Invention finding device, invention finding method, and invention finding program

Abstract

To easily examine an application.SOLUTION: An invention discovery device 100 receives information on the invention. The invention discovery device 100 inputs the received information related to the invention and a command to generate information related to the patent property of the invention to a generation model learned by information related to the disclosed technique, and generates information related to the patent property of the invention related to the information related to the invention. The invention discovery device 100 outputs the generated information on the patent property of the invention.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to an invention discovery device, an invention discovery method, and an invention discovery program. [Background technology]

[0002] Inventions are conceived based on the need to solve or improve upon issues that are difficult or insufficient to realize using already existing disclosed technology (hereinafter referred to simply as "prior art"). One known method for creating new inventions is invention discovery, which involves extracting new technical ideas buried in the information held by developers and giving them form.

[0003] As mentioned above, there are cases where a patent application is considered for a conceived or discovered invention. In such cases, consideration is given to whether the invention is new compared to the prior art (novelty), whether it is something that a person with ordinary skill in the field to which the invention belongs could not easily invent (inventive step), etc.

[0004] In order to study a patent application, it is necessary to search for information on the prior art in order to compare and study the "invention" with the "prior art." Therefore, a conventional technique is known in which, based on superordinate conceptualization information indicating, as a superordinate concept, bibliographic information corresponding to prior document information containing a search keyword transmitted from a user terminal, technology superordinate information is generated to provide a superordinate view of document data, and the generated information is output to a user terminal (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2022-187527 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the above-mentioned conventional techniques have the problem that it is difficult to conduct patent application review. For example, in the conventional techniques, users themselves must select and input search keywords to search for prior art related to the content of the invention, which can make it difficult to obtain information on prior art and review patent applications without imposing a burden. [Means for solving the problem]

[0007] Therefore, in order to solve the above-mentioned problems and achieve the objectives, the invention discovery device of the present invention is characterized by having a reception unit that receives information related to an invention, a generation unit that inputs the information related to the invention received by the reception unit and an instruction to generate information related to the patentability of the invention into a generative model trained using information related to disclosed technology, and an output unit that outputs the information related to the patentability of the invention generated by the generation unit. [Effects of the Invention]

[0008] The present invention has the effect of making it possible to easily carry out application review. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram for explaining the overall processing of the excavation device of the present embodiment. [Figure 2] FIG. 2 is a diagram showing an example of generation of information relating to the patentability of an invention according to this embodiment. [Figure 3] FIG. 3 is a diagram showing the configuration of the invention excavation device according to this embodiment. [Figure 4] FIG. 4 is a table showing an example of prior art information according to this embodiment. [Figure 5] FIG. 5 is a table showing an example of invention information according to this embodiment. [Figure 6]FIG. 6 is a diagram illustrating tsuzumi (registered trademark). [Figure 7] FIG. 7 is a diagram illustrating tsuzumi (registered trademark). [Figure 8] FIG. 8 is a table showing an example of patentability information according to this embodiment. [Figure 9] FIG. 9 is a diagram showing an example of an invention information reception screen according to this embodiment. [Figure 10] FIG. 10 is a diagram illustrating an example of the learning process according to this embodiment. [Figure 11] FIG. 11 is a diagram illustrating an example of the generation process according to this embodiment. [Figure 12] FIG. 12 is a diagram illustrating an example of the generation process according to this embodiment. [Figure 13] FIG. 13 is a diagram showing an example of the reception process and the output process according to this embodiment. [Figure 14] FIG. 14 is a flowchart showing the invention discovery process according to this embodiment. [Figure 15] FIG. 15 is a diagram showing an example of a computer that executes the invention excavation process according to this embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention (hereinafter referred to as "embodiments") will be described with reference to the drawings. Note that the embodiments are not limited to the following description.

[0011] <Overview> Fig. 1 is a diagram illustrating an overall view of the processing of the invention discovery device 100 according to this embodiment. The invention discovery device 100 shown in Fig. 1 is an example of a computer that provides a technology for generating information on the patentability of information related to inventions, such as novelty and inventive step, and outputting the generated information.

[0012] (background) Patent applications may be considered for inventions that are conceived from solving or improving problems that are difficult to solve with prior art, or inventions that are discovered by extracting and giving form to new technical ideas from a vast amount of information on research and development, etc. For example, when considering whether to file a patent application, consideration is given to whether the invention in question is novel and involves an inventive step (hereinafter sometimes referred to as "patentability consideration").

[0013] In the above-mentioned patentability examination, information on prior art related to the invention in question is searched and compared with the invention in question to determine whether the invention is patentable. Therefore, in the examination of patentability, it is necessary to search for information on prior art.

[0014] Therefore, a reference technique is known in which information (technical overview information) that provides an overview of document data corresponding to preceding document information searched based on a search word input by a user is generated and output to a user terminal.

[0015] However, the above-mentioned reference technology requires the user to input search terms, and if the user lacks knowledge or insight, they may not be able to search for appropriate prior art, making it difficult to consider filing a patent application.

[0016] Furthermore, there are cases where inventors themselves do not have time to dig out inventions from the vast amount of information, or where they need advice from experts when digging out inventions or considering patent applications.However, with reference technology, it is difficult to dig out inventions or provide advice by simply generating technology overview information.

[0017] (Processing by invention excavation device 100) Therefore, the invention discovery device 100 of this embodiment outputs information regarding the patentability of an invention, which is used by the user to determine the patentability of the invention, based on a generative model that has been trained in advance using information regarding prior art, etc., and information regarding the invention.

[0018] In the following sections, information related to prior art will be referred to as "prior art information." Information related to inventions, such as research and development concept documents, specifications, idea memos, etc., entered by users, etc., will be referred to as "invention information." Information related to the patentability of inventions that users use to determine the patentability of inventions will be referred to as "patentability information."

[0019] 1, a description will be given of the invention excavation process performed by the invention excavation device 100. First, the invention excavation device 100 receives invention information input by a user or the like ((1) in FIG. 1).

[0020] Next, the invention discovery device 100 generates patentability information ((2) in FIG. 1). For example, the invention discovery device 100 inputs the received invention information and a command to generate patentability information ((2-1) in FIG. 1) into a generation model ((2-2) in FIG. 1) trained using prior art information, and generates patentability information related to the invention information ((2-3) in FIG. 1).

[0021] Then, the invention discovery device 100 outputs the generated patentability information ((3) in FIG. 1). For example, the invention discovery device 100 according to this embodiment generates patentability information such as indicators and comments on novelty and inventive step, and suggestions for extension of the invention, based on the generative model, for the input invention information, and outputs the information to the user ((3-1) in FIG. 1).

[0022] Therefore, the invention discovery device 100 has the effect of enabling a user to easily consider an application even when the user does not have time to discover an invention or consider an application through the above-mentioned processing, or when the user lacks knowledge or expertise.

[0023] <Description of the invention excavation device 100> Next, the invention discovery device 100 according to this embodiment will be described. First, a series of steps in which the invention discovery device 100 according to this embodiment generates patentability information based on a generative model will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the generation of patentability information (information relating to the patentability of an invention) according to this embodiment.

[0024] Figure 2 shows an invention mining device 100 that generates patentability information for invention information input by a user, a terminal device 200 operated by the user, and a server 300 that stores prior art information, etc. that the invention mining device 100 uses to generate the patentability information.

[0025] First, the terminal device 200 transmits invention information input by a user or the like to the invention excavation device 100 ((1) in FIG. 2). Next, the invention excavation device 100 inputs the invention information transmitted by the terminal device 200 into search word extraction LLMs (Large Language Models), which are generative models for extracting search words for searching prior art information from the invention information, and extracts the search words ((2) in FIG. 2).

[0026] The invention discovery device 100 uses the extracted search words to obtain prior art information related to the invention information entered by the user from information stored on an external server 300, etc. ((3) in Figure 2).

[0027] The invention mining device 100 uses the acquired prior art information to train a similarity determination LLM and a patentability extraction LLM as generative models for generating patentability information ((4-1) and (4-2) in FIG. 2). Note that the following sections describe examples in which the invention mining device 100 uses trained generative models, but the invention mining device 100 may also use trained generative models trained by an information processing device other than the invention mining device 100.

[0028] Next, the invention discovery device 100 inputs the input invention information and a command to determine whether or not there is prior art similar to the target invention (similarity) into the trained similarity determination LLM ((5-1) in FIG. 2) ((5-2) in FIG. 2), and performs a similarity determination for the target invention. The invention discovery device 100 then outputs the similarity determination result for the input invention information to the user ((5-3) in FIG. 2). At this time, the invention discovery device 100 can interactively receive additional information from the user, such as information to be used in determining similarity ((5-4) in FIG. 2).

[0029] Next, the invention discovery device 100 inputs the input invention information and a command to extract information (patentability information) about whether the target invention has novelty, inventive step, etc. into the pre-trained patentability extraction LLM ((6-1) in FIG. 2) ((6-2) in FIG. 2), and extracts patentability information about the target invention. The invention discovery device 100 then outputs the extracted results of patentability information about the input invention information to the user ((6-3) in FIG. 2). At this time, the invention discovery device 100 can interactively accept additional information to refine the invention information input by the user based on the extracted results of patentability information ((6-4) in FIG. 2).

[0030] Through the above-described processing, the invention discovery device 100 can generate patentability information for the invention based on prior art information including technical content related to the field corresponding to the invention information input by the user, and output the generated patentability information to the user.

[0031] <Description of the invention excavation device 100> From here, the configuration of the invention excavation device 100 according to this embodiment will be described. Figure 3 is a diagram showing the configuration of the invention excavation device 100 according to this embodiment. As shown in Figure 3, the invention excavation device 100 has a communication unit 110, a storage unit 120, and a control unit 130.

[0032] 3, the invention discovery device 100 may include an input unit such as a keyboard or a mouse for receiving inputs such as operations by an administrator, etc. The invention discovery device 100 may also include a display unit such as a display for displaying to an administrator, etc., the input invention information and additional information, the acquired prior art information, and the generated patentability information, etc.

[0033] (Communication unit 110) The communication unit 110 performs data communication related to the input of patent information input via the terminal device 200 operated by a user. The communication unit 110 also performs data communication related to the output of generated information related to the patentability of an invention. The communication unit 110 is realized by a NIC (Network Interface Card) or the like, and controls communication via a telecommunication line such as a LAN (Local Area Network) or the Internet. The communication unit 110 is connected to a network via a wired or wireless connection as necessary, and can transmit and receive information bidirectionally with the terminal device 200, server 300, etc.

[0034] (Storage unit 120) The storage unit 120 stores data and programs used in various processes by the control unit 130, and various data acquired by the operation of the control unit 130. The storage unit 120 is realized by a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk. As shown in FIG. 3 , the storage unit 120 has a prior art information DB 121, an invention information DB 122, a generative model DB 123, and a patentability information DB 124.

[0035] (Prior art information DB121) The prior art information DB 121 is a database that stores prior art information, which is information on technologies that have already been disclosed, such as papers, literature, and patent publications. An example of the prior art information stored in the prior art information DB 121 will now be described using a table. Fig. 4 is a table diagram showing an example of the prior art information according to this embodiment.

[0036] 4, the prior art information DB 121 stores information on the following items in a table format, etc., in association with "No.", which is information identifying individual data of prior art information: background art, problems to be solved by the invention, means for solving the problems, effects of the invention, modes for carrying out the invention, and drawings. In this embodiment, information contained in "patent publications" is used as prior art information as an example, but this is not particularly limited.

[0037] Here, we will explain the information contained in the above items. The background art includes information explaining the background art leading to the invention described in the patent publication. The problem that the invention aims to solve includes information explaining the problem that the invention described in the patent publication aims to solve, such as problems in the prior art. The means for solving the problem includes information explaining the means for solving the above-mentioned problem. The effect of the invention includes information explaining the effect obtained when the invention is realized by the "means for solving the problem." The form for implementing the invention includes information explaining in detail the "means for solving the problem." The drawings include image information, etc. used to explain the invention.

[0038] As an example shown in Figure 4, the prior art information DB121 can store No. "1," background art "A," problem that the invention aims to solve "B," means for solving the problem "C," effect of the invention "D," form for implementing the invention "E," and drawing "F" in a table format or the like, each associated with each other.

[0039] The alphabets "A to F" included in each of the above-mentioned items exemplify the information of each item stored in the prior art information DB 121. For example, the content (information) of each item such as "Background Art" "A," "Problem to be Solved by the Invention" "B," "Means for Solving the Problem" "C," "Advantages of the Invention" "D," and "Mode for Carrying Out the Invention" "E" may be character string information expressed in a predetermined natural language such as hiragana, kanji, roman letters, or numbers. Furthermore, the item for drawings may be image data or link information to the image data.

[0040] (Invention Information DB122) The invention information DB 122 is a database that stores invention information, such as research and development concept documents, specifications, idea memos, and other information related to inventions input by users, etc. An example of the invention information stored in the invention information DB 122 will now be described using a table diagram. Fig. 5 is a table diagram showing an example of the invention information according to this embodiment.

[0041] As shown in Figure 5, the invention information DB122 stores information on the items of file information, text information, expected patent classification, points that are thought to be new, and additional information in a table format or the like, in association with "No.", which is information for identifying individual data of invention information.

[0042] The information contained in the above items will now be explained. File information includes file data in a specified format, such as text files containing information on invention plans and experimental results, files containing numerical information, image files, video files, and audio files. Text information includes information on the invention entered by the user in text format based on specified items such as "Background Art, Prior Art," "Problems and Issues the Invention Aims to Solve," "Key Points of the Invention for Solving the Problems (Differences from the Prior Art)," "Effects of the Invention," "Specific Invention Contents," and "Processing Flow." The assumed patent classification identifies the field to which the invention belongs, including, for example, classification names, information identifying patent classifications such as the International Patent Classification (IPC) and F-terms. The novelty points include information about the novelty and inventive step of the invention as perceived by the user. Additional information includes information entered by the user based on confirmation items and patentability information output by the output unit 135 (described below).

[0043] As an example shown in Figure 5, the invention information DB122 can store file information "G", text information "H", expected patent classification "I", points that are thought to be new "J", and additional information "K" in association with No. "1" in a table format or the like.

[0044] The letters "G to K" included in each of the above-mentioned items exemplify the information of each item stored in the invention information DB 122. For example, the content (information) of each item, such as text information "H," expected patent classification "I," novel point "J," and additional information "K," may be character string information expressed in a predetermined natural language, such as hiragana, kanji, roman letters, or numbers. Furthermore, file information "G" may be file data in a predetermined format, as described above.

[0045] (Generated model DB123) The generative model DB 123 is a database that stores a large-scale language model, which is a predetermined generative model used by the later-described generator 134. For example, the above-described generative model may be tsuzumi (registered trademark) (see, for example, Reference 1), which is a large-scale language model that has undergone adapter tuning using prior knowledge (prior art information) in a predetermined field.

[0046] (Reference 1): NTT's large-scale language model "tsuzumi",<URL:https: / / www.rd.ntt / research / LLM_tsuzumi.html> ,<Searched on June 5, 2020>

[0047] From now on, tsuzumi (registered trademark) will be taken as an example to explain the generative model used by the inventive excavation device 100 according to this embodiment. Figures 6 and 7 are diagrams for explaining tsuzumi (registered trademark).

[0048] First, the concept of tsuzumi (registered trademark) will be explained using Figure 6. tsuzumi (registered trademark) is a large-scale language model that is compact and energy-efficient, yet achieves the same level of accuracy as ChatGPT, a huge single large-scale language model that consumes a lot of power.

[0049] Tsuzumi (registered trademark) is a large-scale language model that is compact by focusing on a high-quality corpus and supporting only English and Japanese, rather than on the amount of training data. Tsuzumi (registered trademark) also supports fine-tuning for domain specialization, and external data integration that combines search and generative AI (Artificial Intelligence).

[0050] As described above, tsuzumi (registered trademark) is a compact generative model, making it possible to create and operate multiple large-scale language models with specific specialized fields and diverse personalities. Furthermore, by linking compact large-scale language models for each specialized field, such as "medical care," "retail," "construction," "local government," "technology," "travel," "culture," "religion," "art," "education," "finance," "legal affairs," and "banking" based on a specified network, tsuzumi (registered trademark) realizes the formation of generative models that are high-performance, efficient, fault-tolerant, and highly democratic compared to conventional single huge large-scale language models.

[0051] Furthermore, tsuzumi (registered trademark) enables flexible tuning such as the fine tuning described above. Specifically, tsuzumi (registered trademark) enables "tuning by prompt engineering," "full fine tuning," and "adapter tuning" as shown in Figure 7.

[0052] For example, "prompt engineering" shown in Figure 7 (1) is a tuning method that uses prompts with added information about a specific field when setting prompts for the base model. In tuning using prompt engineering, the base model itself is not changed and only the prompts that are set are changed, so the learning cost can be reduced compared to other methods.

[0053] For example, "full fine-tuning" shown in Figure 7 (2) is a tuning method in which the base model is additionally trained or retrained using training data related to a specific field. In full fine-tuning, a tuning model specialized for a target field can be constructed by training the base model using training data for the target field. Therefore, full fine-tuning can improve the accuracy of inference and generation compared to other methods.

[0054] For example, "adapter tuning" shown in Figure 7 (3) is a tuning method that adds a submodule related to a specific field to the base model. Adapter tuning can improve the accuracy of inference and generation by fine-tuning the base model using a submodule related to the target field. Furthermore, because adapter tuning does not require re-learning of the base model, it can improve accuracy while reducing learning costs compared to full fine tuning.

[0055] Tsuzumi (registered trademark) is capable of building a "multi-adapter model" based on the various learning methods described above. As explained using (3) in Figure 7, Tsuzumi (registered trademark) allows fine-tuning of the model by adding adapters (sub-modules) to the base model. Tsuzumi (registered trademark) can also add a combination of one or more adapters to the base model.

[0056] For example, as shown in (4) of Fig. 7, tsuzumi (registered trademark) fine-tunes the foundation model using an adapter specialized for organization A, an adapter specialized for organization B, and an adapter specialized for organization C. As a result, even when data with different characteristics, such as data from organization A, organization B, and organization C, is input, tsuzumi (registered trademark) can accurately perform inference and generation processing based on the model fine-tuned by the adapter specialized for each organization.

[0057] For example, organization A could be the "research laboratory," organization B could be the "sales department," and organization C could be the "entire company." In other words, tsuzumi (registered trademark) can make fine adjustments using an adapter appropriate for the organization, even if the characteristics and granularity of the organization differ.

[0058] As described above, tsuzumi (registered trademark) is small and energy-efficient, and combines small, large-scale language models specialized for specific fields, etc., with flexible tuning, making it possible to achieve both accuracy and cost when performing inference and generation processing compared to conventional huge, large-scale language models.

[0059] In other words, tsuzumi (registered trademark) can be used as an appropriate generation model corresponding to the use, purpose, and target field of the processing, as a generation model for executing specific processing, such as the generation processing of patentable information by the invention excavation device 100 of this embodiment.

[0060] (Patentability Information DB124) The patentability information DB 124 is a database that stores patentability information, which is information relating to the patentability of an invention and is used by a user to determine the patentability of an invention. An example of the patentability information stored in the patentability information DB 124 will now be described using a table. Figure 8 is a table showing an example of the patentability information according to this embodiment.

[0061] As shown in Figure 8, patentability information DB124 stores information on the items of similarity determination, patent completeness, novelty points, inventive step points, extension points, and potentially related prior art in a table format or the like, in association with "No.", which is information for identifying individual data of patentability information.

[0062] The information included in the above items will now be explained. The similarity determination includes information on the presence or absence of prior art similar to the input invention information. The patent completeness index is an index showing the degree to which the invention related to the input invention information is complete in terms of novelty and inventive step. For example, it may be a percentage indicating the degree to which the similarity index, novelty index, inventive step index, etc., meet predetermined standards. The novelty points include information such as the presence or absence of prior art similar to the input invention as a result of comparison with prior art, and comments on the similarities. The inventive step points include information such as comments on whether the input invention has inventive step, and if not, what the problem is. The extension points include information such as comments on suggestions for further expanding the scope of the content included in the invention. The potentially relevant prior art includes the names and predetermined identification information of prior art similar to the input invention.

[0063] As an example shown in Figure 8, the patentability information DB124 can store No. "1", similarity judgment "L", patent completeness "M", novelty point "N", inventive step point "O", extension point "P", and potentially related prior art "Q" in a table format or the like, each associated with the other.

[0064] The letters "L to Q" included in each of the above-mentioned items exemplify the information of each item stored in the patentability information DB 124. For example, the content (information) of each item, such as the similarity determination "L," the patent completeness "M," the novelty score "N," the inventive step score "O," the extension score "P," and the possibly related prior art "Q," may be character string information expressed in a predetermined natural language, such as hiragana, kanji, roman letters, or numbers.

[0065] (control unit 130) Here, we will return to Fig. 3 to continue the explanation. The control unit 130 has an internal memory for temporarily storing programs that define various processing procedures and the like of the invention excavation device 100 and processing data, and is realized by electronic circuits such as a CPU (Central Processing Unit) and an MPU (Micro Processing Unit), and integrated circuits such as an ASIC (Application Specific Integrated Circuit) and an FPGA (Field Programmable Gate Array). As shown in Fig. 3, the control unit 130 has a receiving unit 131, a searching unit 132, a learning unit 133, a generating unit 134, and an output unit 135.

[0066] (Reception Department 131) The receiving unit 131 receives invention information and additional information input via an invention information receiving screen displayed on the terminal device 200 of the user.

[0067] Here, an example of an invention information reception screen displayed on the terminal device 200 operated by the user will be described with reference to Fig. 9. Fig. 9 is a diagram showing an example of an invention information reception screen according to this embodiment. Fig. 9 shows an example of an invention information reception screen ((1) in Fig. 9) displayed on the user's terminal device 200 by the invention excavation device 100.

[0068] In the example described below, it is assumed that the invention information entered into the invention information reception screen displayed on the display unit of the user's terminal device 200 is received by the terminal device 200 and transmitted from the terminal device 200 to the invention excavation device 100.

[0069] The reception unit 131 receives target files such as text files, image files, and audio files (formats: pdf, docx, txt, pptx, xlsx, jpg, wav, etc.) via the "item related to file input ((2) in FIG. 9)" on the invention information reception screen displayed on the display unit of the terminal device 200. For example, when the user selects and operates the "item related to file input ((2) in FIG. 9)", the reception unit 131 can receive the upload of the target file by separately displaying on the terminal device 200 a screen for selecting the file to be uploaded or a screen for drag-and-drop.

[0070] Furthermore, the receiving unit 131 receives invention information input in text form via the "item for inputting text ((3) in FIG. 9)" in the invention information reception screen displayed on the display unit of the terminal device 200. For example, the receiving unit 131 can receive text information that explains background art, conventional art, issues and problems, key points of the invention, effects of the invention, specific contents and processing flow of the invention, etc., input by the user via the terminal device 200 in the "item for inputting text ((3) in FIG. 9)".

[0071] The receiving unit 131 also receives information about the patent classification of the invention input by the user. In this case, the receiving unit 131 receives information about the patent classification of the invention via the "item related to the assumed patent classification (optional) ((4) in FIG. 9)" in the invention information reception screen displayed on the display unit of the terminal device 200. For example, the receiving unit 131 can receive information such as the International Patent Classification (IPC), F-terms, and information expressing the classification of the invention in a predetermined text sentence, which the user inputs via the terminal device 200 into the "item related to the assumed patent classification (optional) ((4) in FIG. 9)."

[0072] The receiving unit 131 also receives information input by the user about the aspects of the invention that the user considers to be novel and inventive. In this case, the receiving unit 131 receives information about the aspects of the invention that the user considers to be novel and inventive via an "item related to points considered to be new (optional) ((5) in FIG. 9)" in the invention information reception screen displayed on the display unit of the terminal device 200. For example, the receiving unit 131 can receive information about the aspects that the user considers to be novel and inventive, such as "completed on a smartphone," input by the user via the terminal device 200 in the "item related to points considered to be new (optional) ((5) in FIG. 9)."

[0073] The reception unit 131 receives invention information including information entered by the user in each field, which is sent from the terminal device 200 when the “Consult ((6) in Figure 9)” button on the invention information reception screen is pressed as a trigger.

[0074] (Search unit 132) Continuing the explanation, returning to Fig. 3, the search unit 132 uses search words extracted from the invention information received by the receiving unit 131 to search for prior art information.

[0075] Specifically, the search unit 132 inputs the invention information into a generative model (search word extraction LLM) that has been trained to extract search words from information input as a generative model, and extracts the search words. The search unit 132 uses the extracted search words to obtain prior art information related to the invention information from an external server 300 or the like.

[0076] Furthermore, the search unit 132 may present to the user search word candidates that satisfy a predetermined condition for the extracted search words, and allow the user to select a plurality of search words.

[0077] For example, the search unit 132 extracts important words that are included multiple times in the input invention information and presents them to the user. Then, the search unit 132 can accept one or more words selected by the user from the presented multiple words as search words to be used in searching prior art information. This allows the invention discovery device 100 (search unit 132) to search prior art information using appropriate search words even when a huge number of search words are extracted.

[0078] (Learning Section 133) The learning unit 133 performs a learning process in advance on a predetermined generative model. Specifically, the learning unit 133 additionally learns the generative model by using, as prior knowledge, prior art information related to the invention searched based on a predetermined search word extracted from the invention information.

[0079] Here, an example of the learning process of the generative model (tsuzumi (registered trademark)) by the learning unit 133 will be described with reference to Fig. 10. Fig. 10 is a diagram for explaining an example of the learning process according to this embodiment. Fig. 10 shows an example of additional learning of the base model (generative model) using prior art information (patent field data).

[0080] 10, the learning unit 133 performs additional learning on the basic model 11 using prior art information (patent field data) searched using search words extracted from the input invention information. Then, the learning unit 133 constructs the adapter tuning model 12 that has been additionally learned (adapter tuned) using the prior art information (patent field data).

[0081] As described above, the accuracy and density of information generated from the generative model are improved by additional learning performed by the learning unit 133. For example, when generating information based on the basic model 11 before additional learning, information such as "Invention A has no novelty ((1-2) in FIG. 10)" is generated (output) in response to an input such as "Determine the novelty of invention A ((1-1) in FIG. 10)."

[0082] On the other hand, when generating information based on the adapter tuning model 12 that has undergone additional learning, in response to an input such as "The novelty of invention A was determined ((2-1) in Figure 10)," information such as "A patent application related to invention A was found ((2-2) in Figure 10)" is generated (output).

[0083] 10, the base model 11 simply outputs "whether or not there is novelty," whereas the adapter tuning model 12, which has undergone additional learning using prior art information (patent field data) by the learning unit 133, outputs "whether or not there is novelty + specific information" or the like as more specialized content that conforms to the patent field. Therefore, the learning unit 133 can construct a generative model for generating more accurate patentability information by additionally learning the base generative model using prior knowledge such as patent field data.

[0084] Furthermore, the learning unit 133 can additionally train the generative model using the results of the examination of the patent application and information on responses to the examination results as prior knowledge. Specifically, the learning unit 133 can additionally train the generative model using, as prior knowledge, information on the notice of reasons for refusal notified by the Japan Patent Office, the applicant's response to the notice of reasons for refusal, and the final examination results such as a patent decision or a rejection decision.

[0085] Furthermore, the above-mentioned "results of the examination" may include the results of appeals such as appeals against the examiner's decision of refusal (appeals against examiner's decision of refusal), the results of appeals related to lawsuits to set aside appeal decisions, etc., against appeals against examiner's decision of refusal, etc. That is, the learning unit 133 can additionally train the generative model using the results of appeals related to appeals against examiner's decision of refusal, etc., and the results of appeals related to lawsuits to set aside appeal decisions, etc., as prior knowledge.

[0086] As the above-mentioned "additional learning," the learning unit 133 can perform pre-learning based on MLM (Masked Language Modeling) using information contained in prior art information such as patent publications, supervised learning using predetermined input information and output information, learning of some parameters based on LoRA (Low-Rank Adaptation), etc.

[0087] (Generation unit 134) Continuing the explanation by returning to Fig. 3, the generation unit 134 generates, based on the trained generative model, comments on similarity, comments on the novelty of the invention, comments on the inventive step of the invention, and the like as patentability information.

[0088] Specifically, the generation unit 134 inputs, as patentability information, invention information and an instruction to generate information on the similarity between the invention and prior art information (hereinafter, sometimes referred to as "similarity information") to the similarity determination LLM, thereby generating the similarity information. Here, the similarity determination LLM is an example of a generative model trained to generate similarity information between an invention and a disclosed technology.

[0089] Furthermore, the generation unit 134 inputs the invention information and instructions for generating information on the novelty of the invention (hereinafter, may be referred to as "novelty information") and information on the inventive step of the invention (hereinafter, may be referred to as "inventive step information") as patentability information into the patentability extraction LLM, thereby generating at least one of the novelty information and the inventive step information related to the invention information. Here, the patentability extraction LLM is an example of a generative model trained to generate novelty information and inventive step information as a generative model.

[0090] The above-mentioned "novelty information" includes, as a predetermined result of comparing the invention with the prior art, information identifying prior art similar to the invention, information expressing in natural language sentences the similar parts and contents between the invention and the prior art, and a predetermined index (indicator of novelty) regarding the similarity between the invention and the prior art.

[0091] Here, the novelty index may be determined based on the similarity of character strings contained in the invention information and the prior art information. For example, the higher the similarity, the more commonalities or similarities there are between the invention and the prior art, resulting in a "lower novelty index." Conversely, the lower the similarity, the fewer commonalities or similarities there are between the invention and the prior art, resulting in a "higher novelty index." For example, the invention discovery device 100 can determine that an invention has novelty if the novelty index exceeds a predetermined threshold. The similarity may be a known similarity index such as minimum edit distance, Jaro-Winkler distance, cosine similarity, word transport distance, or word rotation distance.

[0092] Furthermore, the above-mentioned "inventive step information" includes, as a predetermined result of comparing the invention with the prior art, information identifying the prior art used for comparison with the invention, information expressing in natural language sentences the parts and contents that have similarities and differences when comparing the invention with the prior art, information for expanding the content of the invention, and predetermined indicators of the differences between the invention and the prior art (indicators of inventive step), etc.

[0093] Here, the indicator of inventive step may be, for example, the ratio of the calculated amount of difference to the total amount of invention information generated by the generation unit 134, as an indicator of the difference when comparing invention information with prior art information.

[0094] For example, the generation unit 134 calculates the difference between the prior art and the invention as described above. Then, the generation unit 134 generates, as an index of inventive step, a ratio of the calculated amount of difference to the total amount of invention information. Here, the higher the ratio of the amount of difference, the more the invention differs from the prior art (the more advanced it is), and therefore the "higher the index of inventive step." On the other hand, the lower the ratio of the amount of difference, the more the invention differs from the prior art (the more advanced it is), and therefore the "lower the index of inventive step." For example, the invention discovery device 100 can determine that an invention has an inventive step if the index of inventive step exceeds a predetermined threshold.

[0095] An example of the information generation process by the generation unit 134 will now be described with reference to Fig. 11 and Fig. 12. Fig. 11 and Fig. 12 are diagrams illustrating an example of the generation process according to this embodiment.

[0096] First, an example of generating "similarity information" will be described with reference to Fig. 11. Fig. 11 shows the similarity determination LLM 13 used to generate the similarity information, as well as the input information and output information for the similarity determination LLM 13.

[0097] For example, the generation unit 134 inputs the invention information “A system that allows you to create patent specifications using a smartphone,” the command “Please determine whether the idea is similar to an existing patent,” and the searched prior art information into the similarity determination LLM 13 ((1) in Figure 11).

[0098] The generation unit 134 then performs a similarity determination based on the input information. Specifically, the generation unit 134 calculates a predetermined similarity (for example, a known similarity index such as minimum edit distance, Jaro-Winkler distance, cosine similarity, word transport distance, or word rotation distance) calculated based on the similarity determination LLM. The generation unit 134 then performs a similarity determination based on whether the calculated similarity exceeds a predetermined threshold. For example, the generation unit 134 can determine that the two are "similar" if the calculated similarity exceeds a threshold, and determine that the two are "not similar" if the calculated similarity is equal to or less than the threshold.

[0099] The generating unit 134 generates the result of the similarity determination described above, "similar or dissimilar," and, if similar, the likely related patent, "Japanese Patent Application Laid-Open No. xxxxx," as similarity information ((2) in FIG. 11).

[0100] Next, an example of generating "novelty information" and "inventive step information" will be described with reference to Figure 12. Figure 12 shows the patentability extraction LLM 14 used to extract patentability, as well as the input and output information for the patentability extraction LLM 14.

[0101] For example, the generation unit 134 inputs the invention information "A system that allows the creation of patent specifications using a smartphone," the command "Please extract information on novelty and inventive step," and the retrieved prior art information into the patentability extraction LLM 14 ((1) in Figure 12), and generates novelty information and inventive step information ((2) in Figure 12).

[0102] Specifically, the generation unit 134 generates novelty points (novelty information) such as "The technology related to XX has not been found in existing patents," inventive step points (inventive step information) such as "The technology related to XX may be inventive," and extension points (inventive step information) such as "This is a smartphone technology, but can it also be applied to telephones?" as novelty information and inventive step information.

[0103] The generation unit 134 can generate extension points including synonyms of words related to the invention based on a database of synonyms that associate similar words, etc. The generation unit 134 may also generate synonyms for words related to the invention included in the invention information based on a generation model that is configured to learn and prompt so as to output synonyms for an input word.

[0104] Furthermore, if there is prior art information similar to the invention related to the input invention information in excess of a preset threshold, the generation unit 134 can narrow down the information interactively with the user. For example, the generation unit 134 can generate novelty information and inventive step information in which similar prior art information has been narrowed down based on additional information input by the user in response to an interactive screen displayed by the output unit 135 (described later) that includes content such as "Please enter conditions to further narrow down the discovered prior art" or "Currently, technologies related to ●● and XX have been discovered. Please let us know if there is any technology that is not similar to the invention."

[0105] (output unit 135) 3, the explanation will be continued. The output unit 135 outputs the patentability information generated by the generation unit 134. For example, the output unit 135 outputs similarity information, novelty information, inventive step information, etc., generated as patentability information by the generation unit 134 to the terminal device 200 or the like operated by the user.

[0106] Specifically, when the novelty index included in the novelty information is equal to or less than a predetermined threshold, the output unit 135 outputs to the user the portions of the novelty information in which the invention is similar to the prior art, the patent publication, and a comment on the novelty. For example, the output unit 135 outputs to the user a message included in the novelty information generated by the generation unit 134, such as "XX of the invention may be similar to ●● of JP XXXXX," or "Please let us know if there are any differences from the prior art." Note that the information identifying the patent publication, such as "JP XXXXX," may be a link or the like that directly transitions to the document.

[0107] On the other hand, when the novelty index exceeds a predetermined threshold, the output unit 135 can output a message such as "Technology related to xx was not found in existing patents" included in the novelty information generated by the generation unit 134. Furthermore, when the novelty index exceeds a predetermined threshold, the output unit 135 may select processing such as not displaying a message.

[0108] If the inventive step index included in the inventive step information is equal to or less than a predetermined threshold, the output unit 135 outputs to the user the similarities between the invention and the prior art included in the inventive step information, the corresponding patent publication, and a comment on the inventive step. For example, the output unit 135 outputs to the user a message included in the inventive step information generated by the generation unit 134, such as "There may be no difference between the XX of the invention and the ●● of JP XXXXX," or "Please let us know if there are any differences from the prior art." Note that the information identifying the patent publication, such as "JP XXXXX," may be a link or the like that directly transitions to the document.

[0109] On the other hand, if the inventive step index exceeds a predetermined threshold, the output unit 135 outputs to the user a comment regarding the expansion of the content of the invention, including synonyms, general phrases, etc. to be presented to the user. For example, the output unit 135 can output a message such as "The technology related to xx may involve an inventive step," which is included in the inventive step information generated by the generation unit 134.

[0110] Furthermore, the output unit 135 can output messages about extension points, which are suggestions for changing expressions that lead to limitations on the content of the invention, changing expressions to more general-purpose phrases, etc. For example, the output unit 135 can output messages about extension points such as "Can you change a smartphone into a portable device?" or "This technology is ____, but can it also be applied to ____?"

[0111] Furthermore, the output unit 135 displays a screen to the user prompting the user to input additional information that allows the generation unit 134 to narrow down the search for similar prior art information related to novelty information or inventive step information.

[0112] (An example of reception and output processing) An example of the reception process and output process of the generated patentability information will now be described with reference to Fig. 13. Fig. 13 is a diagram showing an example of the reception process and output process according to this embodiment. Fig. 13 shows an invention information input screen 20, an additional information input screen 21, a similarity information display screen 22, and a patentability information display screen 23. Note that each of the above-mentioned output screens is displayed on the user's terminal device 200, and information input is also performed via the terminal device 200.

[0113] First, the receiving unit 131 receives input of invention information via the invention information input screen 20. If the input invention information does not include content related to predetermined items, or if the number of characters is below a predetermined reference value, the output unit 135 displays the additional information input screen 21 for the user.

[0114] 13 is a screen that suggests input of additional information to the user. For example, the output unit 135 displays a question such as "Please tell me more about the XX technology in your idea" to the user. The receiving unit 131 then receives the additional information input via the input fields on the additional information input screen 21.

[0115] The output unit 135 can display a question to the user using a question template that is pre-registered in a database, etc. The output unit 135 may also generate a question based on a generative model that is configured to execute learning and prompts so as to generate a question that suggests additional input of information according to the input invention information.

[0116] Next, the generation unit 134 generates similarity information based on the similarity determination LLM using the input invention information and additional information. Then, the output unit 135 outputs a display screen (similarity information display screen 22) for the generated similarity information to the user.

[0117] For example, if prior art information similar to the input invention information is present, the output unit 135 outputs to the user a message such as "We have found prior art that appears to be related. Is it similar to your invention?", along with the similarity information display screen 22, which displays prior art identification information, information such as an outline or summary, transition destination information, etc., and buttons for the user to select whether the displayed prior art information is similar to the invention. The receiving unit 131 then receives the user's selection of "similar" or "not similar" and stores it in the storage unit in association with the searched prior art information.

[0118] That is, the invention mining device 100 generates similarity information between an invention and prior art and presents the information to the user in an interactive manner with the user.

[0119] Next, the generation unit 134 generates patentability information based on the patentability extraction LLM using the input invention information and additional information. The output unit 135 then outputs the generated patentability information display screen 23 to the user. For example, the output unit 135 displays patentability information on the patentability display screen 23, such as a patent completeness rate of "68%," a novelty point of "technologies related to xx were not found in existing patents," an inventive step point of "technologies related to xx may be inventive," an extension point of "this is smartphone technology, but can it be applied to telephones as well?", and a potentially related patent, such as "JP Patent Publication No. xxxxx."

[0120] Furthermore, the output unit 135 can display a "brush-up button" on the patentability display screen 23. Pressing the "brush-up button" triggers the invention discovery device 100 to interactively suggest and accept input of additional information from the user. That is, the invention discovery device 100 generates novelty information and inventive step information and presents the information to the user interactively.

[0121] Specifically, when the "brush up button" displayed on the patentability display screen 23 is pressed, the output unit 135 outputs a screen for the user to input additional information, such as the additional information input screen 21 ((1) in Figure 13).

[0122] Next, the receiving unit 131 receives additional information input by the user. Then, the generating unit 134 inputs the additional information input by the user regarding the patentability information output by the output unit 135 and an instruction to generate the patentability information into a generative model to generate updated patentability information.

[0123] As described above, the invention excavation device 100 can repeatedly generate novelty information and inventive step information about an invention through an interactive format with the user and present the information to the user, thereby brushing up the information.

[0124] Note that the questions displayed on the additional information input screen 21 shown in FIG. 13 are merely examples, and for example, the output unit 135 can display questions to the user based on example questions stored in advance in a memory unit or the like.

[0125] Furthermore, when displaying the above-mentioned brush-up button, the output unit 135 determines whether or not a brush-up suggestion is possible for the input invention information. If the output unit 135 determines that a brush-up suggestion is possible, the output unit 135 can display the brush-up button to the user. On the other hand, if the output unit 135 determines that a brush-up suggestion is not possible, the output unit 135 can hide the brush-up button.

[0126] (Terminal device 200) The terminal device 200 is an information processing terminal device operated by a user. Specifically, the terminal device 200 receives invention information from a user, transmits the received invention information to the invention excavation device 100, and receives and displays patentability information and the like from the invention excavation device 100. The type of the terminal device 200 is not particularly limited, and may be, for example, a smartphone, a tablet, a PDA (Personal Digital Assistant), a PC (Personal Computer), a notebook PC, or the like.

[0127] (Server 300) The server 300 is an external information processing device that stores prior art information, etc. The type of information processing device that realizes the server 300 is not particularly limited, and may include, for example, a cloud-based server, an on-premise server, etc.

[0128] (Procedure for Invention Discovery Processing) Next, the procedure of the invention excavation process realized by the invention excavation device 100 according to this embodiment will be described with reference to Fig. 14. Fig. 14 is a diagram showing a flowchart of the invention excavation process according to this embodiment.

[0129] The receiving unit 131 receives invention information (S101). Next, the searching unit 132 extracts search words from the invention information (S102). Then, the searching unit 132 searches prior art information using the extracted search words (S103).

[0130] Here, if the generative model is to be trained (Yes in S104), the training unit 133 trains the generative model using the retrieved prior art information as prior knowledge (S105). On the other hand, if the generative model is not to be trained by the invention mining device 100 (No in S104), for example, if the generative model has been trained in advance by another information processing device, the invention mining device 100 skips the process.

[0131] The generation unit 134 inputs the invention information and the patentability information generation command into the trained generation model to generate patentability information (S106). Subsequently, the output unit 135 outputs the generated patentability information to the user (S107). Then, the invention discovery device 100 ends the process.

[0132] (effect) Next, we will explain the effects of the invention discovery device 100 according to this embodiment. Conventionally, when a user lacks knowledge or insight, it is difficult to search for appropriate prior art, which can make it difficult to consider filing an application. Furthermore, even when an inventor does not have the time to personally discover an invention from a vast amount of information, or when expert advice is needed to discover an invention or consider filing a patent application, conventional technology can make it difficult to present appropriate information.

[0133] Therefore, the receiving unit 131 of the invention excavation device 100 according to this embodiment receives invention information. The generating unit 134 of the invention excavation device 100 inputs the invention information received by the receiving unit 131 and a command to generate patentability information into a generative model trained on prior art information, thereby generating patentability information related to the invention information. The output unit 135 of the invention excavation device 100 outputs the patentability information generated by the generating unit 134.

[0134] Therefore, the invention discovery device 100 according to this embodiment generates patentability information using information such as development concept documents, specifications, idea memos, etc., making it possible to extract technical elements that are likely to be patentable and clarify the differences between prior art and inventions, etc. As a result, the invention discovery device 100 has the effect of making it possible to easily consider filing an application.

[0135] Furthermore, the invention excavation device 100 according to this embodiment achieves predetermined effects by executing the processes described below.

[0136] The generation unit 134 inputs the invention information and an instruction to generate similarity information about the similarity between the invention and prior art as patentability information into a generation model that has been trained to generate similarity information about the similarity between the invention and prior art, thereby generating similarity information.

[0137] As described above, the invention discovery device 100 generates information on whether the input invention information is similar to prior art information, thereby making it easier for users to understand whether there is any technology similar to their own invention and to consider whether to file a patent application for the invention.

[0138] The generation unit 134 inputs the invention information and an instruction to generate at least one of novelty information and inventive step information as patentability information into a generation model trained to generate at least one of novelty information and inventive step information, thereby generating at least one of novelty information and inventive step information related to the invention information.

[0139] As described above, the invention discovery device 100 generates information on whether the input invention information has novelty and inventive step by comparing it with prior art information, thereby making it easier for users to understand for themselves whether the invention has novelty and inventive step, and thereby to consider whether to file a patent application for the invention.

[0140] If the novelty index included in the novelty information is equal to or less than a predetermined threshold, the output unit 135 outputs to the user the portions where the prior art and the invention are similar, the patent publications, and comments regarding the novelty of the invention.

[0141] As described above, the invention discovery device 100 allows the user to easily grasp the similarities between prior art information and an invention, and to easily grasp the related prior art information itself. Furthermore, the invention discovery device 100 not only presents the similarities and related prior art information but also outputs comments in natural language to the user, thereby enabling the user to grasp the outline of the novelty of the invention information they input with little effort.

[0142] When the inventive step index included in the inventive step information is equal to or less than a predetermined threshold, the output unit 135 outputs to the user the portions where the prior art and the invention are similar, the relevant patent publications, and a comment on the inventive step of the invention. When the inventive step index exceeds a predetermined threshold, the output unit 135 outputs to the user a comment on the extension of the content of the invention.

[0143] As described above, the invention discovery device 100 allows the user to easily determine whether the input invention information has an inventive step by comparing it with prior art information, and to easily understand the related prior art information itself.

[0144] Furthermore, the invention discovery device 100 not only presents whether an invention has an inventive step or not and related prior art information, but also outputs comments in natural language to the user, thereby enabling the user to grasp an overview of the inventive step of the invention information they have input with little effort. Also, the invention discovery device 100 presents the user with comments regarding the expansion of the content of the invention, allowing the user to further refine the content of the invention.

[0145] The output unit 135 outputs to the user the patentability information generated by the generation unit 134. The generation unit 134 inputs additional information input by the user regarding the patentability information output by the output unit 135 and an instruction to generate the patentability information into a generative model to generate updated patentability information.

[0146] Specifically, the invention discovery device 100 can use prior art information collected from external servers, etc., and patentability information generated based on input internal, undisclosed information, etc., to conduct invention discovery, application consideration, invention brush-up, etc. in a bouncing-back format with users such as inventors and intellectual property officers. For example, the invention discovery device 100 realizes advice such as "You can apply for a patent if you make XX more specific," which was previously given based on the experience and intuition of intellectual property officers, through an interactive format between the inventor and the device.

[0147] The invention discovery device 100 then clarifies the differences between the invention extracted from input information such as development concept documents, specifications, idea memos, etc. and the prior art, and discovers technical elements that can be used in applications.

[0148] Therefore, the invention discovery device 100 can reduce the work required for the user to discover inventions. The invention discovery device 100 can also identify invention ideas that users tend to overlook and consider appropriate scope of rights. The invention discovery device 100 can also automatically perform invention discovery and application consideration, which have traditionally been performed by users relying on their experience and intuition, and can provide advice to users on "points that should be specified in the invention," etc.

[0149] The "hitting-against-the-wall format" refers to a series of interactive processes between the user and the invention discovery device 100, such as the user inputting invention information into the invention discovery device 100, the invention discovery device 100 generating patentability information according to the inputted invention information and outputting it to the user, and the user then inputting additional information according to the outputted patentability information.

[0150] The search unit 132 searches for prior art information related to the invention based on a predetermined search word extracted from the invention information received by the reception unit 131. Then, the learning unit 133 additionally learns the generative model using the searched prior art information related to the invention as prior knowledge.

[0151] As described above, the invention discovery device 100 searches for prior art information related to the input invention information, and uses the prior art information corresponding to the invention information to perform additional learning on the generative model, thereby constructing a generative model specialized for the input invention information. Therefore, compared to the case where a general-purpose generative model is used, the invention discovery device 100 can generate appropriate patentability information according to the input invention information.

[0152] The learning unit 133 additionally learns the generative model using the examination results for the patent application and information on responses to the examination results as prior knowledge.

[0153] As described above, when searching for prior art information related to input invention information, the invention discovery device 100 can additionally train a generative model using the examination results of the patent application invention, the response results to the examination results, the trial results, and the response results to the trial results.The invention discovery device 100 then constructs a generative model that is specialized for the input invention information and can generate information on the novelty and inventive step related to the patent application examination.Therefore, compared to using a general-purpose generative model, the invention discovery device 100 can generate appropriate patentability information corresponding to the input invention information.

[0154] Furthermore, the invention discovery device 100 generates patentability information based on a generative model, making it possible to present appropriate information to various users who are discovering inventions, considering filing applications, and the like.

[0155] For example, patent attorneys and patent attorney corporations that represent clients in filing applications to the Japan Patent Office require a large amount of man-hours to discover inventions from clients and examine the patentability of the discovered inventions, etc. Therefore, the invention discovery device 100 can improve the efficiency and optimize the work of patent attorneys and patent attorney corporations by automatically generating patentability information used to discover inventions and examine patentability.

[0156] Furthermore, for companies with intellectual property departments, reducing the costs and man-hours involved in patent applications can be an issue. Therefore, the invention discovery device 100 automatically generates patentability information used for invention discovery and patentability examination, thereby enabling the selection and concentration of work based on the judgment of the priority of each case, thereby achieving the effect of reducing costs and man-hours.

[0157] Furthermore, small and medium-sized enterprises, etc., are often unable to file patent applications appropriately due to the high cost burden of patent applications, making it difficult to protect their proprietary technologies. Therefore, the invention discovery device 100 automatically generates patentability information used for invention discovery and patentability review, thereby enabling invention discovery, application review, draft creation, etc. to be done in-house. Therefore, the invention discovery device 100 has the effect of enabling small and medium-sized enterprises, etc., to reduce application costs and labor costs.

[0158] Furthermore, conventionally, when performing a keyword search, there are problems such as not obtaining the intended search results or obtaining unintended search results (noise). Therefore, the invention discovery device 100 according to this embodiment performs a search using appropriate search words obtained based on the search word LLM, and further narrows down prior art information similar to invention information based on the similarity judgment LLM, thereby making it possible to present appropriate prior art information.

[0159] <Modification> Below, modifications realized by the invention excavation device 100 according to this embodiment will be described.

[0160] (Data, etc.) The invention information, additional information, prior art information, similarity information, novelty information, inventive step information, patentability information, search word extraction LLM, search word, similarity determination LLM, patentability extraction LLM, names of functional parts of the invention discovery device 100, steps, processes, names of steps or processes, etc. used in the description of the above embodiments are merely examples and can be changed as desired.

[0161] For example, the prior art information DB121 stores the following items in association with "No.", which is information that identifies individual prior knowledge: background art, problem that the invention aims to solve, means for solving the problem, effect of the invention, form for implementing the invention, and drawings; however, the items stored are not limited to the above-mentioned items.

[0162] For example, the invention information DB122 stores the items of file information, text information, expected patent classification, points that are thought to be new, and additional information, in association with "No.", which is information for identifying individual invention information, but the items stored are not limited to the items mentioned above.

[0163] For example, patentability information DB124 stores information on the items of similarity determination, patent completeness, novelty points, inventive step points, extension points, and potentially related prior art, in association with "No.", which is information for identifying individual patentability information, but the items stored are not limited to the items mentioned above.

[0164] (About "learning") In this embodiment, "learning" includes unsupervised learning, supervised learning, incremental learning, reinforcement learning, fine tuning, adapter tuning, etc., for a given generative model. In addition, the above-mentioned "learning" also includes providing prior knowledge using prompts, input, data collection commands, etc., for the generative model.

[0165] (About generative models) In this embodiment, an example has been described in which the "generative model" is stored in and used by the invention excavation device 100, but this is not limiting. For example, the generative model may be stored in and used by an information processing device or server other than the invention excavation device 100 according to this embodiment.

[0166] In this embodiment, an example using "tsuzumi (registered trademark)" as a generative model has been described, but the invention discovery device 100 can use a large-scale language model such as ChatGPT (registered trademark) as a generative model (see, for example, Reference 2).

[0167] (Reference 2):ChatGPT(OpenAI),<URL:https: / / openai.com / chatgpt> ,<Searched on June 5, 2020>

[0168] (About the threshold) In this embodiment, it has been described that threshold values ​​are used for similarity determination, determination of novelty (novelty determination), and determination of inventive step (inventive step determination). These threshold values ​​may be set to any value by the user, etc. In other words, the threshold values ​​may be appropriately tuned according to the strictness of the determination desired by the user.

[0169] (Flowcharts, etc.) The steps in the flowcharts may be interchanged as long as there is no contradiction, and some steps may not be performed. In addition, conjunctions such as "next," "continue," "further," "at this time," and "on this occasion" used in the explanation of the flowcharts do not limit the order or timing of the execution of the processes in the flowcharts.

[0170] <Hardware configuration> The components of each device shown in the figure are conceptual functional units and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown, and all or part of each device can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc. Furthermore, all or any part of the processing functions performed by each device can be realized by a CPU and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic.

[0171] Furthermore, among the processes described in this embodiment, all or part of the processes described as being performed automatically can also be performed manually using known methods. In addition, the information including the processing procedures, control procedures, specific names, various data, and parameters shown in the drawings can be changed as desired unless otherwise specified.

[0172] <Program> In one embodiment, the various devices constituting the invention discovery device 100 can be implemented by installing an invention discovery program as package software or online software on a desired computer. For example, by executing the invention discovery program on an information processing device, the various devices constituting the invention discovery device 100 can function. The information processing device referred to here includes desktop and notebook personal computers. Other information processing devices also include mobile communication terminals such as smartphones and mobile phones, as well as slate terminals such as PDAs (Personal Digital Assistants).

[0173] 15 is a diagram showing an example of a computer that executes the invention excavation process according to this embodiment. The computer 1000 includes, for example, a memory 1010 and a CPU 1020. The computer 1000 also includes a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These components are connected by a bus 1080.

[0174] The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM 1012. The ROM 1011 stores, for example, a boot program such as a BIOS (Basic Input Output System). The hard disk drive interface 1030 is connected to a hard disk drive 1090. The disk drive interface 1040 is connected to a disk drive 1100. A removable storage medium such as a magnetic disk or optical disk is inserted into the disk drive 1100. The serial port interface 1050 is connected to, for example, a mouse 1110 and a keyboard 1120. The video adapter 1060 is connected to, for example, a display 1130.

[0175] The hard disk drive 1090 stores, for example, an OS (Operating System) 1091, an application program 1092, a program module 1093, and program data 1094. That is, the programs that define the processes of the various devices that make up the invention excavation device 100 are implemented as program modules 1093 in which computer-executable code is written. The program modules 1093 are stored, for example, in the hard disk drive 1090. For example, the program modules 1093 for executing processes similar to the functional configurations of the various devices that make up the invention excavation device 100 are stored in the hard disk drive 1090. The hard disk drive 1090 may be replaced with an SSD (Solid State Drive).

[0176] Furthermore, setting data used in the processing of the above-described embodiment is stored as program data 1094, for example, in the memory 1010 or the hard disk drive 1090. Then, the CPU 1020 reads the program module 1093 or the program data 1094 stored in the memory 1010 or the hard disk drive 1090 into the RAM 1012 as necessary, and executes the processing of the above-described embodiment.

[0177] The program module 1093 and program data 1094 are not limited to being stored in the hard disk drive 1090, but may also be stored in, for example, a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. Alternatively, the program module 1093 and program data 1094 may be stored in another computer connected via a network (such as a LAN or a WAN (Wide Area Network)). The program module 1093 and program data 1094 may then be read by the CPU 1020 from the other computer via the network interface 1070.

[0178] <Other> Although the present embodiment has been described above, the present embodiment is not limited by the descriptions and drawings that form part of the disclosure. In other words, other embodiments, examples, operational techniques, etc. that are made by those skilled in the art based on the present embodiment are all included in the scope of the present embodiment. [Explanation of symbols]

[0179] 100 Invention Excavation Device 110 Communications Department 120 Storage section 121 Prior art information DB 122 Invention Information DB 123 Generative Model DB 124 Patentability information DB 130 control section 131 Reception 132 Search Section 133 Learning Department 134 Generation part 135 Output section 200 Terminal Device 300 servers

Claims

1. a reception unit for receiving information related to the invention; a generation unit that inputs the information about the invention received by the reception unit and an instruction to generate information about the patentability of the invention into a generative model trained using information about the disclosed technology, and generates information about the patentability of the invention related to the information about the invention; an output unit that outputs information regarding the patentability of the invention generated by the generation unit; An inventive excavation device characterized by having:

2. The generation unit Information regarding the patentability of the invention: inputting information about the invention and an instruction to generate information about the similarity between the invention and the disclosed technology into a generative model trained to generate information about the similarity between the invention and the disclosed technology, thereby generating information about the similarity; 2. The excavation device according to claim 1.

3. The generation unit Information regarding the patentability of the invention: inputting the information about the invention and an instruction to generate information about the novelty of the invention or information about the inventive step of the invention into the generative model trained to generate information about the novelty of the invention or information about the inventive step of the invention, thereby generating information about the novelty of the invention or information about the inventive step of the invention related to the information about the invention; 2. The excavation device according to claim 1.

4. The output unit If the novelty index contained in the information on the patentability of said invention is below a predetermined threshold, outputting to the user the patent publications and portions of the disclosed technology that are similar to the invention, as well as comments on the novelty of the invention; 2. The excavation device according to claim 1.

5. The output unit If the indicator of inventive step contained in the information on the patentability of said invention is below a predetermined threshold, outputting to the user the similarities between the disclosed technology and the invention, the relevant patent publications, and comments on the inventive step of the invention; If the indicator of inventive step exceeds a predetermined threshold, outputting comments to the user regarding the extension of the content of the invention; 2. The excavation device according to claim 1.

6. The output unit outputting to a user the information regarding the patentability of the invention generated by the generating unit; The generation unit inputting additional information input by the user regarding the information on the patentability of the invention output by the output unit and an instruction to generate information on the patentability of the invention into the generative model to generate updated information on the patentability of the invention; 6. An excavation device according to claim 1, wherein the excavation device is a device for excavating a material.

7. a learning unit that additionally learns the generative model using, as prior knowledge, information about the disclosed technology related to the invention searched based on a predetermined search word extracted from the information about the invention received by the receiving unit; 6. An excavation device according to claim 1, wherein the excavation device is a device for excavating a material.

8. The learning unit additionally training the generative model using the results of the patent application examination and information regarding responses to the results of the examination as prior knowledge; 8. The excavation device of claim 7.

9. An invention excavation method executed by an invention excavation device, a receiving step of receiving information regarding the invention; a generating step of inputting the information about the invention received in the receiving step and an instruction to generate information about the patentability of the invention into a generative model trained with information about the disclosed technology, thereby generating information about the patentability of the invention related to the information about the invention; an output step of outputting information regarding the patentability of the invention generated by the generation step; The invention relates to an excavation method comprising:

10. a receiving step of receiving information related to the invention; a generating step of inputting the information about the invention received in the receiving step and an instruction to generate information about the patentability of the invention into a generative model trained with information about the disclosed technology, thereby generating information about the patentability of the invention related to the information about the invention; an output step of outputting information regarding the patentability of the invention generated by the generating step; An invention discovery program that runs on a computer.

Images