Intellectual property business support method, intellectual property business support apparatus, and intellectual property business support program

Abstract

To provide an intellectual property support method which can intuitively grasp a patent.SOLUTION: The present invention is directed to an intellectual property business support method by an intellectual property business support system having a processor and a memory. The method includes: acquiring patent information; providing a multi-modal model with the patent information and instruction information for use in generation of patent objects as visual information corresponding to the patent information; acquiring the patent object generated by the multi-modal model; and outputting the patent object to a display apparatus.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to an intellectual property business support method, an intellectual property business support device, and an intellectual property business support program. [Background technology]

[0002] Patent Document 1 describes a graphical display method for electronic document searches. The graphical display method includes receiving at least a first search query and a second search query, and searching an electronic document database using the first search query and the second search query to obtain a first document set based on the first search query and a second document set based on the second search query. The first document set includes a first plurality of electronic documents, and the second document set includes a second plurality of electronic documents. The graphical display method further includes generating and displaying a Venn diagram on a graphic display device, the Venn diagram including a first circle representing the first document set and a second circle representing the second document set. The sizes of the first and second circles reflect the number of electronic documents in the first and second document sets, respectively. The first circle overlaps the second circle in an overlapping area representing common electronic documents present in the first and second document sets. The graphical display method further includes, in response to a user input, showing a separation of the first circle and the second circle on a graphic display device, and generating a first visualization chart from the first circle and a second visualization chart from the second circle, the first visualization chart and the second visualization chart showing the ratio of the first document set to the second document set according to user-defined parameters. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 6185127 Summary of the Invention [Problem to be solved by the invention]

[0004] The present disclosure aims to provide an intellectual property business support method that allows users to intuitively understand patents. [Means for solving the problem]

[0005] The present disclosure provides an intellectual property business support method using an intellectual property business support system having a processor and a memory, wherein the processor acquires patent information, provides the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquires the patent object generated by the multimodal model, The patent object is output to a display device.

[0006] The present disclosure provides an intellectual property business support device having a processor and a memory, wherein the processor acquires patent information, provides the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquires the patent object generated by the multimodal model, and outputs the patent object to a display device.

[0007] The present disclosure provides a computer with a method for acquiring patent information, providing the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquiring the patent object generated by the multimodal model, and An intellectual property business support program is provided that causes the patent object to be output to a display device.

[0008] The present disclosure provides an intellectual property business support method using an intellectual property business support system having a processor and a memory, in which the processor acquires patent information, determines characteristic words related to the patent information or themes related to multiple pieces of patent information, provides the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model, acquires the summary information generated by the language model, and outputs the summary information to a display device.

[0009] The present disclosure provides an intellectual property business support device having a processor and a memory, wherein the processor acquires patent information, determines characteristic words related to the patent information or themes related to multiple pieces of patent information, provides the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model, acquires the summary information generated by the language model, and outputs the summary information to a display device.

[0010] The present disclosure provides an intellectual property business support program that causes a computer to acquire patent information, determine characteristic words related to the patent information or themes related to multiple pieces of patent information, provide the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model, acquire the summary information generated by the language model, and output the summary information to a display device.

[0011] The present disclosure provides an intellectual property business support method using an intellectual property business support system having a processor and memory, which associates a focus setting, indicating that a patent indicated by patent information is a notable patent, with the patent information, and when creating a search formula to be used for searching patents from a patent database, displays information on a search result screen indicating whether the patent for which the focus setting has been made is included in the search results based on the search formula.

[0012] The present disclosure provides an intellectual property business support device having a processor and a memory, wherein the processor associates a focus setting with the patent information to indicate that a patent indicated by patent information is a notable patent, and when creating a search formula to be used for searching patents from a patent database, displays information on a search result screen indicating whether the patent for which the focus setting has been made is included in the search results based on the search formula.

[0013] The present disclosure provides an intellectual property business support program that causes a computer to perform the following: associate a focus setting, indicating that a patent indicated by patent information is a noteworthy patent, with the patent information; and, when creating a search formula to be used for searching patents from a patent database, display on a search result screen information indicating whether the patent for which the focus setting has been set is included in the search results based on the search formula.

[0014] The present disclosure provides an intellectual property business support method using an intellectual property business support system having a processor and a memory, in which the processor registers user information of a user, performs inventor authentication for each patent, receives reaction information showing respect for a patent for which the user is the inventor, and outputs information showing the results of compiling the reaction information to the user who is the inventor of the patent.

[0015] The present disclosure provides an intellectual property business support device having a processor and a memory, wherein the processor registers user information of a user, performs inventor authentication for each patent, receives reaction information showing respect for a patent for which the user is the inventor, and outputs information showing the results of compiling the reaction information to the user who is the inventor of the patent.

[0016] The present disclosure provides an intellectual property business support program that causes a computer to register user information of a user, perform inventor authentication for each patent, receive reaction information showing respect for a patent for which the user is an inventor, and output information showing a result of aggregating the reaction information to the user who is the inventor of the patent. [Effects of the Invention]

[0017] According to the present disclosure, it is possible to provide an intellectual property business support method that allows for an intuitive understanding of patents. [Brief explanation of the drawings]

[0018] [Figure 1] A block diagram showing an example of the configuration of an intellectual property business support system according to the present disclosure. [Figure 2] Flowchart illustrating an example of an intellectual property business support method [Figure 3] A diagram showing an example of a Python program when the processing unit makes an API call to the LLM [Figure 4] Conceptual diagram illustrating a selection screen for language expression data [Figure 5] A diagram illustrating visual representation data generated in response to a prompt input. [Figure 6] A conceptual diagram illustrating a plurality of visual expression data displayed on the user terminal 2. [Figure 7] Conceptual diagram showing examples of patterns for visual expression using animal motifs [Figure 8] Conceptual diagram showing examples of patterns for visual expression of personified characters [Figure 9] Conceptual diagram showing an example of the configuration of an intellectual property business support device [Figure 10] Conceptual diagram showing an example of the configuration of an intellectual property business support device [Figure 11] Flowchart illustrating an example of an intellectual property business support method [Figure 12] A diagram illustrating the reaction of a patent object [Figure 13] A diagram illustrating strong operations on patent objects. [Figure 14] Conceptual diagram showing an example of the configuration of an intellectual property business support device [Figure 15] A diagram illustrating a landscape-like visual representation [Figure 16] Flowchart illustrating an example of an intellectual property business support method [Figure 17] Figure showing an example of output to the user [Figure 18] Figure showing an example of output to the user [Figure 19] A conceptual diagram illustrating a search result UI of the present disclosure. [Figure 20] A conceptual diagram illustrating a search result UI of the present disclosure. [Figure 21] A conceptual diagram illustrating a search result UI of the present disclosure. [Figure 22] Conceptual diagram showing an example of the configuration of an intellectual property business support device [Figure 23] Conceptual diagram showing an example of the configuration of an intellectual property business support device [Figure 24] Flowchart illustrating an example of an intellectual property business support method [Figure 25] FIG. 10 is a diagram illustrating an example of an image captured when verifying the identity of an inventor. [Figure 26] Conceptual diagram showing an example of the configuration of an intellectual property business support device DETAILED DESCRIPTION OF THE INVENTION

[0019] Hereinafter, embodiments of the present disclosure will be described in detail with appropriate reference to the drawings. However, more detailed description than necessary may be omitted. For example, detailed descriptions of well-known matters and redundant descriptions of substantially identical configurations may be omitted. This is to avoid unnecessary redundancy in the following description and to facilitate understanding by those skilled in the art. Note that the accompanying drawings and the following description are provided to enable those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter described in the claims.

[0020] (Embodiment 1) FIG. 1 is a block diagram showing an example of the configuration of an intellectual property business support system 100 according to the present disclosure.

[0021] The intellectual property business support system 100 includes an intellectual property business support device 1, a user terminal 2, an external system 3, and a communication network 4.

[0022] The intellectual property business support device 1, the user terminal 2, and the external system 3 are interconnected so as to be able to communicate with each other via a communication network 4 such as the Internet.

[0023] The intellectual property business support device 1 is typically a server, and includes at least a processor 11 and a storage device 12. The processor 11 reads and executes a program 121, data 122, etc. stored in the storage device 12, thereby executing various information processing performed by the intellectual property business support device 1.

[0024] The storage device 12 stores various programs 121 and data 122. The storage device 12 also stores AI models 123. The Artificial Intelligence (AI) models 123 include various models such as language models, image generation models, multimodal models, and Large Language Models (LLMs) described herein.

[0025] The information stored in the storage device 12 may be stored outside the intellectual property business support device 1. For example, various types of information may be stored on a cloud, and the processor 11 may acquire various types of information from the cloud via the communication network 4. Information may be stored on an external server or the like, and the intellectual property business support device 1 and the external server may be connected to be able to communicate with each other.

[0026] The AI ​​model 123 may be stored outside the intellectual property business support device 1. For example, the AI ​​model 123 may exist on the cloud, and the processor 11 may input information to the AI ​​model 123 from the cloud via the communication network 4 and obtain output information from the AI ​​model 123. The AI ​​model 123 may be implemented on an external server or the like, and the intellectual property business support device 1 and the external server may be connected to be able to communicate with each other.

[0027] The intellectual property business support device 1 may further include other components than those described above, such as an input device and a display.

[0028] The user terminal 2 is a device such as a computer or a mobile terminal used by a user. The user may be referred to as a user. The user terminal 2 includes a processor 21, a storage device 22, an input device 23, and a display device 24.

[0029] The processor 21 reads and executes programs stored in the storage device 22 to execute various information processes performed by the user terminal 2. The input device 23 is, for example, a mouse, keyboard, or touch panel, and accepts information input from the user. The display device 24 is, for example, a display or touch panel, and displays information to the user.

[0030] The external system 3 includes, for example, a patent search system such as J-Platpat, an internet application system, etc. The external system 3 may also be various systems provided by vendors other than the Patent Office.

[0031] FIG. 2 is a flowchart illustrating an example of an intellectual property business support method.

[0032] The processing unit 111 acquires patent data (St101). Patent data refers to various data related to patents, such as bibliographies, specifications, claims, drawings, abstracts, and information on intermediate processing. The patent data may be stored in advance in the storage device 12, or may be downloaded on the spot from a server that stores patent data via an Application Programming Interface (API) or the like.

[0033] The processing unit 111 may acquire patent data from data sources such as official patent databases provided by national or regional patent offices or commercial API services that provide patent information. These data sources are typically accessible through APIs. The processing unit 111 acquires patent data by setting an appropriate endpoint URL, necessary authentication information such as an API key, and an appropriate request format such as GET or POST.

[0034] The processing unit 111 acquires the claim text, which is the text information of the claim, from the patent data (St102).

[0035] The processing unit 111 generates one or more pieces of linguistic expression data based on the claim text (St103).

[0036] The linguistic expression data refers to data such as the description of a claim, the description of an inventive step, the description of a patent specification, and the gain pane of a patent specification. The linguistic expression data is typically text data. In this embodiment, the case where the linguistic data is the description of a claim is exemplified.

[0037] The processing unit 111 generates one or more pieces of visual expression data based on the linguistic expression data (St104).

[0038] The processing unit 111 determines the number of visual expression data that have been generated (St105). If the number of visual expression data that have been generated is 0, the processing in this embodiment ends. If the number of visual expression data that have been generated is 1, the processing proceeds to step St106. If the number of visual expression data that have been generated is 2 or more, the processing proceeds to step St107.

[0039] In step St106, the processing unit 111 displays the visual expression data on the display device 24. Displaying on the display device 24 may mean transmitting the display data to the user terminal 2.

[0040] In step St107, the processing unit 111 causes the display device 24 to display a plurality of visual expression data.

[0041] The user selects which of the displayed visual expression data to use by operating the user terminal 2. The processing unit 111 acquires selection result information indicating the result of the user's selection (St108).

[0042] The processing unit 111 displays the selected visual expression data on the display device 24 (St109).

[0043] An example based on the above flowchart will be described below. Assume that the claim text acquired by the processing unit 111 in step St102 is as follows:

[0044] [Claim text content] an arrangement means for arranging, in the game space, a plurality of virtual objects that are movable within the game space and can be glued to each other; a selection means for selecting, by a user's selection operation, a first object from the plurality of virtual objects arranged in the game space; an adhesion object generation means for generating adhesion objects that indicate respective positions of the first object selected by the selection operation and a second object from the plurality of virtual objects; and an adhesion means for adhering, in response to an adhesion instruction from a user, the first object and the second object at the respective positions indicated by the adhesion objects, wherein the adhesion object generation means changes the respective positions indicated by the adhesion objects in response to a movement or a change in posture of the selected first object in the game space before the adhesion instruction is given.

[0045] The processing unit 111 generates a linguistic expression based on the acquired claim text. The linguistic expression can be generated using an existing language model. For example, the processing unit 111 issues an instruction to the language model as a set of the claim text and an instruction command to the LLM, i.e., a prompt, as follows:

[0046] [Example prompt] Please explain what kind of patent this is based on the patent claims entered. Claim: Please explain in a series of sentences of about 3 to 5 lines, without using bullet points or breaking down into steps. Novelty: Please explain the inventive step of the patent in two to three lines. Claim 1: An information processing system comprising: an arrangement means for arranging in the game space a plurality of virtual objects that are movable within the game space and can be glued together; a selection means for selecting a first object from the plurality of virtual objects arranged in the game space through a user's selection operation; an adhesion object generation means for generating adhesion objects that indicate positions of the first object selected by the selection operation and a second object from the plurality of virtual objects; and an adhesion means for adhering the first object and the second object at the positions indicated by the adhesion objects in response to a user's adhesion instruction, wherein the adhesion object generation means changes the positions indicated by the adhesion objects in response to a movement or a change in posture of the selected first object in the game space before the adhesion instruction is given.

[0047] In the above, for ease of understanding, we have described the case where a human inputs a prompt on the screen of the web version of ChatGPT, but in reality, the processing unit 111 calls the API of LLM to operate. Below are simple examples of expressions and examples of API calls.

[0048] FIG. 3 is a diagram showing an example of a Python program when the processing unit 111 makes an API call to the LLM.

[0049] The program function is call_llm_json, which calls the API internally. Enter instructions to LLM in system_prompt. Enter the claim text of the patent to be processed in patent_input.

[0050] By executing the prompt shown in Figure 3 on the LLM, the following linguistic expression data can be obtained from the LLM:

[0051] { ”Claim”: “This patent relates to an information processing system that handles multiple glueable virtual objects placed in a game space. The user selects an object through a selection operation, generates a glue object, and glues it to another object at a specified position. In particular, it has a function that changes the position of the glue object in real time when the selected object moves or changes its posture.” ”novelty”: “The inventive step of this patent is that the position of the adhesion points changes dynamically when an object in the game space moves. This allows the user to manipulate objects in the virtual space with more flexibility and intuitive operation.” }

[0052] In the above example, the language expression text is output as JSON format data including "claim," which is the description of the claim, and "novelty," which is the description of the inventive step.

[0053] [LLM] LLMs can call APIs provided by major cloud providers such as ChatGPT and Gemini, as well as private LLMs installed on-premise. In addition to large-scale language models (LLMs), small-scale LLMs, or language models (LMs), can also be used. Alternatively, language models scaled down to a level that can be run on a local PC or smartphone can also be used. When using a language model that runs on a local device, programming languages ​​used for mobile apps, such as Java, C++, Kotlin, Rust, Objective-C, and Swift, may be used instead of Python code. However, the input / output information and concepts required to use an LLM are not significantly different from the Python code shown as an example, and can be implemented as appropriate by those skilled in the art.

[0054] When using language models such as LLMs, rather than the deterministic language processing that existed before the advent of language models, their behavior is probabilistic, meaning that the output sentence varies randomly even when the same input and instructions are given. Therefore, it is possible to generate multiple different texts expressed in a language from a single input and a single LLM. Furthermore, LLMs are developed by many companies, and responses vary depending on the model. Therefore, by giving the same instructions to multiple models and obtaining their respective responses, it is possible to generate multiple different texts expressed in a language.

[0055] The processing unit 111 may display the output of a local LLM model for a specific company. To avoid making the user wait, the processing unit 111 may input only claim 1 in the claim text into the LLM and promptly display the generated results, thereby shortening the processing time. In parallel with this, the processing unit 111 instructs the LLM to generate linguistic features based on the content of claims 1 and 2 in the claim text and displays the response on the screen. Similarly, the processing unit 111 instructs the LLM to generate linguistic features based on the content of claims 1 to 3 in the claim text and displays the response on the response screen. By providing information including claims 2 and 3 in the claim text to create a description, an appropriate description can be obtained even when sufficient features cannot be obtained from the description of claim 1 alone. Furthermore, by calling the LLM in parallel, the user's waiting time can be shortened.

[0056] FIG. 4 is a conceptual diagram illustrating an example of a selection screen for linguistic expression data.

[0057] The selection screen is displayed on the display device 24 of the user terminal 2. The original claims are displayed vertically on the display screen displayed by the display device 24. To the right of the claims, linguistic expression data generated based on each claim is displayed. Note that the linguistic expression data may be referred to as descriptions. For example, to the right of claim 1, description A of claim 1, description B of claim 1, and description C of claim 1, which were generated based on claim 1, are displayed. If there are so many candidate descriptions that they cannot all fit on the display screen of the display device 24, the descriptions may be displayed in a line using a method such as a carousel display.

[0058] In the above, it has been described that LLM is used to generate linguistic expression data. However, the processing unit 111 may generate linguistic expression data using a method for extracting feature words, which are important words in a document, such as TF-IDF or BM25. For example, the top five important words may be used as feature words. These feature words may be generated as linguistic expression data. Since feature word extraction using TF-IDF or the like allows for high-speed processing, feature words, i.e., linguistic expression data, may be generated for the entire text of the specification, not just the claim text.

[0059] The above example shows the generation of one or more pieces of linguistic expression data based on the claim text, but one or more pieces of linguistic expression data may also be generated based on the text of the specification. In this case, in step St102, the processing unit 111 acquires text information of the specification from the patent data. In step St103, the processing unit 111 generates one or more pieces of linguistic expression data based on the text information of the specification.

[0060] For example, the processing unit 111 issues instructions to the language model as a set of the specification text and an instruction command to the LLM, that is, a prompt, as shown below.

[0061] Instructions: From this patent specification, please extract the patent claims, technical details, novelty and inventive step of the invention, specific functions and uses, etc., and print them out in a short, concise, and impressive format (approximately 2-4 lines). Information about the inventor and applicant does not need to be included. <<Patent specification>>

[0062] Note that "<<Patent specification>>" refers to the content of the data of the patent specification that has already been obtained. This notation is also used in other parts of this specification.

[0063] As a result of the above instruction, the processing unit 111 acquires the following linguistic expression data:

[0064] response: This patent provides an information processing system for combining multiple objects in a virtual game space. Users can create their own unique game environment by selecting and gluing objects together. This technology, through its novelty and advancement, enhances the creativity and interactivity of the game experience.

[0065] [Generating visual expression data from linguistic expression data] As described above, the processing unit 111 generates one or more pieces of visual expression data based on the linguistic expression data (St104).

[0066] In this embodiment, the generation of the visual representation data includes the following two steps. First step: A step of creating a motif for generating visual expression data based on linguistic expression data (St104-1). Second step: generating visual representation data based on the motif (St104-2).

[0067] [Creating motifs from linguistic expression data (St104-1)] The processing unit 111 receives linguistic expression data as input and instructs the LLM to generate an image. However, if all linguistic expression data is input and an instruction to generate an image is given, it is likely that the resulting image will contain too much information, i.e., a picture with many visual elements that makes it difficult to understand the message without looking at multiple elements. In order to generate simple visual expression data that easily conveys the concept of the patent to humans, the processing unit 111 instructs the LLM on the direction of drawing. The direction of drawing is specified by feature word extraction such as TF-IDF and language processing such as LLM.

[0068] (1) Creation of a directly expressed motif A directly expressed motif means a concrete motif or a literal motif. To create a directly expressed motif, the processing unit 111 inputs a prompt such as the following to the LLM:

[0069] The prompt reads: Below is the patent description. Please describe the features of this patent in one word within the scope of the expression given in the specification. <<Language Expression Data>>

[0070] Note that "linguistic expression data" refers to the content of the acquired linguistic expression data. This notation is also used in other parts of this specification.

[0071] Because language models work probabilistically, feeding the same prompt into an LLM multiple times can result in multiple different responses, such as: Response (characteristic word): Adhesion Response (characteristic word): Dynamic adhesion Response (characteristic word): Dynamic changes in adhesion points

[0072] Alternatively, the processing unit 111 can extract multiple candidates with one LLM call as follows.

[0073] The prompt reads: Below is a patent description. Please describe the features of this patent in one word, within the scope of the expressions described in the specification. Please provide five candidates. Parse this in JSON format. The JSON keys should be 1, 2, 3, 4, 5. <<Language Expression Data>>

[0074] Response (characteristic words): { “1”: “Adhesion”, “2”: “Ease of use”, "3": "Intuitive", “4”: “Innovative”, "5": "Interactive" }

[0075] The processing unit 111 may extract characteristic words by a characteristic word extraction process such as TF-IDF.

[0076] (2) Creation of figurative motifs Figurative motifs are also called metaphors. The processing unit 111 inputs the characteristic words created as described above to the LLM and instructs it to think of a metaphorical motif. In this example, we will explain the case where "adhesion" is created as a characteristic word.

[0077] The prompt reads: Below is the patent description. Please list three animals, living things, and tools that you associate with the word "adhesion," which succinctly describes this patent. <<Language Expression Data>>

[0078] Response (motif): Octopus, chameleon, bee, clay, magnet, adhesive tape, stapler,...

[0079] Alternatively, the processing unit 111 may instruct the LLM to output more types of metaphorical motifs as follows.

[0080] The prompt reads: Below is the patent description. Please list three of the following that are associated with the features, methods, effects, and focus of this patent: animals, creatures, tools, natural phenomena, celestial bodies, historical figures or events from over 100 years ago, seasons, buildings, and colors and shapes. Please parse this in JSON format. The JSON keys are: animals, creatures, tools, phenomenon, celestial, historical, seasons, buildings, colors_shapes. <<Language Expression Data>>

[0081] Response (Metaphor): { "animals":["octopus", "spider", "chameleon"], "creatures":["sticky slime creatures", "telekinetic beings"], "tools":["multi-tool", "glue gun", "3D printer"], "phenomena":["magnetic force", "gravity", "capillary action"], "celestial":["binary star system", "black hole", "nebula"], ”historical”:[”Leonardo da Vinci”,”Isaac Newton”], "seasons":["spring", "autumn", "winter"], "buildings":["modular home", "skyscraper", "castle"], "colors_shapes":["chameleon colors", "geometric tiling", "prism"] }

[0082] Next, step St104 will be described. The processing unit 111 generates one or more pieces of visual expression data based on the linguistic expression data (St104). The visual expression data is typically a still image or a moving image.

[0083] The processing unit 111 generates visual expression data by inputting the following prompt to the LLM, for example:

[0084] The prompt reads: Below is the patent description. Please briefly draw a picture (3D) that represents this patent using an octopus as the motif. <<Language Expression Data>>

[0085] The prompt includes a statement that the visual expression data is to be generated based on the linguistic expression data by specifying the metaphor determined as described above. Figure 5 is a diagram illustrating an example of visual expression data generated in response to the input of the prompt.

[0086] In step St104, the processing unit 111 may generate multiple visual expression data based on the multiple acquired motifs. In the above example, the processing unit 111 specifies a spider as a metaphor after an octopus, and generates visual expression data of the spider. Other metaphors are also specified in sequence, and corresponding visual expression data is generated.

[0087] If multiple visual expression data have been generated, the process proceeds to step St107, where the processing unit 111 displays the multiple visual expression data on the display device 24. The user selects an appropriate one from the displayed visual expression data (steps St108 and St109). Figure 6 is a conceptual diagram illustrating an example of multiple visual expression data displayed on the user terminal 2.

[0088] The processing unit 111 and the display device 24 may have a UI for instructing the addition of a motif or for overwriting a motif. For example, the remaining motifs extracted by the LLM for which visual representation data has not been generated may be displayed on the display device 24, and the user may select a motif to apply from the displayed additional motifs. Alternatively, feature words extracted from the claim text or the text of the specification using an existing feature extraction method such as TF-IDF may be displayed on the display device 24, allowing the user to select an additional motif.

[0089] In the above example, a figurative motif was created from a directly expressed motif, and then the figurative motif was included in the LLM instructions. However, one or more directly expressed motifs and / or one or more figurative motifs may be entered into the LLM together.

[0090] As the AI ​​model 123, a language model compatible with image generation is used. However, it is not limited to this. For example, a model compatible with multiple modalities such as language, image, and voice may be used. A multimodal model such as Google Gemini may be used. A model specialized for image generation such as StableDiffusion may be used.

[0091] Specifying certain rules and providing example patterns will create a sense of unity in the generated images. For this reason, the processing unit 111 may instruct the LLM to generate images based on, for example, the following example patterns. The example patterns may be stored in advance in the storage device 12 in the form of table data or the like.

[0092] Fig. 7 is a conceptual diagram showing an example of a pattern of visual expression with an animal motif, and Fig. 8 is a conceptual diagram showing an example of a pattern of visual expression with an anthropomorphic character.

[0093] An example in which the processing unit 111 instructs the LLM of a pattern example is shown below: The processing unit 111 acquires a pattern example from table data stored in the storage device 12, and instructs the LLM using the pattern.

[0094] The prompt reads: Below is the patent description: Using the technical field and example visual representation patterns as a reference, please briefly draw a picture (3D) that represents this patent. <<The above-mentioned language expression (omitted)>> #Examples of technical fields and visual expression patterns Communication Technology The falcon's futuristic wings and advanced communication equipment attached to its ears represent the advancement of communication technology, which emphasizes speed and efficiency. The deer, representing harmony with nature, is depicted as a symbol of renewable energy and environmental protection. The deer incorporates green motifs and symbols of solar and wind power generation to represent earth-friendly technology.

[0095] The processing unit 111 may supplement the technical information by providing a definition of a classification code, such as an IPC classification or F-term, assigned to the patent. For example, for a patent with an IPC classification of "A63F 13 / 5372," the processing unit 111 adds the definition of the code, "tags a character, object, or location in a game scene, e.g., displays a circle under a character controlled by a player," to the LLM instructions.

[0096] The prompt reads: Below is the patent description: Using the technical field, examples of visual representation patterns, and technical code explanations as a reference, please briefly draw a picture (3D) that represents this patent. <<The above-mentioned language expression (omitted)>> #Examples of technical fields and visual expression patterns Communication Technology The falcon's futuristic wings and advanced communication equipment attached to its ears represent the advancement of communication technology, which emphasizes speed and efficiency. The deer, representing harmony with nature, is depicted as a symbol of renewable energy and environmental protection. The deer incorporates green motifs and symbols of solar and wind power generation to represent earth-friendly technology. #Description of the technical field code of this patent Tagging characters, objects, or locations in a game scene, e.g., displaying a circle under a player-controlled character

[0097] [Gain and Pain Extraction from Patent Specifications] Patent specifications are written in a unique way, making it difficult for those unfamiliar with the language to understand the pains that a patent is intended to solve or the gains that a patent is intended to provide. For example, Japanese patent specifications include "effects of the invention" and "problems that the invention is intended to solve" as descriptions similar to gain pains, but these descriptions do not closely match the gain pains found in business analysis frameworks such as the Value Proposition Canvas. Furthermore, because "effects of the invention" and "problems that the invention is intended to solve" are not included in U.S. specifications, when using a Japanese application as the basic application, it is sometimes hesitant in the U.S. and elsewhere to provide detailed information in these sections.

[0098] Here, we will explain how to extract gains and pains from the entire specification. Gains and pains are types of linguistic expressions.

[0099] The processing unit 111 can extract patent gains by giving the following instructions to the LLM:

[0100] The prompt reads: From this patent specification, please extract the effects (gain, benefit) that the patent is intended to provide and the background, and output it in a short, concise, and impactful manner (about 2 to 4 lines). You do not need to include information about the inventor or applicant. You do not need to include the method of the invention. <<Patent specification>>

[0101] response: This patent aims to enhance user immersion and creativity in video games, allowing players to freely and intuitively manipulate and combine multiple objects within the game. This technology is intended to significantly improve the richness and interactivity of the gaming experience.

[0102] The processing unit 111 can extract the panes of the patent by giving the following instructions to the LLM:

[0103] The prompt reads: From this patent specification, please extract the problem (issue / pain) that the patent is trying to solve and the background to it, and output it in a short, concise, and impactful manner (about 2-4 lines). You do not need to include information about the inventor or applicant. You do not need to include the purpose or method of the invention. <<Patent specification>>

[0104] response: This patent addresses the problem of search accuracy when different terms in document information refer to the same technical element. Patent documents cover the latest technologies and are often created before terminology was standardized, meaning that applicants may use different terms. Furthermore, as terminology changes over time, the same technical element may be expressed using different terms. This reduces the accuracy of machine searches and makes search time and effort intensive, even for experienced searchers.

[0105] [Gain and / or Pain Judgment] There are two types of patents and solutions: problem-solving (those that reduce pain, problems, and discomfort) and value-creating (those that increase gain and benefits). After determining which type it is, displaying only the explanatory information for that type to the user reduces the user's effort in reading. The processing unit 111 can determine which type it is by giving the following instructions to the LLM.

[0106] The prompt reads: Is this patent a Pain Relief Solution (reducing problems, discomforts) or a Gain Creation Solution (creating additional value, benefits, or positive experiences)? answer: pain-relief or gain-creation Please answer in JSON format. The JSON key should be "answer". <<Patent specification>>

[0107] { “answer”:”pain-relief” }

[0108] The flow for determining whether the patent is a gain or a pain is as follows: The processing unit 111 acquires patent data. The processing unit 111 determines whether the patent is a Pain Relief Solution (relieving problems and discomforts) or a Gain Creation Solution (creating additional value, benefits, or positive experiences). For this determination, the processing unit 111 acquires an answer obtained by issuing the above prompt to the LLM, and determines whether the answer is "pain-relief" or "gain-creation." If the determination result is "pain-relief," the processing unit outputs pain information. If the determination result is "gain-creation," the processing unit 111 outputs gain information.

[0109] [Support for spelling variations, synonyms, and similar words] Generally, words have variations in spelling (e.g., katakana, hiragana, and kanji in Japanese), resulting in the existence of synonyms and similar words that have the same or similar meaning but different character strings. Feature word extraction methods such as TF-IDF and BM25 are capable of faster processing than large-scale language models (LLMs). However, these methods are expected to improve accuracy by appropriately processing variations and synonyms. Variations in spelling can often be absorbed by pattern replacement processing. Variations in synonyms can be addressed by using machine learning models that use vector representations of words, such as WORD2VEC. Similar methods, such as BERT and RoBERTa, are also effective for vector representation. These methods enable vector representation that takes context into account, and are expected to absorb synonyms more effectively. For example, in the present disclosure, it is possible to appropriately use a process in which words with very similar vectors are replaced with one of the other words.

[0110] Furthermore, when replacing multiple words with similar meanings, it is important to select words that correspond to the user's technical vocabulary. For example, simple words should be selected for junior high or high school students, and technical terms should be avoided for non-engineers. To achieve this, the system can be configured to input user attribute information (e.g., student, engineer, accounting department, etc.) and automatically select words that correspond to that information.

[0111] Fig. 9 is a conceptual diagram showing an example of the configuration of an intellectual property business support device. Fig. 10 is a conceptual diagram showing an example of the configuration of an intellectual property business support device.

[0112] Figure 9 shows a configuration in which the modules are divided into a language model that accepts prompt input and generates linguistic expressions, a module that checks and processes the linguistic expressions, and an image generation model that accepts linguistic expression input and generates visual expressions.

[0113] Figure 10 shows a configuration in which a single multimodal model performs a series of processes from receiving a prompt input to generating a visual representation.

[0114] [RPG-style use: RPG-style world map display] FIG. 11 is a flowchart illustrating an example of an intellectual property business support method.

[0115] The processing unit 111 acquires patent data (St201). The processing unit 111 acquires a characteristic expression of the patent data (St202). The characteristic expression means the above-mentioned linguistic expression or visual expression.

[0116] The processing unit 111 processes the visual representation of the patent (Step 203). The processing unit 111 places a patent object in the map (Step 204). The patent object means a virtual object corresponding to a feature related to the patent. The processing unit 111 provides an operation means to the user (Step 205).

[0117] Steps St201 and St202 will now be described. As described above, the processing unit 111 creates linguistic and visual representations based on patent information such as claims, invention abstracts, problems, solutions, and inventor information. The linguistic and visual representations are created based on language processing such as LLM and TF-IDF.

[0118] [Processing visual representation of patents] Steps St203 and St204 are described below. If the map is a 3D world and the visual representation is a two-dimensional card, the visual representation may be displayed as is, or it may be converted using a generation AI that converts 2D illustrations into 3D models before displaying. If a patent has entered a status unfavorable to the applicant, such as deemed withdrawal, the visual representation may be displayed in a special way. For example, if the visual representation is an animal or living creature, the processing unit 111 may process the image of the animal or living creature to make it appear dead. If the visual representation is a tool or object, the processing unit 111 may process the image of the tool or object to make it appear broken. These processing steps can be generated by instructing the LLM or image generation AI. Conversely, if the patent expires after 20 years, after having made a sufficient contribution to the industry or the owning company, the visual representation may be processed into an image of a magnificent bronze or stone statue. Whether a sufficient contribution has been made can be determined, for example, by a) having received patent annuities for 20 years, b) having received an invention award from the Japan Patent Office, c) having received licensing income from other companies, or being used in the company's own products or services, or d) having been incorporated into a well-known service or product.The processing unit 111 can obtain information by using the Japan Patent Office's file wrapper API for a) or by scraping the Japan Patent Office's website for b), by linking with the systems and databases of the intellectual property departments of the corporations using this system for c), and by referring to articles on news sites and blog sites that comment on patents for d).

[0119] [Placing patent objects in the map] The processing unit 111 places each patent object on the map based on the above linguistic expressions as well as features such as citation information at the time of requesting examination, applicants, and technology classification symbols such as IPC classifications and F-terms. A patent object is a virtual object that corresponds to a feature related to a patent. Patents are generally displayed individually, not in groups. However, multiple patents with roughly the same inventor and roughly the same invention details, such as divisional applications, may be displayed together as a single object. The spatial location of patents can be somewhat random, as long as they are placed near patents with similar features. A method similar to that used for conventional bird's-eye view patent landscapes is acceptable.

[0120] The processing unit 111 may arrange patent objects on the map that are likely to interest the user closer to the user and those that are likely to interest the user less far away, based on the user's search history, application history, recent business interests, etc. The processing unit 111 may arrange patent objects that are likely to interest the user closer to the user and those that are likely to interest the user less far away, based on the average or typical interests of multiple users belonging to the organization to which the user belongs, such as the search history, application history, recent business interests, etc. The organization refers to, for example, a corporation, department, or organization. The processing unit 111 may calculate a value indicating interest based on the above information.

[0121] [Providing means of operation] The user can use the input device 23 of the user terminal 2 to operate the map and a player character representing the user. The user can freely explore the map via the player character and focus on specific visual elements. When the player character approaches a specific patent object or when the mouse pointer approaches the object, a linguistic description of the patent object is displayed in a pop-up. Alternatively, the linguistic description may be read aloud.

[0122] [Approach Action] When a patent object is selected by the player character touching it, selecting it with the mouse pointer, or touching the touch panel screen, detailed information is displayed. This detailed information includes the application date, registration date, applicant name, inventor name, and the continuation of the claims. When the player character performs an action based on user input (throwing an object, casting a spell, shooting a bow and arrow, etc.), the patent object reacts in a way that matches its visual or linguistic expression. The linguistic features are summary-based, and if the patent object is a living or tool-like object, it may be displayed as speech. If the patent object is a structure, text corresponding to the linguistic features may be written on an explanatory plaque and automatically read aloud.

[0123] [reaction] The user can respond to each displayed patent object in various ways, such as "Good," "Bad," "Interesting," "Threatening," "Unclear," "Difficult," and "Boring." These responses are made via the input device of the user terminal. The processing unit 111 records these responses for each information patent object in the storage device 12. The responses are reflected in the appearance of the patent object.

[0124] For example, if the user responds favorably or appreciatively, an animated video showing the patent object being happy is displayed, whereas if the user responds critically or indifferently, an animated video showing the patent object being disappointed is displayed.

[0125] [A performance showing respect] The system will also include a feature that allows users to express professional respect for outstanding patents, regardless of whether they are from their own company or another company. This is expressed through words and icons such as "respect," "respectful," and "applause." Player characters can animate themselves applauding, saluting, or presenting trophies or medals to patent objects. For a more fantasy-like effect, players can also present angel rings, wings, or stars flying around. This will make outstanding patents easier to identify at a glance, and is expected to help engineers search for prior patents and engineers and entrepreneurs with little patent filing experience find patents they should refer to. The operation of services and software that incorporate these expressions will contribute to the creation of excellent patents both domestically and internationally.

[0126] [Automatically extract interests from behavioral history] The processing unit 111 may automatically extract the user's level of interest from the user's behavioral history for individual patents, such as the frequency with which the user character approaches or converses with patent objects, browsing history and download history in the patent search system, and citation history from patents filed by the user himself, and may reflect this in the reaction of the patent object. FIG. 12 is a diagram illustrating an example of the reaction of a patent object. The reaction of the patent object may be stored in advance in the storage device 12 in the form of table information that pairs conditions with the contents of the reaction. The processing unit 111 refers to the table information and causes the patent object to react according to the conditions.

[0127] [Strong operation] Users can perform operations and change the status of individual patent objects. This visually expresses the user's feelings and interest toward the patent (character), and helps reflect their personal perception and evaluation of the patent. Figure 13 shows an example of strong operations on patent objects.

[0128] [External system integration] The user controls a player character and moves through an RPG-style world map. Operation information, including user intentions regarding patents, such as reactions, respect, and strong operations performed on patent objects through user operations, may be exported to an external system. The processing unit 111 exports the operation information to a patent search system or a patent writing support system, allowing these external systems to use the investigation information as reference information for searches and writing.

[0129] The logical structure of the export data may be a description of operation information, including user intent information for each patent number, as well as the linguistic and visual representations of the patent. The export data may be in any format commonly used in IT systems, such as JSON or XML. Since the export data must be protected from the perspective of security and personal information, both this system and the export destination system must have appropriate authentication and identity verification functions.

[0130] FIG. 14 is a conceptual diagram showing an example of the configuration of the intellectual property business support device 1. As shown in FIG.

[0131] The basic configuration is the same as that shown in Figure 1. In Figure 1, various data and AI models stored in the storage device 12 are shown divided into a personal data area and a public data area. The personal data area stores patent object expression DBs, map data, user history information, etc. The public data area stores patent databases, patent object expression DBs, etc.

[0132] Additionally, if a partner patent search system or patent writing support system provides information on viewing and labeling of each patent, it can be imported into this system in a similar format, allowing the labeling information to be displayed on an RPG-style map. Furthermore, since the visual representation of a patent may contain a large amount of data, it is also possible to include URL information indicating the location where the labeling information is stored.

[0133] [Processing after time has passed, such as re-login] A user explores the map through a player character, and after completing marking and studying, logs out of the intellectual property business support system 100. Suppose the user logs back in to the intellectual property business support system 100 a week or several months later. During the period between logging out and logging back in, changes occur, such as a patent similar to the patent of interest in the map being newly published or expiring. The processing unit 111 may automatically reflect these changes in the patent objects in the map. Patent objects that appear in the map after publication or other events may be animated as if they were being born. For example, they may appear from an egg or jump out of the ground or a lake. Patents that have been filed for divisional applications may be displayed with an animation that resembles a doppelganger. If the strength of a patent is recognized during the logoff period, for example, through a new examiner's citation, the patent object may be enlarged or animated as if it has been powered up.

[0134] [Rewind time, time lapse] An embodiment of the present disclosure provides a time bar for visualizing the creation and disappearance of patents, as well as their rise and fall, over time. When a user operates this time bar, the status of all patent objects displayed on the map changes in tandem. Users can deeply experience the patent filing and strengthening process of industries and competitors in a specific field, just like they would in the real world. This allows users to experience a high level of immersion beyond simply viewing a PDF file. As a result, users are more motivated to create better patents.

[0135] [When the user specifies a single point in time] When the user selects a specific time on the time bar, the processing unit 111 extracts from the patent database and displays patents that were published or registered at the specific time specified by the user (e.g., April 2010). Patents that have been applied for but not yet published at the specified time can also be displayed. A phrase such as "pre-publication" is added to the patent object. In addition, as a visual effect for pre-publication patent objects, they are displayed in a semi-transparent state or dim color, and a lock or key icon is placed nearby.

[0136] [When the user specifies a period] When a user specifies the start and end times of a period on the time bar (e.g., from January 2000 to the end of 2019), the processing unit 111 divides the time range into sections with a resolution of approximately 1 / 10 to 1 / 100, extracts patent information from the patent database for each divided time, and displays patent objects. Like a looped video replay, the strength and status of patents change at each sampling time. Displaying patent objects according to their strength and status makes it easier to understand the industrial contribution of individual patents. The processing unit 111 may also calculate a value indicating the strength of a patent based on, for example, examiner citations, commercial use history, and accumulated professional respect from others. The status refers to a state based on patent application, expiration, or other factors.

[0137] In addition to video viewing, the system may also have a mode in which the user controls a player character and repeatedly changes the time period from the past three or five years to the present. Conventional patent search systems generally provide instantaneous information at the time of the search, but are not suited to clearly extracting patents that have suddenly risen in popularity in the past few years. By automatically displaying patent objects while changing their visual representation without requiring user operation, as in the present disclosure, users can intuitively grasp rapidly rising patents.

[0138] [Landscape-like patent visual representation, examples of landscape objects] In the explanation of this disclosure, examples have been given in which patent objects are visually represented in the form of animals, living things, tools, etc., but they may also be represented in landscapes such as castles and airships. This disclosure is based on expressing the characteristics of each patent, but landscape-like representations are particularly effective when visually representing patents in a certain group or aggregate unit. This is because castles and forests are generally large. Figure 15 is a diagram illustrating an example of a landscape-like visual representation.

[0139] Rather than representing the patent object itself as a landscape, it is possible to place landscape objects such as large structures or terrain within the map, and then place patents that are close to the image conveyed by the landscape objects as patent objects on or within those landscape objects.

[0140] [Patent search improvements] [Explain each patent individually or explain the differences between multiple patents] When multiple patents are found in a patent search, or when similar patents are filed at the same time by other companies, or when similar patents are filed as divisional applications, it can be a lot of work to check the contents of these multiple patents.This disclosure provides a method for checking the contents and differences of patents using LLM.

[0141] FIG. 16 is a flowchart illustrating an example of an intellectual property business support method.

[0142] The processing unit 111 acquires a plurality of patent data (St301). The processing unit 111 determines characteristic words related to the patents from the acquired patent data (St302). The processing unit 111 determines a theme for explaining the patent (St303). The theme can also be expressed as a viewpoint.

[0143] The processing unit outputs the features of the patents based on the determined theme (St304), or performs a comparative analysis based on the determined viewpoint (St305).

[0144] Step St301 will now be described. Assume a situation in which a keyword, key phrase, or patent description text is entered into a patent document search system, and multiple patent documents are returned. Alternatively, in the aforementioned RPG-style patent display system, multiple patents located near each other or near the player character may be returned. In these cases, for example, the processing unit 111 acquires patent data for multiple patents.

[0145] Step St302 will now be described. Whether performing a keyword search, keyphrase search, or document search, the processing unit 111 can extract representative words from the search query string using feature word extraction, such as the aforementioned TF-IDF. Alternatively, the processing unit 111 may use a language model to extract representative words and feature words. In TF-IDF, it is important to appropriately specify the denominator of the target patent when calculating the IDF. For example, the processing unit 111 may collect at least one classification symbol, such as an IPC classification or F-term, assigned to each patent document, extract patents indicated by these classification symbols from a patent database, and use this as the denominator for the TF-IDF. Alternatively, because setting an appropriate denominator each time is costly, the processing unit 111 may calculate the IDF for each section. The "section" here refers to a patent classification section, such as "Section A: Daily Necessities" or "Section H: Electrical."

[0146] Step St303 will now be described. Feature word extraction methods such as TF-IDF can calculate the importance of words. The importance of a word can also be expressed as the rarity of its occurrence frequency. Therefore, the processing unit 111 selects a predetermined number of feature words with high importance from among multiple feature words that indicate multiple perspectives for describing the patent. Alternatively, the processing unit 111 displays multiple feature words on the display device 24 and accepts user input regarding which feature word to select. Alternatively, the user may directly input keywords or phrases indicating a perspective via the input device 23. Perspectives involving comparison, such as "the one with the lowest cost," can also be input.

[0147] Step St304 will now be described. For example, let us consider a case where three patent documents are found as a result of a patent search relating to "refrigerator" and "magnet." The viewpoint determined in step St303 is assumed to be "magnet." In this case, in step St304, the processing unit 111 gives the following instructions to the LLM for each of the three patent documents.

[0148] The prompt reads: Please summarize the following document (patent claim) focusing on "magnets." Please explain in a series of sentences of about two to three lines, without using bullet points or breaking down into steps. Parse it in JSON format. The JSON key should be description. <<Claims of Document 1>>

[0149] Based on the above prompts, the LLM might respond with something like this:

[0150] response: { "description": "The structure related to the rotating partition of the refrigerator is described in detail, but a "magnet" is only mentioned in claim 6. This claim describes a refrigerator design in which a magnet is installed on the inner surface of the outer casing member on the outside air side of the rotating partition, and the magnet surface is protected by covering it with tape." }

[0151] In step St304, the processing unit 111 combines the responses to the LLM instructions executed for the three patent documents as described above, for example, and outputs the combined responses to the user in a table format as shown in Fig. 17. Fig. 17 is a diagram showing an example of output to the user.

[0152] Step St305 will now be described. For example, in the case of a viewpoint such as "the one with the lowest cost," better results can be expected by simultaneously providing multiple patent documents to the LLM. In such a case, the processing unit 111 instructs the LLM as follows.

[0153] The prompt reads: Compare the following three documents (patent claims) and explain the differences in about two lines. Focus especially on the "lowest cost" option. Parse it in JSON format. The JSON keys should be doc1, doc2, and doc3. #Reference 1 <<Claims of Document 1>> #Reference 2 <<Claims of Document 2>> #Reference 3 <<Claims of Document 3>>

[0154] Based on the above prompts, the LLM might respond with something like this:

[0155] response: { ”doc1”: “This patent relates to the structure of a refrigerator, specifically a refrigerator including a double door, a rotating partition, foam insulation, a reinforcing plate, an injection hole and an air vent hole. The focus of this patent is to improve the insulation performance and structural strength, and the cost may be somewhat higher due to the use of insulation and reinforcing plates.” ”doc2”: “This is a patent for a voice interaction device for refrigerators. It includes a wireless communication unit, a microphone, a speaker, and a fixed part that can be attached to the surface or inside of the door. Due to the addition of the voice interaction function, the cost is likely to be higher than in Reference 1. Implementing the voice interaction function will incur additional costs in both hardware and software.” "doc3": "This is a patent for a magnetic sheet that displays emergency contact information. This patent relates to a magnetic sheet with an adhesive back that displays emergency contact information and recovery worker information printed in luminous characters. The product in this patent has a relatively simple structure and is made from low-cost materials, so it is thought to be the most cost-effective compared to references 1 and 2." }

[0156] The processing unit 111 outputs the result to the user in a table format, for example, as shown in Fig. 18. Fig. 18 is a diagram showing an example of output to the user.

[0157] [Local pronoun substitution in LLM output] The LLM response for Patent No. 6252XXX states, "...the product of this patent is considered to have the lowest cost compared to References 1 and 2 because it has a relatively simple structure and is constructed from low-cost materials." This description of "References 1 and 2" is a local expression between the processing unit 111 and the LLM, and if it were output as is, the user may not be able to understand the content. Therefore, the processing unit 111 uses pattern matching and string replacement processing to replace "References 1 and 2" with "Patent No. 6937XXX and Patent No. 7220XXX." There are several other possible variations of the local expression in the LLM response, such as "References 1 and 2," "References 1 and 2," and "References 1 and 2." By listing replacement patterns in advance and defining them as replacement targets, replacement can be performed without omission.

[0158] [Controlling the scope of claims to LLM] Compared to feature word extraction methods such as TF-IDF, language model processing requires a large amount of memory and is slow. It is economical to limit the length of claims to an appropriate length when entering them into the LLM. In the method of the present invention, the processing unit 111 inputs only claims that contain the aforementioned "feature words related to the above patent" and "viewpoint words" into the LLM. If a dependent claim contains a target word, the dependent claim of that claim is also input into the LLM.

[0159] For example, when the only feature word or viewpoint word is "magnet," magnet does not appear in claims 1 or 2 but does appear in claim 3, claim 3 is dependent on claim 2, and claim 2 is dependent on claim 1, processing unit 111 inputs claims 1 to 3 into the LLM.

[0160] [Add description to file name when downloading file] The patent search system has a function to download patent publication PDFs, but the file name often contains only the patent number, which makes it inconvenient to have to open the file in a viewer each time to check the contents. For example, the current file name is in the format "JPB006937XXX-000000.pdf".

[0161] In this disclosure, the language expression may be included as part of the file name so that it can be downloaded. For example, the file name may be something like "JPB006937XXX-000000-Refrigerator structure. In particular, refrigerators including double doors, rotating partitions, foam insulation, reinforcing plates, injection holes, and air vent holes.pdf." This makes it easy to identify the required patent after downloading the PDF.

[0162] Furthermore, the file name may include information about the user's intention regarding the patent described in this disclosure, such as "JPB006937XXX-000000-[Respected]-Refrigerator structure. In particular, refrigerator including double doors, rotating partition, foam insulation, reinforcing plate, injection hole and air vent hole..pdf."

[0163] If the reference is a citation in a notice of decision of refusal or a patent specification, the file name may include information such as citation 1, citation 2, etc., depending on the order in which the list is displayed. The information for the list of citations may be obtained from the patent office's file wrapper API, or it may be extracted from the text information in the notice of decision of refusal or the patent specification using language processing.

[0164] [Improvement of search result UI (Hits / Non-hits, visual representation)] FIG. 19 is a conceptual diagram illustrating a search result user interface (UI) of the present disclosure. The patent search software displays the search formula and the number of hits. To adjust the number of search hits, the search formula and the number of hits are displayed in a table format, one row at a time. Although this differs depending on the person and the situation, the user can stop adjusting the search formula once they have reduced the number of hits to a number that can be visually checked and that includes patents that they do not want to miss.

[0165] In the method of the present disclosure, not only the number of search hits but also a visual representation of the aforementioned patents is displayed in the Hits column, as shown below. Also, not only the number of non-hits but also a visual representation of the aforementioned patents is displayed in the Non-hits column. Note that instead of pairs of Hits and Non-hits, pairs of Retrieved and Non-retrieved, or Hits and Non-hits, etc. may be displayed.

[0166] During the process of tuning the search formula, if the user visually checks some of the patents and identifies patents that should not be overlooked, or if the user knows from prior knowledge that patents should not be overlooked, the user can mark those patents with an icon such as a star. This star is not intended to indicate a "particularly excellent patent" as described elsewhere in this specification, but is an icon to make the patent stand out when tuning the search formula. To avoid confusion with the intention of the entire system, it may be represented by an icon such as a caution (eye), an exclamation point (!), a pin, or a paperclip. Here, we will explain the case of the star icon as an example.

[0167] We want to tune the system so that patents of interest are included in the search hits as much as possible (Hits), so in addition to visually representing patents with a ★ icon, we also display the number of hits for patents of interest next to the ★. We do not distinguish between patents of interest and other patents, and simply display the number of hits next to the Σ, just as before. Other patents include unrelated patents (irrelevant patents) that are in completely different technical or product fields from the one we are focusing on, or patents whose contents we have not looked at and therefore are unknown.

[0168] For example, let's say a user conducting patent research follows the steps below. On their first search, the user enters a few keywords and finds 1,721 patent documents in the patent database. The user then skims through some of the patent documents and finds one patent that is noteworthy (one that they don't want to miss), so they mark that patent document with a noteworthy ★ icon.

[0169] The user then added the keyword "fixed to surface" to reduce the number of hits. When the user searched the patent database again for a second time, he or she intuitively understood that the patents with the star icon had been moved to the "Non-hits" column, because the visual representation of the patent document had changed. Furthermore, the number of star hits showed that the "Hits" column had 0 hits, while the "Non-hits" column had 1, so the user could once again recognize from the number of stars that the search criteria were too strict.

[0170] For the third search, the user changed the search keyword expression from "fixed to surface" to "paste." The number of hits increased slightly, but the user realized that the target patent was now included in the hit range. The total number of hits was only 17, which was relatively small, so the user decided that it would be okay to visually check all of the hits again.

[0171] Furthermore, if you find any noteworthy patents during a visual check of all the patents, you can simply mark the patents with a star.

[0172] For example, as described above, users can mark patents they do not want to miss with a spotlight icon.The user can then easily determine that the patent with the spotlight icon is correctly within the search range based on one or more of the changes in the patent's visual representation, the star icon, and the number of stars displayed.

[0173] In Figure 19, a maximum of three patent visuals are displayed, but the maximum number may be set by the user. Note that the patent visuals referred to here refer to the visual representations generated for a particular patent as described above. Alternatively, all patents of interest may be displayed. The size of each patent visual is not fixed and may be variable. For example, when displaying a large number of patent visuals in the display area, particularly patents of interest, the size of each may be small.

[0174] For patents that do not have patent visual information, a visual representation may be automatically generated in real time by the system of the present disclosure, or may be displayed as an icon representing a general PDF file, etc.

[0175] As shown in the example in Figure 19, when a maximum of three patents are displayed and many hits are found, the featured patents are displayed first. Featured patents can be marked in two levels, such as super featured patents ★★ and regular featured patents ★. If the user distinguishes featured patents in multiple levels, the patent visuals displayed will be those with the highest priority.

[0176] Furthermore, when a large number of patents of interest with the same priority are found, the processing unit 111 displays patents with the highest degree of match to the search query. In the case of a vector search system, a search query is input as text and then vectorized. Documents registered in a patent database are vectorized in advance, and in a vector search system that calculates the degree of match between the vector of the input search query and the vector of a registered document using cosine similarity or the like, the degree of similarity between the search query and each individual patent can be obtained as a numerical value, so the patents can be displayed in descending order of this degree of match.

[0177] In the case of a word search system that allows multiple words to be specified, the similarity of the hit documents can be calculated numerically based on the number of occurrences of the search word contained in the registered documents and the weighting of the importance of the words using TF-IDF, etc., so the documents can be displayed in order of highest similarity.

[0178] [Automatic suggestions for patents to look at] As users experiment with search queries and accumulate markings of patents of interest and irrelevant patents, the system can automatically extract patents that are likely to attract users' attention from other patents by using a machine learning model for document classification. For example, the machine learning model is trained using patent attribute information, text information, and user marking information as training data. Training occurs in real time with each user operation. Next, the probability of all patents being classified as patents of interest is estimated for each search target. The processing unit displays patents with the highest probability (at most a few) by marking them with a question mark (?) or a magnifying glass icon. Users can view the contents of patents proposed by the intellectual property business support system 100 disclosed herein and mark them as interesting ★ or irrelevant × at their own discretion. Each time a user mark is added, the system automatically retrains the model. As training data increases, estimation accuracy improves.

[0179] [Search query input UI] The UI for entering search expressions is similar to that of XLS input. Figure 19 shows the stage after three searches have been completed. When the user wants to perform a fourth search, they would enter the next search query in the row indicated by the arrow in the lower left of Figure 19. In this example, when the cursor for user input moves to the cell next to the arrow, that is, the search expression column, the search expression can be entered.

[0180] FIG. 20 is a conceptual diagram illustrating a search result UI of the present disclosure. FIG. 20 shows the stage after the second search has ended. When the user starts typing where the arrow indicates, the previous search formula may simply be automatically copied. Furthermore, the user may modify characters based on the copied previous search formula. Each time a character is modified, the system may automatically perform a search and update the display in the Hits and Non-Hits fields. Alternatively, a search formula automatically suggested by the intellectual property business support system 100, as described below, may be displayed.

[0181] Instead of directly inputting a system suggestion in the line indicated by the arrow, the processing unit 111 may present multiple search query candidates and display the search execution results, the number of Hits and Non-Hits Σ and ★ in advance. To reduce search waiting time, the suggested search query may be displayed first, and the contents of the Hits and Non-Hits columns may be displayed as soon as the search is completed.

[0182] [UI on the search hit list screen] FIG. 21 is a conceptual diagram illustrating a search result UI of the present disclosure. A search result list screen displayed on the display device 24 allows users to check marks with stars or crosses. FIG. 21 shows an example in which a column of check boxes for adding stars is provided. A column of check boxes for adding crosses may also be added. Alternatively, the UI may allow users to click on a cell to toggle between stars, crosses, and no entry. For patents that have a visual representation of the patent, the visual representation is displayed in the "Visual" column.

[0183] The UI has columns that display the linguistic information extracted by the method of the present invention, and can display Pain only, Gain only, Pain or Gain, Novelty, Inventive Step, Summary, etc. In the example of Figure 21, only Pain is displayed on the far right. The user can toggle between the displayed contents by clicking the column label. Alternatively, multiple columns can be used to display content selected by the user. Of course, the problem to be solved by the invention and the summary, as described in the patent publication, can also be displayed.

[0184] [3 categories: Noteworthy, Garbage, Unknown] Irrelevant patents can also be displayed visually and by counting. For example, irrelevant patents can be marked with a trash can or cross icon. For counting, an X or a downward pointing arrow can be used. Patents that are found to be irrelevant during the process of tuning the search formula can be marked with an irrelevant icon, and the user can tune the search formula so that the number of irrelevant patents is higher in the Non-hits column rather than the Hits column. Reducing the number of irrelevant patents reduces the number of visual checks by the user, saving time.

[0185] 22 is a conceptual diagram showing an example of the configuration of an intellectual property business support device. The UI is implemented in, for example, a user terminal 2. The data area of ​​the storage device 12 stores a patent publication DB that stores information on patent publications, a patent object characteristic expression DB that stores data indicating characteristic expressions of patent objects, and an attention DB that stores information indicating noteworthy points such as stars.

[0186] [Search query modification LLM suggestion] In the example shown in Figure 19, the user changed the expression in the search formula from "fixed to the surface" to "paste." However, the processing unit 111 may instruct the LLM to make suggestions. Since the number of hits decreased too much when conditions were added from the first search (No. 1) to the second search (No. 2), the processing unit 111 instructs the LLM to provide the changed keyword "fixed to the surface" and the content of the target patent and suggest five other keywords or phrases that have a similar meaning to the changed keyword and will result in a search hit for the specified patent. The content of the target patent may be the entire specification or the text of the claims.

[0187] The processing unit 111 then presents the alternative keywords and phrases obtained in response from the LLM to the user. The user simply selects from the presented options, eliminating the need for typing and thinking. Below is an example of an instruction given to the LLM by the processing unit 111. In this example, the LLM is instructed to suggest an alternative query that includes two patents of interest.

[0188] The prompt reads: Please suggest five short words or phrases that have a similar meaning to the search query "fixed to surface" and will return hits for the following patent documents: #Reference 1 <<Claims of Document 1>> #Reference 2 <<Claims of Document 2>>

[0189] [Full LLM suggestions for search queries] This is particularly effective when there are multiple patents of interest, for example, five to ten or more, and the processing unit 111 may also cause the LLM to propose a search query that will result in a search hit for all patents of interest. The following is an example of an instruction from the processing unit 111 to the LLM.

[0190] The prompt reads: Please suggest five words or phrases, as short as possible, that will result in hits for the following patent documents. #Reference 1 <<Claims of Document 1>> #Reference 2 <<Claims of Document 2>> #Reference 3 <<Claims of Document 3>> ...

[0191] [Automatic search query optimization agent] The processing unit 111 may automatically execute a search in the search system using the search query proposed by the LLM, and automatically check whether the target patent of interest is included. If the patent of interest is not included in the search results, the processing unit 111 automatically relaxes the search conditions indicated by the search formula so that the patent of interest is included. If the number of hits for patents other than the patent of interest (the total of irrelevant patents and patents not considered) increases, or if the number of irrelevant patents increases, the processing unit 111 automatically adjusts the search conditions indicated by the search formula to be more restrictive. The processing unit 111 repeats the above process a predetermined number of times or a number of times specified by the user. Alternatively, the processing unit 111 repeats modifying and generating the search formula and executing the search until a target value set by the user is achieved.

[0192] The target value set by the user means, for example, the target hit rate or number of irrelevant patents. The smaller this value, the better.

[0193] The target value set by the user means, for example, the target hit ratio or number of patents other than the patent of interest. The smaller this value, the better.

[0194] The target value set by the user means, for example, the target hit rate or number of patents of interest. The higher this value, the better.

[0195] 23 is a conceptual diagram showing an example of the configuration of the intellectual property business support device 1. The data area of ​​the storage device 12 stores a patent publication DB that accumulates information on patent publications.

[0196] [Respect Network] Conveying the respect and professional respect shown to inventors for outstanding patents can contribute to increased motivation to invent. By making it easy to see which patents have received the most "respect," other users, including those new to patents, can easily find outstanding patents and use them as learning materials. This method creates a positive feedback loop in intellectual property creation.

[0197] 24 is a flowchart illustrating an example of an intellectual property business support method. The processing unit 111 performs user registration (St401). The processing unit 111 registers patent information about the inventor (St402).

[0198] A user logs in to the intellectual property business support system 100 and browses patent information (Step 403). The user may be an inventor or a person who is not an inventor.

[0199] The processing unit 111 acquires information indicating the respectful reactions (St404). The processing unit 111 compiles the information indicating the respectful reactions and displays it on the display device (St405). The processing unit 111 provides feedback to the inventor (St406).

[0200] Steps St401 and St402 will be described. In the intellectual property business support system 100 of the present disclosure, a user first performs identity verification and creates an account. That is, the user performs user registration. User registration requires the submission of personal information. For users who are not inventors and who simply "like" posts, identity verification at the same level as that of general SNS is sufficient. Identity verification at the same level as that of general SNS refers to, for example, verifying the existence of an email address, a phone number, or a credit card.

[0201] [Inventor identity verification] If the inventor is a user, an additional procedure is required to verify that the user is the inventor of a patent. The inventor registers unique number information that identifies the patent, such as the patent application number, in the intellectual property business support system 100. Note that the intellectual property business support system 100 may be referred to as the present system. Once the inventor verification process, which will be described later, is complete and the inventor is recognized as the actual inventor, the patent will be made publicly available within the system for other users to view.

[0202] The processing unit 111 retrieves patent information from databases such as patent offices. This public information includes not only the technical content of the patent, but also data on its practicality, innovativeness, and inventive step, allowing other users to accurately understand its value. In addition to the original patent specification, information on practicality, innovativeness, and inventive step is automatically generated in this system using language processing techniques such as large-scale language models. The methods for generating the above-mentioned linguistic and visual representations described in this disclosure may also be used.

[0203] [Inventor identity verification] The intellectual property business support system 100 provides the following method for a user to prove that they are the inventor of an invention. The user holds a patent certificate printed and issued by the Patent Office and an official photo ID such as a driver's license. The intellectual property business support system 100 simultaneously photographs the user's face, patent certificate, and driver's license. Figure 25 is a diagram showing an example of the captured image when verifying the identity of the inventor.

[0204] The processing unit 111 performs the following match checks, and if all match, determines that the user is the inventor of the present invention. a) The "Inventor (Name)" listed on the patent certificate and the "Name" listed on the driver's license b) The face of the person holding the two certificates and the photograph on the driver's license c) The "patent number" written on the patent certificate and the "patent number" entered by the user in the UI on the registration screen

[0205] Instead of a driver's license, a My Number card may be used. A photograph is taken so that the facial photograph printed on the My Number card is visible. Furthermore, personal authentication using the My Number card may be further performed, such as by reading an IC card and having the person enter a personal identification number known only to the person.

[0206] [Pre-uploading of pre-publication application documents] 26 is a conceptual diagram showing an example of the configuration of an intellectual property business support device. When a user attempts to register as an inventor of a certain patent in the intellectual property business support system 100, the user uploads the application number and information known only to the inventor to the intellectual property business support system 100. Later, when the patent gazette for the application number entered by the user is published by the Japan Patent Office, the processing unit 111 checks whether the information previously uploaded by the user matches the data published by the Japan Patent Office. If there is a sufficient match, the processing unit 111 determines that the user is the actual inventor of the invention.

[0207] The information to be uploaded in advance includes the application document "patent application" submitted to the Patent Office, but if the user's company regulations prohibit uploading the entire application document, it is acceptable to upload only partial information about the invention, such as the invention title, claim 1, abstract, and problem of the invention. In this case, the user is prompted to enter enough information to confirm the match. For example, if the number of characters is small, a UI may be provided that prompts the user to enter more.

[0208] Furthermore, for employees of companies that are reluctant to upload pre-filing information to the public, it may be possible to upload documents that are encrypted so that only the user can decrypt them.

[0209] [Post-publication match check] After the patent document associated with the application number registered by the user is published by the Patent Office, the processing unit 111 obtains the published document from the Patent Office. The document may be obtained from bulk patent data or by using a file wrapper API. The processing unit 111 then checks whether the document matches the information uploaded by the user before publication. If the processing unit 111 determines that the document contains sufficient information that only the inventor would know, it identifies the user as the inventor.

[0210] If an encrypted document is uploaded in advance, for example, after publication, users can be asked to upload or enter a decryption key. If the publication date has passed, users who may have forgotten the key can be prompted to enter the decryption key by email or chat.

[0211] When the user's decryption key is input to the intellectual property business support system 100, the processing unit 111 decrypts the pre-uploaded document. If the decryption is successful and the contents of the pre-uploaded document can be obtained, the subsequent confirmation of the match with the data published by the Japan Patent Office will be the same as described above.

[0212] Information that only the inventor himself / herself knows is not limited to the "patent application," but can also be an argument, an amendment, etc. These are made public as soon as they are submitted to the Patent Office, so it is advisable to wait a few days after registering the invention in the system before submitting them to the Patent Office.

[0213] [Linkage with the Patent Office submission system] The user acquires in advance the user ID in the intellectual property business support system 100, or encrypted information of the user ID obtained by hashing the user ID, from the intellectual property business support system 100. Alternatively, the user may acquire encrypted information of the user ID encrypted with a public key managed by the intellectual property business support system 100.

[0214] Users belonging to a company register their own user ID in the patent management system provided by the company. The company's patent management system then provides the user ID to the patent office acting as the application agent. The patent office then lists the user ID along with the inventor's name in the patent document submitted to the Japan Patent Office. The user ID may be written within the document, or it may be registered outside the document as attribute information when registering the document in the Japan Patent Office online system.

[0215] If a user ID is written in a patent document, the intellectual property business support system 100 can obtain the user ID from the patent disclosure information document and identify the user. If the user ID is hash-encrypted, it searches for a match among the hash-encrypted keys for all users held by the intellectual property business support system 100. If the user ID is encrypted using a public key system, it is decrypted using a private key held only by the intellectual property business support system 100 to obtain the user ID information.

[0216] If the user ID is not included in the patent document, an API is provided on the Patent Office server that returns whether or not the user ID is included in the patent number when the encrypted user ID and patent number are entered at the same time. The intellectual property business support system 100 can confirm the link between the user and the patent by calling this Patent Office API.

[0217] [Applicant's personal identification] Users can register not only users intended for inventors, but also users who are applicants, including companies. The identity authentication of applicant users can also be performed using the method of pre-registering information that only the applicant knows and confirming its consistency after disclosure, as described above. The operations staff or sales staff of the intellectual property business support system 100 may communicate directly with the intellectual property staff or intellectual property system staff of the applicant company, or may conclude a service agreement or NDA.

[0218] [Company applicants' collective certification of employee inventors] It is assumed that the intellectual property officer of the corporate applicant or the patent management system used by the company has some kind of trust relationship with the intellectual property business support system 100 and is in a state where data can be shared. The company may provide the intellectual property business support system 100 with information that lists the patent number, inventor name, and user ID in a predetermined format for each patent, thereby authenticating that the user of the intellectual property business support system 100 is the actual inventor of the patent.

[0219] [User-visited patents] In step St403, a user who has an account registered in the intellectual property business support system 100 can view the registered patents.

[0220] [Send Respectful Reaction] A user can send a "respect" reaction to a patent that they have viewed. This reaction has a meaning that goes beyond a simple "like" and expresses praise for the technological superiority and creativity of the patent. In step St404, the processing unit 111 acquires information indicating the "respect" reaction (St404).

[0221] [Counting and displaying respectful reactions] In step St405, the processing unit 111 compiles information indicating respectful reactions and displays it on the display device 24. The number of reactions is compiled in real time and displayed on the page of each patent. This allows the user to identify popular patents at a glance and intuitively understand how the patent is evaluated within the industry or community.

[0222] [Feedback to inventors] In step St406, the processing unit 111 provides feedback to the inventor (St406). The counting of the respect reactions by the processing unit 111 is an important motivating factor for the inventor. The inventor can confirm the value of his / her creative activities by the number of "respects" his / her patent has collected, which stimulates further inventive activities.

[0223] If the intellectual property business support system 100 is operated on a national scale, for example, if only Japanese citizens are allowed to register as users, the system will be sponsored by the Japan Patent Office and will provide inventors with rewards such as virtual badges, priority display, patent promotion support, etc. These rewards will enhance the status and prestige of inventors and increase the opportunities for their activities and patents to be widely recognized in society.

[0224] The intellectual property business support system 100 can be provided as a global cloud service, with each company signing a service contract and creating a user group for each company. Only users within the company can send respect reactions and view the respect aggregation results. In this case, the intellectual property department of the company acts as a sponsor and provides incentives to inventors.

[0225] Furthermore, in the case of a user group that is larger than a single company, such as the entire country of Japan, there is a concern that users will not be able to freely send words of respect due to consideration for the position of the company or organization to which they belong. Therefore, the intellectual property business support system 100 may be configured so that other users cannot identify who is the sender of a "send." In other words, it is possible to not provide a function to display the sender's real personal name, handle name, etc.

[0226] [Submit patent visual representation] Users can send visual representations of patents created using the methods described in this disclosure, or simply any images, videos, or songs, to inventors. To prevent flaming, the content of the sent content will not be displayed to other users. It can only be viewed by the inventor who sent it. If an inventor likes the sent content, they can select one and designate it as representative content for one of their patents. The designated representative content will be displayed on the display device instead of the patent specification itself, or simultaneously with the patent specification itself. It may also be displayed faded in the background of the specification. Once designated as representative content, it will be displayed in the RPG-style display system and search system described in this specification.

[0227] (Summary of this embodiment) The above description of the present embodiment discloses the following techniques.

[0228] <Technology 1> In one embodiment, in an intellectual property business support method using an intellectual property business support system having a processor and memory, the processor acquires patent information, provides the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquires the patent object generated by the multimodal model, and outputs the patent object to a display device. This allows the patent object, which is visual information, to be acquired and output using a multimodal model.

[0229] <Technology 2> In the intellectual property business support method described in Technology 1, the processor places the patent object on a patent map. This allows the patent object to be placed on the patent map.

[0230] <Technology 3> In the intellectual property business support method described in Technology 1 or Technology 2, the processor changes the visual information of the patent object according to the degree of interest from the user in the patent object. This allows the visual information of a patent object to be changed depending on the degree of interest of the user in the patent object.

[0231] <Technology 4> In one embodiment, an intellectual property business support device having a processor and memory has the processor acquire patent information, provide the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquire the patent object generated by the multimodal model, and output the patent object to a display device. This allows the patent object, which is visual information, to be acquired and output using a multimodal model.

[0232] <Technology 5> In the intellectual property business support device described in Technology 4, the multimodal model is provided by an external system different from the intellectual property business support device, and the processor transmits the patent information and the instruction information to the external system and receives the patent object from the external system. This allows patent objects to be retrieved using multimodal models of external systems.

[0233] <Technology 6> In the intellectual property business support device described in Technology 4, the multimodal model is placed in the memory, and the processor obtains the patent object using the multimodal model placed in the memory. This allows the patent object to be acquired using the multimodal model in the intellectual property business support device.

[0234] <Technology 7> An intellectual property business support program in one embodiment causes a computer to acquire patent information, provide the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model, acquire the patent object generated by the multimodal model, and output the patent object to a display device. This allows the patent object, which is visual information, to be acquired and output using a multimodal model.

[0235] <Technology 8> In one embodiment, an intellectual property business support method using an intellectual property business support system having a processor and a memory includes the steps of: acquiring patent information; determining characteristic words related to the patent information or themes related to multiple pieces of patent information; providing the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model; acquiring the summary information generated by the language model; and outputting the summary information to a display device. This makes it possible to obtain and output summary information of patent information using a language model.

[0236] <Technology 9> In the intellectual property business support method described in Technology 8, the processor acquires a plurality of pieces of patent information, determines characteristic words related to the plurality of pieces of patent information or themes related to the plurality of pieces of patent information, provides the plurality of pieces of patent information and instruction information for generating summary information that is a summary of the plurality of pieces of patent information and comparison information that compares the plurality of pieces of patent information based on the characteristic words or themes for each patent to a language model, acquires the summary information and comparison information generated by the language model, and outputs the summary information and comparison information to a display device. This allows the language model to be used to obtain and output summary information and comparison information.

[0237] <Technology 10> In one embodiment of an intellectual property business support device having a processor and a memory, the processor acquires patent information, determines characteristic words related to the patent information or themes related to multiple pieces of patent information, provides the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model, acquires the summary information generated by the language model, and outputs the summary information to a display device. This allows summary information to be obtained using a language model.

[0238] <Technology 11> In the intellectual property business support device described in Technology 10, the language model is provided by an external system different from the intellectual property business support device, and the processor transmits the patent information and the instruction information to the external system and receives the summary information from the external system. This allows the language model of the external system to be used to obtain summary information.

[0239] <Technology 12> In the intellectual property business support device described in Technology 10, the language model is placed in the memory, and the processor acquires the summary information using the language model placed in the memory. This allows the language model in the intellectual property business support device to be used to obtain summary information of patent information.

[0240] <Technology 13> An intellectual property business support program according to one embodiment causes a computer to acquire patent information, determine characteristic words related to the patent information or multiple themes related to the patent information, provide the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the characteristic words or themes, to a language model, acquire the summary information generated by the language model, and output the summary information to a display device. This makes it possible to obtain and output summary information of patent information using a language model.

[0241] <Technology 14> In one embodiment of an intellectual property business support method using an intellectual property business support system having a processor and memory, the processor associates a focus setting, indicating that a patent indicated by patent information is a notable patent, with the patent information, and when creating a search formula to be used for searching patents from a patent database, displays information on a search result screen indicating whether the patent for which the focus setting has been made is included in the search results based on the search formula. This allows the user to know whether or not the patent for which the focus setting has been made is included in the search results based on the search formula.

[0242] <Technology 15> In the intellectual property business support method described in Technology 14, the processor inputs the search results obtained by the patent search into a model that has undergone machine learning using patent information corresponding to the patent for which the focus setting has been made as learning data, obtains the focus patents extracted by the model, and presents information about the focus patents to the user. This allows us to obtain notable patents using machine learning models.

[0243] <Technology 16> In the intellectual property business support method described in Technology 14, the processor provides the patent for which the focus setting has been made and instruction information for generating search words that may include the patent for which the focus setting has been made to a language model, obtains the search words generated by the language model, and presents the search words to the user. This makes it possible to use a language model to obtain search words that will likely include patents with focus settings in the search results.

[0244] <Technology 17> In the intellectual property business support method described in Technology 16, the processor updates the search word if the patent for which the focus setting has been set is not included in the search results in a patent search based on the generated search word. This allows you to update your search terms if the patents for which you have set a focus are not included in the search results.

[0245] <Technology 18> In one embodiment of an intellectual property business support device having a processor and memory, the processor associates a focus setting with the patent information, indicating that the patent indicated by the patent information is a notable patent, and when creating a search formula to be used for searching patents from a patent database, displays information on a search result screen indicating whether the patent for which the focus setting has been made is included in the search results based on the search formula. This allows the user to know whether or not the patent for which the focus setting has been made is included in the search results based on the search formula.

[0246] <Technology 19> An intellectual property business support program in one embodiment causes a computer to associate a focus setting, indicating that a patent indicated by patent information is a notable patent, with the patent information, and when creating a search formula to be used for searching patents from a patent database, to display on a search result screen information indicating whether the patent for which the focus setting has been made is included in the search results based on the search formula. This allows the user to know whether or not the patent for which the focus setting has been made is included in the search results based on the search formula.

[0247] <Technology 20> In one embodiment, an intellectual property business support method using an intellectual property business support system having a processor and memory includes the processor registering user information of a user, performing inventor authentication for each patent, receiving reaction information indicating respect for a patent for which the user is the inventor, and outputting information indicating the results of compiling the reaction information to the user who is the inventor of the patent. This allows the user who is the inventor of the patent to receive reaction information showing respect.

[0248] <Technology 21> In the intellectual property business support method described in Art 20, in authenticating the inventor for each patent, the processor acquires the patent number entered by the user, acquires the inventor's name based on the patent number, and performs a first check to compare the patent number reflected in the captured image obtained by photographing the patent certificate with the patent number entered by the user; a second check to compare the name and facial image written on the user's identification document reflected in the captured image by photographing the image so that the name and facial image written on the user's identification document and the user's face fit within a single image; and a third check to compare the user's face reflected in the captured image with the facial image written on the identification document reflected in the captured image. If the first check is a match, or if the second check and the third check are a match, the processor authenticates the user as the inventor of the patent. This allows the user to be authenticated as the inventor of the patent.

[0249] <Technology 22> In the intellectual property business support method described in Technology 20, when authenticating the inventor for each patent, the processor registers unpublished patent information that only the inventor can know before the patent information is made public, and after the patent information is made public, it compares the published patent information with the unpublished patent information, and if the two match, authenticates that the user is the inventor of the patent. This allows the user to be authenticated as the inventor of the patent.

[0250] <Technology 23> In the intellectual property business support method described in Technology 20, when authenticating the inventor for each patent, the processor registers a user ID in advance, and files a patent application while linking the user ID to the patent application documents.The processor then compares the user ID contained in the patent information obtained from the Patent Office database after the patent is made public with the user ID that was registered in advance, and if the two match, authenticates that the user is the inventor of the patent. This allows the user to be authenticated as the inventor of the patent.

[0251] <Technology 24> In one embodiment, an intellectual property business support device having a processor and memory has the processor registering user information of a user, performing inventor authentication for each patent, receiving reaction information showing respect for a patent for which the user is the inventor, and outputting information showing the results of compiling the reaction information to the user who is the inventor of the patent. This allows the user who is the inventor of the patent to receive reaction information showing respect.

[0252] <Technology 25> An intellectual property business support program in one embodiment causes a computer to register user information of a user, perform inventor authentication for each patent, receive reaction information showing respect for a patent for which the user is the inventor, and output information showing the results of compiling the reaction information to the user who is the inventor of the patent. This allows the user who is the inventor of the patent to receive reaction information showing respect.

[0253] While the intellectual property business support method according to the present disclosure has been described above with reference to the drawings, it goes without saying that the present disclosure is not limited to such examples. It is clear that a person skilled in the art can conceive of various modifications, alterations, substitutions, additions, deletions, and equivalents within the scope of the claims, and it is understood that these also naturally fall within the technical scope of the present disclosure. Furthermore, the components of the various embodiments described above may be combined in any manner as long as they do not deviate from the spirit of the invention. [Industrial Applicability]

[0254] The present disclosure is useful as an intellectual property business support method that allows users to intuitively understand patents. [Explanation of symbols]

[0255] 1. Intellectual property business support device 2. User terminal 3 External Systems 4. Communication Network 11 processors 111 Processing Unit 12 Storage device 121 Programs 122 Data 123 AI models 21 processors 22 Storage device 23 Input Devices 24 Display device 100 Intellectual Property Business Support System

Claims

1. An intellectual property business support method using an intellectual property business support system having a processor and a memory, The processor: Obtain patent information, providing the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model; obtaining a patent object generated by the multimodal model; outputting the patent object to a display device; Intellectual property business support methods.

2. The processor: placing said patent object on a patent map; The intellectual property business support method according to claim 1.

3. The processor: changing visual information of the patent object according to a degree of interest from a user in the patent object; 3. The intellectual property business support method according to claim 1 or 2.

4. An intellectual property business support device having a processor and a memory, The processor: Obtain patent information, providing the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model; obtaining a patent object generated by the multimodal model; outputting the patent object to a display device; Intellectual property business support device.

5. the multimodal model is provided by an external system different from the intellectual property business support device; The processor: transmitting the patent information and the instruction information to the external system; receiving the patent object from the external system; The intellectual property business support device according to claim 4.

6. the multimodal model is located in the memory; the processor uses the multimodal model located in the memory to obtain the patent object. The intellectual property business support device according to claim 4.

7. On the computer, Obtain patent information, providing the patent information and instruction information for generating a patent object, which is visual information corresponding to the patent information, to a multimodal model; obtaining a patent object generated by the multimodal model; outputting the patent object to a display device; This is an intellectual property business support program that helps you achieve this.

8. An intellectual property business support method using an intellectual property business support system having a processor and a memory, The processor: Obtain patent information, determining characteristic words related to the patent information or a plurality of themes related to the patent information; providing the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the feature words or themes, to a language model; obtaining the summary information generated by the language model; outputting the summary information to a display device; Intellectual property business support methods.

9. The processor: Obtaining multiple pieces of patent information, Determine characteristic words or themes related to multiple patent information, providing the plurality of patent information, summary information that is a summary of the plurality of patent information, and comparative information that compares the plurality of patent information, for each patent, based on the feature words or themes, to a language model; obtaining the summary information and the comparison information generated by the language model; outputting the summary information and the comparison information to a display device; The intellectual property business support method according to claim 8.

10. An intellectual property business support device having a processor and a memory, The processor: Obtain patent information, determining characteristic words related to the patent information or a plurality of themes related to the patent information; providing the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the feature words or themes, to a language model; obtaining the summary information generated by the language model; outputting the summary information to a display device; Intellectual property business support device.

11. the language model is provided by an external system different from the intellectual property business support device, The processor: transmitting the patent information and the instruction information to the external system; receiving the summary information from the external system; The intellectual property business support device according to claim 10.

12. the language model is located in the memory; the processor obtains the summary information using the language model stored in the memory. The intellectual property business support device according to claim 10.

13. On the computer, Obtain patent information, determining characteristic words related to the patent information or a plurality of themes related to the patent information; providing the patent information and instruction information for generating summary information, which is a summary of the patent information, based on the feature words or themes, to a language model; obtaining the summary information generated by the language model; outputting the summary information to a display device; This is an intellectual property business support program that helps you achieve this.

14. An intellectual property business support method using an intellectual property business support system having a processor and a memory, The processor: A notice setting indicating that the patent indicated by the patent information is a noteworthy patent is performed in association with the patent information; When creating a search query to be used for patent searches from a patent database, information indicating whether or not the patent for which the focus setting is set is included in the search results based on the search query is displayed on the search result screen. Intellectual property business support methods.

15. The processor: The search results obtained by the patent search are input to a model that has undergone machine learning using patent information corresponding to the patent for which the focus setting is set as learning data. Obtaining noteworthy patents extracted by the model, presenting information about the patent of interest to the user; The intellectual property business support method according to claim 14.

16. The processor: providing a language model with patents for which the attention setting is performed and instruction information for generating search words that will allow the patents for which the attention setting is performed to be included in search results; Obtaining search words generated by the language model; presenting the search words to a user; The intellectual property business support method according to claim 14.

17. The processor: In a patent search based on the generated search word, if the patent for which the focus setting is performed is not included in the search results, the search word is updated. The intellectual property business support method according to claim 16.

18. An intellectual property business support device having a processor and a memory, The processor: A notice setting indicating that the patent indicated by the patent information is a noteworthy patent is performed in association with the patent information; When creating a search query to be used for patent searches from a patent database, information indicating whether or not the patent for which the focus setting is set is included in the search results based on the search query is displayed on the search result screen. Intellectual property business support device.

19. On the computer, A notice setting indicating that the patent indicated by the patent information is a noteworthy patent is performed in association with the patent information; When creating a search query to be used for patent searches from a patent database, information indicating whether or not the patent for which the focus setting is set is included in the search results based on the search query is displayed on the search result screen. This is an intellectual property business support program that helps you achieve this.

20. An intellectual property business support method using an intellectual property business support system having a processor and a memory, The processor: Register the user's user information, We authenticate inventors for each patent, receiving reaction information indicating respect for a patent for which the user is an inventor; outputting information indicating the result of aggregating the reaction information to the user who is the inventor of the patent; Intellectual property business support methods.

21. The processor: In the inventor certification for each patent, a first comparison step of acquiring a patent number input by a user, acquiring an inventor's name based on the patent number, and comparing the patent number captured in an image obtained by capturing an image of the patent certificate with the patent number input by the user; a second comparison in which an image is captured so that the name and face image written on the user's identification document and the user's face are included in one image, and the acquired inventor name is compared with the name written on the identification document that appears in the captured image; performing a third comparison in which the face of the user reflected in the captured image is compared with a facial image of an identification document reflected in the captured image; if the first match is a match, or if the second match and the third match are a match, authenticating the user as an inventor of the patent; The intellectual property business support method according to claim 20.

22. The processor: In the inventor certification for each patent, Before the patent information is made public, unpublished patent information that only the inventor knows is registered in advance, After the patent information is made public, the published patent information is compared with the unpublished patent information, and if the two match, the user is authenticated as the inventor of the patent. The intellectual property business support method according to claim 20.

23. The processor: In the inventor certification for each patent, Please register your user ID in advance. By filing a patent application with the user ID linked to the patent application documents, the user ID included in the patent information obtained from the Patent Office database after the patent is published is compared with the user ID registered in advance, and if the two match, the user is authenticated as the inventor of the patent. The intellectual property business support method according to claim 20.

24. An intellectual property business support device having a processor and a memory, The processor: Register the user's user information, We authenticate inventors for each patent, receiving reaction information indicating respect for a patent for which the user is an inventor; outputting information indicating the result of aggregating the reaction information to the user who is the inventor of the patent; Intellectual property business support device.

25. On the computer, Register the user's user information, We authenticate inventors for each patent, receiving reaction information indicating respect for a patent for which the user is an inventor; outputting information indicating the result of aggregating the reaction information to the user who is the inventor of the patent; This is an intellectual property business support program that helps you achieve this.

Images