as text data, it becomes easy to extract information from the material SDSs. Then, based on the material SDSs composed of text data, a composition table of the composite raw material is generated that includes the information for the composite raw material's SDS in an easily recognizable manner. Therefore, it becomes possible to efficiently generate SDSs for composite raw materials.

[0010] In SDS generation systems relating to other aspects, the text conversion unit may obtain a text-based SDS by converting a material SDS that is not composed of text data into image data and extracting text from the image data.

[0011] From this perspective, by converting material SDSs, which are not composed of text data, into image data, it becomes possible to extract consistent text.

[0012] In SDS generation systems relating to other aspects, the text generation unit may structure the text data contained in the material SDS and the text generation SDS, which are composed of text data, based on items.

[0013] From this perspective, structuring the text-based SDS makes it easier to process the information for subsequent composition table generation.

[0014] In SDS generation systems relating to other aspects, common items are associated with one or more items in the material SDS for each material, and the composition table generation unit may associate the attribute values ​​associated with the items in each material SDS with the common items associated with those items.

[0015] From this perspective, it becomes possible to generate a composition table in which the attribute values ​​of each material are appropriately associated with common items.

[0016] In SDS generation systems relating to other aspects, the composition table generation unit may extract attribute values ​​associated with items corresponding to common items from the material SDS or text-based SDS of each material, and input instruction information that instructs the large-scale language model to associate the extracted attribute values ​​with the material and the common item, thereby causing the large-scale language model to output a composition table.

[0017] From this perspective, by inputting appropriate instructional information into a large-scale language model and obtaining the composition table output from the model, a composition table of composite raw materials can be easily generated.

[0018] In SDS generation systems relating to other aspects, the instruction information may include information indicating the correspondence between items in the material SDS and common items.

[0019] From this perspective, in a large-scale language model, attribute values ​​that were associated with each item in the material SDS for each material are appropriately associated with the corresponding common items.

[0020] In SDS generation systems relating to other aspects, the composition table generation unit may refer to dictionary information that defines the correspondence between items in material SDS and common items based on material SDS composed of text data and text-based SDS, and associate the attribute values ​​associated with each item in each material SDS with the common items corresponding to those items.

[0021] From this perspective, by referencing dictionary information that defines the correspondence between items in material SDS and common items, the attribute values ​​associated with each item in the material SDS for each material can be accurately linked to the corresponding common items in the large-scale language model.

[0022] In other aspects of an SDS generation system, the SDS generation unit may further include an SDS generation unit that generates an SDS of composite raw materials based on at least a composition table, wherein the SDS is composed of multiple information items, and the unit generates descriptive information to be described in relation to each information item by at least referring to the composition table, and inputs SDS creation instruction information that at least includes descriptive instructions that instruct the SDS to include the generated descriptive information in relation to each information item into the SDS, thereby causing the large-scale language model to output an SDS of composite raw materials containing the descriptive information.

[0023] From this perspective, by inputting appropriate SDS creation instruction information into a large-scale language model and having it refer to at least the composition table to generate the information to be included in each information item of the SDS, it becomes possible to easily and efficiently generate SDSs for composite raw materials.

[0024] In the SDS generation system related to other aspects, by inputting GHS classification instruction information including at least a predetermined GHS determination condition for determining whether a chemical substance corresponds to a GHS classification item for each predetermined GHS classification item related to danger and harmfulness, and a GHS determination instruction for instructing to determine that a composite raw material corresponds to the GHS classification item when the information included in the composition table corresponds to the GHS determination condition, into a large language model, a GHS classification unit that causes the large language model to output the GHS classification item to which the composite raw material corresponds is further provided, and the SDS generation unit may include information indicating the GHS classification item to which the composite raw material corresponds in the SDS of the composite raw material.

[0025] According to such an aspect, by inputting GHS classification instruction information including GHS determination conditions and GHS determination instructions into a large language model, it is possible to easily determine the GHS classification of the composite raw material and include the determined GHS classification in the SDS.

Advantages of the Invention

[0026] According to one aspect of the present disclosure, it is possible to efficiently generate the SDS of the composite raw material.

Brief Description of the Drawings

[0027] [Figure 1] It is a block diagram showing an example of the configuration of the SDS generation system according to the embodiment and the functional configuration of the SDS generation device. [Figure 2] It is a hardware block diagram of the SDS generation device according to the embodiment. [Figure 3] It is a diagram schematically showing an example of the material SDS information stored in the material SDS information storage unit. [Figure 4] It is a diagram schematically showing the processing content for the material SDS prior to the processing of generating the composition table. [Figure 5] It is a diagram showing an example (part) of the instruction information (prompt) input into the large language model for generating the composition table. [Figure 6] It is a diagram showing an example of the dictionary information stored in the dictionary information storage unit. [Figure 7] This figure shows some other examples of instruction information (prompts) that are input into a large-scale language model for generating composition tables. [Figure 8] This figure schematically shows an example of a generated composition table. [Figure 9] This figure shows some examples of instruction information (prompts) that are input into a large-scale language model for determining GHS classification. [Figure 10] This figure shows some examples of instruction information (prompts) that are input into a large-scale language model for generating SDS. [Figure 11] This flowchart shows an example of the content of the SDS generation method in the SDS generation system according to the embodiment. [Modes for carrying out the invention]

[0028] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In the description of the drawings, the same or equivalent elements are denoted by the same reference numerals, and redundant descriptions are omitted.

[0029] Figure 1 is a block diagram showing an example of the configuration of an SDS generation system and the functional configuration of an SDS generation device according to an embodiment. The SDS generation system 1 is a system that generates Safety Data Sheets (SDS) for composite raw materials (products) consisting of multiple materials. An SDS is a document that contains information on the hazards of chemical substances, issued by businesses when providing chemical substances and products containing chemical substances, and has 16 items (information items).

[0030] Generally, the 16 items that make up an SDS are as follows:

[0031] (1) Chemical substance and company, (2) Summary of hazards, (3) Composition and component information, (4) First aid measures, (5) Fire prevention measures, (6) Leakage measures, (7) Handling and storage precautions, (8) Exposure prevention and protection measures, (9) Physical and chemical properties, (10) Stability and reactivity, (11) Hazard information, (12) Environmental impact information, (13) Disposal precautions, (14) Transportation precautions, (15) Applicable laws and regulations, (16) Other information.

[0032] The SDS generation system 1 is configured, for example, to include an SDS generation device 10. The server 30 is a computer having a Large Language Model (LLM) md, which will be described in detail later. The Large Language Model md is configured to include a user interface (UI) for interaction with the user and may be an interactive AI that enables text chat or voice chat with the user. Examples of such Large Language Model md include ChatGPT (GPT(registered trademark)-3.5, etc.), GPT-4, GPT-4o, GPT-4V, PaLM2, etc. In the example shown in Figure 1, the SDS generation device 10 and the server 30 are shown as separate computer devices, but the SDS generation device 10 and the server 30 may be configured as a single device.

[0033] Terminal T is operated by the user and is configured to communicate with the SDS generation device 10. Based on user input, Terminal T provides instructions to the SDS generation device 10, presents information received from the SDS generation device 10, etc. The user may also perform input operations directly through the input interface of the SDS generation device 10 without going through Terminal T, or may refer to and recognize information output from the output interface of the SDS generation device 10.

[0034] As shown in Figure 1, the SDS generation device 10 includes functional units 11-16 configured by a processor 101, a material SDS information storage unit 21, and a dictionary information storage unit 22. Each functional unit will be described later.

[0035] Figure 2 shows an example of the hardware configuration of a computer 100 that constitutes an SDS generation device 10 according to an embodiment. As an example, the computer 100 includes a processor 101, a main memory 102, an auxiliary memory 103, and a communication control device 104 as hardware components. The computer 100 that constitutes the SDS generation device 10 may further include an input device 105 such as a keyboard, touch panel, or mouse, and an output device 106 such as a display.

[0036] The processor 101 is a computing unit that executes the operating system and application programs. Examples of processors 101 include CPUs (Central Processing Units) and GPUs (Graphics Processing Units), but the type of processor 101 is not limited to these. For example, the processor 101 may be a combination of dedicated circuits. These dedicated circuits may be programmable circuits such as FPGAs (Field-Programmable Gate Arrays), or other types of circuits.

[0037] The main memory 102 is a device that stores programs for realizing the SDS generation device 10, calculation results output from the processor 101, and the like. The main memory 102 is composed of, for example, at least one of ROM (Read Only Memory) and RAM (Random Access Memory).

[0038] The auxiliary storage device 103 is generally a device capable of storing a larger amount of data than the main memory 102. The auxiliary storage device 103 is composed of a non-volatile storage medium such as a hard disk or flash memory. The auxiliary storage device 103 stores the SDS generation program P1 for enabling the computer 100 to function as an SDS generation device 10, as well as various other data.

[0039] The communication control device 104 is a device that performs data communication with other computers via a communication network. The communication control device 104 is composed of, for example, a network card or a wireless communication module.

[0040] Each functional element of the SDS generation device 10 is realized by loading the corresponding SDS generation program P1 onto the processor 101 or main memory 102 and having the processor 101 execute the program. The SDS generation program P1 contains code to realize each corresponding functional element. The processor 101 operates the communication control device 104 according to the SDS generation program P1 and performs data reading and writing in the main memory 102 or auxiliary storage device 103. Through such processing, each corresponding functional element of the computer is realized.

[0041] The SDS generation program P1 may be provided permanently recorded on a tangible, non-temporary recording medium such as a CD-ROM, DVD-ROM, or semiconductor memory. Alternatively, the SDS generation program P1 may be provided via a communication network as a data signal superimposed on a carrier wave.

[0042] Referring again to Figure 1, the SDS generation device 10 functionally comprises a material information acquisition unit 11, a text conversion unit 12, a composition table generation unit 13, a GHS classification unit 14, an SDS generation unit 15, and an SDS output unit 16. These functional units 11 to 16 may be configured in a single SDS generation device 10, as shown in Figure 1, or they may be distributed across multiple devices. Furthermore, the material SDS information storage unit 21 and the dictionary information storage unit 22 may be configured in the SDS generation device 10, as shown in Figure 1, or they may be configured as other devices accessible from the SDS generation device 10.

[0043] The material information acquisition unit 11 acquires material SDSs, which are SDSs for multiple materials constituting the composite raw material. Specifically, the material information acquisition unit 11 acquires information about the composite raw material (such as information about the materials constituting the composite raw material) based on user input, for example, and acquires material SDSs for the materials based on the acquired information about the composite raw material. In this embodiment, the material information acquisition unit 11 may acquire material SDSs that have been previously stored in the material SDS information storage unit 21.

[0044] Figure 3 is a schematic diagram showing an example of material SDS information stored in the material SDS information storage unit 21. As shown in Figure 3, the material SDS information storage unit 21 stores material SDS information SS. The material SDS information SS includes SDS (ss1, ss2, ss3, ...) for various materials. The material information acquisition unit 11 extracts and acquires the SDS of the materials constituting the composite raw material from the material SDS information storage unit 21. In addition to material SDS, the information acquired by the material information acquisition unit 11 may also include information on the composition of each material.

[0045] Furthermore, the data format of material SDSs is not necessarily standardized and may differ depending on the company or organization creating the SDS. Therefore, the data format of material SDSs pre-stored in the material SDS information storage unit 21 is not necessarily text data suitable for subsequent data processing.

[0046] The text conversion unit 12 extracts text from the non-text material SDS if the material SDS for the composite raw materials is not composed of text data, i.e., if it is in a format that makes text extraction impossible, such as image data or PDF format, and obtains a text-converted SDS consisting of the extracted text.

[0047] Figure 4 is a schematic diagram showing the processing of material SDS prior to the composition table generation process. Specifically, the text conversion unit 12 may obtain a text-converted SDS(ts) by extracting text from the material SDS(ss) using known OCR processing. Alternatively, the text conversion unit 12 may obtain a text-converted SDS(ts) by converting a material SDS(ss) that does not consist of text data into image data ps, and then extracting text from the image data ps. Converting a material SDS(ss) that does not consist of text data into image data ps enables the extraction of consistent text.

[0048] The text conversion unit 12 may further structure the material SDS(ss), which was composed of text-converted SDS(ts) and text data, in a predetermined format, and provide the material SDS converted into structured data to the composition table generation unit 13 for generating the composition table. By structuring the material SDS in the format of text data in this way, subsequent information processing for composition table generation becomes easier.

[0049] The composition table generation unit 13 generates a composition table for composite raw materials based on the material SDS(ss) or text-based SDS(ts) for each of the multiple materials, which are composed of text data. The composition table is used to generate the SDS for composite raw materials and is information that associates the attribute values ​​of multiple items associated with each material with that material, as well as with each of the common items common to all materials. In other words, the composition table is a table in which the common items are used as columns, and the attribute values ​​associated with each item in each material are mapped to each column.

[0050] The material SDS for each material does not necessarily associate the attribute values ​​of that material with common items; rather, they may be associated with items specific to each material SDS. In light of this, the composition table generation unit 13 associates the attribute values ​​associated with items in each material SDS with the common items associated with those items, based on the correspondence between one or more items in each material's SDS and common items. This makes it possible to generate a composition table in which the attribute values ​​of each material are appropriately associated with common items.

[0051] Specifically, the composition table generation unit 13 inputs instruction information for generating a composition table into the large-scale language model md, thereby causing the large-scale language model md to output a composition table. The instruction information includes instructions to extract attribute values ​​associated with items corresponding to common items from the material SDS or text-based SDS of each material, and to associate the extracted attribute values ​​with the material and the common items.

[0052] Figure 5 shows an example (partial) of prompts, which are instruction information input into a large-scale language model for generating a composition table. The prompt PT1 exemplified in Figure 5 includes a generation instruction i11 that instructs the generation of a composition table, correspondence information r1, common item information i12 that shows common items in the composition table, etc. The generation instruction i11 includes an instruction to generate a composition table in the large-scale language model md. The correspondence information r1 includes information that shows the correspondence between items in the material SDS and common items.

[0053] Specifically, the generation instruction i11 includes an instruction to refer to the correspondence information r1 to recognize the correspondence between items in the material SDS and common items, associate the attribute values ​​associated with the corresponding items in the material SDS with the common items listed in the common item information i12, and generate a table with common items as columns.

[0054] When such a prompt PT1 is input to the large-scale language model md, the attribute values ​​that were associated with each item in the material SDS for each material are appropriately associated with the corresponding common items in the large-scale language model.

[0055] The composition table generation unit 13 may refer to dictionary information and associate the attribute values ​​associated with each item in each material SDS with the common items corresponding to those items. In this embodiment, the composition table generation unit 13 may refer to dictionary information pre-stored in the dictionary information storage unit 22.

[0056] Figure 6 shows an example of dictionary information stored in the dictionary information storage unit 22. The dictionary information is information that defines the correspondence between items in material SDS and common items based on SDS composed of text data and text-based SDS. As shown in Figure 6, the dictionary information is information that associates common items with items used in each material SDS and the attribute values ​​associated with those items.

[0057] The correspondence between common items and individual items corresponds to the correspondence information r1 of prompt PT1 shown in Figure 5, and is information that absorbs fluctuations in the items corresponding to common items. The dictionary information may be generated, for example, based on existing material SDS.

[0058] Dictionary information is generated, for example, by identifying a predetermined section in an existing material SDS, extracting keywords from the identified section, and, if the extracted keywords are not yet registered in the generated dictionary information, registering the extracted keywords in the dictionary information along with their association with common items. The extraction of keywords from the material SDS and the association (classification) of keywords with common items may be performed, for example, by a pre-trained machine learning model (e.g., a text classification model).

[0059] Figure 7 shows an example (partial) of instruction information (prompts) input to a large-scale language model for generating a composition table when dictionary information is referenced. The prompt PT2 exemplified in Figure 7 includes a generation instruction i21 that instructs the generation of a composition table, common item information i22 that indicates common items in the composition table, etc. The generation instruction i21 includes instructions to generate a composition table for the large-scale language model md and to reference dictionary information.

[0060] Specifically, the generation instruction i21 includes an instruction to recognize the correspondence between items in the material SDS and common items by referring to dictionary information, associate the attribute values ​​associated with the corresponding items in the material SDS with the common items listed in the common item information i22, and generate a table with common items as columns. In such a prompt PT2, the correspondence information r1 that was included in prompt PT1 is no longer needed.

[0061] Figure 8 is a schematic diagram showing an example of a generated composition table. The composition table generation unit 13 generates a composition table TC, for example, as shown in Figure 8. As shown in Figure 8, the composition table TC is, for example, a table-formatted data, in which attribute values ​​corresponding to each common item such as product name, chemical name, CAS number, SDS issuance year, flammable liquid, etc. are associated with each material constituting the composite raw material. The composition table TC thus generated is used to generate the SDS for the composite raw material.

[0062] The GHS classification unit 14 determines the GHS classification items to be included in the SDS of the composite raw material. Specifically, the GHS classification unit 14 inputs GHS classification instruction information into the large-scale language model md, causing the large-scale language model md to output the GHS classification items to which the composite raw material belongs.

[0063] Figure 9 shows an example (part of it) of GHS classification instruction information (prompts) that are part of the prompts for generating an SDS and are input into a large-scale language model for determining the GHS classification. The GHS classification instruction information includes at least a GHS determination instruction and may further include GHS determination conditions.

[0064] The prompt PT3 shown in Figure 9 constitutes an example of GHS classification instruction information, including a GHS determination instruction i31 and a GHS determination condition i32. Furthermore, prompt PT3 includes a GHS item r3 listing the GHS classification items.

[0065] The GHS determination condition i32 includes predetermined conditions for determining whether a chemical falls under a GHS classification item for each predetermined GHS classification item related to hazards. The GHS determination instruction i31 is information that instructs the system to determine whether the composite raw material falls under the GHS classification item if the information contained in the composition table TC meets the GHS determination condition i32. The GHS classification unit 14 inputs prompt PT3 to the large-scale language model md, causing the large-scale language model md to output the GHS classification item to which the composite raw material belongs.

[0066] The SDS generation unit 15 generates an SDS for the composite raw materials based at least on the composition table. Specifically, the SDS generation unit 15 outputs the SDS for the composite raw materials to the large-scale language model md by inputting SDS creation instruction information into the large-scale language model md.

[0067] Figure 10 shows an example (part of it) of SDS creation instruction information (prompts) to be input into the large-scale language model md for SDS generation. As mentioned above, an SDS consists of multiple (for example, generally 16) information items, and the SDS creation instruction information includes at least a description instruction that instructs the generation of description information to be written in relation to each information item by referring at least to the composition table TC, and to include the generated description information in the SDS in relation to each information item.

[0068] The prompt PT4 shown in Figure 10 constitutes an example of SDS creation instruction information and includes description instructions i41 and i42. As shown in Figure 10, description instruction i41 includes the instruction to refer to the composition table TC and the product name of the composite raw material. Description instruction i42 includes information to be taken into consideration by the large-scale language model md in order to generate the description information for each information item based on the composition table TC. Description instruction i42 also includes an instruction to include the GHS classification items and their corresponding symbols determined by the GHS classification unit 14 in the SDS.

[0069] The SDS generation unit 15 outputs an SDS for composite raw materials, including the information for each information item, to the large-scale language model md by inputting the prompt PT4.

[0070] In this way, by inputting appropriate SDS creation instruction information into the large-scale language model md and having it refer to at least the composition table to generate the information to be included in each information item of the SDS, it becomes possible to easily and efficiently generate SDSs for composite raw materials. Note that the large-scale language model md used for generating the composition table, determining the GHS classification, and generating the SDS may be the same, or they may be different models constructed using different machine learning methods.

[0071] The SDS output unit 16 outputs the SDS of the composite raw materials generated by the SDS generation unit 15. The output method is not limited and may include transmission to a predetermined device (e.g., terminal T), display by a predetermined display means (e.g., the display of terminal T), or storage in a predetermined storage means.

[0072] Figure 11 is a flowchart showing an example of the contents of the SDS generation method in the SDS generation system 1 according to this embodiment. The SDS generation method is executed when the SDS generation program P1 is loaded into the processor 101 and the program is executed, thereby realizing each of the functional units 11 to 16.

[0073] In step S1, the material information acquisition unit 11 acquires material SDSs, which are SDSs for multiple materials constituting the composite raw material. In step S2, the text conversion unit 12 determines whether the material SDS acquired in step S1 includes material SDSs that are not composed of text data. If it is determined that material SDSs that are not composed of text data are included, the process proceeds to step S3. On the other hand, if it is not determined that material SDSs that are not composed of text data are included, the process proceeds to step S4.

[0074] In step S3, the text conversion unit 12 extracts text from the material SDS, which is not text data, using OCR technology or the like, and obtains a text-converted SDS consisting of the extracted text.

[0075] In step S4, the composition table generation unit 13 generates a composition table of composite raw materials based on the material SDS(ss) or text-based SDS(ts) for each of the multiple materials, which are composed of text data.

[0076] In step S5, the GHS classification unit 14 determines the GHS classification item to which the composite raw material belongs, based on at least the generated composition table.

[0077] In step S6, the SDS generation unit 15 generates an SDS for the composite raw materials based on at least the composition table and the determined GHS classification items.

[0078] In step S7, the SDS output unit 16 outputs the SDS of the composite raw materials generated by the SDS generation unit 15.

[0079] According to the SDS generation system 1, SDS generation device 10, SDS generation method, and SDS generation program P1 of this embodiment described above, the material SDSs of multiple materials contained in the composite raw material are acquired as text data, making it easy to extract information from the material SDSs. Based on the material SDS composed of text data, a composition table of the composite raw material is generated that includes information for the composite raw material's SDS in an easily recognizable manner. Therefore, it becomes possible to efficiently generate SDSs for composite raw materials.

[0080] The present invention has been described in detail above based on its embodiments. However, the present invention is not limited to the above embodiments. The present invention can be modified in various ways without departing from its spirit.

[0081] The gist of this disclosure is as follows: [1] to

[10] .

[0082] [1] A Safety Data Sheet (SDS) generation system for generating Safety Data Sheets (SDS) for composite raw materials consisting of multiple materials, A material information acquisition unit that acquires material SDSs, which are SDSs for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion unit obtains a text SDS consisting of text extracted from the material SDS which is not text data. A composition table generation unit that generates a composition table of composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of multiple items associated with each material with the material and with each of the common items common to all materials, An SDS generation system equipped with the following features.

[0083] [2] The text conversion unit converts the material SDS, which is not composed of text data, into image data, and obtains the text-converted SDS by extracting text from the image data. [1] The SDS generation system described above.

[0084] [3] The text conversion unit structures the text data contained in the material SDS, which is composed of text data, and the text data contained in the converted SDS, based on the items. The SDS generation system described in [1] or [2].

[0085] [4] Each of the aforementioned common items corresponds to one or more items in the material SDS for each material. The composition table generation unit associates the attribute values ​​associated with each material SDS item with the common item corresponding to that item. An SDS generation system as described in any one of the items [1] to [3].

[0086] [5] The composition table generation unit extracts attribute values ​​associated with items corresponding to the common items from the material SDS or the text SDS of each material, and inputs instruction information to the large-scale language model that instructs it to associate the extracted attribute values ​​with the material and the common items, thereby causing the large-scale language model to output the composition table. The SDS generation system described in [4].

[0087] [6] The instruction information includes information indicating the correspondence between the items in the material SDS and the common items. The SDS generation system described in [5].

[0088] [7] The composition table generation unit refers to dictionary information that defines the correspondence between the items in the material SDS and the common items based on the material SDS composed of text data and the text-based SDS, and associates the attribute values ​​associated with each item in each material SDS with the common items corresponding to those items. The SDS generation system described in [5].

[0089] [8] An SDS generation unit that generates an SDS of the composite raw material based on at least the composition table, wherein the SDS is composed of a plurality of information items, and the unit generates description information to be described in relation to each information item by referring at least to the composition table, and inputs SDS creation instruction information that includes at least a description instruction that instructs to include the generated description information in the SDS in relation to each information item, to a large-scale language model, thereby causing the large-scale language model to output an SDS of the composite raw material including the description information. An SDS generation system as described in any one of items [1] to [7].

[0090] [9] The system further comprises a GHS classification unit that, by inputting GHS classification instruction information into a large-scale language model, the large-scale language model outputs the GHS classification item to which the composite raw material belongs, by inputting GHS classification instruction information that includes at least predetermined GHS determination conditions for determining whether a chemical falls under a GHS classification item for each predetermined GHS classification item related to hazards, and GHS determination instructions that instruct the system to determine whether the composite raw material falls under the GHS classification item if the information contained in the composition table falls under the GHS determination conditions, the large-scale language model outputs the GHS classification item to which the composite raw material belongs. The SDS generation unit includes information indicating the GHS classification item to which the composite raw material belongs in the SDS of the composite raw material. [8] The SDS generation system described.

[0091]

[10] A method for generating an SDS, performed by an SDS generation system that includes at least one processor and generates a safety data sheet for a composite raw material consisting of multiple materials, A material information acquisition step of acquiring material SDS which are SDS for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion step is performed to obtain a text SDS consisting of text extracted from the material SDS, which is not composed of text data. A composition table generation step for generating a composition table of composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of a plurality of items associated with each material with that material and with each of the common items common to all materials, A method for generating SDS having the following characteristics.

[0092]

[11] An SDS generation program for causing a computer to function as an SDS generation system for generating safety data sheets for composite raw materials consisting of multiple materials, A material information acquisition step of acquiring material SDS which are SDS for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion step is performed to obtain a text SDS consisting of text extracted from the material SDS, which is not composed of text data. A composition table generation step for generating a composition table of composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of a plurality of items associated with each material with that material and with each of the common items common to all materials, An SDS generation program that causes the aforementioned computer to execute. [Explanation of symbols]

[0093] 1...SDS generation system, 10...SDS generation device, 11...Material information acquisition unit, 12...Text conversion unit, 13...Composition table generation unit, 14...GHS classification unit, 15...SDS generation unit, 16...SDS output unit, 21...Material SDS information storage unit, 22...Dictionary information storage unit, P1...SDS generation program, T...Terminal, TC...Composition table.

Claims

1. A Safety Data Sheet (SDS) generation system for generating Safety Data Sheets (SDS) for composite raw materials consisting of multiple materials, A material information acquisition unit that acquires material SDSs which are SDSs for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion unit obtains a text SDS consisting of text extracted from the material SDS, which is not text data. A composition table generation unit that generates a composition table of the composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of a plurality of items associated with each material with the material and with each of the common items common to all materials, An SDS generation system equipped with [the following features].

2. The text conversion unit converts the material SDS, which is not composed of text data, into image data, and obtains the text-converted SDS by extracting text from the image data. The SDS generation system according to claim 1.

3. The text conversion unit structures the material SDS, which is composed of text data, and the text data included in the text conversion SDS, based on the items. The SDS generation system according to claim 1 or 2.

4. Each of the aforementioned common items corresponds to one or more items in the material SDS for each material. The composition table generation unit associates the attribute values ​​associated with each material SDS item with the common item corresponding to that item. The SDS generation system according to claim 1.

5. The composition table generation unit extracts attribute values ​​associated with items corresponding to the common items from the material SDS or the text SDS for each material, and inputs instruction information to the large-scale language model that instructs it to associate the extracted attribute values ​​with the material and the common items, thereby causing the large-scale language model to output the composition table. The SDS generation system according to claim 4.

6. The instruction information includes information indicating the correspondence between the items in the material SDS and the common items, The SDS generation system according to claim 5.

7. The composition table generation unit refers to dictionary information that defines the correspondence between the items in the material SDS and the common items based on the material SDS composed of text data and the text-based SDS, and associates the attribute values ​​associated with each item in each material SDS with the common items corresponding to those items. The SDS generation system according to claim 5.

8. An SDS generation unit that generates an SDS of the composite raw material based on at least the composition table, wherein the SDS is composed of a plurality of information items, and the unit generates description information to be described in relation to each information item by referring at least to the composition table, and inputs SDS creation instruction information that includes at least a description instruction that instructs to include the generated description information in the SDS in relation to each information item, thereby causing the large-scale language model to output an SDS of the composite raw material including the description information. The SDS generation system according to claim 1.

9. The system further comprises a GHS classification unit that inputs GHS classification instruction information, which includes at least predetermined GHS determination conditions for determining whether a chemical falls under a GHS classification item for each predetermined GHS classification item related to hazards, and a GHS determination instruction that instructs the system to determine whether the composite raw material falls under a GHS classification item if the information contained in the composition table falls under the GHS determination conditions, into a large-scale language model, thereby causing the large-scale language model to output the GHS classification item to which the composite raw material belongs. The SDS generation unit includes information indicating the GHS classification item to which the composite raw material belongs in the SDS of the composite raw material. The SDS generation system according to claim 8.

10. A method for generating an SDS, which is performed by an SDS generation system comprising at least one processor and generating a safety data sheet for a composite raw material consisting of multiple materials, A material information acquisition step to acquire material SDSs which are SDSs for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion step is performed to obtain a text SDS consisting of text extracted from the material SDS, which is not composed of text data. A composition table generation step for generating a composition table of composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of a plurality of items associated with each material with the material and with each of the common items common to all materials, A method for generating an SDS (Software Data Sheet) having the following characteristics.

11. An SDS generation program for causing a computer to function as an SDS generation system for generating safety data sheets for composite raw materials consisting of multiple materials, A material information acquisition step to acquire material SDSs which are SDSs for the aforementioned multiple materials, If the material SDS is not composed of text data, a text conversion step is performed to obtain a text SDS consisting of text extracted from the material SDS, which is not composed of text data. A composition table generation step for generating a composition table of composite raw materials based on the material SDS or the text-based SDS, which is composed of text data for each of the plurality of materials, wherein the composition table is used for generating the SDS of the composite raw materials and is information that associates the attribute values ​​of a plurality of items associated with each material with the material and with each of the common items common to all materials, An SDS generation program that causes the aforementioned computer to execute.

Images