Support system for writing notice of reasons for refusal, and support method for writing notice of reasons for refusal

Abstract

A system that assists in creating notices of reasons for refusal. [Solution] This is a support system for creating office action notices, which has an input unit, a processing unit, and an output unit. The input unit accepts the content of the invention as input. The processing unit generates office action notices for the invention based on the content of the invention using a generative AI model that has learned from data including office action notices and patent documents previously created by the Patent Office. The output unit outputs the generated office action notices.

Description

[Technical Field]

[0001] The present invention relates to a system and a method for supporting the creation of a notice of reasons for refusal. [Background technology]

[0002] When an inventor files a patent application, it is not uncommon for a notice of reasons for refusal to be issued during the examination stage. Most notices of reasons for refusal are due to similarity to prior art, lack of inventive step, or incomplete descriptions. In response, the applicant prepares amendments and arguments, which incurs time and financial costs. Patent Document 1 discloses a system that supports patent searches. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7154645 Summary of the Invention [Problem to be solved by the invention]

[0004] However, it is difficult to determine what specific reasons for refusal will be issued based solely on the results of a prior art patent search. [Means for solving the problem]

[0005] An example of the present invention is a support system for creating a notice of refusal having an input unit, a processing unit, and an output unit, wherein the input unit accepts the content of an invention as input, the processing unit generates a notice of refusal for the invention based on the content of the invention using a generative AI model that has learned data including notices of refusal and patent publications previously created by the Japan Patent Office, and the output unit outputs the generated notice of refusal.

[0006] Problems, configurations, and effects other than those described above will become apparent from the following description of the embodiments. [Brief explanation of the drawings]

[0007] [Figure 1] 1 is a diagram illustrating an overview of a system according to an embodiment of the present invention. [Figure 2] FIG. 2 is a diagram showing a processing flow of a system according to an embodiment of the present invention. [Figure 3] 1 is a diagram illustrating an overview of hardware of a system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0008] A system (a support system for preparing notices of reasons for refusal) to which one embodiment of the present invention is applied will be described below with reference to the drawings. For the sake of convenience, the following embodiments will be divided into multiple sections or embodiments, but unless otherwise specified, they are not unrelated to each other, and one is related to the other as a partial or complete modification, detail, supplementary explanation, etc.

[0009] Furthermore, in the following embodiments, when referring to the number of elements (including the number, numerical value, amount, range, etc.), unless otherwise specified or when it is clearly limited to a specific number in principle, it is not limited to that specific number and may be more or less than the specific number.

[0010] Furthermore, it goes without saying that in the following embodiments, the components (including element steps, etc.) are not necessarily essential unless otherwise specified or unless they are clearly considered essential in principle.

[0011] Similarly, in the following embodiments, when referring to the shapes, positional relationships, etc. of components, etc., it is intended to include those that are substantially similar or similar to those shapes, etc., unless otherwise specified or when it is considered that this is clearly not the case in principle. This also applies to the above numerical values ​​and ranges.

[0012] In addition, in all the drawings for explaining the embodiments, the same components are generally designated by the same reference numerals, and repeated explanations thereof will be omitted.

[0013] <System Overview> FIG. 1 is a schematic diagram of a system 1 according to this embodiment.

[0014] The system 1 of this embodiment is a computer system in which a plurality of computers are connected via a network, and includes a management device 10, a user terminal 20, an external information providing device 30, and the like.

[0015] The management device 10 is a server device that plays a central role in the present system 1. The management device 10 collects information from the user terminal 20 and the external information providing device 30, stores, processes, edits, etc. the information, and transmits the processed and edited information to the user terminal 20 and the external information providing device 30. The management device 10 also accepts instructions from the user terminal, collects, processes, generates, etc. the information according to the instructions, and returns the results to the user terminal 20.

[0016] The management device 10 has a processing unit 11, an input unit 12, an output unit 13, a storage unit 14, a communication unit 15, etc. The processing unit 11 is a functional unit that performs various processes (data analysis, information generation, model control, etc.) within the management device 10. The input unit 12 is a functional unit that accepts input from an operator, etc., and includes, for example, a GUI or a command line interface. The output unit 13 is a functional unit that outputs information, and includes a display, an audio output device, etc. The storage unit 14 is a functional unit that stores various information (input data, generation results, model parameters, etc.) and computer programs. The communication unit 15 is a functional unit that transmits and receives information to and from other devices via a network.

[0017] As described above, the management device 10 is responsible for collecting, processing, and controlling information in the system 1, and functions as a server device that provides services to the user terminals 20.

[0018] The user terminal 20 is a terminal device used by a user. The user terminal 20 has functional units such as a processing unit 21, an input unit 22, an output unit 23, a storage unit 24, and a communication unit 25. The input unit 22 accepts operations from a user, the output unit 23 displays or outputs generated information and responses, and the processing unit 21 executes a user interface and local processing. The storage unit 24 also stores user settings, temporary data, computer programs, etc., and the communication unit 25 communicates with the management device 10, etc.

[0019] The user terminal 20 is equipped with input devices such as a keyboard, a voice input device, and a touch panel to receive various inputs, as well as various sensors (for example, a camera that captures images of the external environment, a temperature sensor, a position sensor (GPS), a speed sensor, an angle sensor, a gyro sensor, a humidity sensor, etc.). The user terminal 20 also has a display, a speaker, etc. as output devices. The user terminal 20 can be configured as a PC, a tablet computer, a smartphone, etc.

[0020] The external information providing device 30 is a device that plays a central role in the intelligent functions of the present system 1 by storing and providing AI models, related information, external databases (DBs), and the like.

[0021] The external information providing device 30 is a device that collects and stores information required to realize the functions provided by the present system 1. In particular, it complementarily stores information that the management device 10 and the user terminal 20 do not have and provides it as needed. Some external information providing devices 30 function as data servers equipped with databases, while others have generative AI models and function as generative AI service servers. The latter external information providing device 30 includes an AI model and has the function of generating information and returning it to the management device 10 based on instructions (prompts, etc.) from the management device 10. The AI ​​model is, for example, generative AI that performs natural language generation, image generation, summarization, translation, etc., and includes large-scale language models (LLMs) and the like.

[0022] The external information providing device 30 may be configured as a single device or multiple devices, which may operate independently or in a cooperative manner.

[0023] <Processing flow in this system> FIG. 2 is a diagram showing the basic processing flow performed in this system.

[0024] Information processing in this system 1 begins based on input from a user terminal and consists of a flow that works in conjunction with a management device and multiple external information providing devices (generation AI and DB) to generate, complement, process, and output information in stages.

[0025] First, the input unit 22 of the user terminal 20 receives an instruction from the user (e.g., a request to create a notice of reasons for refusal, etc.) (step S101). The content of this instruction is transmitted to the management device 10 via the processing unit 21 of the user terminal (step S102). The processing unit 11 of the management device 10 analyzes the received instruction content (S103) and identifies information required to execute the instruction. If the information exists in the external information providing device (DB) 30, a request to acquire the information is sent to the external information providing device (DB) 30 (S104). In response, the external information providing device (DB) 30 returns the requested information to the management device (S105). In the support service for creating notices of reasons for refusal, the necessary information is electronic data of application documents (e.g., claims, specifications, drawings) describing the invention, and this information is set in advance in the program of the storage device. Alternatively, the processing unit 11 may instruct the generation AI to identify the necessary information based on user instructions and obtain the response, thereby identifying the necessary documents. If the necessary information can be obtained from the user terminal 20, the processing unit 11 of the management device 10 sends a transmission request to the user terminal 10 and obtains information (e.g., patent application documents for the target invention) returned from the user terminal 10. If the external information providing device 30 stores the necessary information (e.g., patent application documents for the target invention, related prior art documents, Japan Patent Office examination standards, examination guidelines, examination statistics by technical field, etc.), the processing unit 11 of the management device 10 sends a transmission request to the user terminal 10 and obtains information (e.g., patent application documents for the target invention, related prior art documents, Japan Patent Office examination standards, examination guidelines, examination statistics by technical field, etc.). The patent application documents for the invention in question do not have to be in a formal format, but may simply contain information disclosing the invention.

[0026] Thereafter, the processing unit 11 of the management device 10 creates a prompt suitable for processing using the generation AI based on the user instructions and the acquired information (S106), and transmits it to the external information providing device (generation AI) 30 (S107).

[0027] Before generating information based on the received prompt, the external information providing device (generation AI) 30 first analyzes the prompt (S108). If it determines based on the analysis result that additional information is required, the external information providing device (generation AI) 30 itself requests information from another external information providing device (DB) 30 (S109).

[0028] Upon receiving the request, the external information providing device (DB) 30 searches for the required information and returns it to the generation AI device (S110). Based on the prompt and the acquired additional information, the generation AI then executes the final generation process (S111) and transmits the generation results (the generated rejection notice, etc.) to the management device 10 (S112).

[0029] The processing unit 11 of the management device 10 receives the generated results and processes and formats them as necessary to construct output data suitable for presentation to the user. The constructed data is transmitted to the user terminal (S113) and provided to the user in the form of display, audio output, etc. by the output unit 23 of the user terminal 20 (S114).

[0030] If the user terminal 20 receives an additional instruction from the user regarding the provided content, the process is repeated from S101. For example, when options are presented and the selection of one of the options is received, the user terminal 20 transmits the content to the processing unit 11 of the management device 10. The processing unit 11 passes the content to the external information providing device (generation AI) 30 in the form of a prompt, and sends the answer content back to the user terminal 20 to be output. This allows the user to obtain the desired answer with the feeling that they are interacting with the system 1.

[0031] In this way, this system is configured so that starting from user input, the management device 10 and multiple external information providing devices (generation AI and DB) 30 work together to share processing, realizing step-by-step and dynamic information generation and output. The management device 10 may have a generative AI model having the same function as the generative AI model provided in the external information providing device (generative AI) 30. Then, the processing unit 11 may use the generative AI model to generate desired information.

[0032] <System hardware configuration> FIG. 3 is a diagram showing an example of the hardware configuration of the management device 10, the user terminal 20, and the external information providing device 30. As shown in FIG.

[0033] Each device 300 includes an input device 301, a processor 302, a storage 303, a memory 304, a display 305, a communication device 306, and a bus 307 connecting the devices together.

[0034] The input device 301 is any of various input devices such as a keyboard, mouse, touch panel, etc. The input unit is realized by the input device 301 and the processor 302. The processor 302 is an arithmetic unit such as a CPU or GPU, and executes processing according to a program recorded in the memory 304 or the storage 303. The processing unit of each device realizes its respective function by the processor 302 executing the program.

[0035] The memory 304 is a storage device such as a RAM (Random Access Memory) or a flash memory, and functions as a storage area from which programs and data are temporarily read. The storage 303 is a writable and readable storage device.

[0036] The display 305 serving as an output device is, for example, a display device such as a liquid crystal display, an organic EL display, etc. The output device may also include a speaker and a control interface for other devices.

[0037] The communication device 306 is an interface for establishing a communication connection with an external device. For example, the communication device 306 performs wireless communication using an antenna that can use a predetermined radio wave (e.g., 5 GHz band, 2.4 GHz band, etc.) to establish a connection with the maintenance management device 100 according to the Wi-Fi standard. The communication unit 340 of each device is realized by the communication device 306 and the processor 302.

[0038] The management device 10 is required to have particularly high-performance processing capabilities, and is preferably configured with multiple CPU cores, a GPU, sufficient memory capacity, and high-speed storage.

[0039] The user terminal 20 has lightweight processing and display functions, and is configured with emphasis on portability and operability. If necessary, it may be equipped with a camera, a group of sensors, a touch panel, an audio input / output device, and the like.

[0040] The external information providing device 30 has an execution environment for an AI model, and therefore is configured to include computing resources such as a GPU and a TPU (Tensor Processing Unit) optimized for AI processing, as well as a large-capacity storage device. It may be constructed in a cloud environment, in which case it may be implemented using a virtual server or a distributed processing system.

[0041] The configuration of these devices can be expanded or reduced depending on the functions provided by the system 1. It is also possible to add other peripheral devices and interfaces to each device as needed.

[0042] Note that each component of the system 1 (such as the management device 10 and the external information providing device 30) does not necessarily have to be configured as a single physical device. These components may be implemented in a distributed manner by multiple devices connected via a network such as the Internet. Furthermore, each functional unit (such as a processing unit, a storage unit, and a communication unit) may be constructed on the cloud as a virtualized resource, and the actual physical device on which the virtual resource operates does not need to be located in a specific country or region.

[0043] For example, the functions of the management device 10 may be realized in a distributed manner by multiple virtual servers on a cloud platform, and even if part of the processing is performed in an overseas data center, this does not deviate from the technical scope of the present invention.

[0044] Therefore, the present invention is not limited to a physical configuration (hardware configuration or location) but is specified based on a logical functional configuration.

[0045] The claims are written in a technical configuration format that clarifies the relationships between components and the order of processing, etc., in order to satisfy the description requirements of the Patent Act. The system according to the present invention may be realized in a cloud environment or a distributed processing environment, and some or all of the processes described in the claims may be executed by physical devices or servers located outside Japan. However, if the results of these processes are provided to a user in Japan and the user enjoys the substantial effects (benefits) of the services provided by the system, the use of such a system may be considered to be equivalent to the implementation of the present invention in Japan.

[0046] Therefore, even if the configuration or processing described in the claims is partially implemented overseas, as long as the technical idea of ​​the invention realized by said configuration is embodied in Japan, it should be interpreted as falling under the "implementation" under the Patent Act.

[0047] <Details of this system (support system for creating notices of reasons for refusal)> The purpose of this system is to enable users to easily obtain documents (simulated office action) that are similar to actual office action. This allows users to predict possible office action during the invention or pre-filing stage and adjust their application strategy, thereby avoiding unnecessary amendments and increasing the chances of obtaining a stronger patent.

[0048] The user of the user terminal is a person who receives support in preparing a notice of reasons for refusal, such as an inventor, a patent applicant, a patent attorney, an intellectual property consultant, a person playing the role of an examiner, etc. Although the user is not a real examiner, since the user uses the system as if they were an examiner, the user of the system will hereinafter be referred to as an "examiner (playing the examiner)."

[0049] This system has the function of having a generation AI analyze the technical content of application documents such as claims and specifications from the perspective of a Patent Office examiner and automatically generate a notice of reasons for refusal in accordance with the Patent Examination Guidelines. In order to support appropriate judgments on the novelty and inventive step of an application, this system must accurately and quickly obtain prior art information such as cited documents. For this reason, the system may be configured to obtain publicly known technical information, including cited documents, via communication means with an external information provider.

[0050] The external information providing device can be, for example, a patent document search database (J-PlatPat, Espacenet, etc.), a private prior art search service, or a document providing API that aggregates this information. The processing unit 11 analyzes the document data acquired from the device syntactically and semantically and uses the results when comparing the claims with the technical configuration.

[0051] The information that can be transmitted by the external information providing device preferably includes the full text of the cited document, an abstract, a publication number, an application date, the name of the invention, the structure of the invention, drawings, chemical formulas, numerical data, etc., and this information is also used as auxiliary information for comparison of component units and numerical conversion, etc.

[0052] In addition, the external information providing device 30 may have the function of providing multiple candidate citation documents all at once, and the processing unit may be configured to input them sequentially or selectively into the generation AI, so as to select the document that is most suitable for the reason for rejection.

[0053] Furthermore, in this system, the generation AI can also be configured to be provided by the external information providing device 30. That is, the processing unit 11 of the management device of this system may be configured to use an externally located generation AI function through an interface such as an API via the cloud or a communication network.

[0054] Such generative AI may include, for example, general-purpose language models such as ChatGPT (OpenAI) and Gemini (Google), or it may utilize API-based services provided by private companies that provide generative AI specialized for patent practice (e.g., domain-specific models optimized for patent document analysis).

[0055] With this configuration, the system can output documents that comply with the format, structure, and examination standards of patent documents while taking advantage of the flexible language generation capabilities of general-purpose AI. Furthermore, by selectively utilizing specialized patent generation AI, it is possible to achieve higher accuracy in extracting technical configurations, sorting out causal relationships between components, and outputting standard phrases for reasons for refusal.

[0056] Furthermore, the processing unit 11 may have a means of accessing multiple generation AI services in the external information providing device, and may be configured to dynamically select an appropriate generation AI depending on the application (component extraction, summary generation, comparative evaluation, notification creation, etc.).

[0057] Next, the main functions of this system will be described.

[0058] <Description of distinctive features> This system has the following representative functions 1 to 16. Functions 1 to 7 are functions that promote understanding between the user (examiner) and the AI, and bridge the gap in understanding between the two. Function 8 is a function that checks the consistency of application documents. Functions 9 to 16 are functions that create notices of reasons for refusal that are more useful. The functions of this system are not limited to these. 1. A function that breaks down the invention into components, actions, effects, etc., and presents to the examiner how AI understands the claims. 2. A function that allows AI to summarize claims and allows users (examiners) to review and revise the summary. 3. A function that generates multiple candidates (interpretation patterns) for claim interpretation, allowing examiners to select and indicate the best one 4. A function that automatically generates confirmation questions for the user (examiner) regarding the syntax of claims, assisting in understanding confirmation. 5. A function that allows AI to generate hypothetical working examples based on claims, and allows the user (examiner) to check their understanding 6. When the user (examiner) modifies the wording of a claim, the AI ​​provides feedback on the improvement in understanding. 7. A function that detects claims that may be difficult to understand during examination by analyzing their syntax 8. A function to evaluate the consistency between claims, specifications, and drawings, and to identify violations of written requirements 9. Function to output notice of reasons for refusal due to insufficient description requirements 10. A function to output notices of reasons for refusal in both cases where violations of description requirements are strictly applied and where they are loosely applied 11. A function that uses documents provided by an external cited document research system to generate a comparison table with claims, and judges and visualizes whether or not a document has novelty or inventive step in accordance with patent examination standards. 12. A function that performs conversion to calculate comparable values ​​when the units or notation formats of numerical values ​​differ between cited documents and claims. A function that presents the results of the conversion and the prerequisites used for the conversion to the examiner, allowing them to confirm their validity and provide instructions for correction. 13. Ability to adjust the criteria for determining inventive step (Article 29, Paragraph 2) to be stricter or lenient, and output a notice of reasons for refusal in each case. 14. A function that automatically adjusts the content and severity of reasons for refusal according to the examination trends for each application field (mechanical, electrical, chemical, etc.) 15. A function that allows you to set the patent allowance rate arbitrarily and output a notice of refusal that corresponds to the applicable rate. 16. A simple mode that generates a notice of rejection even when some of the application documents (specification, claims, etc.) are missing.

[0059] <Function 1. A function that breaks down the invention into components, actions, effects, etc. and presents them to the examiner, showing how AI understood the claims> The processing unit 11 has the function of analyzing the content of the claims of an application using a generation AI, automatically extracting and organizing the components of the claims, the actions or functions of each component, and the technical effect of the invention as a whole, and visually presenting how the examiner should understand the claims.

[0060] In general, the expression of claims varies in level of abstraction and structure depending on the technical field and the inventor's expression policy, and it is often difficult for examiners to accurately grasp the intention of the configuration by reading the claim once. In response to this, this function aims to support examiners' technical understanding by mechanically and logically breaking down and clarifying the understanding of the claim. The processing unit 11 receives as input the text information of the claims contained in the application documents, and automatically separates and classifies the following elements using natural language processing and grammatical analysis.

[0061] (1) Component: a noun phrase corresponding to a technical means or physical element (2) Action / function: A verb phrase that describes the process or action that the component performs. (3) Technical effect: The purpose, advantage, or result achieved by a structure or combination of structures. (4) Relationships between components: Linkages, causal relationships, flows, dependency structures, etc. between the components The processing unit 11 organizes these analysis results into a hierarchical structure or a table format, and outputs a list of components and their descriptions, the role of each component, the overall effect of the invention, as well as the data flow and functional dependencies between the elements.

[0062] At this time, in order to extract technical semantic structures from the wording contained in the claims, the processing unit 11 is configured to provide the generation AI with analysis target information including natural language data obtained from the application documents, and to instruct it to perform structural classification and semantic relevance estimation processing based on a predetermined analysis perspective.

[0063] The "analysis viewpoint" includes at least the technical configuration, operation (action), purpose or effect, and their related structure, and the processing unit 11 has the function of controlling and instructing the generation AI to output based on these viewpoints.

[0064] Furthermore, the processing unit 11 controls the processing by presenting the following instruction sentence (prompt) to the generated AI. "You are an AI that breaks down and visualizes patent claims in a way that is easy for examiners to understand. Based on the wording of the claims, 1) A list of technical components (with a brief description of each element); 2) The specific functions and actions of each component 3) The effect achieved by the invention; 4) Print the relational structure (cause-effect or means-end) of the components. Furthermore, if the processing unit 11 determines that any of the configurations, actions, or effects extracted by the AI ​​is highly abstract and may hinder the examiner's understanding, it may be configured to warn the examiner that "there is an unclear expression."

[0065] This function allows examiners to quickly grasp the technical scope and key points of the invention described in the claims, reducing the risk of misunderstandings and misinterpretations in the early stages of examination. It also provides applicants and attorneys with information to confirm whether the AI's understanding is consistent with the invention's intent, making it useful for correcting misunderstandings and considering supplementary explanations. Furthermore, this function constitutes a technical feature of the present invention in that it uses the natural language understanding capabilities of the generative AI to handle context-dependent expressions, object omissions, and the extraction of functional relationships, which were difficult to achieve using conventional rule-based or simple syntactic analysis methods. <Example output> The following is an example in which the processing unit 11 analyzes the content of claim 1, breaks it down into constituent elements, functions, effects, and inter-configuration relationships, and outputs the results. <Claim> "A wearable health management device comprising a temperature sensor that detects the temperature of the skin surface, a determination means that determines whether the user has a physical condition abnormality based on the output from the temperature sensor, and a notification means that outputs a warning in accordance with the determination result." <Components and explanations> Temperature sensor: A means for detecting the temperature of the skin surface - Judgment method: A method for judging abnormalities in physical condition based on the output of the temperature sensor Notification method: A method for outputting a warning depending on the judgment result <Functions of each component> Temperature sensor → temperature measurement · Judgment method → ​​Abnormality judgment by comparison with standard Notification method → ​​Alert output (sound, vibration, light, etc.) <Technical effect> -Real-time health monitoring -Early detection and prevention of health risks -Continuous measurement possible while wearing the device <Relationships between components> Temperature Sensor ↓(Temperature data) Judgment means ↓(Judgment result) Notification means ↓(Warning) user <Points to check (Notes from AI)> If the "criteria for determining abnormal physical condition" are not clearly stated in the specification, attention must be paid to the description requirements. - The disclosure of the specific output means (LED, vibration motor, etc.) of the "notification means" may be too abstract.

[0066] <Function 2. AI summarizes claims, and users (examiners) can review and revise the summary> The processing unit 11 of the present invention has a function of using a generation AI to summarize the technical content of claims included in application documents and outputting the summary in a format that can be confirmed and modified by the examiner.

[0067] Claims are usually written in a formalized style to clarify their legal effect, but this can make it difficult to intuitively grasp the overall technical essence and structure of the invention. This function summarizes the claim information in natural language and displays it in a format that is easy for examiners to understand, thereby quickly supporting technical understanding at the early stages of examination.

[0068] The processing unit 11 extracts the important components, functions, and purposes from the claim text and then instructs the generating AI to output a concise summary that maintains the technical intent. This summary is adjusted to be an informal and neutral description of the technical gist of the invention.

[0069] The processing unit 11 is further configured to output the summary in an editable form for the examiner, allowing the examiner to add his or her own understanding or supplementary notes, or to link with an interface that corrects discrepancies with the AI's understanding.

[0070] Furthermore, the processing unit 11 may be configured to switch the abstract level (simple / detailed) depending on the field of application and the complexity of the claims when generating the abstract, thereby enabling flexible display according to the reader's examination strategy and level of proficiency.

[0071] The generated AI is given prompts containing the following instructions: "Please write an abstract that briefly summarizes the structure and function of the claims below, allowing the examiner to intuitively understand the key points of the invention. Please write in a general technical explanation style, rather than in legal terms, so that the examiner can easily grasp the content of the invention." This summary function allows users to confirm the AI's understanding and determine the direction of the examination at an early stage. It also provides a means for AI to verify, from a third-party perspective, whether the wording of the claims matches the intention.

[0072] Note that this summary process is not a legal interpretation of claims, but is intended to assist in understanding the technical intent, and is different from the format of the "abstract" prescribed in the Patent Act.

[0073] Below is an example in which the processing unit 11 outputs a summary based on the claims using a generation AI, and then presents the summary in a format that allows the examiner to review and revise the content. <Claim 1> "A driving assistance device for a vehicle, comprising a radar sensor that measures the distance to a vehicle ahead, a warning means that issues a warning in accordance with the distance, and a display means that notifies the driver of the operating status of the warning means." <Summary output by generative AI (standard mode)> The present invention relates to a driving assistance device mounted on a vehicle, which has a function of measuring the distance to a vehicle ahead using a radar sensor and issuing an alarm based on the result. It also has a display means so that the driver can visually confirm the status of the issued alarm. This makes it possible to assist in collision avoidance and alert the driver. <Example of corrections made by examiner (editing within the confirmation UI)> *The following is an example in which the user (examiner) made some corrections to the AI ​​output.

[0074] (Before modification) "It has the ability to issue an alert based on the results."

[0075] (After correction) "This includes a configuration that issues an alarm such as a sound or vibration depending on the result."

[0076] (Additional notes) "Regarding the display means, it is understood that their role is to visually notify the driver whether or not a warning has been issued." <Final confirmed abstract (after examiner's revisions)> The present invention relates to a driving assistance technology for vehicles, and is equipped with a configuration that measures the distance to a vehicle ahead using a radar sensor and issues an alarm using sound, vibration, etc. depending on the distance. It also has a display means that visually notifies the driver whether an alarm has been issued, thereby contributing to alerting the driver and preventing collisions.

[0077] As shown in this output example, this function starts with the abstract created by the AI ​​generator and adds adjustments based on the examiner's knowledge, making it possible to understand the claims more clearly and accurately. This also contributes to improving the accuracy of initial examination decisions and linked outputs with other functions (e.g., creating notices of reasons for refusal, presenting candidate interpretations, etc.).

[0078] <Function 3. Generate multiple candidates (interpretation patterns) for claim interpretation, allowing the user (examiner) to select and indicate them> The processing unit 11 has the function of using a generation AI to present multiple different interpretation patterns, taking into account the possibility of ambiguity or ambiguity in the interpretation of technical configurations of the wording described in the claims.

[0079] The wording of claims may have multiple possible technical interpretations depending on the inventor's intentions, the field's practice, and the syntactic structure. In response to this, this function uses AI to perform syntactic and semantic parsing processing assuming multiple interpretations, and by presenting different interpretation candidates in parallel, it helps examiners make appropriate selections or suggestions.

[0080] The processing unit 11 clearly indicates the extracted components, differences in operational flow, differences in technical effects, etc. for each candidate, and outputs them in a format that can be compared by the examiner. At the same time, comments on the differences between the candidates created by AI are also presented.

[0081] The processing unit 11 may also be configured to accept an input operation by the examiner to select and confirm the interpretation that the examiner considers most appropriate from the presented candidates.The examiner may also input his or her own corrections or additions, and record them in a format that can be used for reevaluation.

[0082] When there is a possibility that the interpretation of the technical structure of a claim is not unambiguous, the processing unit 11 is configured to present input information including the natural language structure of the claim to the generation AI and instruct it to output multiple technical interpretation candidates.

[0083] At this time, the processing unit 11 instructs the generation AI to include, as elements to be included in the interpretation candidates, components, the flow of processing, technical effects, and the relationship between components, and controls the AI to explicitly separate and output these elements.

[0084] In addition, the processing unit 11 may be configured to request the generation AI to additionally explain the differences from a technical perspective for the plurality of output interpretation candidates.

[0085] An example of a prompt for the generation AI is as follows. "Regarding the following claim, detect locations where there may be differences in interpretation in terms of syntax and meaning, and present a plurality of interpretation candidates (ways of reading the configuration) corresponding to each. Also, show how the configuration, operation, and technical effects differ for each interpretation candidate."

[0086] With this function, the examiner can pre-examine the ambiguous interpretation of the claim, preventing the incorrect issuance of rejection reasons due to incorrect understanding and the disagreement of views with the applicant.

[0087] In addition, it may be configured such that the interpretation selected by the examiner is linked and processed so as to be reflected in subsequent other functions (such as description requirement judgment, inventiveness judgment, etc.), thereby ensuring the consistency of the examination.

[0088] <Output example> The following is an output example in which the processing unit 11 analyzes the claim and presents a plurality of reading candidates for a configuration where the interpretation may vary. <Target claim> "An image processing apparatus comprising: display means for detecting an object from input image data and displaying information corresponding to the object." <Output of interpretation candidates by AI

[0089] (Interpretation pattern 1) · Configuration: Object detection means + display means · Processing flow: Image → Object detection → Search for relevant information → Display Effect: UI control that switches the display content based on the object Note: Mainly for object detection, and information is assumed to be obtained from an external database

[0090] (Interpretation pattern 2) Configuration: Integrated configuration in which the image display means detects and displays the object Processing flow: Image → Object recognition and display are processed in an integrated manner Effect: Automatic object analysis during real-time display Supplementary Note: The origin of the "information corresponding to the object" is unclear, and the separation of the detection means and the display means is unclear.

[0091] (Interpretation pattern 3) Configuration: Object detection is a separate process, and the display method is simply a UI Processing flow: The object is recognized externally or through pre-processing, and the device simply displays the information. Effect: The terminal is dedicated to displaying information, and recognition is assumed to be performed on the cloud, etc. Supplement: Only the configuration of the display means may be the invention of the present application.

[0092] <AIによるコメント> The relationship between "display means for displaying information corresponding to the object" is unclear, and it is not clear whether the entire process from "detection" to "display" is completed within a single device, which means there are multiple possible interpretations. <Example of examiner selection operation> [ ] Pattern 1 is adopted [R] Pattern 2 is adopted [ ] Check box mouth Pattern 3 is used [ ] Enter your own interpretation (→Enter in the text box)

[0093] In this way, by providing multiple technical interpretations for the wording of claims and allowing the examiner to select or modify the most appropriate interpretation, the accuracy and efficiency of the examination will be significantly improved.

[0094] <Function 4. AI automatically generates confirmation questions for the user (examiner) regarding the syntax of claims, helping to confirm understanding> When the expression contained in a claim is abstract or the relationship between elements is unclear, the processing unit 11 has the function of automatically extracting points that the examiner should confirm in order to correctly understand the content of the claim, and outputting corresponding confirmation questions using a generation AI.

[0095] This function analyzes the syntactic structure of claims, detecting technical ambiguity, ambiguity, abbreviations, unclear dependencies, etc., and then presents corresponding confirmation questions in the form of "Does this statement mean XX?" or "What is the relationship between elements A and B?" to help examiners confirm their understanding.

[0096] The processing unit 11 uses the results of the syntax and semantic analysis of the claims to prioritize extracting points that may be misunderstood by the examiner, and instructs the generation AI to generate confirmation questions to encourage clarification of the technical meaning of those points.

[0097] At this time, the processing unit 11 is configured to provide the generation AI with the following higher-level conceptual information and control instructions:

[0098] (1) Natural language text of claims to be analyzed (2) The location and classification of detected syntactic ambiguities (e.g., modifier relationships, missing objects, etc.) (3) The format of the question to be output (it must be a question and must be phrased to confirm understanding) (4) Add and display the background of each question (which configuration it relates to). The processing unit 11 may also be configured to display the generated questions in a list format in cooperation with a UI that allows the examiner to select whether the question is appropriate or not and whether confirmation is required or not.

[0099] Examples of prompts for the generative AI include: "Regarding the following claims, in order to assist in understanding the configuration, please generate confirmation questions for each ambiguous expression or omitted meaning that the examiner may need to confirm. The questions should be output from perspectives such as configuration, relationship, operation flow, and meaning of terms, and presented in a list." With this function, when reading the claims, the examiner can consciously confirm the important points in advance, making it possible to improve the efficiency of the examination and prevent misunderstandings.

[0100] In addition, these confirmation questions can also be utilized as materials for improving the description and considering supplementary descriptions by providing feedback to the applicant or the agent.

[0101] The following is an output example where the processing unit 11 analyzes the syntax of the claims and automatically generates confirmation questions to be presented to the examiner using the generation AI.

[0102] <Target claim (Claim 1)> "A sensing device comprising: a sensor that acquires environmental information; a detection means that detects a danger based on the output of the sensor; and a control means that performs control according to the danger." <Confirmation question output by AI (in list form)> Q1: "What specific events or states does the “danger ” refer to? Among the information detected by the sensor, what is used to determine danger?" Q2: "What specific control actions are assumed for the “control according to the danger ”? For example, are examples such as warnings, stops, and avoidance actions included?" Q3: "In addition to general information such as temperature, humidity, and illuminance, does the “environmental information ” also include sounds, vibrations, position information, etc.?" Q4: "Are there any data conversion processes or pre - processing steps (such as filtering) between the sensor and the detection means? If so, is the configuration also included in the technical scope of this application?" Q5: "Are the 'detection means' and 'control means' to be understood as separate components, or are they part of a processing circuit that operates in an integrated manner?" <Example of interface display for examiners> [R] I would like to confirm Question 1 with the applicant. [ ] Question 2 does not need to be confirmed [R] Save Question 3 as a memo for internal review [R] Question 4 will be used as a reference for determining the description requirements [ ] Delete question 5

[0103] As shown in this output example, this function allows examiners to clearly recognize "what is ambiguous" and "what they need to check" when reading claims, which can lead to technical alignment with the applicant's intentions and guidance on how to write the claims as needed.

[0104] <Function 5. AI generates hypothetical examples based on claims, allowing users (examiners) to check their understanding> The processing unit 11 has the function of using a generation AI to imagine how the technical configuration of a claim described in an application document can be realized as a real embodiment, and presenting the resulting imagined example.

[0105] This function uses AI to supplement the output of specific configurations and operations to make it easier to visualize actual technical devices, processing procedures, UIs, and configuration examples for abstract or structurally complex claims, enabling examiners to gain a more realistic understanding of the technical content of the invention.

[0106] The processing unit 11 analyzes the components, operation details, processing order, input / output relationships, etc. in the claims, and then instructs the generation AI to output an assumed embodiment corresponding to the claims. At this time, the processing unit 11 is configured to give the following information and control instructions to the generated AI:

[0107] (1) Input: The full text of the claim and its component parts (2) Output format: Summary description of the embodiment, specific configuration examples, operation scenarios (3) Control: Cover all aspects so that the components are reflected, and if possible, include the user perspective and environmental settings (4) Presentation of usage examples: Also instruct to illustrate examples of typical usage scenarios in light of the technical application field Examples of prompts given to generative AI are, for example, as follows. "Regarding the following claims, please show examples of embodiments assumed based on the technical configuration. Specifically describe how each component is implemented as what parts or processes and how it operates in what order. Also give examples of what usage scenarios it is used in." With this function, the examiner can have a technical actual image corresponding to the claim, and is assisted in grasping the configuration, understanding the effect, etc. in the judgment of the description requirements and the judgment of inventiveness.

[0108] <Output example> <Target claim (Claim 1)> "A pedestrian detection system comprising a camera that images the front of a vehicle, a detection means for detecting pedestrians from the camera image, and a warning means for alerting a driver based on the detection result." <Assumed embodiment generated by AI> This system uses a wide-angle camera mounted on the upper part of the windshield of an automobile to acquire in real time the video of the traveling direction of the vehicle.

[0109] This video data is transmitted to an in-vehicle processor and analyzed by an algorithm (e.g., an object detection model based on deep learning) that performs detection based on the posture, movement, and contour features of pedestrians.

[0110] When a pedestrian is detected and it is determined that there is a possibility of intersecting the vehicle's path, an audio warning is issued from the in-vehicle speaker, and at the same time, an icon of "Pedestrian Attention" is displayed on the display on the dashboard.

[0111] This configuration enables the driver to detect the presence of pedestrians early even in situations where it is difficult for the driver to see (at night, in blind spots, etc.), thereby supporting safe driving. <Application example> Used to detect pedestrians running out into the street when driving in urban areas or around schools Standard equipment for applications where safety is of paramount importance, such as elderly assistance vehicles - Works with obstacle avoidance functions as part of autonomous driving technology This output example allows examiners to grasp the concrete system behavior from the abstract claim language, and more clearly understand the feasibility of implementing technical features, the interrelationships between components, and how technical challenges are addressed.

[0112] <Function 6. When the user (examiner) modifies the wording of a claim, the AI ​​provides feedback on the improvement in understanding> When the examiner performs operations such as amending, supplementing, or simplifying the wording of a claim, the processing unit 11 has a function of evaluating the degree of improvement in syntactic and semantic understanding using a generative AI based on a comparison of the wording before and after the amendment, and presenting the results to the examiner as feedback.

[0113] This function helps examiners confirm whether changes made to claims that are unclear and that have been made to eliminate the possibility that the AI ​​may have misunderstood the claims have actually improved the AI's technical understanding.

[0114] The processing unit 11 provides the AI ​​with the claim text before and after the amendment, and instructs it to compare the extracted results of the components, actions, and effects for each, as well as the technical summary, and to output the results including the degree of agreement, differences, and improvements in clarity.

[0115] At this time, the processing unit 11 is configured to give the following inputs and instructions to the generating AI: (1) Input: Claim before amendment, Claim after amendment (2) Output: A comparison table of the technical interpretations (structure, functions, effects) of each AI (3) Evaluation items: Degree of component consistency, clarity of logical structure, clarity of causal relationship, reduction of abstraction level, etc. (4) Comprehensive evaluation: Whether the understanding has been improved by the amendment and a summary comment on the reason Prompt examples given to the generative AI may consider the following formats.

[0116] "Regarding the claims before and after the following amendments, technically understand and decompose each of them, and compare the extraction results of the composition, function, and effect. Also, evaluate how much the understanding accuracy of the AI has been improved by the amendment and explain the reason." With this function, the examiner can know more deeply how the AI understands the claims and confirm whether their own amendments are technically beneficial, contributing to the improvement of the accuracy of examination support. <Output example> <Claim before amendment> "A sensing device comprising a sensor for measuring temperature and processing means for performing processing according to the output of the sensor." <Claim after amendment> "A sensing device comprising a temperature sensor for measuring the outside air temperature and processing means for controlling the rotation speed of a cooling fan based on the output of the temperature sensor." <Comparison of component extraction by AI> Components (before amendment): · Sensor: Measures temperature · Processing means: Performs processing (unclear) Components (after amendment): · Temperature sensor: Measures outside air temperature · Processing means: Controls the rotation speed of the cooling fan Clarity of operation: Before amendment: The function of the processing means is abstract, and the object and the control target are unclear. After amendment: The processing target (cooling fan) and the operation (rotation control) are clearly stated. Extraction of technical effects: Before amendment: Unclear or unable to judge [[ID= forty-six]]After amendment: Temperature stabilization by fan control according to the environment can be read. <Evaluation of Understanding Improvement by AI> · After the correction, the relationship between components became clear, and the causal flow of "temperature → rotation control" was explicitly shown, making technical understanding much easier. · Since the actual operation content performed by the processing means was explicitly shown, it became easier to extract the differences from the prior art, and the possibility of contributing to the determination of inventive step increased. <AI Comprehensive Evaluation Comment> "It can be evaluated that the correction improved the specificity of the configuration and the clarity of the processing content, and clarified the technical framework of the invention. A significant improvement in understanding during examination was recognized." Thus, by providing a feedback function for the understanding results of AI, the clarity of the description and the quality of examination support are greatly improved.

[0117] <Function 7. Function to detect the possibility of being judged difficult to understand in examination by syntactic analysis of claims> The processing unit 11 is provided with a function of analyzing the syntax of the claims included in the application documents, detecting portions where the expression of the claims is judged to be unclear, illogical, or overly abstract, or portions that are likely to make it difficult for an examiner to understand, and outputting them as a notice.

[0118] This function extracts portions that may violate the description requirements (especially clarity) of Article 36, Paragraph 6, Item 2 of the Patent Law through syntactic analysis by AI in advance, thereby assisting in grasping descriptions that are difficult for an examiner to read and making an early correction determination. The processing unit 11 performs syntactic and semantic checks including the following viewpoints through the cooperation of the syntactic analysis engine and the generative AI

[0119] (1) Uncertainty of dependency (2) Existence of ambiguous subjects, objects, and modifiers (3) Logical leaps or omissions (e.g., means for performing B using A is unclear) (4) Excessive abstract expressions, functional expressions with unclear uses (5) Redundancy or circular description of components

[0120] At this time, the processing unit 11 gives the following high-level instructions to the generative AI. · Input: Claim text · Analysis targets: Sentence structure, semantic consistency, clarity of composition · Output: Extraction of parts estimated to be difficult to understand, description of reasons, direction of improvement · Classification: Assign categories (such as grammatical structure, ambiguity of terms, leap in composition, etc.) for each pointed-out item An example of the prompt given to the generative AI is as follows. "Regarding the following claims, detect parts that are considered difficult to read from a grammatical and logical perspective, and present them in a list with reasons and classifications (e.g., unclear dependency, unclear function, etc.) for each." With this function, the examiner can objectively and structurally analyze the description of the claims, accurately identify parts that are difficult to understand, and appropriately point out violations of clarity and request amendments from the applicant. <Output example> <Target claim (Claim 1)> "An information processing apparatus comprising means for processing image data and control means for optimizing by providing information to a user." <Syntactic analysis results and list of pointed-out items by AI> [Pointed-out part]: "Control means for optimizing by providing information" [Classification]: Ambiguity of functional expression / leap of purpose and means [Reason]: What is to be "optimized" is not specified, and the causal relationship between "providing information" and "optimization" is unclear. The target and specific operations of the control means cannot be read. [Pointed-out part]: "Providing information to a user" [Classification]: Redundancy of composition / ambiguity of subject [Reason]: The means of providing information is unclear, and the content and method of provision are not described, so the relationship with the party "user" is ambiguous as a technical configuration. [Points to be noted]: "Means for processing image data" [Classification]: Excessive abstraction [Reason]: The specific content of the "processing" (conversion, compression, analysis, etc.) is completely unknown, making it difficult to identify the technical configuration. <AIによるコメント(まとめ)> The description of the "control means" in this claim is ambiguous as a technical configuration, and the logical structure of the purpose and means is not clear. - Because verbs such as "process" and "provide" are ambiguous, it is desirable to clearly state the processing content and purpose in order to clarify the technical content. - As there is a possibility of a violation of the clarity requirement, guidance to the applicant for correction should be considered.

[0121] This function allows the AI ​​to thoroughly examine the structure of the text and present a comprehensive and classified list of important "difficult-to-read passages" that are important to examiners.

[0122] <Function 8: Evaluate the consistency between claims, specifications, and drawings, and extract violations of written requirements> The processing unit 11 evaluates the technical consistency between the descriptions in the claims, specifications, and drawings contained in the application documents, and has the function of extracting and pointing out areas that may violate the description requirements, particularly when the structure of the claims does not correspond to or are not consistent with the specification or drawings.

[0123] This function will assist examiners in efficiently and accurately checking the consistency of application documents from the perspective of the description requirements of Article 36, Paragraph 6, Item 1 (detailed description of the invention) and Item 6 (consistency with drawings) of the Patent Act.

[0124] The processing unit 11 detects and verifies whether there is a corresponding description in the specification for each component described in the claims, and whether there is a drawing number or illustration corresponding to the configuration, and automatically extracts the possibility of missing descriptions or inconsistencies. At this time, the processing unit 11 is configured to provide the following information and control instructions to the generated AI:

[0125] (1) Input information: Claim text, full specification, drawing description, and symbol list (2) Comparison viewpoint: correspondence between claim structure and description in the specification, correspondence between claim structure and drawing elements (3) Output: Check for consistency, identify corresponding areas, and list of items that may be incomplete (4) Classification: omissions (no description in the specification or drawings), misprints (inconsistencies in code numbers), ambiguity in the scope of application, etc. The generated AI can be given prompts such as: For each element included in the following claims, please check the corresponding parts in the specification and drawings, and extract any elements where the correspondence cannot be clearly confirmed, where the description may be insufficient, or where the symbols or descriptions do not match the drawings. This function will enable examiners to comprehensively check application documents from the perspective of description requirements, improving the accuracy of pointing out incomplete descriptions or requesting corrections from applicants. It can also be used as a pre-check function for applicants, helping them consider corrections or supplements.

[0126] The following is an example of an output in which the processing unit 11 evaluates the correspondence between the elements of the claim and the specification and drawings, and extracts and presents items that are determined to have a problem with consistency. <Claim 1> "An information processing device comprising: an operation means for accepting user input; a calculation means for executing processing based on the user input; and a display means for displaying the processing results." <Elements of the specification and drawings> [Figure 1]...Display unit 101, calculation unit 102, memory 103 [Figure 2] Example of operation screen (component names not included) [Description]...There is a detailed explanation of the display unit. The calculation process is briefly mentioned, but there is no explicit description of the "operation means." <Consistency Evaluation Output by AI> [Component]: Operating means [Evaluation]: Insufficient corresponding description in both the drawings and the specification [Reason for pointing out]: The specific configuration corresponding to the "operating means" specified in the claims is not described in the drawings or the specification. For example, the names and symbols of input means such as keyboards, touch panels, mice, etc. are unknown. [Classification]: Lack of description (no description or illustration of corresponding configuration) [Component]: Arithmetic means [Evaluation]: Illustrated as "arithmetic unit 102" in the drawings. The description of the processing by the "arithmetic unit" in the specification is brief, and the correspondence with the "processing based on user input" described in the claims is unclear. [Reason for pointing out]: Difficult to determine whether the processing content matches the claims. [Classification]: Unclear description (ambiguous correspondence) [Component]: Display means [Evaluation]: Described and illustrated as "display unit 101" in both the specification and the drawings. Clearly corresponding. [Classification]: Good correspondence (consistent in both description and illustration) <AI Comment (Summary)> ·Regarding the "operating means" in the claims, no specific description or illustration corresponding to the entire application documents was found, so there may be a violation of the description requirements based on Article 36(6)(i) of the Patent Law. ·Regarding the arithmetic means, the relationship between the requirement of "processing based on user input" in the claims and the processing content in the drawings and the specification is unclear, and there is a lack of technical consistency. ·Regarding the display means, it is judged that the consistency is good. With this output example, it becomes possible to visualize the correspondence between each component of the claims and the specification and the drawings, and support the examiner to grasp early the violation of the description requirements. It is also useful for the applicant to check the description of the application documents in advance.

[0127] <Function 9: Function to output notice of rejection reason for lack of description requirements> The processing unit 11 is configured to acquire the specification, claims, drawings, and abstract entered as application documents relating to the invention, and evaluate the correspondence and consistency of the components described in these documents using a generation AI.

[0128] Specifically, the processing unit 11 extracts the words, technical functions, and interrelationships of each component described in the application documents, and analyzes the semantic correspondence between the drawings, specifications, and claims using a large-scale language model (LLM) capable of natural language processing. This makes it possible to determine whether the technical objects match based on the context, not just based on the match of words and the consistency of drawing numbers, but also in cases where synonyms, variations in expression, and demonstrative terms (e.g., "this component," "the means," etc.) are included.

[0129] In this process, the processing unit 11 inputs the following information to the generation AI (stores it in memory): - The full specification as filed (with paragraph numbers) - Claims (with claim numbers) - A table of the components shown in the drawings and their corresponding drawing numbers - A set of evaluation rules based on past examination practices and examination standards (guidelines for determining description requirements) Then, the processing unit 11 gives the generation AI a prompt such as the following to cause it to perform a consistency evaluation and a determination of the possibility of a violation of description requirements. <Example prompt> You are a generative AI assisting patent examiners. Based on the application documents below (specification, claims, and drawing information), please identify the semantic correspondence between each element and identify any areas that may not comply with the description requirements. ·Details: <0001> The present invention relates to <0002> As shown in FIG. 1, configuration A includes: · Claims: <Claim 1> A mechanical device comprising configuration A and configuration B. · drawing: Figure 1: 100: Configuration A, 110: Configuration B · Instructions: 1. Please list the names and functions of the components described in the specification. 2. Determine whether there is a semantic correspondence between the above configuration and the elements described in the claims and drawings. 3. If there are any elements in the claims that are not shown in the drawings, please point them out. 4. If the claim contains an element that is not described in the specification, please point it out. 5. If there is a possibility of incompleteness in the information, please print out a mock example of a notice of refusal in accordance with the format used by the Patent Office.

[0130] Based on the analysis results obtained from the generation AI, if a component not described in the drawings is included in the claims or specification, or if it is determined that there is a discrepancy in the technical relationship between the components, the processing unit 11 determines that this may fall under the category of ``matters not described in the detailed description of the invention'' as defined in Article 36, Paragraph 6, Item 1 of the Patent Act, and generates pointed out information that can be output as a violation of the description requirements.

[0131] In addition, since the generation AI has learned the patterns of points raised based on past notices of reasons for refusal and examination guidelines documents, the processing unit 11 generates a draft notice of reasons for refusal with sample text for the points raised, and supports the examiner in making decisions on corrections and amendments.

[0132] In this way, by utilizing generative AI to evaluate semantic consistency and extract violations of written requirements, which were difficult to detect using conventional rule-based processing, the support system for writing office action notices of the present invention enables a high level of automation and support for examination practices.

[0133] Below is an example of the output produced by function 9: <Function 1: Output example> Integrity assessment report (example) Target application: Patent application 202X-123456 Output date and time: April X, 2025, 13:24 Examiner in Charge: (Unspecified) Evaluation mode: Description requirement check mode (Article 36, Paragraph 6, Item 1) <Consistency judgment result> Item: Correspondence between elements of claims and drawings Evaluation result: Discrepancy Comments and supplement: The "rotational support member" in claim 1 does not have a corresponding symbol or component name on the drawings (Figs. 1 to 3). Item: Drawing structure and specification Evaluation result: Partial discrepancy Point raised: The component "120: Rotating shaft holder" shown in Figure 2 is not specifically explained in the specification (only mentioned in the <Brief description of the drawings>). Item: Terminology consistency Evaluation result: Discrepancy The “rotary support member” in the claim is described in several different ways in the specification, such as “rotary holder” and “support component,” making the technical consistency unclear.

[0134] <Automatically generated Office Action Draft> Regarding the invention of claim 1 of this application, the specification does not contain a detailed description of the structure corresponding to the "rotation support member," nor can the drawings contain any reference symbols or names corresponding to said structure. <Classification of findings (classification and evaluation by AI)> Article 36, Paragraph 6, Item 1 (Insufficient detailed explanation) Possibility of application: High Recommended Action: - Add a description in the specification that clearly indicates the specific structure and function of the "rotation support member" Add corresponding symbols to drawings or clarify component names · The terminology has been standardized to "rotation support member" or the correspondence is indicated in parentheses. <Automatically generated Office Action Draft> "Regarding the invention of claim 1 of the present application, the specification does not contain a detailed description of the structure corresponding to the "rotation support member," nor can the reference symbol or structure name corresponding to said structure be found on the drawings. This makes it difficult to understand the technical content of the configuration of the invention, and it is deemed that the invention does not comply with the description requirement of "no detailed explanation of the invention" as stipulated in Article 36, Paragraph 6, Item 1 of the Patent Act. This output example can also function as a draft of a rejection notice that examiners can quote directly, and is also intended to be used as a checklist for evaluating the consistency of elements.

[0135] <Feature 10. A function to output notices of reasons for refusal in both cases where violations of description requirements are strictly applied and where they are loosely applied> The processing unit 11 has a configuration that can control the application level (evaluation level) of the examination standards when evaluating whether the application complies with the description requirements stipulated in Article 36 of the Patent Act based on the specification, claims, drawings, etc. contained in the application documents for the invention.

[0136] That is, the processing unit 11 selects either the "strict mode" or the "lenient mode," or both simultaneously, depending on the tendency of judgments in examination practice regarding the description requirements based on Article 36 of the Patent Act (for example, Paragraph 6, Items 1 to 3), and performs a process of comparing and presenting the output contents of the rejection notice based on each examination standard.

[0137] In strict mode, defects such as missing elements, failure to state technical effects, and insufficient explanation of the function of the configuration are evaluated from the perspective of a relatively high likelihood of being deemed non-compliant under the examination standards. On the other hand, in lenient mode, even if an application contains variations in expression or abstractness within the range permitted by normal amendments or practical practices, it is not immediately deemed to be a violation, and a judgment is made taking into account the discretionary scope of the examination.

[0138] The processing unit 11 is configured to use a generation AI that can switch between and apply multiple evaluation criteria, and controls the evaluation and draft output of rejection notices based on stricter and lenient criteria by providing the generation AI with prompts including contextual instructions with parameters such as the following, along with the application documents.

[0139] For example, a generator AI might be given prompts of the form:

[0140] "You are a generation AI assisting examiners at the Japan Patent Office. Please evaluate the following application documents (specification and claims) to determine whether they violate the written description requirements under Article 36 of the Patent Act, changing the application level of the examination standards.

[0141] In Mode A, the examination criteria should be strictly applied, and evaluation should be rigorously conducted on unclear components, lack of technical effect, inconsistency in terminology, etc.

[0142] In Mode B, take into consideration the practical tolerance range and do not judge as a violation anything that is likely to be corrected.

[0143] For each mode, print the following information:

[0144] 1) The description requirements that were found to be in violation and the relevant sections 2) Explanation of the reason for the violation 3) Draft of Notice of Refusal 4) Comments on possible amendments (optional) The processing unit 11 outputs the evaluation results of the written requirements for each mode and the corresponding draft of the notice of reasons for refusal in parallel based on the output results of the generation AI, allowing the applicant or examiner to relatively understand the degree of strictness with which the contents of the application are likely to be rejected.

[0145] Furthermore, if the items pointed out in both modes are the same, it can be judged that there is an objectively high possibility of rejection, whereas if there are points pointed out only in the strict mode, it can be assumed that the case can be avoided by amendment or supplementary explanation.

[0146] With this configuration, the processing unit 11 can utilize the generation AI to simultaneously present multiple examination perspectives for a single application, and can visualize risks related to clarity and description requirements in three dimensions.

[0147] <Example output> Below is an example in which the processing unit 11 uses the generation AI to evaluate the description requirements for the same claim and specification in both "strict mode" and "lenient mode," and then outputs the results for comparison. Applicable claim: <Claim 1> "An image recognition device comprising: image processing means for analyzing an input image; discrimination means for discriminating the type of subject using the feature amount extracted by the image processing means; and notification means for issuing a notification in accordance with the discrimination result obtained by the discrimination means." Subject statement: Paragraph <0015> ~ <0023> There is a related statement in Evaluation Mode A (Strict Enforcement): Regarding the element "notification means," the specification does not state any specific examples of the configuration of said means (e.g., buzzer, LED, audio output device, etc.), and there is no technical explanation as to the mechanism by which said means notifies. Therefore, it may be determined that there is no description corresponding to the detailed description of the invention.

[0148] Furthermore, the algorithm and processing conditions for the "identification of the type of subject" performed by the "identification means" are not clearly stated, and no information is provided that would enable a person skilled in the art to implement the means.

[0149] <Points to be pointed out>: Patent Law Article 36, Paragraph 6, Item 1 (Incomplete detailed description of the invention) <Draft of Notice of Refusal (excerpt)>: "In the specification of the present application, the technical configuration and operation of the 'notification means' and 'identification means' recited in claim 1 are unclear, and therefore the invention is deemed not to have been described in detail." Evaluation mode B (lenient): With regard to the element "notification means," the general description in the specification can be interpreted as being within the scope of easily envisioning the adoption of notification means, such as an alarm sound or visual output, and therefore it can be judged that a person skilled in the art would be able to understand it even without a detailed description of the configuration.

[0150] Furthermore, with regard to the discrimination of the type of subject in the "discrimination means," it can be interpreted from the context of the entire specification as assuming general image classification processing (e.g., feature extraction + use of a classifier), and there is room for interpretation that it does not violate the description requirements under the Patent Act.

[0151] <Points to be noted>: None (determined to meet requirements) <Output>: No notice of reasons for refusal. However, the following suggestions will be output as optional supplementary information. <Supplementary suggestion>: "For notification means and determination means, specific configuration examples or processing flows are illustrated or paragraphed. <0016> <0017> By adding this information, we hope to improve the details. Comparison results: In strict mode, the lack of technical support for the elements is pointed out as a violation of the description requirements, while in lenient mode, the elements are deemed to be understandable within the scope of common general knowledge of a person skilled in the art, and no reason for refusal is output. This allows applicants to flexibly assess the risks of the description content of the application and consider the need for amendments or additions.

[0152] In this way, the processing unit 11 can evaluate multiple examination perspectives for a single application in a three-dimensional manner and output the comparison results by using the generation AI's context understanding and switching control of legal judgment criteria.

[0153] The switching between the evaluation levels (strict mode and lenient mode) is not necessarily limited to a binary one, and the processing unit 11 may be configured to set the evaluation level of the description requirements in multiple stages (for example, levels 1 to 5). Here, level 1 is the standard for applying the description requirements most leniently, and the higher the level number, the stricter the standard for evaluating the description contents in light of the examination standards. This enables a gradual risk assessment of the application contents and enables output control according to the standards desired by the examiner or applicant.

[0154] The evaluation level may be set by the processing unit 11 dynamically receiving an instruction from a system user, for example, by allowing an applicant or examiner to specify a desired evaluation level via a user interface. Furthermore, the processing unit 11 may be configured to apply a specific mode or evaluation level as a default based on a predetermined initial setting value.

[0155] <11. A function that uses documents provided by an external cited document research system to generate a comparison table with claims, and judges and visualizes whether or not a document has novelty or inventive step in accordance with patent examination standards. The processing unit 11 has a function of organizing the structural correspondence between the cited documents and the claims as a comparison table based on the cited documents provided from an external literature research system, and determining and visualizing the presence or absence of novelty (Article 29, paragraph 1 of the Patent Act) or inventive step (Article 29, paragraph 2 of the Patent Act) based on the comparison table in accordance with the patent examination standards.

[0156] This function, based on the fact that comparative examination with cited documents is essential in examination practice, aims to automatically organize and output matching and differences for each structure, clarify technical differences, and provide materials that will help in making judgments.

[0157] The processing unit 11 breaks down the technical components described in the claims into element units, extracts the corresponding descriptions in the cited documents, and classifies and evaluates whether each element is "explicitly described," "suggested," or "not described" in the cited documents.

[0158] At this time, the processing unit 11 is configured to provide the following information and instructions to the generating AI: (1) Input: Claim structure breakdown information, full text or key excerpts of cited documents (2) Processing content: Search for the description corresponding to each component, judge semantic identity and similarity, and classify (match, similarity, no description) (3) Output format: Comparison table for each element, highlighting of differences, simple judgment results regarding novelty / inventive step (4) Criteria: Based on the patent examination standards, consider whether the disclosure of the cited document is identical to the structure of the claim or whether it is obvious to a person skilled in the art. Examples of prompts that can be given to the generating AI include: "Compare the claims below with the disclosures contained in the cited documents and determine whether each element is described, not described, or implied in the documents. Print out the results in the form of a comparison table and evaluate the overall novelty and inventive step. If there is a possibility of a violation of inventive step, please print out a mock rejection notice in the format used by the Japan Patent Office." This function allows examiners to visually grasp the relationship between cited documents and claims, and clearly organize technical differences and the basis for judgments. Applicants can also use it as material for comparison with anticipated cited examples.

[0159] Alternatively, the processing unit 11 has the function of using cited documents provided by an external literature research system to compare the structures of the claimed invention with those of the inventions described in the cited documents, generating a comparison table for each component, and determining the presence or absence of novelty or inventive step based on the comparison results in accordance with the patent examination guidelines, and finally automatically generating text that conforms to the format of the Japan Patent Office's notice of reasons for refusal.

[0160] The purpose of this function is to systematically organize technical comparisons between claims and cited documents from the perspectives of whether the structure is the same or different, and whether it is explicit, suggestive, or absent, and to quickly and accurately output notices of reasons for refusal based on the comparisons.

[0161] The processing unit 11 breaks down the elements of the claims using natural language processing, etc., searches and extracts corresponding descriptions in the cited documents using a generation AI, and then evaluates whether each element is disclosed in the cited documents in accordance with the patent examination guidelines, and generates a comparison table.

[0162] Furthermore, based on the comparison table, the processing unit 11 applies a predetermined judgment logic (e.g., if all components match, it lacks novelty; if some components are missing, it moves on to an inventive step judgment), derives a legal evaluation using AI, and then outputs a text formatted in the Patent Office's notification format.

[0163] At this time, the processing unit 11 is configured to provide the following information and control instructions to the generated AI: (1) Input: Claim element breakdown results, full text or summary of cited documents, output format (JPO format) (2) Processing: Classification of correspondences by component (matching / similar / unspecified), evaluation of the novelty and inventive step of the component, and formatting and output of the reason for the judgment (3) Output: A completed document based on the JPO's standard rejection notice, with the relevant items (cited documents, structure, reasons for judgment, etc.) automatically inserted. An example prompt for the generation AI is designed as follows: "Compare the following claims with the inventions described in the cited documents, element by element, and determine whether they are novel or involve an inventive step. Based on your findings, please print out a notice of refusal in accordance with the format of the notice of refusal issued by the Japan Patent Office." This function enables examiners to efficiently carry out the entire examination process, from comparing each configuration to drafting a notice, and also enables applicants to plan countermeasures based on the notice output by the AI.

[0164] <Example output> <Claim 1> "A vending machine comprising: a product selection button; a dispensing means for dispensing a product corresponding to the button; a storage means for storing inventory information; and a control means for determining whether or not to dispense the product based on the inventory information." <Reference (JP 2020-123456 A)> - Product selection button available -Available to dispense products according to selection No mention of inventory information storage function - There is no mention of control over whether or not payouts are possible. <Component Comparison Table> Elements of the claim Contents of the cited document Judgment Product selection button Description Match Payment method stated Match Storage method for storing inventory information No corresponding description Not specified Control method to determine whether or not payment is possible No corresponding description Not specified <Determination of inventive step> The cited document does not describe a configuration for controlling dispensing based on inventory management, and it cannot be said that such a configuration could be easily arrived at from the cited invention. Therefore, it cannot be recognized that the invention of claim 1 of the present application could have been easily made by a person skilled in the art based on the cited documents. <Draft Notice of Reasons for Refusal (JPO Format)> [Notice of Reasons for Refusal (Draft)] The invention according to claim 1 of the present application falls under Article 29, paragraph 1 or 2 of the Patent Act and is therefore unpatentable for the following reasons: References JP 2020-123456 A (hereinafter referred to as "Reference 1") Comparison results Cited Document 1 discloses a product selection button and a configuration for dispensing products in accordance with the selection, which coincides with the configuration described in Claim 1 of the present application. On the other hand, the cited document 1 does not disclose any configurations corresponding to the "storage means for storing inventory information" and the "control means for determining whether or not to dispense" described in claim 1 of the present application. Judgment Regarding the undescribed configuration of the invention according to claim 1 of the present application, even taking into consideration the disclosure of cited document 1 and the well-known art, it is not recognized that a person skilled in the art could have easily arrived at the configuration. Therefore, the invention of claim 1 of the present application cannot be said to have been easily made by a person skilled in the art based on cited document 1, and does not fall under Article 29, paragraph 2 of the Patent Act.

[0165] (Note: There are no reasons for refusal, so the application is considered to be a patent.) In this way, based on the results of the comparison and legal judgment on a component-by-component basis, a draft notification in accordance with the format used by the Japan Patent Office is ultimately output, thereby achieving rationalization and transparency in examination practice.

[0166] <Function 12: When the units or notation formats of numerical values ​​differ between cited documents and claims, this function converts them to calculate comparable values. This function presents the results of the conversion process and the prerequisites used for the conversion to the examiner, allowing them to check their validity and provide instructions for correction.> The processing unit 11 has a function to convert numerical ranges or characteristic values ​​described in claims into mutually comparable numerical values ​​when the unit systems or notation formats differ between the numerical ranges or characteristic values ​​described in the claims and the corresponding numerical values ​​described in the cited documents.

[0167] For example, if a claim is written in "weight percentage (wt%)" and a cited reference is written in "mole percentage (mol%)," a logical and quantitative conversion is performed based on supporting information such as the molecular weight and density of the component in question, and the two are converted into a comparable state using the same unit system.

[0168] In addition, the processing unit 11 also has a validity presentation function that explicitly records auxiliary information (such as molecular weight, temperature, pressure, and concentration conditions) that is a prerequisite for the conversion process, allowing the examiner to confirm, approve, or correct the content.

[0169] The processing unit 11 is configured to provide the following information and control instructions to the generating AI:

[0170] (1) Input: Numerical information in claims, corresponding numerical information in cited documents, auxiliary information (molecular weight, specific gravity, known database) (2) Processing: Numerical unit conversion, formula construction, and result verification (3) Output 1 (corresponding to function 12): Comparable converted value pairs (claim value vs. converted cited reference value) (4) Output 2 (corresponding to function 13): Display of parameters, assumptions, reference data used for conversion, and their reliability For the generated AI, you can use prompts like the following: "In order to make the 'wt%' in the claim and the 'mol%' in the cited document comparable, please convert using the molecular weight of the component in question. Also, please print out the molecular weight and conditions used for the conversion and submit them in a form that can be confirmed by the examiner." This function eliminates formal differences resulting from differences in numerical formats, enabling accurate assessment of novelty or inventive step based on substantive comparison. It also prevents the conversion process from becoming a black box, enabling transparent examination.

[0171] <Example output> <Description of the Claim> "...the component A is contained in an amount ranging from 2.0 to 5.0 wt%." <References> "...Component A in the present invention is contained in the range of 0.10 to 0.26 mol%." <Conversion processing result by processing unit 11 (output of function 12)> Molecular weight of target component A: 98.1 g / mol Total sample mass assumption: 100 g (standard) mol% → wt% conversion formula: wt% = (mol% × molecular weight) / denominator conversion factor (based on a standard sample mass of 100 g) → 0.10 mol% × 98.1 ÷ 1,000 = 0.0981 wt% → 0.26 mol% × 98.1 ÷ 1,000 = 0.255 wt% Conversion result: The concentration of component A in the cited reference is 0.10-0.26 mol% → approximately 0.098-0.255 wt% <Comparison judgment> Claim range: 2.0 to 5.0 wt% -Range after citation conversion: 0.098~0.255 wt% → The scope is non-overlapping and novel (the structure is clearly different) <Presenting conversion premise information to examiner (output of function 13)> Molecular weight used: 98.1 (see PubChem) Total mass assumed at the time of conversion: 100g (standard conversion conditions) Effects of other additives and dilution solvents: Negligible (not specified) Reliability of conversion process: Medium (limited conditions apply due to unknown component status) Confirmation request: If necessary, applicants can be asked to submit supplementary information on component status or environmental conditions. This output example makes it possible to determine substantial differences through conversion even for components that are difficult to compare formally due to differences in units and notation formats, and ensures transparency of the basis for conversion, thereby increasing the objectivity and reliability of the review.

[0172] <Feature 13. Adjust the criteria for determining inventive step (Article 29, Paragraph 2 of the Patent Act) to be stricter or lenient, and output a notice of reasons for refusal in each case> The processing unit 11 is configured to be able to switch the application level of the examination standards when evaluating whether or not an invention satisfies the inventive step requirement stipulated in Article 29, paragraph 2 of the Patent Act based on the content of the invention described in the application documents.

[0173] In other words, the processing unit 11 can set a mode or level that allows the strictness of the evaluation to be adjusted according to the instructions of the examiner or applicant for each evaluation element in determining inventive step, such as "extraction of differences from cited inventions," "commonality of problems," "citability," and "whether or not the invention is easily arrived at."

[0174] In the stricter mode, the existence of well-known art based on the cited documents, the similarity of the problem to be solved, or the possibility of combining with other documents are broadly evaluated, and there is a tendency to deny the inventive step relatively strictly. On the other hand, in the lenient mode, there is a tendency to emphasize factors that are favorable to denying ease of arrival, such as differences in the problem to be solved, the difficulty of substituting elements, and differences in technical fields, and there is a tendency to evaluate the inventive step positively.

[0175] The processing unit 11 receives as input not only application documents (claims, specifications, drawings, etc.) but also a group of cited documents (e.g., obtained from an external search support system or a research database), and performs a logical evaluation of the technical differences and ease of arrival of the invention based on these documents.

[0176] The evaluation process is carried out by a generating AI, and determines whether or not an invention involves an inventive step based on factors set out in the examination guidelines, such as the correspondence between the elements of the invention and the cited documents, the commonality of the problems, the possibility of substituting means, adjacent technical fields, and the presence or absence of design matters.

[0177] In this case, the generating AI is given a prompt that includes contextual instructions such as: "You are a generative AI assisting patent office examiners. Please evaluate whether or not the invention involves an inventive step (Article 29, Paragraph 2 of the Patent Act) based on the application documents and cited documents below. The evaluation criteria change depending on the mode. In Mode A, the evaluation should be based on criteria that make it easier to deny inventive step, taking into consideration a wide range of factors, such as well-known technologies, similarity of the problem, and the possibility of design changes. In Mode B, emphasis should be placed on the differences in the tasks and the uniqueness of the structure, and evaluation should be in favor of progressiveness. For each mode, print the following information: 1) Clearly state the differences from the cited references 2) Judgment on whether or not the invention was easily arrived at and its grounds 3) Step-by-step explanation of evaluation according to the screening criteria (3 steps) 4) Draft of Notice of Reasons for Refusal (by mode) Based on the output of the generation AI, the processing unit 11 outputs the results of the inventive step judgment evaluated for each mode or level and a draft notice of reasons for refusal based on the results, and provides this as information that the examiner or applicant can compare and consider.

[0178] This configuration makes it possible to visualize how the assessment of inventive step for the same application varies depending on the strictness of the application of examination guidelines, and also supports the development of strategies for amendments, drafting opinions, and responding to refusals.

[0179] The mode setting for the inventive step evaluation is not limited to a binary value, but may be configured to set evaluation levels in stages (for example, from level 1 to level 5). Here, level 1 is the standard that makes it easier to affirm the inventive step (the lenientest), and the higher the level number, the stricter the evaluation standard that denies the inventive step.

[0180] The switching of the mode or the setting of the level may be configured to be dynamically accepted and set by the processing unit 11 based on an input instruction from a system user, for example, by allowing an applicant or an examiner to specify a desired evaluation level via a user interface. Furthermore, the processing unit 11 may be configured to apply a specific mode or evaluation level as a default based on a predetermined initial setting value.

[0181] <Example output> The following is an example in which the processing unit 11 uses the generation AI to evaluate the inventive step of the same claim and cited documents in "Mode A (strict evaluation)" and "Mode B (relaxed evaluation)," and then outputs the evaluation results and a draft notice of reasons for refusal for comparison. Applicable claim: <Claim 1> "A mobile information terminal comprising a touch panel input unit disposed on the front surface and control means for switching operation modes based on information operated via the touch panel input unit." Reference 1: "A control device for switching operation modes by touching the bottom edge of the screen in a touch-screen mobile device." Reference 2: "A control device for a mobile device that uses mode selection buttons. There is no mention of a function to switch modes via touch input." Evaluation Mode A (Strict Enforcement): Similarities in the elements and the problem to be solved are recognized between the invention of claim 1 and D1. The technical idea of ​​touch panel operation and mode switching is common, and the fact that the specific configuration of the control means is not illustrated is also similar to the description in D1. Furthermore, the button-based mode selection function described in Cited Document 2 differs only in the form of the input means, and can be interpreted as a replaceable configuration from the perspective of the technical field and the problem at hand.

[0182] Therefore, the structure of claim 1 could have been easily arrived at by a person skilled in the art by combining cited document 1 with cited document 2, and is therefore deemed to lack the inventive step defined in Article 29, paragraph 2 of the Patent Act.

[0183] <Draft Office Action (Mode A)>: "The invention of claim 1 of the present application could have been easily arrived at by a person skilled in the art by applying the well-known mode selection technology described in cited document 2 to the technology described in cited document 1, and therefore does not involve an inventive step as defined in Article 29, paragraph 2 of the Patent Act." Evaluation mode B (relaxed application): The “touch panel input unit arranged on the front” and the “control means for switching modes” in claim 1 can be construed as not merely a substitution of components, but as including creative features related to the design concept and operability of the user interface. The cited document 1 is limited in that it uses touch on a specific area (bottom edge) of the screen as a trigger, and does not mention free mode operation using the entire screen as in the present application. In addition, the cited document 2 assumes physical button operation, and does not suggest the benefits that can be obtained by introducing a touch panel (for example, intuitive operation or reduction in physical parts).

[0184] Therefore, the invention in question differs from the cited invention in terms of the problem, configuration, and effect, and is not easily arrived at. It is therefore evaluated as involving an inventive step.

[0185] <Output (Mode B)>: It is determined that a notice of reasons for refusal is not necessary. However, the following reference output is available. <Reference suggestions>: "By clearly stating that the technical features of the claimed invention are manifested in terms of operability and design concept, it is possible to further strengthen the argument for inventive step." Comparison results: In Mode A, a draft notice of reasons for refusal is output due to lack of inventive step, while in Mode B, the inventive step is affirmed due to differences in structure and unique effects, and no notice is output. In this way, by switching between evaluation modes, it is possible to quantitatively grasp the likelihood of fluctuations in examination results for the same application content, which is effective as material for considering application strategies.

[0186] In this way, the processing unit 11 can control the strictness of the evaluation criteria for determining the inventive step, thereby supporting risk analysis of the application contents and the determination of appropriate amendments and assertions. <Feature 14. Automatically adjusts the content and severity of reasons for refusal according to the examination trends for each application field (mechanical, electrical, chemical, etc.)> The processing unit 11 has a function to adjust the content of the notice of reasons for refusal and the strictness of the judgment criteria in a manner appropriate to the technical field to which the applied invention belongs, by reflecting the tendency of reasons for refusal being pointed out during examination and the actual application of examination standards.

[0187] It is known that there are differences in the examination trends regarding description requirements, inventive step, and clarity depending on the field of application. For example, in the field of chemistry, strict scrutiny is placed on the specific description of constituent substances and feasibility of implementation, while in the field of software, emphasis tends to be placed on clarifying the structure and logical description of actions and effects.

[0188] The processing unit 11 acquires information on the technical field of the invention specified by the user or by identifying the technical field of the invention from the contents of the application documents (specification, claims, drawings, abstract, etc.), and compares it with a database of examination practices, past notices of reasons for refusal, and trends in the application of examination standards related to that field to select an appropriate examination standard model.

[0189] Based on the selected model, the processing unit 11 evaluates the application documents and generates a draft notice of rejection using prompts that clearly indicate the field-specific judgment perspectives for the generating AI.

[0190] For example, in the field of chemistry, the following instructions might be given:

[0191] "You are a generative AI assisting examiners specializing in chemistry at the Japan Patent Office. Please evaluate the chemical structures described in the application documents from the following perspectives in light of Article 36, Paragraph 4, Item 1 (enablement requirement) and Paragraph 6, Item 1 (detailed description requirement) of the Patent Act. -Is the manufacturing method of the compound clear? -Is the correspondence between the structural formula and physical property data described? -Is the reproducibility of the effect in a specific range of values ​​(e.g., wt%, mol%, temperature, pressure, etc.) clearly stated? -Are there any ingredients included in the claims that are not listed in the specification? Based on this, please output the points you have pointed out and the reasons for any suspected violations, as well as a draft of the notice of reasons for refusal in accordance with the examination guidelines. On the other hand, in the software field, prompts that include the following aspects are used:

[0192] "You are a generation AI assisting an examiner specializing in information processing at the Japan Patent Office. Please evaluate the following items regarding the information processing invention described in the application documents, focusing on Article 36, Paragraph 6, Item 2 (clarity requirement) and Article 29, Paragraph 2 (inventive step) of the Patent Act.

[0193] -Is the processing flow in the claims clear? · Are expressions such as "processing means," "judgment unit," and "output module" specific? ·Are the tasks, structure, and effects technically consistent? -Is there a description that cannot be interpreted as simply automating business procedures? In this way, the processing unit 11 generates a draft of the notice of reasons for refusal based on the evaluation viewpoint and expression style according to the field of invention, thereby realizing highly effective feedback in line with examination practice. The automatic determination of the technical field may be based on keywords contained in the application documents, JPO classifications (FI / F terms, etc.), IPC or CPC classification codes, or user input selections.

[0194] In addition, the processing unit 11 can improve the accuracy of outputting notices of reasons for refusal optimized for each field by using, as learning data, correlation information between patterns of reasons for refusal that have frequently occurred in the past in each technical field and the examination pass rate.

[0195] <Example output> The following are examples of cases in which the processing unit 11 adjusts the examination viewpoints and judgment criteria according to the characteristics of the application field, and then outputs a draft notice of reasons for refusal using the generation AI based on those viewpoints. Two examples are shown for the fields of chemistry and software. <Example 1: Chemistry field (application related to the manufacturing method of organic compounds)> Applicable claim: <Claim 1> "A method for producing an organic compound represented by formula (I), the method comprising the step of mixing raw material A and raw material B and carrying out a heating reaction at a temperature in the range of 40°C to 60°C." Description of the specification (abstract): Although the reaction process is described, there is no information about the yield or physical properties of the target product. The specific solvent name and reaction time are also not given. <Processing results (indications based on the chemical field model)>: Issue: Although the reaction temperature range (40-60°C) is stated, the specification does not state the basis for realizing the effects of the present invention (e.g., high yield, selectivity, etc.) within this range. Regarding the synthesis of constitutive formula (I), analytical data (NMR, IR, etc.) corresponding to the structural characteristics of the target substance are not described, and reproducibility and feasibility are not guaranteed. <Applicable article>: Patent Law Article 36, Paragraph 4, Item 1, Paragraph 6, Item 1 <Draft Notice of Reasons for Refusal (Chemical Field)>: "In the specification of the present application, there is no description in the specification supporting the effect of producing the organic compound described in claim 1. Therefore, it is not recognized that the invention is possible for a person skilled in the art to carry out the invention, and it is determined that the invention does not comply with Article 36, Paragraph 4, Item 1 of the Patent Act." <Example 2: Software field (application related to business processing support systems)> Applicable claim: <Claim 1> "A business process support system comprising: a processing means for analyzing input business information and automatically generating work instructions for a person in charge; and a display means for displaying the work instructions." Description of the specification (abstract): There is no flow chart of the processing method. The description of how the analysis is performed is vague. There are no examples of the displayed content. <Processing results (indications based on the software domain model)>: Issue: The algorithms for the "analysis" and "automatic generation of work instructions" processes are not described, and the specific processing steps are unclear. The "display method" is too general, making it difficult to understand what should be displayed, to whom, and how. The claims are abstract and insufficient to define the technical scope of the invention. <Applicable Provisions>: Patent Act Article 36, Paragraph 6, Item 2 (clarity), Article 29, Paragraph 2 (inventive step) <Draft Notice of Reasons for Refusal (Software)>: "The specification does not contain sufficient information to grasp the technical content of the invention regarding the functions and configuration of the 'processing means' and 'display means' recited in claim 1 of the present application, making it difficult to identify the invention. Therefore, the invention is deemed not to comply with Article 36, Paragraph 6, Item 2 of the Patent Act. Furthermore, the issue of presenting information based on business processing is widely found in existing technologies, and the configuration in question is deemed to be easily conceivable by a person skilled in the art. Therefore, the invention is not in compliance with Article 29, Paragraph 2 of the Patent Act either." Comparison supplement: As shown in this output example, the processing unit 11 controls the detection of reasons for refusal and the creation of notification drafts based on the examination perspective and required level of expression for each technical field, enabling flexible output in line with examination practices.

[0196] <Feature 15. The ability to set a patent allowance rate and output a notice of refusal appropriate to the rate> The processing unit 11 is configured to receive a desired patent assessment rate (e.g., 50%, 70%, 90%, etc.) for the invention related to the application from the system user, or to generate a notice of reasons for refusal that matches the assessment rate based on a preset assessment rate.

[0197] Here, the "patent allowance rate" refers to the likelihood that the invention will be granted a patent in practice, with a higher value indicating fewer obstacles in the examination process, and conversely, a lower value indicating a higher likelihood of rejection due to issues such as lack of inventive step or written requirements. The processing unit 11 adjusts and outputs the severity, content, number of pointed out items, and strength of expression of the notice of reasons for refusal to match the desired allowance rate.

[0198] The processing unit 11 receives application documents (claims, specifications, drawings, etc.) and cited or comparative documents as input, and uses a generation AI to perform a comprehensive evaluation of inventive step, clarity, and written requirements. At this time, the generation AI is given prompts including contextual instructions such as:

[0199] "You are a generation AI assisting patent examiners. Please draft a notice of reasons for refusal based on the application documents and cited documents below. Please note that the target patent allowance rate for this application is 70%, so please output comments that are consistent with the applicable probability.

[0200] - If there is a significant lack of inventive step or a violation of the description requirements, clearly state it. Minor defects will be pointed out to the extent possible, - The application as a whole should be in a state where there is a good chance that it will be granted a patent through amendments and arguments. Furthermore, when the appraisal rate is set low (for example, 30%), the processing unit 11 controls the generation AI to intentionally incorporate the following strong rejection factors. - Denial of inventive step by combining multiple documents - Extraction of unclear structure and violation of Article 36, Paragraph 6 -Insufficient technical evidence based on non-correspondence between the structure and the effect On the other hand, if the assessment rate is set high (e.g., 90%), only minor defects (such as inconsistent punctuation, missing symbols, or the use of mildly abstract language) will be pointed out, and a notice of refusal will be generated for issues that can be easily addressed through amendments or written opinions.

[0201] The processing unit 11 has a control configuration for scoring the points of concern so as to be consistent with the assessment rate, and adjusting the number of points of concern, weighting, and description style so that the score matches the expected rejection risk.

[0202] In addition, the designation of the allowance rate can be used by applicants for preliminary evaluation, by agents for strategic design, or to assist examiners in determining whether or not to allow allowance, and can be applied flexibly depending on the purpose of evaluation.

[0203] The assessment rate may be set in stages, and the severity of the content and expression of the notice of reasons for refusal may be changed in conjunction with the magnitude of the numerical value. The processing unit 11 may set the assessment rate by explicit input from the system user, or may be configured to set it automatically based on a predetermined initial setting value.

[0204] <Example output> The following is an example in which the processing unit 11 uses the generation AI to specify different patent allowance rates (90%, 70%, 40%) for the same claim, and outputs a notice of rejection that matches those allowance rates. Applicable claim: <Claim 1> "An air conditioning control device with a temperature sensor, comprising a control means for automatically adjusting the airflow volume and direction according to the indoor temperature, and a storage means for recording changes in the outdoor temperature." Reference 1: Air conditioning equipment that adjusts airflow according to the room temperature Reference 2: Air conditioning equipment with a logger function that records outside temperatures <When the assessment rate is set to 90%> Evaluation policy: In principle, only minor deficiencies that can be avoided by correction are pointed out. - No major issues were found with regard to inventive step Issue: The structure of the "control means" described in the specification is somewhat abstract, and it would be desirable to describe more clearly how "wind direction adjustment" is performed. - There are some inconsistencies in the correspondence between the drawings and the claims, so it is recommended to clarify the reference symbols. <Draft Notice of Refusal (90%)>: "The configuration of the 'control means' described in the specification is abstract, so in order to facilitate understanding of the invention, it is desirable to provide a supplementary explanation of the specific operation of the control. In addition, we would like to request that the reference relationships between the drawings and the claims be adjusted to improve consistency." Remarks: There is a high possibility that amendment will lead to a patent being granted. <When the assessment rate is set to 70%> Evaluation policy: - Although it can be addressed by amendment, attention is needed to some technical aspects regarding feasibility and inventive step. Issue: -Regarding the "automatic wind direction adjustment" configuration, there are insufficient examples of the correspondence with specific sensor signals and mechanisms, and no detailed explanation is given. - The type of data to be recorded and the storage format for the "storage means for recording" are not clearly stated, making it difficult for those skilled in the art to understand. <Draft Notice of Refusal (70%)>: "Regarding the 'control means for automatically adjusting the wind direction' recited in claim 1, the specification does not specifically describe the configuration or operation of the control mechanism, making it difficult to implement the invention. In addition, the handling of the outside temperature data recorded in the 'storage means' is unclear, and a more detailed explanation is required." Remarks: This can be addressed by submitting a statement of opinion or making minor amendments. <When the assessment rate is set to 40%> Evaluation policy: - Point out content that is likely to be rejected in practice from the perspective of both inventive step and description requirements Issue: The cited document 1 discloses air volume control and describes a control that is functionally similar to air direction adjustment. The recording method is shown in Reference 2, and it is technically easy to combine the two documents. - The description of "wind direction adjustment" is abstract and insufficient, and feasibility is not ensured. <Draft Notice of Refusal (40%)>: "The invention of claim 1 of the present application is recognized as one that a person skilled in the art could have easily arrived at by combining the airflow control device described in D1 with the outdoor temperature recording means described in D2, and therefore does not involve an inventive step as defined in Article 29, Paragraph 2 of the Patent Act. Furthermore, the specification does not provide a detailed description of the specific means for 'adjusting the airflow direction,' and therefore does not meet the requirement of a detailed description as defined in Article 36, Paragraph 6, Item 1 of the Patent Act." Remarks: The possibility of rejection is high, and the application contains structural problems that are difficult to address through amendment. Comparison results: As shown in this output example, the processing unit 11 adjusts the severity and depth of the grounds for rejection to match the desired allowance rate, providing simulation results according to the likelihood of the application being approved. This provides applicants and their attorneys with important information for making practical decisions, such as amendment strategies, adjustments to the scope of claims, and the submission of preliminary arguments. <Feature 16. Simple mode for generating an Office Action even when some of the application documents (description, claims, etc.) are missing> The processing unit 11 is equipped with a "simple mode (prediction mode)" that automatically generates a draft of a notice of rejection based on some of the input information, even when any of the specification, claims, drawings, and abstract of the set of documents related to the application have not been input or are in the preparation stage.

[0205] This mode meets the needs of those who want to understand in advance the potential risk of rejection during examination, even when only the claims or technical description have been provisionally prepared during the application preparation or invention consideration stage.

[0206] Specifically, when only a claim is input, the processing unit 11 uses a generation AI to analyze the description of the elements, functions, and effects of the claim, and automatically estimates the possibility of a violation of the description requirements or a lack of inventive step in light of past rejection cases in examination practice.

[0207] In addition, if only the specification is input, the system may be configured to generate anticipated claims based on the purpose, problem, configuration, and effects of the invention, and virtually output reasons for rejection of those claims.

[0208] In this case, the generation AI is given prompts to instruct it to process the context to compensate for any deficiencies in the input information, resulting in the following output:

[0209] "You are a generative AI tasked with evaluating inventions before they are filed for patents. Below is an outline or claim of an invention.

[0210] Although it is still in the early stages and the specification and drawings are not yet complete, please use this information to predict possible reasons for rejection (description requirements, clarity, inventive step) and print out the items that the examiner may point out.

[0211] If necessary, please also suggest technical comparisons with the cited documents and supplementary information. Based on the output of the generation AI, the processing unit 11 presents the applicant or inventor with information that visualizes the risk of future rejection, while clearly indicating that it is not a formal notice of reasons for refusal. This provides effective feedback for reviewing the structure and description of the application documents at the preparation stage.

[0212] Furthermore, in the simple mode, the processing unit 11 may be configured to automatically create virtual drawings, configurations with drawing numbers, operation flows, etc. by having the generation AI assume and complete elements that have not been input (drawings, configuration diagrams, etc.), and to perform trial evaluations based on them.

[0213] The simple mode is activated by the system user (applicant, agent, in-house engineer, etc.), and allows the predictive capabilities of the system of the present invention to be utilized even when all formal application documents are not yet ready.

[0214] <Example output> The following is an example of the processing unit 11 using the generation AI to predict and output a notice of reasons for refusal when part of the application documents are missing (only claims / only description) in "simple mode." <Case 1: When only the claim is entered> Input details (claim): <Claim 1> "A medical information processing device comprising: a determining means for analyzing vital information acquired from a patient and determining whether or not there is an emergency; and a notifying means for notifying the result of the determination." Evaluation results (simple mode): Abstract processing expressions such as "analysis," "judgment," and "notification" are used, and it is unclear what logic or mechanism is used to realize these means. In particular, the “determination of urgency” involves a medical judgment, so it is expected that disclosure of the algorithm or evaluation criteria will be necessary. - As corresponding drawings and specifications have not been entered, there is a high possibility that the structure will be evaluated as not being sufficiently supported. Possible clauses to be pointed out: ·Article 36, Paragraph 6, Item 2 of the Patent Act (clarity) Patent Act Article 36, Paragraph 4, Item 1 (Enablement) Predicted output (draft reasons for refusal): "The invention of claim 1 may not satisfy the clarity requirement stipulated in Article 36, Paragraph 6, Item 2 of the Patent Act, because the details of the configuration or function of each means are unclear and, in particular, no technical support is provided for the method of determining whether there is an urgency. In addition, there may be insufficient information for a person skilled in the art to carry out the invention." Additional suggestions: "By adding to the specification the standard values, measurement targets, evaluation algorithms, etc. that the assessment means refer to, it is expected that the feasibility and clarity will be improved." <Case 2: When only the statement is entered> Input details (description summary): The present invention relates to automatic sorting of packages in a logistics warehouse. The camera captures images of the parcel and identifies its size, weight, and destination code. The device is configured to distribute packages to multiple conveyors based on the recognition results. - Automating sorting processes reduces work time. Assumed claim by processing unit 11 (automatically generated): "A logistics system for sorting packages, comprising: an imaging means for imaging packages; a discrimination means for discriminating attributes of the packages based on the images; and a transport means for sorting the packages according to the attributes." Possible reasons for refusal: If the means for "identifying attributes" or "assigning" are not specifically stated, there is a possibility that inventive step and feasibility may become an issue. -There are many similar configurations in existing logistics systems, so there is a risk that it may be deemed to be easily conceived. Predicted output (draft reasons for refusal): "The envisioned invention comprises components for imaging, discrimination, and transport, but each component may be merely a combination of publicly known technical elements, which may negate the inventive step stipulated in Article 29, Paragraph 2 of the Patent Act. Furthermore, if the specification does not provide sufficient technical details regarding the discrimination process, it may violate the description requirement." Additional suggestions: "Patentability and enforceability are expected to be improved by including specific examples of image processing techniques and machine learning methods used for discrimination, as well as distribution control algorithms, in the specification." Even when input information is limited, the processing unit 11 can predict reasons for refusal and output improvement proposals by utilizing the context understanding, technical knowledge, and past examination cases of the generation AI. This allows for practical and useful feedback to be obtained even during pre-application review.

[0215] The above describes typical functions of a support system for writing notices of reasons for refusal. It is not necessary to provide all of the above functions; a combination of some of the above functions may be implemented. For example, of the 16 functions, the system may be configured to provide at least two, three, four, three to five, or four to eight of them. For example, the system may be configured to provide only a function for evaluating consistency with drawings, a function for syntactic analysis, and a function for determining inventive step, or a function for combining a function for generating a comparison table with documents and a conversion processing function, depending on the application and the subject of examination.

[0216] In addition, the system may be configured to dynamically and selectively execute specific functions depending on the field of the application under examination, the status of missing documents, the examiner's operational policy, etc. This allows for flexible operation specialized to support the examiner's judgment.

[0217] The following aspects are also included in the present invention. [Form 1] A support device for creating a notice of reasons for refusal, Using generative AI, (1) A function to break down the claimed invention into its constituent elements, actions, effects, etc., in order to aid users in understanding the claims of the application; (2) A function that presents a draft abstract of the claim and allows the user to modify it; (3) A function that generates multiple interpretations of the claim and allows selection or indication; (4) A function to generate confirmation questions for the syntax of the claim to aid in understanding; (5) A function to generate hypothetical examples based on the claims and to confirm the level of understanding; (6) A function to update and present the AI's understanding in response to amendments to claims by the user; (7) A function to detect expressions that are deemed difficult to understand by syntactically analyzing claims; (8) A function to evaluate the consistency between the drawings, specifications, and claims included in an application and to identify violations of written requirements; (9) A function to output a notice of reasons for refusal based on insufficient description requirements; (10) A function to output multiple notices of reasons for refusal with varying application standards for violations of description requirements, (11) A function to generate a comparison table with claims using documents provided by an external cited document research system, and visualize the presence or absence of novelty or inventive step; (12) A function to perform conversion processing according to differences in numerical units and notation formats between claims and cited documents, and to present the conversion results and prerequisites; (13) A function to switch between standards for determining inventive step and output notices of reasons for refusal based on each standard. (14) A function to automatically adjust the content and severity of reasons for refusal based on the examination trends for each application field; (15) A function to set the patent allowance rate arbitrarily and output a notice of reasons for refusal according to the set value. and (16) A function to provide a simplified mode for generating a notice of rejection even when some application documents are missing; A support device for creating a notice of reasons for refusal characterized by having at least one of the functions above. [Form 2] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit of the system for supporting the creation of a notice of reasons for refusal is characterized by having functions (1) and (7). [Form 3] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (2) and (3). [Form 4] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (4) and (5). [Form 5] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (6) and (7). [Form 6] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (8) and (9). [Form 7] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (9) and (10). [Form 8] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (11) and (12). [Form 9] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (12) and (13). [Form 10] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (14) and (15). [Form 11] In the system for supporting the creation of notice of reasons for refusal according to aspect 1, The processing unit is characterized by having functions (15) and (16). [Explanation of symbols]

[0218] 1: System 10: Management device 20: User terminal 30: External information providing device 11, 21, 31: Processing section 11 12, 22, 32: Input section 13, 23, 33: Output section 14,24,34: Storage part 15, 25, 35: Communications Department

Claims

1. A support system for creating a notice of reasons for refusal having an input unit, a processing unit, and an output unit, the input unit accepts the content of the invention as an input; The processing unit generates a simulated notice of refusal for the invention based on the content of the invention using a generation AI model that has learned data including notices of refusal and patent documents previously created by the Patent Office, the output unit outputs the generated notice of reasons for refusal, The processing unit has a function of generating hypothetical examples based on the claims using the generation AI model and outputting the results so that a user can confirm the results. A support system for creating notices of reasons for refusal.

2. A support system for creating notice of reasons for refusal having an input unit, a processing unit, and an output unit, the input unit accepts the content of the invention as an input; The processing unit generates a simulated notice of refusal for the invention based on the content of the invention using a generation AI model that has learned data including notices of refusal and patent documents previously created by the Patent Office, The output unit outputs the generated notice of reasons for refusal. The processing unit has a function of applying strict or lenient criteria for determining whether a patent application violates written requirements using the generative AI model, and outputting notices of reasons for refusal for both patterns; In the strict application mode, the examination criteria are strictly applied, and instructions are given to strictly evaluate unclear elements, lack of technical effect, and inconsistency in terminology. In the loose application mode, instructions to the generative AI model are included that take into account practical tolerances and do not judge violations for cases where there is a high possibility of correction. A support system for creating notices of reasons for refusal characterized by:

3. A support system for creating notice of reasons for refusal having an input unit, a processing unit, and an output unit, the input unit accepts the content of the invention as an input; The processing unit generates a simulated notice of refusal for the invention based on the content of the invention using a generation AI model that has learned data including notices of refusal and patent documents previously created by the Patent Office, The output unit outputs the generated notice of reasons for refusal. The processing unit has a function of taking into account the patent allowance rate using the generating AI model and outputting a notice of reasons for refusal according to the allowance rate, The instructions to the generative AI model include instructions to specify multiple different patent grant rates and output reasons for rejection that are consistent with each grant rate. A support system for creating notices of reasons for refusal.

4. A support system for creating notice of reasons for refusal having an input unit, a processing unit, and an output unit, the input unit accepts the content of the invention as an input; The processing unit generates a simulated notice of refusal for the invention based on the content of the invention using a generation AI model that has learned data including notices of refusal and patent documents previously created by the Patent Office, The processing unit uses the generative AI model to (1) A function to break down the claimed invention into its constituent elements, actions, effects, etc., in order to assist users in understanding the claims of the application; (2) A function to present a draft summary of the claim and allow the user to modify it; (3) A function to generate multiple interpretations of the claim and enable selection or indication; (4) A function to generate confirmation questions for the syntax of the claim to aid comprehension; (5) A function to generate hypothetical examples based on the claims and to confirm the level of understanding; (6) A function to update and present the AI's understanding in response to amendments to claims by the user; (7) A function to detect expressions that are deemed difficult to understand by syntactically analyzing claims; (8) A function to evaluate the consistency between the drawings, specifications, and claims included in an application and to extract violations of written requirements; (9) A function to output a notice of reasons for refusal based on insufficient description requirements; (10) A function to output multiple notices of reasons for refusal with varying application standards for violations of description requirements, (11) A function to generate a comparison table with claims using documents provided by an external cited document research system and visualize the presence or absence of novelty or inventive step; (12) A function to perform conversion processing according to differences in units and notation formats of numerical values ​​between claims and cited documents, and to present the conversion results and prerequisites; (13) A function to switch between standards for determining inventive step and output a notice of refusal based on each standard. (14) A function to automatically adjust the content and severity of reasons for refusal based on the examination trends for each application field; (15) A function to arbitrarily set the patent allowance rate and output a notice of reasons for refusal according to the set value. and (16) A function to provide a simplified mode for generating a notice of rejection even when some application documents are missing; A support system for creating notices of reasons for refusal, characterized by having at least four or more functions including the function (5) above.

5. A method for supporting the creation of a notice of reasons for refusal in a system having an input unit, a processing unit, and an output unit, comprising: the input unit accepts the content of the invention as an input; The processing unit generates a notice of reasons for refusal for the invention based on the content of the invention using a generative AI model that has learned data including notices of reasons for refusal and patent documents previously created by the Patent Office, the output unit outputs the generated notice of reasons for refusal, The processing unit generates a hypothetical example based on the claims using the generative AI model and outputs it so that the user can confirm its contents. A method for supporting the preparation of a notice of reasons for refusal.

Images