Patent application creation system and patent examination system

Abstract

This patent application creation system (11) comprises: a separation unit (31) which, from a description of an advantage resulting from differences between the prior art and an invention, separates common features that are common to the prior art and the invention and which outputs common-feature data (39c) specifying the common features; a requirement description unit (32) which acquires the common-feature data (39c) and which describes functions of the common features and the relationship between the common features on the basis of an inference model; a presentation unit (41) which acquires common-requirement data (42) specifying the functions and the relationship and which presents the functions and the relationship so as to be perceivable by human senses; and a claim description unit (33) which, in response to the presentation of the functions and relationship, acquires association data (44a) describing the relationship between the common features and the differences and which describes claims on the basis of the common-requirement data (42) and the association data (44a).

Description

Patent application document preparation system and patent examination system

[0001] The present invention relates to a patent application document preparation system and a patent examination system using machine learning.

[0002] Patent Document 1 discloses a system for preparing patent application documents based on machine learning such as deep learning. A rule-based algorithm is combined with machine learning. A format of a patent specification is set in the rule-based algorithm. When claims are input to the system, the patent application documents are output according to an inference model. The claims are created by a person.

[0003] Japanese Patent Publication No. 2020-510270

[0004] In deep learning, it is presumed that patent publications are used as learning data. AI (Artificial Intelligence) reads description patterns from a large number of patent publications in accordance with the format of patent specifications. The AI constructs an inference model based on the patterns. The inference model forms an algorithm that generates descriptions of a patent specification from input claims. The descriptions of the patent specification must accurately reflect the technology and the patent requirements. As a result, the inference model becomes complicated. If the inference model becomes complicated, the language generation processing is delayed, and the load on the executing processor increases.

[0005] An object of the present invention is to provide a patent application document preparation system and a patent examination system that can reduce the load on a processor.

[0006] According to a first embodiment of the present invention, a patent application document creation system is provided, comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common items common to the prior art and the invention from the description of the advantages and outputs common item data that identifies the common items; a requirements description unit that acquires the common item data and describes the functions of the common items and the relationships between the common items based on an inference model; a presentation unit that acquires common requirements data that identifies the functions and relationships and presents the functions and relationships in a way that can be perceived by human senses; and a claim description unit that, in response to the presentation of the functions and relationships, acquires linking data describing the relationships between the common items and the differences and describes claims based on the common requirements data and the linking data.

[0007] In the patent application document creation system, a framework is established for describing claims. The separation unit separates common elements based on descriptions of differences and advantages according to the framework. The requirements description unit describes the functions of the common elements and the relationships between them according to the framework. The claim description unit describes the claims based on the functions of the common elements and the relationships between them, while linking differences to the common elements according to the framework. The AI ​​(artificial intelligence) only needs to complete standard language generation tasks such as separating common elements, describing functions, and describing relationships. The establishment of complex inference models based on learning from a large number of patent publications is avoided. The processing burden of language generation is reduced. Processing is accelerated. Parallel processing is accelerated. The number of processing cases per processor increases. Moreover, because it does not rely on complex inference models based on learning from a large number of patent publications, fluctuations in language generation are minimized. Accurate descriptions are established with standard language generation tasks.

[0008] According to a second embodiment of the present invention, a patent application document creation system is provided, comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common terms common to the prior art and the invention from the description of the advantages and outputs common term data that identifies the common terms; a requirements description unit that acquires the common term data and describes the functions of the common terms and the relationships between the common terms based on an inference model; and a claim description unit that acquires common requirements data that identifies the functions and relationships and describes claims by combining the differences with the common terms based on the description of the advantages.

[0009] In a patent application document creation system, a framework is established for describing claims. The separation unit separates common elements based on descriptions of differences and advantages according to the framework. The requirements description unit describes the functions of the common elements and the relationships between them according to the framework. The claim description unit combines the differences with the common elements according to the framework to describe the claims. The AI ​​(artificial intelligence) only needs to complete standard language generation tasks such as separating common elements, describing functions, and describing relationships. The establishment of complex inference models based on learning from a large number of patent publications is avoided. The processing burden of language generation is reduced. Processing is accelerated. Parallel processing is accelerated. The number of processing cases per processor increases. Moreover, because it does not rely on complex inference models based on learning from a large number of patent publications, fluctuations in language generation are minimized. Accurate descriptions are established with standard language generation tasks.

[0010] According to a third embodiment of the present invention, a patent examination system is provided comprising: an acquisition unit that receives advantage data from an input device describing advantages arising from the differences between the prior art and the invention; an analysis unit that compares the advantages with the nearest technology and identifies the difference between the nearest technology and the advantages; a superiority determination unit that acquires difference data and advantage data that identify the difference, identifies the effect arising from the difference separated from the nearest technology based on an inference model, compares the effect with the advantages and determines whether the effect is within expectations; and an examination unit that, upon acquiring determination data that recognizes the effect as within expectations, denies the inventive step of the invention.

[0011] In the patent examination system, a framework is established for the examination of patent applications. The analysis unit identifies the differences between the nearest technology and its advantages according to the framework. The advantage determination unit identifies the effects resulting from the differences separated from the nearest technology according to the framework. The effects are compared with the advantages. Generally, the effects of the differences do not contribute to satisfying the requirement of inventive step (non-obviousness). If an unexpected effect is found in the advantages, the inventive step (non-obviousness) of the invention is affirmed. Thus, the inventive step (non-obviousness) of the invention is judged by the advantage determination unit. Objectivity of the examination is ensured. Machine learning is used in the examination of inventive step (non-obviousness). The AI ​​only needs to perform standard language generation tasks such as identifying differences and their effects, and comparing advantages and effects. The establishment of complex inference models based on learning from patent publications and examination result notices (notices of reasons for rejection) is avoided. The processing burden of inference is reduced. Processing is accelerated. Parallel processing is accelerated. The number of cases processed per processor increases.

[0012] A patent examination system is provided comprising: an acquisition unit that receives advantage data from an input device describing advantages arising from differences between the prior art and the invention; an analysis unit that compares the advantages with the nearest technology and identifies the difference between the nearest technology and the advantages; a feasibility determination unit that acquires difference data and advantage data that identify the difference, describes a mechanism from the difference to the advantages based on an inference model, and determines whether the mechanism is feasible; and an examination unit that, upon acquiring determination data that acknowledges the feasibility of the mechanism, affirms the inventive step of the invention.

[0013] In the patent examination system, a framework is established for the examination of patent applications. The analysis unit identifies the differences between the nearest technology and its advantages according to the framework. The feasibility determination unit describes the mechanism from the differences to the advantages according to the framework. If the mechanism is valid, the advantage is judged to be a reasonable claim. Furthermore, if an unexpected effect is found in the advantage, the inventive step (non-obviousness) of the invention is affirmed. Even if an unexpected effect is found in the advantage, if the mechanism is not valid, the advantage is judged to be an unreasonable claim, and the inventive step (non-obviousness) of the invention is denied. Thus, the inventive step (non-obviousness) of the invention is determined by the feasibility determination unit. Objectivity of the examination is ensured. Machine learning is used in the examination of inventive step (non-obviousness). The AI ​​only needs to complete standard language generation tasks such as identifying differences and their effects and describing mechanisms. The establishment of complex inference models based on learning from patent publications and examination result notices (notices of reasons for rejection) is avoided. The processing burden of inference is reduced. Processing is accelerated. Parallel processing is accelerated. The number of transactions processed per processor will increase.

[0014] According to a fifth embodiment of the present invention, a patent examination system is provided comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common items common to the prior art and the invention from the description of the advantages and outputs common item data that identifies the common items; a search unit that acquires the common item data, searches a prior art database based on the common items, and extracts prior art containing the common items; an analysis unit that, based on the advantage data, identifies the prior art with the largest range of common items as the nearest prior art; and a communication unit that acquires materials including the nearest prior art from the prior art database and transmits the materials to the source of the difference data and the advantage data.

[0015] Commonalities correspond to the prior art itself, as described in the advantages data. Prior art can be effectively extracted if the differences and the advantages arising from those differences are understood. The researcher's burden is reduced because they only need to input the differences and advantages when extracting the nearest prior art. Since the prior art is narrowed down to one, the amount of data transmitted from the communication unit is minimized. The processing load on the processor is reduced during transmission. Access time is significantly reduced because the researcher only needs to verify one nearest prior art. The number of simultaneous accesses increases. The size of the hardware can be reduced.

[0016] The patent examination system may include a second search unit that searches the prior art database based on the difference between the nearest prior art and the advantages, and extracts prior art containing the difference; and a second communication unit that obtains materials containing the prior art containing the difference from the prior art database and transmits them to the source of the difference data and the advantages data. Generally, in the examination of inventive step (non-obviousness), one or more prior art pieces are combined with the nearest prior art. If an unexpected effect is found in the difference, the inventive step (non-obviousness) of the invention is affirmed. The combination of the nearest prior art and the prior art containing the difference provides material for judging inventive step (non-obviousness). Here, since advantages are considered when searching the prior art database, the accuracy of the judgment material is improved. The burden of judgment is reduced. Since the prior art is narrowed down to those directly related to the judgment of inventive step (non-obviousness), access time is significantly reduced. The number of simultaneous accesses increases. The size of the hardware can be reduced.

[0017] As described above, according to the form of disclosure, a patent application document preparation system and a patent examination system that can reduce the processor load are provided.

[0018] This is a block diagram schematically showing the configuration of a patent application document preparation system according to the first embodiment of the present invention. This is a block diagram schematically showing the configuration of a patent application document preparation system according to the second embodiment of the present invention. This is a block diagram schematically showing the configuration of a patent examination system according to the third embodiment of the present invention. This is a conceptual diagram showing the pyramidal structure of the embodiment. This is a conceptual diagram showing the relationship between the pyramidal structure and paragraphs.

[0019] One embodiment of the present invention will be described below with reference to the attached drawings.

[0020] Figure 1 schematically shows the configuration of a patent application document creation system 11 according to a first embodiment of the present invention. The patent application document creation system 11 includes a generation unit 12 that generates a patent application document from defined input values ​​according to a framework. Generally, a patent application document includes an application form, claims including one or more claims, a specification (or description), drawings, and an abstract. The generation unit 12 incorporates an interface 15 that is connected to an input / output device 14 via, for example, a network 13. The input / output device 14 includes, for example, a personal computer 14a, a smartphone 14b, or a tablet computer. In the personal computer 14a, a keyboard 16a, touchpad, mouse, microphone, etc., are used for inputting text and operations. Images are displayed on the screen of the display panel 16b for outputting the results. In the smartphone 14b or tablet computer, a touchscreen panel 17 and microphone are used for inputting text and operations. Images are displayed on the screen of the touchscreen panel 17 for outputting the results. Interface 15 is connected to network 13 via communication port 18. For example, a personal computer 14a is connected to network 13 via communication port 19. For example, a smartphone 14b is connected to network 13 via communication port 21. Communication ports 18 and 19 establish a communication path between the personal computer 14a and the generation unit 12. Communication ports 18 and 21 establish a communication path between the smartphone 14b and the generation unit 12. Depending on the establishment of the communication path, text data, image data such as photographs and graphics, audio data, and other data are exchanged. The language of the text data is not limited to one. The text data may support multiple languages ​​with the assistance of machine translation.

[0021] The generation unit 12 includes a separation unit 31 that identifies common elements common to the prior art and the invention, a requirements description unit 32 that describes the function of the common elements and the relationships between the common elements based on an inference model, and a claim description unit 33 that links the differences between the prior art and the invention to the common elements and describes the claims based on the function of the common elements and the relationships between the common elements. The separation unit 31 is connected to a first acquisition unit 34 that acquires difference data describing the differences between the prior art and the invention, and a second acquisition unit 35 that acquires advantage data describing the advantages arising from the differences. The difference data consists of text data input from, for example, an input / output device 14. Here, the differences are expressed in single words. The differences should reflect the inventor's subjective opinion. The differences may also be expressed in sentences containing two or more words. The advantage data consists of text data input from, for example, an input / output device 14. The advantages include common elements that describe the constituent elements of the prior art and difference elements (feature elements or hidden elements) that describe the differences between the prior art and the invention. The advantages should describe the differences and their relationship to conventional technologies.

[0022] The separation unit 31 separates common elements common to the prior art and the invention from the description of advantages, based on the difference data and advantage data. In separating the common elements, the separation unit 31 generates prompts (language input values) for the AI ​​(artificial intelligence) 37. The prompts instruct the AI ​​to remove differences and actions linked to those differences from the description of advantages. Based on the Large-Scale Language Model (LLM), the AI ​​37 identifies common elements that describe the constituent elements of the prior art and difference elements that describe the differences between the prior art and the invention. The separation unit 31 generates common element data to identify the common elements. The separation unit 31 stores the difference data 39a, advantage data 39b, and common element data 39c in the storage device 38.

[0023] The requirements description unit 32 describes the functions of common elements and the relationships between common elements based on the common element data 39c. In describing the common elements, the requirements description unit 32 generates prompts (language input values) for the AI ​​37. The AI ​​37 identifies the functions of common elements and the relationships between common elements based on a large-scale language model. The requirements description unit 32 generates common requirements data that identifies the functions of common elements and the relationships between common elements. The common requirements data consists of, for example, text data. The requirements description unit 32 is connected to a presentation unit 41 that presents the functions of common elements and the relationships between common elements in a way that can be perceived by human senses. The presentation unit 41 generates image data based on the common requirements data 39c. The image data displays the functions of common elements and the relationships between common elements on the display screen. The requirements description unit 32 stores the common requirements data 42 in the storage device 38.

[0024] A third acquisition unit 43 is connected to the claim description unit 33 to acquire linking data that describes the relationships between common elements and differences. The linking data consists of text data input from, for example, the input / output device 14. The linking data is generated by the input / output device 14 in response to the presentation of common requirement data 42. The claim description unit 33 stores the linking data 44a in the storage device 38. The claim description unit 33 describes the claims based on the common requirement data 42 and the linking data 44a. When describing the claims, the claim description unit 33 generates prompts (language input values) for the AI ​​37. The AI ​​37 generates the claims based on a large-scale language model. When generating the claims, the functions of the common elements, the relationships between the common elements, and the relationships between the common elements and differences are combined. The claim description unit 33 generates claim data 44b that represents the claims. The claim data 44b consists of, for example, text data. The claim data 44b is stored, for example, in the storage device 38. An output unit 45 is connected to the claim description unit 33. The output unit 45 outputs claim data 44b to the network 13. Based on the claim data 44b, the claim can be displayed on the display screen.

[0025] Here, the generation unit 12 is built, for example, within a cloud computer CL connected to the internet. Interface 15 provides a user interface for each generation unit 12. The AI ​​37 may be built in common for multiple generation units 12. A generation unit 12 may be assigned to each user ID. Multiple users may be assigned to a single generation unit 12. In a single generation unit 12, parallel processing may be performed for multiple differences identified by the first acquisition unit 34. In the generation unit 12, processing operations other than those of the storage device 38 are provided by software programs. The network 13 includes the internet, a virtual private network (VPN), a local area network (LAN), and other networks. Via the network 13, the generation unit 12, the AI ​​37, and the input / output device 14 form a patent application document creation system 11.

[0026] Next, the operation of the patent application document creation system 11 will be explained. Currently, the string "quantity" is entered in the differences field. The string "In this food, mucuna bean powder can be easily consumed anywhere, in the same way as honey" is entered in the advantages field. The input / output device 14 generates difference data describing the text data of "quantity," and advantages data describing the text data of "In this food, mucuna bean powder can be easily consumed anywhere, in the same way as honey." The difference data is transmitted to the first acquisition unit 34. The advantages data is transmitted to the second acquisition unit 35. Here, no difference items are specified in the advantages data. For example, a VPN is established between the communication ports 19 and 21 of the input / output device 14 and the communication port 18 of the generation unit 12.

[0027] The separation unit 31 acquires difference data 39a and advantage data 39b. Based on the difference data 39a and advantage data 39b, the separation unit 31 generates commonalities data 39c. The separation unit 31 generates prompts to identify commonalities and differences. Since the string "quantity" is not included in the description of advantages, the separation unit 31 writes the strings "food", "honey", "mucuna bean powder", "anywhere", and "consumed" into the commonalities data 39c.

[0028] The requirements description unit 32 receives common item data 39c from the separation unit 31. The requirements description unit 32 creates common requirements data 42 based on the common item data 39c. The requirements description unit 32 generates prompts to identify the function of the common items and the relationships between the common items. In the AI ​​37, for example, the action "Mucuna bean powder is transported" is recognized in response to the common item "anywhere". Furthermore, the requirement "Mucuna bean powder is placed in a container" is grasped from the action "Mucuna bean powder is transported". The requirement "Mucuna bean powder is mixed with honey" is set in response to the common items "honey" and "Mucuna bean powder". The requirement "The container has an opening that can be opened and closed" is set in response to the common item "ingested". The text data for "Mucuna bean powder is placed in a container", the text data for "Mucuna bean powder is mixed with honey", and the text data for "The container has an opening that can be opened and closed" are written to the common requirements data 42.

[0029] The presentation unit 41 obtains common requirement data 42 from the requirements description unit 32. The presentation unit 41 generates image data based on the common requirement data 42. The image data is transmitted to the corresponding input / output device 14. The input / output device 14 displays the common requirement data "Mucuna bean powder is contained in a container," "Mucuna bean powder is mixed into honey," and "The container has an opening that can be opened and closed." The presentation unit 41 prompts the operator of the input / output device 14 to input linked data. The screen only needs to display a text data input box. The operator writes the string "Relationship between common items and differences" into the input box from the input / output device 14. For example, when the string "Mucuna bean powder is mixed into honey in an amount that maintains the fluidity of the honey" is written into the input box, the text data "Mucuna bean powder is mixed into honey in an amount that maintains the fluidity of the honey" is incorporated into the linked data 44a.

[0030] The claim description unit 33 obtains common requirement data "mucuna bean powder is contained in a container", "mucuna bean powder is mixed into honey", and "the container has an opening that can be opened and closed with a lid" from the presentation unit 41, and obtains linked data "mucuna bean powder is mixed into honey in an amount that maintains the fluidity of the honey" from the third acquisition unit 43. The claim description unit 33 describes the claim based on the common requirement data 42 and linked data 44a. Claim data "a food comprising a container having an opening that can be opened and closed with a lid, and mucuna bean powder contained in the container and mixed into honey in an amount that maintains the fluidity of the honey" is generated.

[0031] In the patent application document creation system 11, a framework is established for describing claims. The AI ​​37 only needs to complete standard language generation tasks such as separating common elements, describing functions, and describing relationships. The establishment of a complex inference model based on learning from a large number of patent publications is avoided. The processing burden of language generation is reduced. Processing is sped up. Parallel processing is accelerated. The number of processing cases per processor increases. Moreover, because it does not rely on a complex inference model based on learning from a large number of patent publications, fluctuations in language generation are eliminated to the greatest extent possible. Accurate descriptions are established with standard language generation tasks. Although patent publications only contain patent specifications that satisfy the legal patentability requirements, it is thought that there is variation in the judgment of examiners and judges regarding patentability requirements. It is impossible to completely eliminate subjectivity from the judgments of examiners and judges. An inference model based on learning from patent publications inevitably includes variations in judgment. When an inference model based on learning from patent publications is used, fluctuations in language generation inevitably occur.

[0032] Figure 2 schematically shows the configuration of a patent application document creation system 11a according to a second embodiment of the present invention. The generation unit 12a includes a claim description unit 47 that describes claims by combining common items with differences, instead of the claim description unit 33 of the first embodiment. When describing claims, the claim description unit 47 acquires difference data 39a, advantage data 39b, and common requirement data 42. Based on the description of advantages, differences are combined with common items. When combining, the claim description unit 47 generates a prompt (language input value) for the AI ​​37. The prompt instructs the combination of common items and differences according to the analysis of advantages. The AI ​​37 generates claims based on a large-scale language model. In the generation unit 12a, the presentation unit 41 and the third acquisition unit 43 of the first embodiment are omitted. The claim description unit 47 describes claims by combining the functions of common items, the relationships between common items, and the relationships between common items and differences. The claim description unit 47 generates claim data 48 that represents the claims. The claim data 48 consists of, for example, text data. The claim data 48 is stored, for example, in a storage device 38. The output unit 45 outputs the claim data 48 to the network 13. Based on the claim data 48, the claim can be displayed on the screen of the display.

[0033] For example, the advantage "In this food, mucuna bean powder can be easily consumed anywhere in the same way as honey" is achieved only when the difference "quantity" is combined with the common requirements "mucuna bean powder is contained in a container," "mucuna bean powder is mixed with honey," and "the container has an opening that can be opened and closed with a lid." Claim description section 47 derives the relationship between the common items "food," "honey," "mucuna bean powder," "anywhere," and "consumed" and the difference "quantity" as "mucuna bean powder is mixed with honey in an amount that maintains the fluidity of the honey." Claim data "a food comprising a container having an opening that can be opened and closed with a lid, and mucuna bean powder contained in the container and mixed with honey in an amount that maintains the fluidity of the honey" is generated.

[0034] Figure 3 schematically shows the configuration of a patent examination system 51 according to a third embodiment of the present invention. The patent examination system 51 includes an examination unit 52 that examines the inventive step (non-obviousness) of an invention related to a patent application according to a single framework. The examination unit 52 incorporates an interface 15 that is connected to an input / output device 14 via a network 13, for example. The input / output device 14 includes, for example, a personal computer 14a, a smartphone 14b, a tablet computer, etc. A communication path is established between the personal computer 14a and the examination unit 52 through the operation of communication ports 18 and 19. A communication path is established between the smartphone 14b and the examination unit 52 through the operation of communication ports 18 and 21. Depending on the establishment of the communication path, text data, image data such as photographs and graphics, audio data, and others are exchanged. The language of the text data is not limited to one. The text data may support multiple languages ​​with the assistance of machine translation.

[0035] The examination unit 52 includes a separation unit 31 that identifies common elements common to the prior art and the invention, an analysis unit 53 that compares the advantages with the nearest technology and identifies the differences between the nearest technology and the advantages, a superiority determination unit 54 that identifies the effects resulting from the differences separated from the nearest technology and compares the effects with the advantages to determine whether the effects are within expectations, and a feasibility determination unit 55 that describes the mechanism from the differences to the advantages and determines the feasibility of the mechanism. The separation unit 31 is connected, as described above, to a first acquisition unit 34 that acquires difference data describing the differences between the prior art and the invention, and a second acquisition unit 35 that acquires advantage data describing the advantages resulting from the differences. The advantages include common elements that describe the constituent elements of the prior art and difference elements (feature elements or concealed elements) that describe the differences between the prior art and the invention. The advantages describe the relationship between the differences and the prior art. The separation unit 31 separates the common elements common to the prior art and the invention. The separation unit 31 generates common item data to identify common items.

[0036] The analysis unit 53 selects the closest prior art from the prior art. In making the selection, the analysis unit 53 obtains proximity technology data from the search unit 57. The search unit 57 searches the prior art database 58 based on commonalities. In doing so, the search unit 57 obtains commonality data 39c. In response to the search, the search unit 57 extracts prior art containing the commonalities. The search unit 57 generates proximity technology data based on the extracted prior art. The proximity technology data may include all (one or more) prior art containing the commonalities. The proximity technology data is composed of, for example, text data. The search unit 57 stores the proximity technology data 61a in the storage device 59. The prior art database 58 registers patent publications (including published applications) and papers that were published before the filing of the application.

[0037] The analysis unit 53 compares the advantages of each prior art. In the comparison, the analysis unit 53 generates a prompt (language input value) for the AI ​​37. The prompt instructs the AI ​​37 to extract the one prior art that has the largest range of commonalities from among the prior art included in the proximity art data. The AI ​​37 identifies one prior art based on a large-scale language model. The analysis unit 53 generates nearest technology data that identifies the nearest prior art. The nearest technology data consists of, for example, text data. The nearest technology data 61b is stored, for example, in the storage device 59. The analysis unit 53 generates difference data that identifies the difference. In the description of the difference, the relationship between the difference and the nearest prior art is omitted. The difference is separated from the nearest prior art. The difference data 61c is stored, for example, in the storage device 59.

[0038] The analysis unit 53 is connected to the first communication unit 62. The first communication unit 62 generates first notification data. The first notification data identifies unconfirmed differences. The first notification data consists of, for example, text data. The first communication unit 62 outputs the first notification data to the network 13. Based on the first notification data, a string of characters prompting a reconsideration of the differences and advantages is displayed on the display screen. For example, the string of characters "Substantially no different from the prior art" is displayed on the display screen. Here, the first communication unit 62 attaches materials including the most recent prior art to the first notification data. The first communication unit 62 obtains materials from, for example, the prior art database 58.

[0039] The superiority determination unit 54 obtains supplementary technology data from the search unit 57 in order to identify the effects that arise from the difference. The search unit 57 searches the prior art database 58 based on the difference. In the search, the search unit 57 obtains difference data 61c. In response to the search, the search unit 57 extracts prior art that contains the difference. The search unit 57 generates supplementary technology data based on the extracted prior art. The supplementary technology data may include all (one or more) prior art that contains the difference. The supplementary technology data is composed of, for example, text data. The search unit 57 stores the supplementary technology data 63a in the storage device 59.

[0040] The superiority determination unit 54 generates a prompt (language input value) for the AI ​​37 when comparing the advantages. The prompt instructs the AI ​​to determine whether there is a match or mismatch between the effect and advantage derived from the difference. Based on a large-scale language model, the AI ​​37 identifies the effect of the difference for each individual prior art included in the complementary technology data 63a. If the effect and advantage match, the superiority determination unit 54 generates superiority data (determination data) that identifies the expected (common) effect. If the effect and advantage do not match, the superiority determination unit 54 generates superiority data that identifies the superiority of the advantage. The superiority data consists of, for example, text data. The superiority data 63b is stored, for example, in the storage device 59.

[0041] A second communication unit (examination unit) 64 is connected to the superiority determination unit 54. The second communication unit 64 generates second notification data. The second notification data identifies descriptions that negate inventiveness (non-obviousness). The second notification data consists of, for example, text data. The second communication unit 64 outputs the second notification data to the network 13. Based on the second notification data, a string of characters prompting a reconsideration of the differences and advantages is displayed on the display screen. For example, the string of characters "It's nothing more than a commonplace effect" is displayed on the display screen. The second communication unit 64 attaches materials containing prior art that includes the differences to the second notification data. The second communication unit 64 obtains materials from, for example, the prior art database 58.

[0042] The feasibility determination unit 55 generates a prompt (language input value) for the AI ​​37 when determining the feasibility of the mechanism. The prompt instructs the AI ​​37 to analyze the mechanism from the differences to the advantages. Based on a large-scale language model, the AI ​​37 operates (makes function) the differences combined with common terms and confirms the advantages that are generated therefrom. If the mechanism is feasible, the feasibility determination unit 55 generates feasibility data (judgment data) that identifies the feasibility of the mechanism. If the mechanism is not feasible, the feasibility determination unit 55 generates feasibility data that identifies the non-feasibility of the mechanism. The feasibility data consists of, for example, text data. The feasibility data 65 is stored, for example, in the storage device 59.

[0043] A third communication unit (examination unit) 66 is connected to the patentability determination unit 55. The third communication unit 66 generates third notification data. The third notification data specifies whether or not the mechanism has an inventive step. The third notification data consists of, for example, text data. The third communication unit 66 outputs the third notification data to the network 13. If the patentability data 65 denies the existence of the mechanism, a string of characters prompting a reconsideration of the differences and advantages is displayed on the display screen based on the third notification data. For example, the string of characters "The advantage does not exist based on those differences" is displayed on the display screen. If the patentability data 65 affirms the existence of the mechanism, the third notification data specifies a description that affirms inventive step (non-obviousness). Based on the third notification data, the string of characters "Patent granted" is displayed on the display screen.

[0044] The patent examination system 51 includes a gazette issuing unit 67 that issues patent gazettes. Similar to the aforementioned generation unit 12, the gazette issuing unit 67 describes claims based on difference data and advantage data. The gazette issuing unit 67 enriches the claims to describe the embodiments of the product. As shown in Figure 4, in describing the embodiments, the gazette issuing unit 67 decomposes the product according to a pyramid structure. The constituent elements of the product are grouped by function under superordinate concepts and intermediate concepts. As shown in Figure 5, functions and relationships are described for each constituent element in one paragraph.

[0045] The pyramid structure is constructed following three principles. Principle [1] stipulates that "within each individual group, ideas must always be ideas of the same type". Principle [2] stipulates that "at any hierarchy, an idea must always be a summary of the ideas of its subordinate group". Principle [3] stipulates that "within each individual paragraph, ideas must always be ordered according to logic".

[0046] Here, the examination unit 52 is constructed, for example, in a cloud computer CL connected to the Internet. A plurality of users are assigned to one examination unit 52. In the examination unit 52, parallel processing is executed for the plurality of differences identified by the first acquisition unit 34. In the examination unit 52, processing operations other than those of the storage device 59 are provided by software programs. The network 13 includes the Internet, a virtual private network (VPN), a local area network (LAN), and other networks. Via the network 13, the examination unit 52, the AI 37, and the input / output device 14 form the patent examination system 51.

[0047] Next, the operation of the patent examination system 51 will be described. Now, the character string "amount" is input as the difference. The character string "In this food, Mucuna bean powder can be easily ingested anywhere in the same manner as ingesting honey" is input as the advantage. The separation unit 31 acquires difference data and advantage data in the same manner as described above. The separation unit 31 writes the character strings "food", "honey", "Mucuna bean powder", "anywhere", and "ingested" into the common term data.

[0048] The analysis unit 53 instructs the search unit 57 to generate proximity technology data 61a. The search unit 57 acquires common term data 39c. The prior art database 58 is searched in accordance with the common term keywords "food", "honey", "mucuna bean powder", "anywhere", and "ingested". The discovered prior art is written into the proximity technology data 61a. The analysis unit 53 generates difference data 61c based on the proximity technology data 61a and the advantage data 39b. The analysis unit 53 generates a prompt for comparing the advantages against each individual prior art. The AI 37 identifies the closest prior art from the proximity technology data 61a. For example, when "mucuna bean powder mixed into honey" is set as the closest prior art, "in an amount that maintains the fluidity of honey" is written into the difference data 61c. In the difference data 61c, the difference is described separately from the closest prior art. Since a difference has been confirmed between the advantage and the closest prior art, the first communication unit 62 suspends output of the first notification data. If no difference is detected between the closest prior art and the advantage, the first communication unit 62 outputs the first notification data. On the screen of the display panel 16b (or the touch screen panel 17) of the input / output device 14, the message "Substantially no difference from the prior art" is displayed.

[0049] The common elements correspond to the prior art itself, as described in the advantages data 39b. If the differences and the advantages arising from those differences are understood, the prior art can be extracted effectively. The researcher's burden is reduced because they only need to input the differences and advantages when extracting the nearest prior art. Since the prior art is narrowed down to one, the amount of data transmitted from the first communication unit 62 is reduced to a minimum. The processing load on the processor is reduced during the transmission process. Since the researcher only needs to check one nearest prior art, the access time is significantly reduced. The number of simultaneous accesses increases. The size of the hardware can be reduced. Generally, keywords and classification codes are used when searching the prior art database during the examination of a patent application. Often, a large number of prior art items are extracted. The number of prior art items is not narrowed down very much. The researcher is forced to check all of the extracted prior art items. If the number is large, the processing load on the processor increases when transmitting the data. Since the researcher has to look at each item one by one, the access time increases considerably. A large size of hardware is required for simultaneous access.

[0050] Once the generation of differential data 61c is confirmed, the superiority determination unit 54 instructs the search unit 57 to generate supplementary technology data 63a. The search unit 57 retrieves the differential data 61c. The prior art database 58 is searched according to the difference "in an amount that maintains the fluidity of honey". The discovered prior art is written to the supplementary technology data 63a. The superiority determination unit 54 generates superiority data 63b. The superiority determination unit 54 generates a prompt to compare the advantages with the effect of the difference. For example, if the prior art discloses "a powder that dissolves in honey", the AI ​​37 compares the advantage "roasted bean powder as is" with the effect "dissolves and maintains fluidity". Since the effect and advantage do not match, the superiority of the advantage is identified in the superiority data 63b. Since the superiority of the advantage has been confirmed, the second communication unit 64 withholds output of the second notification data. If any prior art discloses "a powder mixed into honey in an amount that maintains fluidity," then the effects and benefits will coincide. The second communication unit 64 outputs the second notification data. The input / output device 14 displays "It's nothing more than a commonplace effect" on the screen of the display panel 16b (or touchscreen panel 17).

[0051] The superiority determination unit 54 identifies the effects resulting from the differences separated from the nearest technology according to the framework. The effects of the differences are compared with the advantages. Generally, the effects of the differences do not contribute to satisfying the requirement of inventive step (non-obviousness). If an unexpected effect is found in the advantages, the inventive step (non-obviousness) of the invention is affirmed. Thus, the inventive step (non-obviousness) of the invention is judged by the superiority determination unit 54. Objectivity of the examination is ensured. Machine learning can be used in the examination of inventive step (non-obviousness). In this case, the AI ​​37 only needs to perform standard language generation tasks such as identifying the differences and their effects, and comparing the advantages and effects. The establishment of complex inference models based on learning from patent publications and examination result notices (notices of reasons for rejection) is avoided. The processing burden of inference is reduced. Processing is sped up. Parallel processing is accelerated. The number of processing cases per processor increases. Generally, when exploring the use of AI in the examination of patent applications, it is expected that the AI ​​will be trained on materials such as patent publications, examination result notices (notices of reasons for rejection), and the examination process. The AI ​​will read patterns of patent requirements from a large amount of material. The AI ​​will build an inference model based on these patterns. An extremely complex inference model is expected. Even so, judgments on patent requirements will vary. Fluctuations will occur in the examination results. The reliability of the examination cannot be guaranteed.

[0052] Generally, in the examination of inventive step (non-obviousness), the nearest prior art is combined with one or more prior art pieces. If an unexpected effect is found in the difference, the inventive step (non-obviousness) of the invention is affirmed. The combination of the nearest prior art and the prior art containing the difference provides material for determining inventive step (non-obviousness). Here, since advantages are considered when searching the prior art database 58, the accuracy of the material for judgment is improved. The burden of judgment is reduced. Since the prior art is limited to that which is directly related to the determination of inventive step (non-obviousness), the access time is significantly reduced. The number of simultaneous accesses increases. The size of the hardware can be reduced.

[0053] When the superiority of the advantage is confirmed in the superiority data 63b, the feasibility determination unit 55 generates feasibility data 65 based on the nearest prior art data 61b, difference data 61c, and advantage data 39b. The feasibility determination unit 55 generates a prompt to determine the feasibility of the mechanism from the difference to the advantage. In AI 37, the operation (function) of the difference is confirmed on the nearest prior art. For example, when the difference "in an amount that maintains the fluidity of honey" is combined with the common item "Mucuna bean powder is mixed with honey", the advantage "Mucuna bean powder can be easily consumed anywhere in the same way as honey" is established. Since the establishment of the mechanism is confirmed, the third communication unit 66 outputs third notification data that specifies the patent grant. In the input / output device 14, "Patent Granted" is displayed on the screen of the display panel 16b (or touchscreen panel 17). If the difference "the amount used to maintain the fluidity of the honey" is used to identify the advantage "the calories in the honey are suppressed," then, since the calories in the honey are unrelated to the amount of mucuna bean powder, the feasibility determination unit 55 generates feasibility data to identify that the mechanism is not feasible. The third communication unit 66 outputs third notification data. The input / output device 14 displays "Based on the difference, the advantage is not valid" on the screen of the display panel 16b (or touchscreen panel 17).

[0054] The validity determination unit 55 describes the mechanism from the difference to the advantage according to the framework. If the mechanism is valid, the advantage is judged to be a reasonable claim. Furthermore, if an unexpected effect is recognized in the advantage, the inventive step (non-obviousness) of the invention is affirmed. Even if an unexpected effect is recognized in the advantage, if the mechanism is not valid, the advantage is judged to be an unreasonable claim. The inventive step (non-obviousness) of the invention is denied. Thus, the inventive step (non-obviousness) of the invention is judged by the validity determination unit 55. Objectivity of the examination is ensured. Machine learning can be used in the examination of inventive step (non-obviousness). The AI ​​37 only needs to complete standard language generation tasks such as identifying the difference and its effect and describing the mechanism. The establishment of complex inference models based on learning from patent publications and examination result notices (notices of reasons for rejection) is avoided. The processing burden of inference is reduced. Processing is sped up. Parallel processing is accelerated. The number of processing cases per processor increases.

[0055] If the establishment of the mechanism is confirmed by the feasibility data 65, the publication issuing unit 67 issues a patent publication. The publication issuing unit 67 describes an embodiment of the product based on the claim description "a food comprising a container having an opening that can be opened and closed with a lid, and mucuna bean powder contained in the container and mixed with honey in an amount that maintains the fluidity of the honey." For example, the container is disassembled into a body and a lid. The body is partitioned into a storage space and an opening. The publication issuing unit 67 can instruct the requirements description unit 32 to describe the embodiment.

[0056] 11...Patent application document creation system, 11a...Patent application document creation system, 31...Separation unit, 32...Requirements description unit, 33...Claim description unit, 34...First acquisition unit, 35...Second acquisition unit (acquisition unit), 39a...Difference data, 39b...Advantage data, 39c...Common item data, 41...Presentation unit, 42...Common requirements data, 44a...Linking data, 47...Claim description unit, 51...Patent examination system, 53...Analysis unit, 54...Superiority determination unit, 55...Financial feasibility determination unit, 57...Search unit and second search unit, 62...Communication unit (first communication unit), 64...Examination unit and second communication unit (second communication unit), 66...Examination unit (third communication unit).

Claims

1. A patent application document creation system comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common items common to the prior art and the invention from the description of the advantages and outputs common item data that identifies the common items; a requirements description unit that acquires the common item data and describes the functions of the common items and the relationships between the common items based on an inference model; a presentation unit that acquires common requirements data that identifies the functions and relationships and presents the functions and relationships in a way that can be perceived by human senses; and a claim description unit that, in response to the presentation of the functions and relationships, acquires linking data describing the relationships between the common items and the differences and describes claims based on the common requirements data and the linking data.

2. A patent application document creation system comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common items common to the prior art and the invention from the description of the advantages and outputs common item data that identifies the common items; a requirements description unit that acquires the common item data and describes the functions of the common items and the relationships between the common items based on an inference model; and a claim description unit that acquires common requirements data that identifies the functions and relationships and describes claims by combining the differences with the common items based on the description of the advantages.

3. A patent examination system comprising: an acquisition unit that receives advantage data from an input device describing the advantages arising from the differences between the prior art and the invention; an analysis unit that compares the advantages with the nearest technology and identifies the difference between the nearest technology and the advantages; a superiority determination unit that acquires difference data and advantage data that identify the difference, identifies the effect arising from the difference separated from the nearest technology based on an inference model, compares the effect with the advantages and determines whether the effect is within expectations; and an examination unit that, upon acquiring determination data that recognizes the effect as within expectations, denies the inventive step of the invention.

4. A patent examination system comprising: an acquisition unit that receives advantage data from an input device describing the advantages arising from the differences between the prior art and the invention; an analysis unit that compares the advantages with the nearest technology and identifies the difference between the nearest technology and the advantages; a feasibility determination unit that acquires difference data and advantage data that identify the difference, describes the mechanism from the difference to the advantages based on an inference model, and determines whether the mechanism is feasible; and an examination unit that, upon acquiring determination data that acknowledges the feasibility of the mechanism, affirms the inventive step of the invention.

5. A patent examination system comprising: a first acquisition unit that acquires difference data describing the differences between the prior art and the invention; a second acquisition unit that acquires advantage data describing the advantages arising from the differences; a separation unit that, based on the difference data and the advantage data, separates common items common to the prior art and the invention from the description of the advantages and outputs common item data that identifies the common items; a search unit that acquires the common item data, searches a prior art database based on the common items, and extracts prior art containing the common items; an analysis unit that, based on the advantage data, identifies the prior art with the largest range of common items as the nearest prior art; and a communication unit that acquires materials including the nearest prior art from the prior art database and transmits the materials to the source of the difference data and the advantage data.

6. A patent examination system according to claim 5, comprising: a second search unit that searches the prior art database based on the difference between the nearest prior art and the advantages, and extracts prior art containing the difference; and a second communication unit that obtains materials containing the prior art containing the difference from the prior art database and transmits them to the source of the difference data and the advantages data.

Images