Information processing method and information processing device

Abstract

Provided is an information processing device that contributes to smooth execution of design work. According to the present invention, design information including a design drawing such as a circuit diagram and a specification is accepted, after which a language model is caused to generate design commands based on the design information. Design is performed on the basis of the design commands, and design results are acquired. The language model is caused to extract any defect included in the design results and estimate the cause of the defect. Literature is acquired by performing a search on the basis of the defect and cause. The language model is caused to generate human resource search information on the basis of the acquired literature. The human resource search information includes the skills required to resolve the defect, the time required to resolve the defect, etc. A search for a human resource suitable for resolving the defect is performed on the basis of the human resource search information. Defect-related information including the defect, the cause, and the retrieved literature is transmitted to the retrieved human resource to request resolution of the defect.

Description

Information processing method and information processing apparatus

[0001] One aspect of the present invention relates to an information processing device. Another aspect of the present invention relates to an information processing method using an information processing device. Furthermore, one aspect of the present invention relates to an information processing system including an information processing device.

[0002] Furthermore, one aspect of the present invention is not limited to the above-mentioned technical field. The technical field of one aspect of the invention disclosed herein relates to a product, a method, or a method of manufacture. Alternatively, one aspect of the present invention relates to a process, a machine, a manufacture, or a composition of matter. More specifically, examples of the technical fields of one aspect of the present invention disclosed herein include information processing devices, semiconductor devices, memory devices, methods for driving them, or methods for manufacturing them.

[0003] In the circuit design of semiconductor devices, Electronic Design Automation (EDA) tools, which automate the design process, are used. Automating the design process ensures development speed, product standards, and safety standards. Patent Document 1 discloses a method for automating circuit design.

[0004] In recent years, language models using artificial neural networks (ANN: Artificial Neural Network, hereinafter also simply referred to as neural networks) have been actively developed, and in particular, large language models (LLM: Large Language Model) have attracted attention. A large language model is a natural language processing model trained using a large amount of data. A large language model can realize, for example, a dialogue model that provides responses to user instructions. Non-Patent Document 1 discloses GPT-4 (registered trademark) (Generative Pre-trained Transformer 4) as a large language model, and also discloses ChatGPT as a dialogue model. Other examples of large language models include LaMDA (Language Model for Dialogue Applications), PaLM (Pathways Language Model), Llama2, and Llama3.

[0005] Japanese Unexamined Patent Application Publication No. 2024-170423

[0006] Summary of ChatGPT / GPT-4 Research and Perspective Towards the Future of Large Language Models, Yiheng Liu et al. (Submitted on 4 Apr 2023), [online], Internet <URL: https: / / arxiv.org / abs / 2304.01852>

[0007] In design work such as circuit design, advanced skills are required of designers, and the required skills also cover a wide range. For example, in the case of troubleshooting work, different skills are required depending on the type of failure, etc. In addition, although it is necessary to perform various tasks in design work, the aptitude for each task differs depending on the designer. For example, in the case of troubleshooting work, some designers are suitable for solving short-term problems, while others are suitable for solving long-term problems. From the above, in order to proceed with work smoothly, managers of design work need to appropriately assign work according to the skills and aptitude of each person.

[0008] One aspect of the present invention aims to provide an information processing device, an information processing method, and an information processing system that contribute to the smooth execution of design work. Alternatively, one aspect of the present invention aims to provide an information processing device, an information processing method, and an information processing system that support design work. Alternatively, one aspect of the present invention aims to provide an information processing device, an information processing method, and an information processing system that can improve the performance of semiconductor devices. Alternatively, one aspect of the present invention aims to provide an information processing device, an information processing method, and an information processing system that can reduce design costs. Alternatively, one aspect of the present invention aims to provide an information processing device, an information processing method, and an information processing system that can shorten the design period. Alternatively, one aspect of the present invention aims to provide a novel information processing device, an information processing method, and an information processing system.

[0009] Furthermore, the description of these problems does not preclude the existence of other problems. Moreover, one aspect of the present invention does not need to solve all of these problems. Other problems will naturally become apparent from the description in the specification, drawings, claims, etc., and it is possible to extract other problems from the description in the specification, drawings, claims, etc.

[0010] One aspect of the present invention is an information processing method comprising the first to seventh steps, wherein in the first step, design information comprising design drawings and specifications is received; in the second step, a first prompt for generating design commands based on the design information is created and sent to a language model; in the third step, design is performed based on the design commands and the design results are obtained; in the fourth step, a second prompt for extracting defects included in the design results and inferring the causes of the defects is created and sent to a language model; in the fifth step, a graph is created based on the design drawings; in the sixth step, a first embedding representation is obtained by performing graph embedding on the graph; and in the seventh step, information is obtained by searching for literature based on the defects, causes, and the first embedding representation.

[0011] Alternatively, in the above embodiment, the system may have eighth to tenth steps, wherein in the eighth step, a third prompt is created and transmitted to the language model for generating personnel search information to be used when searching for personnel suitable for resolving the problem, based on acquired literature; in the ninth step, personnel are searched based on the personnel search information; and in the tenth step, problem-related information, including the problem, its cause, and literature, is transmitted to the searched personnel.

[0012] Alternatively, in the above embodiment, the document comprises a main text, a documentary figure, and a second embedded representation obtained by performing graph embedding on the documentary figure, and in the seventh step, the defects and causes are compared with the main text, and the first embedded representation is compared with the second embedded representation.

[0013] Alternatively, in the above embodiment, one or both of the first embedding representation and the second embedding representation may be obtained using a graph neural network.

[0014] Alternatively, one aspect of the present invention comprises a first to tenth step, in the first step receiving design information including design drawings and specifications; in the second step creating a first prompt for generating design commands based on the design information and sending it to a language model; in the third step performing design based on the design commands and obtaining the design results; in the fourth step creating a second prompt for extracting defects included in the design results and inferring the causes of the defects and sending it to a language model; and in the fifth step obtaining multiple documents by searching based on the defects and causes. The information processing method involves, in the sixth step, obtaining similarity scores for defects and causes for each of the multiple documents; in the seventh step, extracting one or more documents for talent search from the multiple documents based on the similarity scores; in the eighth step, creating a third prompt to generate talent search information used when searching for personnel suitable for resolving the defect, based on the talent search documents, and sending it to a language model; in the ninth step, searching for personnel based on the talent search information; and in the tenth step, sending defect-related information, including defects, causes, and talent search documents, to the found personnel.

[0015] Alternatively, in the above embodiment, in the sixth step, a fourth prompt for calculating similarity may be created and sent to the language model.

[0016] Alternatively, one aspect of the present invention is an information processing device having a receiving unit, an output unit, and a processing unit, wherein the receiving unit has the function of receiving design information comprising design drawings and specifications, the output unit has the function of supplying a first prompt and a second prompt to a language model, respectively, and the processing unit has the function of creating a first prompt for generating design commands based on design information, performing design based on the design commands and obtaining design results, creating a second prompt for extracting defects included in the design results and inferring the causes of the defects, creating a graph based on the design drawings, obtaining a first embedding representation by performing graph embedding on the graph, and obtaining literature by searching based on defects, causes, and the first embedding representation.

[0017] Alternatively, in the above embodiment, the output unit may have a function to supply a third prompt to the language model, the output unit may have a function to output defect-related information including defects, causes, and literature, and the processing unit may have a function to create a third prompt for generating personnel search information used when searching for personnel suitable for resolving the defect based on acquired literature, a process for searching for personnel based on the personnel search information, and a process for transmitting defect-related information to the searched personnel via the output unit.

[0018] Alternatively, in the above embodiment, the document may include a main text, a documentary figure, and a second embedded representation obtained by performing graph embedding on the documentary figure, and the processing unit may have a function to compare the defect and its cause with the main text, and to compare the first embedded representation with the second embedded representation.

[0019] Alternatively, in the above embodiment, the processing unit may have a function to acquire one or both of the first and second embedding representations using a graph neural network.

[0020] Alternatively, in the above embodiment, the personnel search information may include the skills required to resolve the problem and the time required to resolve the problem.

[0021] Alternatively, in the above embodiment, the design drawing may be a circuit diagram.

[0022] According to one aspect of the present invention, an information processing device, an information processing method, and an information processing system can be provided that contribute to the smooth execution of design work. Alternatively, according to one aspect of the present invention, an information processing device, an information processing method, and an information processing system can be provided that support design work. Alternatively, according to one aspect of the present invention, an information processing device, an information processing method, and an information processing system can be provided that can improve the performance of semiconductor devices. Alternatively, according to one aspect of the present invention, an information processing device, an information processing method, and an information processing system can be provided that can reduce design costs. Alternatively, according to one aspect of the present invention, an information processing device, an information processing method, and an information processing system can be provided that can shorten the design period. Alternatively, according to one aspect of the present invention, a novel information processing device, an information processing method, and an information processing system can be provided.

[0023] Furthermore, the description of these effects does not preclude the existence of other effects. Moreover, one aspect of the present invention does not necessarily have to possess all of these effects. Other effects will become clear from the description in the specification, drawings, claims, etc., and it is possible to extract other effects from the description in the specification, drawings, claims, etc.

[0024] Figure 1 is a schematic diagram showing an example of an information processing system. Figure 2 is a block diagram showing an example of an information processing system. Figure 3 is a schematic diagram showing an example of a literature list. Figure 4 is a schematic diagram showing an example of a personnel list. Figure 5 is a flowchart showing an example of an information processing method. Figure 6A is a schematic diagram showing an example of design information. Figure 6B is a schematic diagram showing an example of a response statement. Figure 7 is a schematic diagram showing an example of an information processing method. Figure 8 is a schematic diagram showing an example of personnel search information. Figure 9 is a flowchart showing an example of an information processing method. Figure 10 is a flowchart showing an example of an information processing method.

[0025] Embodiments will be described in detail with reference to the drawings. However, it will be readily apparent to those skilled in the art that the present invention is not limited to the following description, and that its form and details can be modified in various ways without departing from the spirit and scope of the present invention. Accordingly, the present invention is not to be interpreted as being limited to the contents of the embodiments shown below. In the configuration of the invention described below, the same reference numerals are used in common across different drawings for the same parts or parts having similar functions, and repeated descriptions are omitted.

[0026] In the drawings attached to this specification, the components are classified by function and shown as independent blocks in block diagrams. However, in reality, it is difficult to completely separate the components by function, and one component may be involved in multiple functions.

[0027] In this specification and drawings, when the same reference numeral is used for multiple elements, and especially when it is necessary to distinguish them, the reference numeral may be accompanied by an identifying numeral such as "_1", "[n]", or "[m,n]". Furthermore, when describing a common matter for multiple elements with identifying numerals, or when it is not necessary to distinguish them, the identifying numeral may be omitted.

[0028] In this specification, the terms "first" and "second" may be used for convenience to understand the technical content or to identify each component. Therefore, the terms "first" and "second" do not limit the number of components. Nor do the terms "first" and "second" limit the order of the components. Furthermore, the terms "first" and "second" or the identification codes used in this specification may not correspond to the terms or identification codes in the claims of this patent.

[0029] (Embodiment) This embodiment describes an information processing system according to one aspect of the present invention. This embodiment also describes an information processing device included in the information processing system and an information processing method using the information processing system.

[0030] One aspect of the present invention relates to an information processing device used when performing design work, such as circuit design work. The information processing device according to one aspect of the present invention has a function to perform processing using an automated design tool such as an EDA tool. The automated design tool can perform design automatically based on design commands. In this specification, the automated design tool is defined as having a function to generate design results based on design commands. The information processing device according to one aspect of the present invention can acquire design results using the automated design tool.

[0031] In one embodiment of the present invention, an information processing system generates design commands based on design information using a language model. The design information includes design drawings and specifications. For example, when designing a circuit, the design drawings can be circuit diagrams. Circuit diagrams can be represented using hardware description languages ​​such as Verilog, SystemVerilog, VHDL, and SystemC. Furthermore, descriptions using hardware description languages ​​can be converted into netlists. Netlists may also be used as design drawings. For example, when designing a logic circuit, a netlist representing the connection relationships of logic gates included in the logic circuit can be used as a design drawing. Thus, design drawings may be in the form of text data or image data.

[0032] The design results include information indicating defects. For example, if a section that should be electrically insulated is short-circuited, the results include information indicating the short-circuited section. In one embodiment of the present invention, the information processing system extracts defects included in the design results and infers the cause of the defects using a language model. Subsequently, the information processing system according to one embodiment of the present invention searches for literature such as papers based on the defects and their causes. As a result, the information processing system according to one embodiment of the present invention obtains literature related to the defects and their causes. The literature to be searched can be stored, for example, in a database owned by the information processing system according to one embodiment of the present invention.

[0033] After obtaining the literature, talent search information is generated using a language model based on that literature. This talent search information is used to find individuals who can be requested to resolve a problem. For example, the talent search information includes the skills required to resolve the problem. It also includes, for example, the time required to resolve the problem.

[0034] An information processing device according to one aspect of the present invention searches for personnel suitable for resolving a problem based on personnel search information. Specifically, it searches for personnel suitable for resolving the problem from among the personnel included in a personnel list. The personnel list includes information that identifies each person, such as their name. The personnel list also includes information such as each person's skills, aptitude, and contact information.

[0035] Aptitude indicates, for example, whether someone is suited to solving short-term or long-term problems. For instance, someone who can complete many tasks in a short period is considered suited to solving short-term problems. On the other hand, someone who can handle difficult tasks, even if it takes some time, and who can solve problems that many others cannot, is considered suited to solving long-term problems. Contact information could be, for example, an email address.

[0036] An information processing device according to one embodiment of the present invention searches for a person suitable for resolving a problem and then transmits problem-related information to the contact information of the found person. This problem-related information includes, for example, the problem, its cause, and the literature found. The found person is then requested to resolve the problem.

[0037] As described above, in one embodiment of the present invention, after performing automated design, the resolution of defects can be requested from suitable personnel. For example, the resolution of the defects can be requested from personnel who possess the necessary skills and high aptitude for the task. Furthermore, literature and other materials related to the defects can be provided to such personnel. Thus, the information processing system in one embodiment of the present invention can contribute to the smooth execution of design work.

[0038] <Example of Information Processing System Configuration 1> Figure 1 is a schematic diagram showing an example of the configuration of an information processing system according to one aspect of the present invention. The information processing system according to one aspect of the present invention comprises an information processing device 10 and an information processing device 40. The information processing device 10 and the information processing device 40 are connected via a network 30 and can transmit and receive data.

[0039] The information processing device 10 has a function to perform processing using an automated design tool. An example of an automated design tool is an EDA tool. By using an EDA tool, for example, circuit design, specifically semiconductor circuit design, can be performed automatically. The automated design tool automatically performs design based on design commands and generates design results. The information processing device 10 can acquire the design results using the automated design tool.

[0040] The information processing device 40 has a function to perform processing using a language model. The language model has a function to generate a response sentence based on a prompt. A prompt can be described as an input sentence that causes the language model to perform a desired action.

[0041] A prompt is supplied, for example, from the information processing device 10 to the information processing device 40. The response sentence generated by the language model is supplied, for example, from the information processing device 40 to the information processing device 10. The information processing device 10 may also have a function to perform processing using the language model.

[0042] Furthermore, in one embodiment of the present invention, the information processing system may be configured so that a user can input information by directly operating the information processing device 10, or, as shown in Figure 1, the information processing device 10 may be configured so that information can be input using an information terminal 20 connected via a network 30.

[0043] The following describes an example configuration of the information processing device 10, the information processing device 40, the information terminal 20, and the network 30.

[0044] <Example Configuration of Information Processing Apparatus 10> Figure 2 is a block diagram showing an example configuration of the information processing apparatus 10. The information processing apparatus 10 includes a reception unit 110, a storage unit 120, a processing unit 130, an output unit 140, and a transmission path 150. Note that Figure 2 also shows an information terminal 20 and an information processing apparatus 40.

[0045] [Reception Unit 110] The reception unit 110 has a function of receiving data from outside the information processing apparatus 10. The reception unit 110 has a function of receiving data from the information terminal 20. The reception unit 110 has a function of receiving data representing, for example, a response sentence from the information processing apparatus 40.

[0046] The reception unit 110 has a function of supplying received data to one or both of the storage unit 120 and the processing unit 130 via the transmission path 150. As the reception unit 110, devices such as a wired communication port, a wireless communication port, and an optical communication port can be used, for example.

[0047] [Storage Unit 120] The storage unit 120 has a function of storing programs executed by the processing unit 130. Furthermore, the storage unit 120 may have a function of storing data created by the processing unit 130 (e.g., calculation results, analysis results, and inference results), data received by the reception unit 110, and the like.

[0048] The storage unit 120 includes a database 121 and a database 122. Documents are stored in the database 121. A talent list is stored in the database 122.

[0049] Figure 3 is a schematic diagram showing an example of documents 210 stored in the database 121. As shown in Figure 3, a plurality of documents 210 are stored in the database 121. The documents 210 are documents related to an object for which the information processing apparatus 10 performs automatic design. For example, if the information processing apparatus 10 has a function of performing automatic design of semiconductor circuits, the documents 210 are documents related to semiconductor circuit design.

[0050] Document 210 can be a technical document, such as a research paper. Alternatively, Document 210 may be a journal, book, patent document, utility model document, etc. Examples of patent documents include published patent gazettes and patent publications. Examples of utility model documents include utility model publications. Note that Document 210 is not limited to publicly known documents. For example, a document accessible only within a specific company may be considered Document 210. For example, a technical report, a technical work report, etc., may also be considered Document 210.

[0051] Document 210 comprises a main text 211, a date 212, a reference 213, and a drawing 215. The main text 211, the date 212, and the reference 213 are presented in text. The drawing 215 is also referred to as the document drawing.

[0052] The main text 211 includes, for example, the issues considered, research methods, research results, and discussion. For example, if reference 210 discloses a method for resolving defects that occur in design, the main text 211 may include the nature of the defect, the cause of the defect, and the method for resolving the defect. Figure 3 shows an example where the main text 211 includes the sentences, "This study revealed that the cause of defect aaa is bbb," and "Defect aaa was resolved by using the ccc method."

[0053] The year, month, and day 212 indicates, for example, the submission date of reference 210 if it is a submitted paper. In the example shown in Figure 3, it is indicated as "Received: ddmm yyyy" that reference 210 was submitted on yyyy / mm / dd.

[0054] Reference 213 lists the literature consulted, for example, in the writing of the paper. Reference 213 is identified by, for example, the author's name, publication year, publication name, etc. Note that Reference 213 may also be included in the main text 211.

[0055] Figure 215 is used, for example, to clearly illustrate the research. The main text 211 may include a description of Figure 215. In the example shown in Figure 3, Figure 1 is shown as Figure 215. Note that Reference 210 may include multiple Figures 215. For example, if Reference 210 includes two Figures 215, then Reference 210 will include Figure 1 and Figure 2.

[0056] Figure 3 shows a circuit diagram, specifically an example of Figure 215, or more precisely, Figure 1. The circuit diagram shown in Figure 3 illustrates a logic circuit. This circuit diagram shows the connection relationships of the logic gates included in the logic circuit.

[0057] Reference 210 may include the title, author, abstract, etc. These are presented in text. These may or may not be included in the main text 211.

[0058] Figure 4 is a schematic diagram showing an example of a personnel list 220 stored in the database 122. Multiple personnel information entries 221 are registered in the personnel list 220. Figure 4 shows personnel information entry 221[1] and personnel information entry 221[2].

[0059] Personnel information 221 represents information about the person performing the design. For example, if the information processing device 10 has a function to automatically design semiconductor circuits, personnel information 221 is information about the person performing the semiconductor circuit design. For example, if the design is performed in-house, personnel information 221 can be information about the employees of the department that performs the design. If the design is outsourced, personnel information 221 can be information about the person in charge at the outsourcing company.

[0060] Personnel information 221 includes information that identifies each person. In Figure 4, the name is shown as the information that identifies each person. Specifically, personnel information 221[1] shows information about a person whose name is "William Smith". Personnel information 221[2] shows information about a person whose name is "Mary Brown". Personnel information 221 may also include, for example, an identification number assigned to each person as information that identifies each person. Personnel information 221 may also include the department name, job title, etc., as information that identifies each person.

[0061] Figure 4 shows an example where the items represented by personnel information 221 include skills, aptitude, and contact information. Skills indicate, for example, the tasks that can be performed. For example, it may indicate the names and functions of tools that can be used.

[0062] Figure 4 shows an example where the item "skills" in personnel information 221[1] includes "ddd" and "e1e1e1". It also shows an example where the item "skills" in personnel information 221[2] includes "hhh" and "e2e2e2". As shown in Figure 4, skills can be shown, for example, in a bulleted list. "Skills" may also indicate the level of proficiency in each task. For example, proficiency can be indicated as A, B, and C in descending order of level. For instance, for a given tool, proficiency level C could be used for simple operations, proficiency level B for being able to use it proficiently, and proficiency level A for being able to use it effectively in any situation.

[0063] Aptitude indicates characteristics related to design work. For example, it can indicate whether someone is suited to solving short-term problems or long-term problems. For instance, someone who can complete many tasks in a short period of time can be considered suited to solving short-term problems. On the other hand, someone who can handle difficult tasks, even if it takes some time, and can solve problems that many others cannot solve, can be considered suited to solving long-term problems. In the example shown in Figure 4, the item "Aptitude" included in personnel information 221 [1] indicates that the person is suited to solving short-term problems. Also, the item "Aptitude" included in personnel information 221 [2] indicates that the person is suited to solving long-term problems.

[0064] The contact information indicates the recipient when sending information. For example, the contact information can be an email address. In the example shown in Figure 4, the "Contact Information" item in personnel information 221[1] is f1f1f1@ggg.com. Similarly, the "Contact Information" item in personnel information 221[2] is f2f2f2@ggg.com.

[0065] [Processing Unit 130] The processing unit 130 has the function of performing calculations, analyses, inferences, and other processing using data supplied from either or both of the receiving unit 110 and the storage unit 120. The processing unit 130 can supply the created data (for example, calculation results, analysis results, and inference results) to either or both of the storage unit 120 and the output unit 140. The processing unit 130 also has the function of creating prompts. The processing unit 130 may also have the function of performing processing using a language model.

[0066] The processing unit 130 has the function of acquiring data from the storage unit 120. The processing unit 130 may also have the function of recording or registering data in the storage unit 120.

[0067] The processing unit 130 has the function of performing processing using the automatic design tool described above. The processing unit 130 also has the function of acquiring design results using the automatic design tool. The design results can be stored in the storage unit 120.

[0068] The processing unit 130 may use artificial intelligence (AI) for at least some of its processing. In this case, the information processing device 10 preferably uses a neural network model. The neural network model is implemented by a circuit (hardware) or a program (software).

[0069] In this specification, the term "neural network model" refers to any model that mimics the neural network of a living organism, determines the strength of connections between neurons through learning, and possesses problem-solving capabilities. A neural network model has an input layer, an intermediate layer (hidden layer), and an output layer.

[0070] In this specification and other documents, when describing neural network models, the process of determining the connection strength (also called weight coefficient) between neurons from existing information is sometimes referred to as "learning."

[0071] In this specification and other documents, the process of constructing a neural network model using connection strengths obtained through learning and deriving new conclusions from it may be referred to as "inference."

[0072] One example of a neural network model used by the information processing device 10 is the graph neural network (GNN) model. Here, a graph has nodes and edges, and two nodes are connected by edges. This represents the data structure. For example, when representing a circuit diagram as a graph, the components of the circuit can be represented by nodes, and the connection relationships between the components can be represented by edges. For example, when representing a logic circuit diagram as a graph, logic gates can be represented by nodes, and the connection relationships between logic gates can be represented by edges.

[0073] The information processing device 10 can perform graph embedding on a graph using a GNN model and obtain an embedding representation. In the embedding representation, the graph can be represented by low-dimensional vectors.

[0074] [Output Unit 140] The output unit 140 has the function of outputting at least one of the calculation results, analysis results, and inference results from the processing unit 130 to the outside of the information processing device 10. For example, a wired communication port, a wireless communication port, or an optical communication port can be used as the output unit 140.

[0075] For example, the output unit 140 has the function of supplying data indicating prompts, etc., to the information processing device 40. The output unit 140 also has the function of supplying data to the information terminal 20.

[0076] [Transmission Line 150] The transmission line 150 has the function of transmitting data. Data can be transmitted and received between the receiving unit 110, the storage unit 120, the processing unit 130, and the output unit 140 via the transmission line 150. As the transmission line 150, for example, a bus line on the motherboard, a wired communication cable, or an optical communication cable can be used.

[0077] <Example Configuration of Information Processing Device 40> The information processing device 40 can process received data and transmit the processing results. For example, it can perform calculations and other processing using data supplied from the information processing device 10. The information processing device 40 can also supply the processing results to the information processing device 10. This reduces the computational burden on the information processing device 10.

[0078] As described above, the information processing device 40 can perform processing using language models. For example, it can perform processing using language models such as BERT and T5. In addition, the information processing device 40 can perform processing using models that utilize language models (such as text generation models and dialogue models).

[0079] As described above, the language model generates a response sentence based on the prompt. For example, the prompt created by the processing unit 130 is supplied to the language model of the information processing device 40 via the output unit 140. The response sentence generated by the language model is supplied via the receiving unit 110 to, for example, one or both of the storage unit 120 and the output unit 140.

[0080] Furthermore, the information processing device 40 can perform processing using a general-purpose language processing model that can handle various natural language processing tasks.

[0081] The information processing device 40 is a large computer such as a server computer or a supercomputer. Preferably, the information processing device 40 also has the functionality of a parallel computer. By using the information processing device 40 as a parallel computer, for example, large-scale calculations necessary for artificial intelligence learning and inference can be performed.

[0082] Furthermore, the information processing device 40 is a computer with higher processing power compared to the information processing device 10. For example, if both the information processing device 10 and the information processing device 40 have the functionality of parallel computers, the information processing device 40 will have higher processing power than the information processing device 10 and will be able to perform large-scale calculations. Also, for example, if both the information processing device 10 and the information processing device 40 can perform processing using models that utilize language models, the information processing device 40 will be able to perform processing using larger-scale models compared to the information processing device 10.

[0083] Furthermore, service providers are not necessarily required to own the information processing device 40 themselves. For example, a service provider can utilize some of the services provided by other businesses using the information processing device 40.

[0084] <Example Configuration of Information Terminal 20> The information terminal 20 can receive data input by a user of an information processing system according to one aspect of the present invention. The information terminal 20 can also present data output by the information processing system according to one aspect of the present invention to the user by displaying it on the display unit of the information terminal 20. Alternatively, the information terminal 20 can present data output by the information processing system according to one aspect of the present invention to the user by printing it on the printing unit of the information terminal 20.

[0085] Furthermore, the information terminal 20 can supply data received from the user to the information processing device 10. The information terminal 20 can also present data supplied from the information processing device 10 to the user.

[0086] Furthermore, the information terminal 20 can supply data created based on data received from the user to the information processing device 10. The information terminal 20 can also present data created based on data supplied by the information processing device 10 to the user.

[0087] For example, dedicated application software, a web browser, etc., are installed on the information terminal 20. The user can access the information processing device 10 through either of these. As a result, the user can enjoy services using an information processing system according to one embodiment of the present invention, even if the information terminal 20 is, for example, a computer with lower processing power than the information processing device 10.

[0088] The information terminal 20 can also be referred to as a client computer or the like. In any case, the information terminal 20 is an information terminal device used by a user of an information processing system according to one embodiment of the present invention.

[0089] For example, a desktop computer 20a, a notebook computer 20b, a smartphone 20c, or a tablet computer 20d can be used as the information terminal 20. The tablet computer 20d can also be used as a notebook computer by connecting it to a casing 21 with a keyboard.

[0090] Network 30: Network 30 connects the information processing device 10 and the information processing device 40. Network 30 also connects multiple information terminals 20 to the information processing device 10. This enables the transmission and reception of input data and processed data between the two. It also allows for the distribution of the information processing load.

[0091] The following describes an information processing method using the information processing system shown in Figures 1 and 2. Specifically, the operation method of the information processing device 10 will be described.

[0092] In one aspect of the present invention, an information processing method generates design commands using a language model and obtains design results based on the design commands. Subsequently, defects included in the design results are extracted, and the causes of the defects are inferred using the language model. Then, based on the defects, causes, etc., a literature search is performed from, for example, the database 121 shown in Figure 2. Based on the retrieved literature, personnel search information is generated using the language model. The personnel search information includes the skills required to solve the defects, the time required to solve the defects, etc. Based on the personnel search information, personnel suitable for solving the defects are searched from the personnel list 220 shown in Figure 4. For example, the defects, causes, and retrieved literature are sent to the contact information of the retrieved personnel, and a request is made to solve the defects.

[0093] <Information Processing Method 1> Figure 5 is a flowchart showing an example of an information processing method according to one aspect of the present invention, and more specifically, it is a flowchart showing an example of how the information processing device 10 operates.

[0094] When the information processing method according to one embodiment of the present invention is "started", in step S101 of Figure 5, the reception unit 110 receives design information 230. The design information 230 includes a design drawing 231 and specifications 233.

[0095] Figure 6A is a schematic diagram showing an example of design information 230. As mentioned above, design information 230 includes a design drawing 231 and specifications 233. The design drawing 231 is a drawing that represents the design object. For example, if the information processing device 10 has a function to automatically perform circuit design, the design drawing 231 can be a circuit diagram. In the example shown in Figure 6A, the design drawing 231 shows an OR circuit 235a, a first NOT circuit 235b, a NAND circuit 235c, and a second NOT circuit 235d as logic gates. The design drawing 231 can also be, for example, a text representation of a drawing that represents the design object. Thus, the design drawing 231 may be text data or image data.

[0096] As mentioned above, circuit diagrams can be represented using hardware description languages ​​such as Verilog, SystemVerilog, VHDL, and SystemC. Furthermore, descriptions using hardware description languages ​​can be converted into netlists. The design drawing 231 can be, for example, a netlist. In this case, the netlist corresponds to the text representation of the drawing that shows the design described above. In the example shown in Figure 6A, a netlist representing the connection relationships of the OR circuit 235a, the first NOT circuit 235b, the NAND circuit 235c, and the second NOT circuit 235d can be used as the design drawing 231.

[0097] Specification 233 is the performance required for the design. In the example shown in Figure 6A, specification 233 includes "minimum machining dimension is x nm".

[0098] Next, in step S102 of Figure 5, the processing unit 130 creates a first prompt and transmits it to the information processing device 40 via the output unit 140. Specifically, the first prompt is transmitted to the language model of the information processing device 40. The first prompt is a prompt for generating design commands based on design information 230. The first prompt contains design information 230. The first prompt also contains instructions such as, "Generate design commands to design the circuit shown in the following circuit diagram to meet the specifications. Assume that 'def' (name of the tool) from abc Corporation will be used as the EDA tool for the design."

[0099] The information processing device 40 uses a language model to generate a first response statement based on a first prompt and supplies it to the receiving unit 110. The information processing device 10 then obtains the first response statement. The first response statement includes design commands. Here, it is preferable to pre-train the language model with information about automated design tools such as EDA tools, for example, through fine-tuning. Specifically, it is preferable to pre-train the language model with information about automated design tools that the information processing device 10 can use, for example, through fine-tuning.

[0100] Next, in step S103 of Figure 5, the processing unit 130 performs the design based on the design command described above. For example, the design is performed using an EDA tool. As a result, the information processing device 10 acquires the design results. The design results include the automatically designed object. For example, when performing layout design, the design results include the circuit layout.

[0101] Automatically designed structures may contain defects. For example, there may be short circuits in areas that should be electrically insulated. Also, some of the specifications specified in Specification 233 may not be met. If an automatically designed structure contains defects, the design results described above will include information indicating the defects. For example, if there is a short circuit in an area that should be electrically insulated, the results will include information indicating the location of the short circuit. Also, if some of the specifications specified in Specification 233 are not met, the results will include information indicating the specifications that are not met.

[0102] Next, in step S104 of Figure 5, the processing unit 130 creates a second prompt and transmits it to the information processing device 40 via the output unit 140. Specifically, the second prompt is transmitted to the language model of the information processing device 40. The second prompt is for extracting defects included in the design results described above and for inferring the cause of those defects. The second prompt contains the design results described above. The second prompt may also contain the design information 230 shown in Figure 6A. The second prompt may also contain instructions such as, "The design results are shown below. Extract the defects and infer their causes."

[0103] The information processing device 40 uses a language model to generate a second response statement based on the second prompt and supplies it to the receiving unit 110. As a result, the information processing device 10 obtains the second response statement.

[0104] Figure 6B is a schematic diagram showing an example of a response statement 240 that can be a second response statement. The response statement 240 has information 241 and information 243. Information 241 is information indicating a malfunction. Information 243 is information indicating the cause of the malfunction as inferred by the language model. In the example shown in Figure 6B, information 241 includes "wire a and wire b are short-circuited" as the malfunction, and information 243 includes "kkk" as the cause of the malfunction. Note that information 241 can include multiple malfunctions. In this case, the cause may be indicated for each of the multiple malfunctions, or multiple malfunctions may be grouped together and indicated as a single cause.

[0105] After performing step S101 as shown in Figure 5, steps S102 to S104, as well as step S105, are performed. In step S105, the processing unit 130 creates a graph 251 based on the design drawing 231. In other words, the processing unit 130 converts the design drawing 231 into a graph 251.

[0106] Figure 7 is a schematic diagram showing an example of a design drawing 231 and a graph 251. In the graph 251 shown in Figure 7, nodes are represented by ellipses and edges by arrows. Nodes represent components included in the design drawing 231. For example, if the design drawing 231 represents a logic circuit, nodes represent logic gates. Two nodes representing connected components are connected by edges. In the example shown in Figure 7, the starting point of the arrow represents the output terminal of the logic gate represented by the node. The ending point of the arrow represents the input terminal of the logic gate represented by the node. Note that Figure 7 shows an example where graph 251 is a directed graph, but it may also be an undirected graph. If graph 251 is an undirected graph, edges can be represented by line segments without arrows.

[0107] In the example shown in Figure 7, the nodes included in graph 251 are shown as nodes 253a, 253b, 253c, and 253d. Node 253a represents an OR gate 235a, node 253b represents a first NOT gate 235b, node 253c represents a NAND gate 235c, and node 253d represents a second NOT gate 235d. In addition, in the example shown in Figure 7, nodes 253a and 253c are connected by an edge 254a that starts at node 253a and ends at node 253c. Furthermore, nodes 253b and 253c are connected by an edge 254b that starts at node 253b and ends at node 253c. Moreover, nodes 253c and 253d are connected by an edge 254c that starts at node 253c and ends at node 253d.

[0108] Following step S105 shown in Figure 5, in step S106, the processing unit 130 performs graph embedding on graph 251. This obtains an embedding representation. This embedding representation represents graph 251 as a low-dimensional vector. This embedding representation can be obtained, for example, using a GNN model. An example of a GNN model is the Graph Convolution Network (GCN) model. The processing unit 130 may also obtain the embedding representation without using a GNN. For example, the embedding representation of graph 251 may be obtained using node2vec, DeepWalk, LINE, ChebNet, etc.

[0109] Steps S105 and S106 can be performed in parallel with steps S102 to S104. Alternatively, steps S105 and S106 may be performed after steps S102 to S104. Or, steps S105 and S106 may be performed after step S101 and before steps S102 to S104.

[0110] After performing steps S104 and S106, in step S107 of Figure 5, the processing unit 130 searches for documents stored in the database 121 shown in Figure 2 based on the information 241 and 243 shown in Figure 6B, and the embedded representation. Specifically, the processing unit 130 searches for documents 210 shown in Figure 3 based on the malfunction represented by information 241, the cause represented by information 243, and the embedded representation.

[0111] Here, embedding representations are obtained in advance for each of the drawings 215 included in reference 210. For example, embedding representations are obtained in advance for drawing 215 that represents a circuit diagram. For example, if drawing 215 represents a logic circuit, the logic gates and the connection relationships between the logic gates are detected by image recognition. Image recognition can be performed using, for example, a neural network model. Next, the processing unit 130 creates a graph based on the results of the image recognition. Then, graph embedding is performed on the graph to obtain the embedding representation. This embedding representation can be obtained in the same way as in step S106, and can be obtained using, for example, a GNN model. Then, in step S107, the processing unit 130 compares the embedding representation obtained in step S106 with the embedding representation obtained based on drawing 215.

[0112] In this specification, the embedded representation obtained by the processing unit 130 in step S106 may be referred to as the first embedded representation. Furthermore, the embedded representation obtained by the processing unit 130 based on drawing 215 may be referred to as the second embedded representation. The second embedded representation can be stored in the database 121 shown in Figure 2. In this case, the document 210 shown in Figure 3 may have the second embedded representation in addition to the main text 211, date 212, references 213, drawing 215, etc. For example, the embedded representation obtained based on drawing 215 may be called the first embedded representation, and the embedded representation obtained by the processing unit 130 in step S106 may be called the second embedded representation.

[0113] In step S107, for example, information 241 and information 243 are used as keywords to perform a text search of document 210. For example, information 241 and information 243 are compared with the main text 211 shown in Figure 3. Also, for example, the processing unit 130 calculates the similarity of the second embedding representation to the first embedding representation. This similarity can be obtained using cosine similarity, covariance, unbiased covariance, Pearson's product-moment correlation coefficient, etc., and the use of cosine similarity is particularly preferred. Then, the processing unit 130 searches for and obtains document 210 that contains a drawing 215 that includes, for example, the defect represented by information 241 and the cause represented by information 243, and whose similarity is greater than or equal to a predetermined value. If information 241 contains multiple defects, the processing unit 130 can search for and obtain document 210 that describes at least one of the multiple defects. Alternatively, the processing unit 130 may search for and obtain document 210 for each of the multiple defects. In this case, the reference document 210 obtained by the processing unit 130 may differ for each malfunction.

[0114] As a result, the processing unit 130 can obtain, for example, documents 210 related to the problem represented by information 241, specifically documents 210 that can be used as reference for resolving the problem. In the information processing method shown in Figure 5, the documents obtained by the processing unit 130 in step S107 may be referred to as documents for personnel search.

[0115] In step S107, by searching for document 210 using the first and second embedded representations, document 210 can be searched while taking the drawings into consideration. This prevents the search from including document 210 that is only loosely related to the design automatically generated in step S103.

[0116] Next, in step S108 of Figure 5, the processing unit 130 creates a third prompt and transmits it to the information processing device 40 via the output unit 140. Specifically, the third prompt is transmitted to the language model of the information processing device 40. The third prompt is a prompt for generating personnel search information 260 based on the document 210 obtained in step S107.

[0117] Figure 8 is a schematic diagram showing an example of personnel search information 260. The personnel search information 260 shown in Figure 8 includes information 241, information 243, and information 261. Information 241 represents a problem, as described above. Information 243 represents the cause of the problem, as described above.

[0118] Information 261 represents information about resolving a problem. Information 261 can include the same items as those included in the personnel information 221 shown in Figure 4. Figure 8 shows an example in which information 261 includes information 261a and information 261b.

[0119] In the example shown in Figure 8, information 261a represents the skills required to resolve the problem. Information 261b represents the time required to resolve the problem. In the example shown in Figure 8, the skills required to resolve the problem include "ddd". It also shows an example where the time required to resolve the problem is "y days". If information 241 in Figure 6B represents multiple problems, information 261 can be shown for each of those problems. For example, for each of the multiple problems, the skills required to resolve the problem and the time required to resolve the problem can be shown. Alternatively, multiple problems can be grouped together and shown as a single piece of information 261.

[0120] The time required to resolve the problem can be calculated, for example, based on the submission date of document 210 obtained in step S107 above and the publication date of the reference document 213 disclosed in document 210. For example, the longer the difference between the submission date of document 210 and the publication date of the reference document 213 disclosed in document 210, the longer it can be inferred that it took a long time to resolve the problem disclosed in document 210. For example, the longer the difference between the submission date of document 210 and the publication date of the most recently published document among the reference documents 213 disclosed in document 210, the longer it can be inferred that it took a long time to resolve the problem disclosed in document 210. The third prompt can include, for example, the correspondence between the publication name and the publication date. This allows the publication date of reference document 213 to be identified even if, for example, only the publication year is indicated in reference document 213 and the publication month and date are not indicated.

[0121] The time required to resolve the problem does not need to be specified in concrete terms. For example, it may be indicated as long-term or short-term. Alternatively, it may be indicated using three categories such as long-term, medium-term, and short-term. Furthermore, it may be indicated using four or more categories.

[0122] The third prompt may include, for example, information 241, information 243, and document 210 obtained in step S107 described above. The third prompt may also include a command statement such as, "Based on the document, estimate the skills required to resolve the following problem and the time required to resolve the problem. Please list the skills in bullet points." The third prompt may also include, for example, a list of skills. This can prevent skills not included in the personnel list 220 shown in Figure 4 from being included in information 261a. Including a description of the skills in the third prompt may make it easier for the language model to estimate the skills suitable for resolving the problem. Furthermore, the third prompt may include, for example, how to express the time required to resolve the problem and how to calculate that time.

[0123] It is preferable to perform fine tuning on the language model in advance so that it can represent, for example, the time required to resolve a bug with high accuracy. For example, fine tuning can be performed using data points that include the publication date of the document, the publication date of reference 213 disclosed in the document, and the time required to resolve the bug. This fine tuning can be performed by supervised learning with the time required to resolve the bug as a label.

[0124] Furthermore, for example, information 261a may indicate the level of proficiency required for the skills. For example, by performing fine-tuning, the language model may be able to indicate the level of proficiency required for the skills. For example, fine-tuning can be performed using data points that include a document relating to the object for which the information processing device 10 performs automatic design, and skills and their proficiency levels. If the information processing device 10 has a function to perform automatic design of semiconductor circuits, the document in question is a document relating to the design of semiconductor circuits. The document may include, for example, the content of defects that occur in the design, the causes of the defects, and methods for resolving the defects. Fine-tuning can be performed by supervised learning with skills and their proficiency levels as labels.

[0125] The information processing device 40 uses a language model to generate a third response statement based on a third prompt and supplies it to the receiving unit 110. As a result, the information processing device 10 obtains the third response statement. The third response statement includes the personnel search information 260 shown in Figure 8.

[0126] Next, in step S109 of Figure 5, the processing unit 130 searches for personnel suitable for resolving the problem represented by information 241 based on the personnel search information 260. Specifically, the processing unit 130 searches for personnel suitable for resolving the above-mentioned problem from the personnel list 220 shown in Figure 4, based on the information 261 contained in the personnel search information 260.

[0127] In the example shown in Figure 8, for example, a search is conducted for personnel who possess all the skills represented by information 261a. Furthermore, for example, if the period represented by information 261b is less than a predetermined value, it is considered short-term; if it is greater than or equal to the predetermined value, it is considered long-term. A search is then conducted for personnel suitable for solving short-term problems or long-term problems, respectively. Then, for example, personnel who are found through both the search based on information 261a and the search based on information 261b can be considered suitable for solving the problem represented by information 241. In the following explanation, we will assume that the personnel represented by personnel information 221[1] shown in Figure 4, namely "William Smith," was found, or in other words, was found through the search.

[0128] Furthermore, even if a person does not possess some of the skills represented by information 261a, they may still be included in the search results. For example, a person who possesses a predetermined number or more of the skills included in information 261a and whose aptitude matches that of information 261b may be included in the search results as a person suitable for resolving the problem represented by information 241. Also, for example, if a person possesses all of the skills represented by information 261a, they may still be included in the search results as a person suitable for resolving the problem represented by information 241, even if their aptitude does not match that of information 261b.

[0129] Furthermore, if the personnel information 221 shown in Figure 4 includes skill proficiency, the proficiency may be reflected in the search results. For example, even if a person does not possess some of the skills represented by information 261a, if that person possesses those skills and has high proficiency in those skills, that person may be included in the search results as a suitable person for resolving the problem represented by information 241. Also, if a person possesses the skills represented by information 261a and has high proficiency in those skills, they may be included in the search results as a suitable person for resolving the problem represented by information 241, even if their suitability does not match information 261b.

[0130] If information 241 contains multiple defects, and different information 261 is provided for each of these defects, personnel can be searched for for each defect. In this case, the personnel found through the search may differ for each defect.

[0131] Next, in step S110 of Figure 5, the processing unit 130 transmits defect-related information to the personnel found in step S109. The defect-related information includes information 241 and 243 shown in Figure 6B, etc., and document 210 acquired by the processing unit 130 in step S107. The processing unit 130 transmits the defect-related information to the contact information of the personnel found in step S109, for example. Specifically, the processing unit 130 transmits the defect-related information to the contact information via, for example, the output unit 140. In this case, the output unit 140 has the function of outputting the defect-related information. In the example shown in Figures 4 and 8, the defect-related information can be transmitted to the contact information "f1f1f1@ggg.com" of personnel information 221[1].

[0132] As a result, the person found in step S109 can be requested to resolve the problem indicated by information 241. Furthermore, relevant literature and other materials can be provided to the person. In the examples shown in Figures 4 and 8, "William Smith" can be requested to resolve the problem indicated by information 241, and relevant literature and other materials can be provided. With this, the information processing method according to one embodiment of the present invention is "completed". Note that if the person found in the search differs for each problem, the problem-related information transmitted to each person can also be made different for each person. Specifically, for each person, problem-related information can be transmitted, including information 241 indicating the problem to be resolved, information 243 indicating the cause of the problem, and relevant literature 210.

[0133] As described above, in one embodiment of the present invention, after automatic design, the resolution of defects can be requested from suitable personnel. For example, the resolution of a defect can be requested from personnel who possess the necessary skills and high aptitude for solving the defect. Furthermore, relevant literature can be provided to such personnel. Therefore, the person requested to resolve the defect can resolve it while referring to the literature.

[0134] As described above, an information processing system, information processing apparatus, and information processing method according to one embodiment of the present invention can contribute to the smooth execution of design work. Furthermore, an information processing system, information processing apparatus, and information processing method according to one embodiment of the present invention can support design work. In addition, when the design drawing 231 represents a semiconductor circuit, an information processing system, information processing apparatus, and information processing method according to one embodiment of the present invention can improve the performance of the semiconductor device. Furthermore, an information processing system, information processing apparatus, and information processing method according to one embodiment of the present invention can reduce design costs. Moreover, an information processing system, information processing apparatus, and information processing method according to one embodiment of the present invention can shorten the design period.

[0135] Furthermore, after resolving the problem, a design command may be created, and the processing unit 130 may again acquire the design results based on that design command. The design command may be created by the person who was asked to resolve the problem, or it may be generated by the language model. As a result, the person who was asked to resolve the problem may not have to complete the design. Therefore, design costs may be reduced. Also, the design period may be shortened.

[0136] <Information Processing Method 2> Figure 9 is a flowchart showing an example of an information processing method according to one aspect of the present invention, different from that shown in Figure 5. In Figure 9, steps that differ from those in the information processing method shown in Figure 5 are indicated by thick borders. Steps that are the same as those in the information processing method shown in Figure 5 are denoted by the same reference numerals.

[0137] When the information processing method shown in Figure 9 is "started", steps S101 to S104 are performed. After step S104, in step S107a, the processing unit 130 searches for the document 210 shown in Figure 3 based on the information 241 and information 243 shown in Figure 6B. As a result, the processing unit 130 can search for and obtain the document 210 that contains the defect represented by information 241 and the cause represented by information 243. In step S107a, the processing unit 130 obtains multiple documents 210. Step S107a differs from step S107 shown in Figure 5 in that it does not perform a search based on embedded expressions. Since a search based on embedded expressions is not performed, steps S105 and S106 shown in Figure 5 are not performed in the information processing method shown in Figure 9.

[0138] Next, in step S111, the processing unit 130 obtains the similarity of each of the acquired documents 210 to information 241 and information 243. For example, the processing unit 130 creates a fourth prompt and transmits it to the information processing device 40 via the output unit 140. The fourth prompt is specifically transmitted to the language model of the information processing device 40. The fourth prompt is a prompt for calculating the similarity of the document 210 to information 241 and information 243. The fourth prompt includes document 210, information 241, and information 243. Specifically, the fourth prompt includes a first text which is the text contained in document 210, and a second text which represents information 241 and information 243. As described above, the fourth prompt includes multiple documents 210. Therefore, the fourth prompt includes multiple first texts. The first text may include only the main text 211 shown in Figure 3 from the text contained in reference 210, or it may include text other than the main text 211. The fourth prompt has a command statement such as, "For each of the following references, calculate the similarity to the following defects and causes."

[0139] The information processing device 40 uses a language model to generate a fourth response statement based on the fourth prompt and supplies it to the receiving unit 110. As a result, the information processing device 10 obtains the fourth response statement. The fourth response statement includes the similarity of each of the multiple documents 210 to information 241 and information 243. In other words, the fourth response statement includes the similarity of each of the multiple first texts to the second text.

[0140] By using a language model to calculate the similarity between information 241 and information 243 of document 210, the similarity can be calculated taking into account the meaning represented by the first text contained in document 210, and the meaning represented by the second text representing information 241 and information 243. Note that the similarity between information 241 and information 243 of document 210 may be obtained without using a language model. For example, the similarity may be obtained using a neural network model other than a language model.

[0141] Next, in step S112, the processing unit 130 extracts one or more documents 210 from the multiple documents 210 obtained in step S107a. The extraction of documents 210 is performed based on the similarity score described above. The extracted documents are designated as documents for talent search. For example, documents 210 with a similarity score of a predetermined value or higher can be designated as documents for talent search. Alternatively, a predetermined number of documents 210, counted from the highest similarity score, can be designated as documents for talent search.

[0142] Next, in step S108a, the processing unit 130 creates a third prompt and transmits it to the information processing device 40 via the output unit 140. Specifically, the third prompt is transmitted to the language model of the information processing device 40. The third prompt created in step S108a differs from the third prompt created in step S108 shown in Figure 5 in that it is a prompt for generating the personnel search information 260 shown in Figure 8 based on the personnel search literature.

[0143] Next, steps S109 and S110 are performed. Here, the defect-related information includes information 241 and information 243 shown in Figure 6B, etc., as well as literature for personnel search. With this, the information processing method shown in Figure 9 is "completed".

[0144] In the information processing method shown in Figure 9, a language model is used to obtain the similarity between document 210 and information 241 and information 243, and documents for personnel search are extracted based on this similarity. As described above, by calculating the similarity between document 210 and information 241 and information 243 using a language model, the similarity can be calculated taking into account the meaning represented by the first text contained in document 210, and the meaning represented by the second text representing information 241 and information 243. Therefore, compared to, for example, the case where all of document 210 acquired by the processing unit 130 in step S107a are used as documents for personnel search, and the case where the similarity described above is calculated without using a language model, it is possible to suppress the selection of documents 210 that are not related to the design automatically designed in step S103 as documents for personnel search.

[0145] <Information Processing Method 3> Figure 10 is a flowchart showing an example of an information processing method according to one embodiment of the present invention, different from Figures 5 and 9. In Figure 10, steps that are different from both the steps in the information processing method shown in Figure 5 and the steps in the information processing method shown in Figure 9 are indicated by thick borders. The same reference numerals are used for the same steps as in the information processing method shown in Figure 5 and the same steps as in the information processing method shown in Figure 9.

[0146] When the information processing method shown in Figure 10 is "started", steps S101 to S106 are performed. As mentioned above, steps S102 to S104 and steps S105 and S106 can be performed in parallel. After steps S104 and S106 are performed, in step S107b, the processing unit 130 searches for the document 210 shown in Figure 3 based on the information 241 and information 243 shown in Figure 6B, and the embedded representation. In step S107b, similar to step S107a, the processing unit 130 obtains multiple documents 210. On the other hand, step S107b differs from step S107a shown in Figure 9 in that it performs a search based on the embedded representation, similar to step S107 shown in Figure 5.

[0147] Next, steps S111, S112, S108a, S109, and S110 shown in Figure 9 are performed in order. This completes the information processing method shown in Figure 10. The information processing method shown in Figure 10 can be described as a combination of the method shown in Figure 5 and the method shown in Figure 9. The information processing method shown in Figure 10 effectively suppresses the retrieval of literature 210, which is only loosely related to the design automatically generated in step S103.

[0148] <Example of Information Processing System Configuration 2> Below, a detailed example of the configuration of an information processing system according to one embodiment of the present invention will be described. Specifically, a configuration example of the storage unit 120, the processing unit 130, and the network 30 that is not shown in <Example of Information Processing System Configuration 1> will be described.

[0149] The memory unit 120 has at least one of volatile memory and non-volatile memory. Examples of volatile memory include DRAM (Dynamic Random Access Memory) and SRAM (Static Random Access Memory). Examples of non-volatile memory include ReRAM (Resistive Random Access Memory), PRAM (Phase Change Random Access Memory), FeRAM (Ferroelectric Random Access Memory), MRAM (Magnetoresistive Random Access Memory), and flash memory. Furthermore, the storage unit 120 may have at least one of NOSRAM (registered trademark) and DOSRAM (registered trademark). The storage unit 120 may also have a recording media drive. Examples of recording media drives include hard disk drives (HDD) and solid state drives (SSD).

[0150] NOSRAM is an abbreviation for "Nonvolatile Oxide Semiconductor Random Access Memory." NOSRAM is a memory in which the memory cell is a 2-transistor type (2T) or 3-transistor type (3T) gain cell, and the transistors are transistors that use metal oxide in the channel formation region (also called OS transistors). OS transistors have an extremely small current flowing between the source and drain when off, i.e., a leakage current. By utilizing the characteristic of extremely low leakage current, NOSRAM can be used as a non-volatile memory by holding a charge corresponding to the data within the memory cell. In particular, NOSRAM can read the stored data without destroying it (non-destructive read), making it suitable for computational processing that repeatedly performs a large number of data read operations. Because the data capacity of NOSRAM can be increased by stacking them, it can be used as a large-scale cache memory, main memory, storage memory, etc., to improve the performance of semiconductor devices.

[0151] DOSRAM is an abbreviation for "Dynamic Oxide Semiconductor RAM," and refers to RAM with a 1T (transistor) 1C (capacitance) type memory cell. DOSRAM is a DRAM formed using OS transistors and is a memory that temporarily stores information sent from the outside. DOSRAM is a memory that takes advantage of the small off-current of OS transistors.

[0152] In this specification, "metal oxide" refers to an oxide of a metal in a broad sense. Metal oxides are classified into oxide insulators, oxide conductors (including transparent oxide conductors), oxide semiconductors (also called oxide semiconductors or simply OS), etc. For example, when a metal oxide is used in the semiconductor layer of a transistor, that metal oxide may be called an oxide semiconductor.

[0153] The metal oxide in the channel-forming region preferably contains indium (In), and for example, indium oxide is preferred. When the metal oxide in the channel-forming region is an indium-containing metal oxide, the carrier mobility (electron mobility) of the OS transistor is increased. Furthermore, the metal oxide in the channel-forming region is preferably an oxide semiconductor containing element M. Element M is preferably at least one of aluminum (Al), gallium (Ga), and tin (Sn). Other elements applicable to element M include boron (B), silicon (Si), titanium (Ti), iron (Fe), nickel (Ni), germanium (Ge), yttrium (Y), zirconium (Zr), molybdenum (Mo), lanthanum (La), cerium (Ce), neodymium (Nd), hafnium (Hf), tantalum (Ta), and tungsten (W). However, multiple elements mentioned above may be combined as element M. Element M is, for example, an element with a high bond energy with oxygen. For example, it is an element whose bonding energy with oxygen is higher than that of indium. Furthermore, the metal oxide containing the channel-forming region is preferably a metal oxide containing zinc (Zn). Zinc-containing metal oxides may be more prone to crystallization.

[0154] The metal oxides present in the channel-forming regions are not limited to indium-containing metal oxides. For example, the metal oxides present in the channel-forming regions may be zinc-tin oxides, gallium-tin oxides, or other metal oxides that do not contain indium but contain zinc, gallium, or tin.

[0155] The processing unit 130 may, for example, have an arithmetic circuit. The processing unit 130 may, for example, have a central processing unit (CPU). Furthermore, the processing unit 130 may have a graphics processing unit (GPU).

[0156] The processing unit 130 may have a microprocessor such as a DSP (Digital Signal Processor). The microprocessor may be implemented using a PLD (Programmable Logic Device) such as an FPGA (Field Programmable Gate Array) or FPAA (Field Programmable Analog Array). The processing unit 130 may also have a quantum processor. The processing unit 130 can perform various data processing and program control by interpreting and executing instructions from various programs using the processor. Programs that can be executed by the processor are stored in at least one of the processor's memory area and the storage unit 120.

[0157] The processing unit 130 may have main memory. The main memory includes at least one of volatile memory such as RAM (Random Access Memory) and non-volatile memory such as ROM (Read Only Memory). The main memory may also include at least one of the above-mentioned NOSRAM and DOSRAM.

[0158] For RAM, for example, DRAM or SRAM is used, and a memory space is virtually allocated and used as a workspace for the processing unit 130. The operating system, application programs, program modules, program data, lookup tables, etc., stored in the storage unit 120 are loaded into RAM for execution. These data, programs, and program modules loaded into RAM are directly accessed and manipulated by the processing unit 130.

[0159] ROM can store BIOS (Basic Input / Output System), firmware, and other data that does not require rewriting. Examples of ROM include mask ROM, OTPROM (One Time Programmable Read Only Memory), and EPROM (Erasable Programmable Read Only Memory). Examples of EPROMs include UV-EPROM (Ultra-Violet Erasable Programmable Read Only Memory), which allows data to be erased by ultraviolet irradiation, EEPROM (Electrically Erasable Programmable Read Only Memory), and flash memory.

[0160] The processing unit 130 may have either or both an OS transistor and a transistor having silicon in its channel formation region (Si transistor).

[0161] The processing unit 130 preferably has an OS transistor. Because the OS transistor has an extremely small off-current, using the OS transistor as a switch to hold the charge (data) that has flowed into a capacitive element that functions as a memory element ensures that the data can be retained for a long period of time. If at least one of the registers and cache memory of the processing unit 130 has this characteristic, the processing unit 130 can be operated only when necessary, and in other cases the information of the previous processing is saved to the memory element and the processing unit 130 can be turned off. In other words, normally-off computing becomes possible, and the power consumption of the information processing system can be reduced.

[0162] Network 30 can, for example, use the Internet, which is the foundation of the World Wide Web (WWW), as a global network. Network 30 can also use a local network. Furthermore, an intranet or extranet can be used as network 30. In addition, PAN (Personal Area Network), LAN (Local Area Network), CAN (Campus Area Network), MAN (Metropolitan Area Network), WAN (Wide Area Network), GAN (Global Area Network), etc., can be used as network 30.

[0163] When a service provider using an information processing method according to one aspect of the present invention and a user enjoying the service belong to the same organization, such as a company, it is preferable that data transmission and reception between the information terminal 20 and the information processing device 10 be performed, for example, using a network established within that organization. This allows for more secure data transmission and reception between the information terminal 20 and the information processing device 10 compared to transmission via the Internet. Furthermore, it prevents the leakage of confidential information within the organization to the outside.

[0164] When performing wireless communication, communication protocols or technologies that can be used include communication standards such as the 4th generation mobile communication system (4G), 5th generation mobile communication system (5G), and 6th generation mobile communication system (6G), or specifications standardized by the IEEE (Institute of Electrical and Electronics Engineers), such as Wi-Fi® and Bluetooth®.

[0165] 10: Information processing device, 20: Information terminal, 20a: Desktop computer, 20b: Notebook computer, 20c: Smartphone, 20d: Tablet computer, 21: Enclosure, 30: Network, 40: Information processing device, 110: Reception unit, 120: Storage unit, 121: Database, 122: Database, 130: Processing unit, 140: Output unit, 150: Transmission line, 211: Main text, 212: Date, 213: References, 215: Drawings, 220: Personnel list, 221[1]: Personnel information, 2 21[2]: Personnel information, 221: Personnel information, 230: Design information, 231: Design drawings, 233: Specifications, 235a: OR circuit, 235b: First NOT circuit, 235c: NAND circuit, 235d: Second NOT circuit, 240: Response statement, 241: Information, 243: Information, 251: Graph, 253a: Node, 253b: Node, 253c: Node, 253d: Node, 254a: Edge, 254b: Edge, 254c: Edge, 260: Personnel search information, 261: Information, 261a: Information, 261b: Information

Claims

The process comprises steps 1 through 7, In the first step described above, design information having design drawings and specifications is received, In the second step described above, a first prompt is created and sent to the language model for generating a design command based on the design information. In the third step described above, design is performed based on the design command, and the design results are obtained. In the fourth step, a second prompt is created and sent to the language model for extracting defects included in the design results and inferring the cause of the defects. In the fifth step described above, a graph is created based on the design drawing, In the sixth step described above, a first embedding representation is obtained by performing graph embedding on the graph, An information processing method for obtaining literature by searching based on the malfunction, the cause, and the first embedded expression in the seventh step. 　 In claim 1, The process comprises steps 8 through 10, In the eighth step, a third prompt is created and transmitted to the language model for generating personnel search information to be used when searching for personnel suitable for resolving the problem, based on the acquired literature. In the ninth step described above, the personnel are searched based on the personnel search information, An information processing method that transmits defect-related information, including the defect, the cause, and the literature, to the person who was found in the tenth step. 　 In claim 1, The aforementioned document comprises a main text, a documentary figure, and a second embedding representation obtained by performing graph embedding on the documentary figure. An information processing method comprising, in the seventh step, comparing the defect and the cause with the main text, and comparing the first embedded expression with the second embedded expression. 　 In claim 1, The first embedding representation is an information processing method obtained using a graph neural network. 　 In claim 3, The first embedding representation and the second embedding representation are information processing methods obtained using a graph neural network, respectively. 　 The process comprises a first to a tenth step, In the first step described above, design information having design drawings and specifications is received, In the second step described above, a first prompt is created and sent to the language model for generating a design command based on the design information. In the third step described above, design is performed based on the design command, and the design results are obtained. In the fourth step, a second prompt is created and sent to the language model for extracting defects included in the design results and inferring the cause of the defects. In the fifth step described above, based on the malfunction and the cause, multiple documents are obtained by searching, In the sixth step, the similarity between each of the multiple documents and the defects and their causes is obtained. In the seventh step described above, based on the similarity, one or more documents for talent search are extracted from the plurality of documents, In the eighth step, a third prompt is created and transmitted to the language model for generating personnel search information used when searching for personnel suitable for resolving the problem, based on the personnel search literature. In the ninth step described above, the personnel are searched based on the personnel search information, An information processing method that transmits to the searched personnel the defect, the cause, and defect-related information including the literature for searching for the personnel in the tenth step. 　 In claim 6, An information processing method comprising, in the sixth step, creating a fourth prompt for calculating the similarity and sending it to the language model. 　 In claim 2 or claim 6, The aforementioned personnel search information includes the skills required to resolve the problem and the time required to resolve the problem, as well as an information processing method. 　 In any one of claims 1 to 7, The aforementioned design drawing is an information processing method for a circuit diagram. 　 It has a reception unit, an output unit, and a processing unit. The aforementioned reception unit has the function of receiving design information, which includes design drawings and specifications. The output unit has the function of supplying a first prompt and a second prompt to the language model, respectively. The aforementioned processing unit, A process for creating the first prompt for generating design commands based on the design information, The process involves performing a design based on the aforementioned design command and obtaining the design results. A process for extracting defects included in the design results and creating a second prompt for inferring the cause of the defects, A process to create a graph based on the aforementioned design drawings, The process involves obtaining a first embedding representation by performing graph embedding on the aforementioned graph, An information processing device having the function of performing a process to retrieve literature by searching based on the aforementioned malfunction, the aforementioned cause, and the first embedded representation. 　 In claim 10, The output unit has the function of supplying a third prompt to the language model. The output unit has a function to output malfunction-related information including the malfunction, the cause, and the literature. The aforementioned processing unit, A process for creating a third prompt to generate personnel search information used when searching for personnel suitable for resolving the aforementioned problem, based on the acquired documents, A process to search for the aforementioned personnel based on the aforementioned personnel search information, An information processing device having the function of transmitting the defect-related information to the searched personnel via the output unit. 　 In claim 10, The aforementioned document comprises a main text, a documentary figure, and a second embedding representation obtained by performing graph embedding on the documentary figure. The processing unit is an information processing device having the function of comparing the malfunction and the cause with the main text, and comparing the first embedded expression with the second embedded expression. 　 In claim 10, The processing unit is an information processing device having the function of acquiring the first embedding representation using a graph neural network. 　 In claim 12, The processing unit is an information processing device having the function of acquiring the first embedding representation and the second embedding representation, respectively, using a graph neural network. 　 In claim 11, The aforementioned personnel search information includes the skills required to resolve the problem and the time required to resolve the problem, as well as the information processing device. 　 In any one of claims 10 to 15, The aforementioned design drawing is a circuit diagram of an information processing device.

Images