Patent analysis program, patent analysis method, and patent analysis system

Abstract

By presenting users with highly accurate patent maps, the system streamlines the process of reading patent documents. [Solution] The patent analysis system 10 performs the following steps: obtains a patent list from artificial intelligence 14 that includes the usage forms of substances extracted from patent documents; creates a patent map that visualizes the quantitative distribution of patent documents in the patent list using the usage forms and other indicators contained in the patent documents; and outputs the patent map to the user terminal 12.

Description

Technical Field

[0001] This disclosure relates to a patent analysis program, a patent analysis method, and a patent analysis system.

Background Art

[0002] For grasping trends in technological development, obtaining technical information, etc., databases accumulating patent documents are utilized. Conventionally, since analysts input search conditions into the database to search for patent documents and read the huge number of extracted patent documents, a great deal of man-hours have been required.

[0003] To improve the efficiency of reading patent documents, a system that uses artificial intelligence to assist in reading has been proposed (see, for example, Patent Document 1). The system extracts necessary portions from patent documents according to a task specified by the user, and automatically generates a prompt corresponding to the task based on the extracted portions and the query input by the user. Then, the system sends the prompt to artificial intelligence and receives a sentence as the answer content from the artificial intelligence.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] Creating a patent map requires mapping the features of patent documents onto the map's coordinate system. However, because the terminology used in patent documents varies depending on the document, mapping the features of each document is difficult. This is especially true in materials science patents, where the variability in terminology is greater than in other technical fields. As a result, obtaining a patent map with the accuracy required by users has been challenging. [Means for solving the problem]

[0007] This disclosure provides a patent analysis program that solves the above problems. The patent analysis program is a patent analysis program for analyzing patent documents, and causes at least one processor to perform the following steps: obtain a patent list including the forms of use of substances extracted from patent documents from artificial intelligence; create a patent map that visualizes the quantitative distribution of patent documents in the patent list using the forms of use and other indicators contained in the patent documents; and output the patent map to a user terminal.

[0008] With respect to the above-mentioned patent analysis program, it is preferable to further cause at least one processor to perform the step of obtaining from the artificial intelligence the properties of raw materials or materials relating to the substance contained in the patent documents of the patent list as other indicators, by sending a prompt to the artificial intelligence.

[0009] With respect to the above-mentioned patent analysis program, it is preferable to have at least one processor perform the following steps: acquire a characteristic value condition that is a user-specified condition and is the value of the characteristic; and output to the user terminal the quantitative distribution of patent documents among the patent lists that satisfy the characteristic value condition.

[0010] In the patent analysis program described above, in the step of outputting the quantitative distribution of patent documents that satisfy the characteristics value conditions, it is preferable to overlay an element indicating the quantity of patent documents that satisfy the characteristics value conditions on each element of the patent map.

[0011] With respect to the above-mentioned patent analysis program, it is preferable to cause at least one processor to perform the step of obtaining, as other indicators, terms of the problem of the invention contained in the patent documents of the patent list by sending a prompt to the artificial intelligence.

[0012] With respect to the above-mentioned patent analysis program, it is preferable to have at least one processor perform the following steps: accept the user's selection of elements of the patent map; create a list of patent documents associated with the elements; and output the list to the user terminal.

[0013] With respect to the above-described patent analysis program, it is preferable to have at least one processor perform the steps of extracting patent documents that satisfy the search conditions received from the user terminal from a database of stored patent documents, and creating a search results list using the extracted patent documents, and in the step of obtaining the patent list, to obtain the patent list that includes the forms of use of the substance included in the search conditions, using the patent documents included in the search results list as the population.

[0014] With respect to the above-described patent analysis program, it is preferable to have at least one processor perform a step of outputting a list of patent documents that satisfy the characteristics of raw materials or materials related to the substance to the user terminal, before the step of obtaining the patent list including the forms of use of the substance.

[0015] This disclosure provides a patent analysis method that solves the above problems. The patent analysis method is a patent analysis method for analyzing patent documents, wherein at least one processor performs the steps described above.

[0016] This disclosure provides a patent analysis system that solves the above problems. The patent analysis system comprises at least one processor and memory, wherein the at least one processor performs the steps described above.

Advantages of the Invention

[0017] According to the present disclosure, by presenting a highly accurate patent map to the user, the user's work of reading and understanding patent documents can be made more efficient.

Brief Description of the Drawings

[0018] [Figure 1] It is a diagram showing an outline of the configuration of the first embodiment of the patent analysis system of the present disclosure. [Figure 2] It is a diagram showing the hardware configuration of the information processing apparatus of the embodiment. [Figure 3] It is a diagram showing an example of a search query, a search result list, and a patent list of the embodiment. [Figure 4] It is a flowchart showing the processing procedure of the patent analysis method of the embodiment. [Figure 5] It is a flowchart showing the procedure for transitioning from the patent map to the list in the embodiment. [Figure 6] It is a diagram showing an example of a patent map for each usage form. [Figure 7] It is a diagram showing an example of a patent map. [Figure 8] It is a flowchart showing the processing procedure of the patent analysis method of the second embodiment. [Figure 9] It is a diagram explaining the input of the condition of the characteristic value of the embodiment. [Figure 10] It is a diagram showing an example of a patent map in the embodiment. [Figure 11] It is a flowchart showing the processing procedure of the patent analysis method in the modification example.

Modes for Carrying Out the Invention

[0019] The patent analysis program, patent analysis method, and patent analysis system of the present disclosure will be described. The present disclosure is not limited to the embodiments, but is defined by the claims, and is intended to include all modifications within the meaning and scope equivalent to the claims.

[0020] [First Embodiment] A first embodiment of a patent analysis program, a patent analysis method, and a patent analysis system will be described.

[0021] As shown in Figure 1, the patent analysis system 10 comprises a patent analysis server 11 and a user terminal 12. The patent analysis server 11 and the user terminal 12 are configured to send and receive data to and from each other via a network 15. The patent analysis server 11 is configured to send and receive data to and from the patent document database 13 (hereinafter referred to as the patent document DB 13) and the artificial intelligence 14 via the network 15.

[0022] <Patent Analysis Server 11> The patent analysis server 11 is a server that analyzes patent documents. The patent analysis server 11 stores the patent analysis program in memory.

[0023] The patent analysis server 11 may send web data for screen display to the user terminal 12. Such web data may include markup data consisting of HTML (Hyper Text Markup Language), CSS (Cascading Style Sheets) data, and script data consisting of JavaScript (registered trademark). The user terminal 12 generates a screen based on this web data using an installed web browser. The screen includes UI elements such as various buttons, pull-down menus, and text boxes.

[0024] The patent analysis server 11 retrieves patent documents from the patent document database 13. The patent document database 13 is a database that stores patent documents such as published patent gazettes, patent publications, and publicly released patent gazettes.

[0025] As an example, the patent analysis server 11 uses the search query received from the user terminal 12 to extract patent documents from the patent document database 13 that match the search query. The patent analysis server 11 then generates a search results list using the extracted patent documents.

[0026] The patent analysis server 11 sends and receives data with the artificial intelligence 14 using an API (Application Programming Interface). <Artificial Intelligence 14> The artificial intelligence 14 connected to the patent analysis server 11 includes a model that includes a set of parameters determined through learning, etc. Using the model, the artificial intelligence 14 performs processing such as evaluation, classification, and judgment based on input provided from an external source, as well as generation processing. For example, the artificial intelligence 14 may include a language processing model such as a large language model (LLM) or a small language model (SLM) that has been pre-trained on a large dataset. Furthermore, the artificial intelligence 14 may include a Retrieval-Augmented Generation (RAG) system that combines the language processing model with a search function that searches for external information such as patent documents, documents other than patent documents, and websites. In addition, the artificial intelligence 14 may include a machine translation system that translates a foreign language (first language) into Japanese (second language).

[0027] <User terminal 12> The user terminal 12 is a device used by a user performing patent analysis. An example of the user terminal 12 is a smartphone, tablet, or personal computer.

[0028] <Hardware Configuration> Although the patent analysis server 11 and the user terminal 12 differ in specifications and performance, they have similar hardware configurations. Therefore, the hardware configurations of the patent analysis server 11 and the user terminal 12 will be described as an information processing device.

[0029] As shown in Figure 2, the information processing device 100 includes a processor 101, a memory 102, and a communication interface 103. The information processing device 100 may also include an input device 105 and an output device 106.

[0030] Processor 101 is a processor that performs one or more of the control processes disclosed herein. Processor 101 is at least one processing circuit. For example, processor 101 is composed of at least one of the following: CPU (Central Processing Unit), GPU (Graphics Processing Unit), NPU (Neural Processing Unit), etc. Processor 101 is not limited to performing software processing for all of the processing it performs. For example, processor 101 may have dedicated hardware circuits (e.g., application-specific integrated circuits: ASICs) that perform hardware processing for at least a portion of the processing it performs. That is, processor 101 is a circuit that includes one or more processors that operate according to a computer program (software), one or more dedicated hardware circuits that perform at least a portion of the various processing, or a combination thereof.

[0031] Memory 102 stores various programs. Each program contains at least one instruction. Memory 102 consists of, for example, main memory and auxiliary memory. Examples of memory 102 include ROM, RAM, and hard disks. Memory 102 includes any available recording media that can be accessed by a general-purpose or dedicated computer. Memory 102 is connected to the processor 101 via bus 104.

[0032] The patent analysis program is stored in the memory 102 of the patent analysis server 11. The memory 102 of the patent analysis server 11 also stores data necessary for patent analysis, such as search queries, search result lists, and patent lists, as well as data of the patent analysis results.

[0033] The memory 102 of the user terminal 12 stores a program related to a web browser. Alternatively, the memory 102 of the user terminal 12 may store an application that analyzes patent documents in cooperation with the patent analysis server 11. Furthermore, the memory 102 of the user terminal 12 may store at least a portion of search queries, search result lists, and patent lists.

[0034] The communication interface 103 is an interface that establishes a communication path with other devices via the network 15 and performs data transmission and reception. The communication interface 103 consists of hardware, software, or a combination thereof.

[0035] Network 15 may include the Internet or a LAN (Local Area Network). Network 15 may also include a core network and multiple base stations. Network 15 is a third-generation communication system such as W-CDMA or CDMA2000, LTE, a fourth-generation communication system, or a fifth-generation communication system.

[0036] The user terminal 12 also includes at least an input device 105 and an output device 106. The patent analysis server 11 may omit these. The input device 105 includes a touch panel, keyboard, mouse, operation buttons, etc.

[0037] The output device 106 consists of at least one device. An example of the output device 106 is the display device 107. The display device 107 is a device capable of displaying a screen, such as a liquid crystal display or an organic EL display. The processor 101 is connected to the display device 107 via a display controller. In addition to the display device 107, the output device 106 may also include a microphone and a speaker.

[0038] <Data structure> Figure 3 schematically shows the configuration of the search query 20, search result list 21, and patent list 22 stored in the memory 102 of the patent analysis server 11.

[0039] <Search Query 20> Search query 20 is a query received from user terminal 12. Search query 20 is a query to extract desired patent documents from patent document database 13. Search query 20 is at least linked to search result list 21. Search query 20 includes a search ID and search conditions. The search ID is an identifier uniquely assigned to the search task. The search conditions are conditions specified by the user. For example, the search conditions include data for searching patent documents, such as the period in which the filing date of the patent document is included, the period in which the publication date of the patent document is included, keywords included in the patent document, the fields in which the keywords are described, and whether or not the patent is registered.

[0040] <Search Results List 21> Search result list 21 is a list of patent documents extracted from patent document database 13 based on search query 20. In this disclosure, search result list 21 is a list of patent documents relating to substances. Substances include all articles such as raw materials, materials, or products. Products are formed from raw materials or materials. Materials are formed from raw materials. For example, a polypropylene sheet, which is a material, is formed from polypropylene, which is a raw material, and the sheet is molded to form a tray, which is a product. Search result list 21 is a list of tens to hundreds, and in some cases thousands, of patent documents, and includes multiple forms of use for substances. Furthermore, the forms of use for substances in material science fields such as organic materials and inorganic materials are diverse.

[0041] The search results list 21 stores records of extracted patent documents for each search ID. Each patent document record includes data such as the publication number, application number, inventor, applicant, application date, invention title, and International Patent Classification (IPC) of the extracted patent document. Each patent document record is linked to patent documents stored in the patent document database 13.

[0042] <Patent List 22> Patent list 22 stores records of extracted patent documents for each search ID. Patent list 22 is linked to at least search result list 21. Each patent document record includes at least bibliographic information, usage, characteristics, and problems. It may also include an abstract and claims.

[0043] Bibliographic information includes basic patent information such as publication number and filing date. This bibliographic information is the same as the data included in search result list 21. The usage patterns are those of the substance extracted by artificial intelligence 14 from patent documents. These usage patterns can also be described as the substance's applications or application fields.

[0044] The characteristics are properties of the substance extracted by the artificial intelligence 14 from patent documents, and are physical properties of the substance. The patent list 22 stores the name of the characteristic and its characteristic value. For example, the properties are linked to the substance specified in the search query 20. For instance, if the search query 20 includes the substance name "polyethylene," the artificial intelligence 14 extracts the properties of "polyethylene" (raw material) and the material formed from "polyethylene." For example, melt index (MI) and melting point are extracted as properties of polyethylene, and the stretch ratio is extracted as a property of the material formed from polyethylene, and these are stored in the patent list 22. The problem represents the "problem of the invention" extracted by artificial intelligence 14 from patent documents.

[0045] <Patent analysis method> The procedure for the patent analysis method described herein will now be explained. Note that the order of each step may be rearranged to the extent that no inconsistencies arise. Furthermore, multiple steps may be combined into a single step.

[0046] The patent analysis server 11 functions as a first processing unit that obtains a patent list 22, including the forms of use of substances extracted from patent documents, from artificial intelligence 14 by executing a patent analysis program stored in memory 102 by the processor 101; a second processing unit that creates a patent map that visualizes the quantitative distribution of patent documents in the patent list 22 using the forms of use and other indicators contained in the patent documents; and a third processing unit that outputs the patent map to the user terminal 12.

[0047] The following describes each step performed by the patent analysis server 11. For example, the user performs the input operation for a search query 20. The user terminal 12 sends a request for the patent list 22 to the patent analysis server 11. As shown in Figure 4, the patent analysis server 11 obtains the search results list 21 (step S1).

[0048] The patent analysis server 11 extracts patent documents from the patent document database 13 that match the search criteria of the search query 20. Then, the patent analysis server 11 obtains a search results list 21 using the bibliographic information of the extracted patent documents.

[0049] At this time, the patent analysis server 11 may send the search results list 21 to the user terminal 12, and the user terminal 12 may output the received search results list 21 to the display device 107. The patent analysis server 11 extracts the forms of use of the substance from the patent documents included in the search results list 21, and then has the artificial intelligence 14 extract the characteristics and problems for each form of use (step S2).

[0050] The patent analysis server 11 targets the patent documents included in the search results list 21 and has the artificial intelligence 14 extract the usage forms described in those patent documents. It should be noted that, before receiving a request from the user terminal 12, the extraction of usage forms of materials, extraction of characteristics, and extraction of problems may have already been performed when the search results list 21 was created in step S1.

[0051] The patent analysis server 11 automatically generates prompts for extracting the usage forms, properties, and challenges of a substance. The prompts may include instructions for a task to extract usage forms, instructions for a task to extract properties for each usage form, and instructions for a task to extract challenges. Alternatively, the patent analysis server 11 may include the instructions for these tasks in a separate prompt and send it. Here, we will describe an embodiment in which the instructions for the above tasks are included together in a prompt. The patent analysis server 11 transmits the patent documents and prompts included in the search results list 21 to the artificial intelligence 14. Alternatively, the patent analysis server 11 may transmit only the search results list 21 and not the patent documents to the artificial intelligence 14, and the artificial intelligence 14 may retrieve the patent documents included in the search results list 21 from the patent document database 13.

[0052] The prompt includes instructions for extracting the form of use of the substance. The prompt may also include specifying the description field from the patent document for extracting the modes of use. For example, the specification of the description field may specify at least one of the following: the "technical field" field, the International Patent Classification, the "modes for carrying out the invention," or the "claims." In the "claims," ​​if there is a claim with a limited use (use-limited claim), such as "packaging material containing polyethylene as described in any one of claims 1 to 10," the prompt may instruct the system to extract the use described in that claim.

[0053] The prompt may also include a description of the usage. Furthermore, the prompt may include examples of phrases (idioms) describing the usage, exclusion conditions, and specifications for the output format. For example, part of a prompt may include a sentence written in natural language (natural language description) like the following. The following prompt is a summary, and various conditions and output formats may be added.

[0054] Please extract the "utilization forms" of polyethylene from the patent documents. Furthermore, please cluster the patent documents using the extracted usage patterns. "Usage form" refers to the specific fields in which polyethylene is used. Specifically, this includes items described as "polyethylene for ~," "polyethylene made ~," "polyethylene used as ~," etc. (Exclusion conditions) Uses of polyethylene solely for testing, measurement, observation, analysis, or verification purposes are excluded. (output format) Please output the extracted usage patterns, removing any duplicates, and include a field to describe the usage patterns used as the basis for the extraction.

[0055] The prompt may include instructions to integrate (create a higher-level concept for) multiple uses extracted from different patent documents, or multiple uses extracted from the same patent document. In this case, examples of uses may be provided to indicate the level of classification of the uses. Examples of uses include "packaging materials," "batteries," "medical and sanitary products," "automobiles," "containers," "food storage bags," "electrical insulating materials," "cushioning materials," "waterproof sheets," "thermal insulation materials," and "agricultural films." If examples of uses are included in the prompt, the prompt may also include instructions to encourage the generation of new uses, such as "If your use does not fit the examples of uses, please generate a new use." It may also include examples of integration, such as "Integrate sheets and films into sheets."

[0056] If multiple uses of a single substance are described in the patent document, the prompt may include instructions to extract a predetermined number of uses or all of the uses from among those uses. The selection criteria for selecting a predetermined number of uses may include that they are the uses that appear most frequently in the patent document, that they address problems or have effects described in the patent document, or that they are uses described in the "claims."

[0057] Alternatively, the prompt may include instructions to select a representative use from among the multiple uses of a single substance described in the patent document. The selection criteria for the representative use may include being the use that appears most frequently in the patent document, or being the use described in the problems or effects described in the patent document.

[0058] The artificial intelligence 14 extracts usage patterns based on the instructions for extracting usage patterns included in the prompt. The artificial intelligence 14 also clusters patent documents that have the same usage pattern for the substance using the extracted usage patterns. For example, artificial intelligence 14 vectorizes patent documents based on the forms of use of the substances described in each patent document. Artificial intelligence 14 generates clusters to which patent documents with similar vectors belong. Clusters are assigned labels indicating the forms of use. For example, patent documents that describe "storage bags," "freezer bags," "food storage bags," "food storage bags," and "food container bags" as forms of use may be included in a cluster labeled "food storage bags."

[0059] If multiple usage patterns are extracted from a single patent document, the patent document may be classified into multiple clusters. The patent analysis server 11 obtains a patent list 22, including usage patterns, from the artificial intelligence 14. At this time, the patent list 22 does not include characteristics or problems.

[0060] The prompt also includes instructions for extracting the name of the raw material or substance and its properties for each extracted usage form. The raw material or substance may be the substance itself included in the search query 20, or it may be a constituent element such as an organic compound, resin, polymer, or inorganic substance that makes up that substance.

[0061] The prompt may include instructions to extract characteristics from the entire text of the patent document, or the prompt may include a specification of the section of the patent document from which to extract characteristics. The prompt may include instructions to extract a predetermined number of properties or all of the properties from among the properties described in the patent document for a single substance. The selection criteria for selecting a predetermined number of properties may include properties that appear most frequently in the patent document, or properties that are described in the problems or effects described in the patent document.

[0062] Alternatively, the prompt may include instructions to select a representative characteristic from among the multiple characteristics described in the patent document for a single raw material or material. Selection criteria for the representative characteristic may include being the characteristic that appears most frequently in the patent document, or being a characteristic described in the problems or effects described in the patent document.

[0063] The prompt may include a description of the properties. The prompt may also include an exclusion condition, such as excluding composite materials (containing both the target substance and untargeted substances), like a sheet with an aluminum layer laminated onto a polyethylene sheet. Furthermore, the prompt may include instructions for separating and extracting the properties of the raw materials from those of the materials.

[0064] The prompt may include examples of properties. Examples of properties include "stretch ratio," "melt index (MI)," "melting point," "oxygen permeability," "water vapor permeability," "Young's modulus," "haze," "melt index," "melting point," and "water vapor permeability." The examples may also be presented separately for properties per raw material and properties per material.

[0065] The prompt may include instructions for integrating properties having the same units, and instructions for unifying units. For example, "film thickness" expressed in "μm" in a patent document and "thickness" expressed in "cm" in other patent documents may be integrated into "thickness".

[0066] Based on instructions for extracting characteristics contained in the prompt, the artificial intelligence 14 clusters patent documents having the same type of characteristics using terminology related to the characteristics described in the patent documents.

[0067] For example, artificial intelligence 14 vectorizes patent documents based on the characteristics described within them. It then generates clusters to which patent documents with similar vectors belong. Artificial intelligence 14 also assigns characteristic labels to these clusters. If multiple characteristics are extracted from a single patent document, that patent document may be classified into multiple clusters.

[0068] The prompt also includes instructions to extract problems from patent documents for each extracted usage pattern. Instructions for extracting problems may include specifying the fields in the patent document that describe the "problem to be solved by the invention," the "mode for carrying out the invention," and the International Patent Classification, etc., as sources for extracting problems.

[0069] The prompt may also specify a phrase and instruct the system to extract issues from that phrase. The prompt may include instructions to broaden the problem to a higher-level concept. The prompt may also include examples of problems. Examples of problems include "recycling properties" and "heat resistance." If examples of problems are included in the prompt, the prompt may also include instructions to encourage label generation, such as "If your problem does not fit the examples, please generate a problem other than the examples."

[0070] The prompt may include instructions to extract a predetermined number of properties or all of the properties from among the properties described in the patent document for a single substance. The selection criteria for selecting a predetermined number of properties may include properties that appear most frequently in the patent document, or properties that are described in the problems or effects described in the patent document.

[0071] Alternatively, the prompt may include instructions to select a representative problem from among the problems described in the patent document if multiple problems are described therein. The selection criteria for a representative problem may include that it is the characteristic that appears most frequently in the patent document and that it is described in the designated fields such as "Problem to be Solved by the Invention" or "Abstract".

[0072] Based on instructions to extract the issues contained in the prompt, the artificial intelligence 14 clusters patent documents that have the same or similar issues, using terminology related to the issues described in the patent documents. For example, artificial intelligence 14 vectorizes the problem terms described in the patent documents. Then, artificial intelligence 14 classifies the patent documents into clusters based on the similarity of the vectors. Artificial intelligence 14 also assigns problem labels to these clusters. If multiple issues are identified from a single patent document, that patent document may be classified into multiple clusters.

[0073] By extracting the forms of use of a substance from patent documents in this way, and then extracting the characteristics and challenges of the raw material or material for each form of use, the accuracy of the patent map described later can be improved. In other words, even if there is variation in the expression of the characteristics and challenges extracted for each form of use, it is presumed that they have a common meaning. Suppose there are multiple patent documents for the same substance, "polyethylene," where the challenge is "heat resistance." If the form of use is an automobile part, the property is required to withstand prolonged use in a harsh high-temperature environment, and the characteristics satisfy that challenge. If the form of use is a packaging material, the property is required to withstand medium to high temperatures for a relatively short period of time, and the characteristics satisfy that challenge. If the form of use is a packaging material, the property is required to withstand medium to high temperatures for a relatively short period of time, and the characteristics satisfy that challenge. By extracting challenges and characteristics for each form of use, even if there is variation in the expression within the patent documents, "heat resistance" in one form of use is the heat resistance required in other forms of use, and extremely different heat resistance can be excluded. For this reason, even if characteristics and challenges with varying expressions are integrated, the amount of noise included will be reduced. Furthermore, the clusters of characteristics and clusters of challenges extracted for each usage pattern exclude patent documents with extremely different content, resulting in a higher proportion of similar patent documents.

[0074] Furthermore, by extracting characteristics and challenges for each usage pattern, characteristics and challenges specific to that usage pattern are identified. As a result, the number of characteristics and challenges items is reduced. This, in turn, suppresses the complexity of the patent map.

[0075] The patent analysis server 11 obtains a patent list 22, including usage patterns, characteristics, and problems, from the artificial intelligence 14 (step S3). At this time, the patent analysis server 11 may create the patent list 22 by receiving usage patterns, characteristics, and problems from the artificial intelligence 14 in stages. The patent analysis server 11 may send the patent list 22 to the user terminal 12. The user terminal 12 may output the patent list 22 to the display device 107. The user terminal 12 may also send a request to the patent analysis server 11 to create a patent map 30 based on the user's operation to create a patent map.

[0076] The patent analysis server 11 creates a patent map that visualizes the quantitative distribution of patent documents in the patent list 22 (step S4). The patent analysis server 11 can output multiple types of patent maps with different axes. One type of patent map is a patent map based on problems and characteristics for each predetermined usage form. Other patent maps include a patent map based on problems and usage forms, and a patent map based on usage forms and characteristics. As an example, the patent analysis server 11 generates a bubble chart.

[0077] When the patent analysis server 11 creates a patent map for each predetermined usage mode, it calculates the number of patent documents belonging to both a cluster of one problem and a cluster of one characteristic for each predetermined usage mode. The patent analysis server 11 repeatedly calculates the quantity for each item of the problem and each item of the characteristic, and creates a bubble chart based on the calculation results. For example, in a bubble chart with the characteristic cluster on the horizontal axis and the problem cluster on the vertical axis, it calculates the number of patent documents at the intersection of the characteristics and the problem. Then, the patent analysis server 11 creates bubbles corresponding to the quantity of patent documents. The creation method is the same for other patent maps. The patent analysis server 11 may have already created patent maps for each usage mode, patent maps of usage mode and characteristics, and patent maps of usage mode and problems before receiving a creation request from the user terminal 12.

[0078] The patent analysis server 11 sends the patent map data to the user terminal 12, thereby causing the user terminal 12 to output the patent map (step S5). The patent analysis server 11 may send the generated bubble chart to the user terminal 12 as an image file such as a PNG file. Alternatively, the patent analysis server 11 may send chart data for generating the bubble chart to the user terminal 12, and the user terminal 12 may display the bubble chart. For example, the chart data includes the bubble's position (x, y) and size "r".

[0079] The user terminal 12 outputs a patent map using data for displaying the patent map. The user terminal 12 can display a type of patent map specified by the user. At this time, the user may first display a patent map of problem-usage modes and a patent map of usage modes-characteristics. Then, the user may identify usage modes with large bubbles in these maps, specify those usage modes, and display a characteristics-problem map for each usage mode.

[0080] <Displaying the list of items to transition from the patent map> Referring to Figure 5, the display of the list of patent documents transitioning from the patent map will be explained. Here, we will explain the case where the user terminal 12 displays the characteristics-problem map for each usage mode, but other patent maps may also be displayed. The user terminal 12 accepts the selection of elements for the patent map (step S10). The user selects a bubble displayed on the patent map. The user terminal 12 sends the characteristic identifier and the issue identifier associated with the bubble to the patent analysis server 11.

[0081] The patent analysis server 11 creates a list of patent documents corresponding to the elements selected by the user based on the received characteristic identifier and problem identifier (step S11). For example, if the user selects a bubble where the characteristic is "stretch ratio" and the problem is "strength", the patent analysis server 11 creates a patent list 22 consisting of patent documents belonging to the cluster where the characteristic is "stretch ratio" and the problem is "strength".

[0082] The patent analysis server 11 sends the patent list 22 to the user terminal 12, causing the user terminal 12 to output the patent list 22 (step S12). The user terminal 12 displays the received patent list 22. The patent list 22 displayed here is a selection of the patent list 22 shown in Figure 3, narrowed down to match the user's requirements in terms of usage, characteristics, and problems.

[0083] <Patent Map> An example of the patent map 30 will be described with reference to Figures 6 and 7. The patent map 30A in Figure 6 is a map for one application. For example, the application is "packaging material". The horizontal axis of the patent map 30A represents the first indicator, which is the properties, and the vertical axis represents the second indicator, which is the problem. The properties include items such as "stretch ratio", "melt index (MI)", "melting point", "oxygen permeability", and "water vapor permeability". The problems include items such as "heat resistance", "adhesion", "stretchability", "barrier properties", and "strength" such as tear strength. Bubbles 31 are displayed at the intersections of properties and problems. For example, the size of the bubble 31 indicates the number of patent documents where the problem is "heat resistance" and the property is "stretch ratio", the number of patent documents where the problem is "adhesion" and the property is "stretch ratio", etc.

[0084] Figure 7 shows patent map 30B, where the horizontal axis represents usage patterns and the vertical axis represents problems, and patent map 30C, where the horizontal axis represents usage patterns and the vertical axis represents characteristics. These patent maps 30B and 30C allow users to understand which usage patterns have the largest number of patent documents. By checking these patent maps 30B and 30C, users can get an idea of ​​the usage patterns they want to investigate.

[0085] [Effects of the First Embodiment] (1-1) The patent analysis server 11 extracts characteristics and problems for each form of use of the substance extracted from the patent documents. The extracted characteristics and problems are narrowed down to those corresponding to the form of use of the substance. As a result, when characteristics and problems with varying expressions are integrated, the amount of noise included is reduced. This improves the accuracy of the patent map 30 created using the forms of use, and also allows for more efficient reading of the patent documents.

[0086] (1-2) The patent analysis server 11 uses artificial intelligence 14 to extract the properties of raw materials or substances related to the patent documents in the patent list 22 as another indicator. This makes it possible to integrate properties that should be integrated, even if there are variations in the expression and units of the properties, and to exclude properties that should not be integrated into a single property. This further improves the accuracy of the patent map 30.

[0087] (1-3) The patent analysis server 11 uses artificial intelligence 14 to extract the problems of inventions described in the patent documents of the patent list 22 as another indicator. By having artificial intelligence 14 extract the problems described in the patent documents, it becomes possible to integrate problems that should be integrated and to exclude problems that should not be integrated into a single problem. This improves the accuracy of the patent map 30.

[0088] (1-4) The patent analysis server 11 accepts the user's selection of bubbles 31, which are elements of the patent map 30. The patent analysis server 11 also creates a patent list 22 associated with the bubbles 31 and outputs this list to the user terminal 12. As a result, the user only needs to read the patent list 22, which has been narrowed down by usage patterns and other indicators, thus reducing the effort required for reading. In addition, the operation to display the patent list 22 corresponding to the bubbles 31 can be made intuitive.

[0089] (1-5) The patent analysis server 11 extracted patent documents from the patent document database 13 that met the search criteria received from the user terminal 12, and created a search results list 21 using the extracted patent documents. The patent analysis server 11 also obtained a patent list 22 that includes the forms of use of the substances included in the search criteria, using the patent documents included in the search results list 21 as the population. Therefore, the user can avoid the trouble of entering the names of the substances to extract the forms of use separately from the search query 20.

[0090] [Second Embodiment] A second embodiment of the patent analysis program, patent analysis method, and patent analysis system of this disclosure will be described. In this embodiment, the configuration differs from the first embodiment in that, after displaying the patent map 30, the quantitative distribution of patent documents corresponding to the characteristic value conditions is displayed. Hereinafter, parts similar to those in the first embodiment are denoted by the same reference numerals and their detailed descriptions are omitted.

[0091] Referring to Figure 8, the procedure for displaying the quantitative distribution of patent documents that satisfy the characteristic value conditions will be explained. The patent analysis server 11 accepts the characteristic value conditions specified by the user (step S20).

[0092] For example, the user specifies a characteristic item displayed in the patent map 30. As shown in Figure 9, for example, if the user specifies "Melt Index (MI)" as a characteristic item, the user terminal 12 displays an input field 32 for entering that value. The user enters at least one of the lower limit and upper limit values ​​of the Melt Index into the input field 32. The characteristic value entered in the input field 32 is the condition for the characteristic value. The user terminal 12 sends the condition for the characteristic value to the patent analysis server 11. In addition to the input field 32, the lower limit and upper limit values ​​may also be entered in a sidebar that is always displayed on the browser screen.

[0093] The patent analysis server 11 may also display items on the patent map 30 in a different manner from items for which a numerical range can be specified. In the example in Figure 9, items for which a numerical range can be specified are underlined. The patent analysis server 11 may predetermine whether or not a numerical range is described for each characteristic item, targeting patent documents corresponding to the patent map 30. The patent analysis server 11 then changes the display manner for items in patent documents that contain a numerical range.

[0094] When a characteristic is specified, the patent analysis server 11 retrieves patent documents that match the condition of the characteristic value (step S21). The patent analysis server 11 automatically creates a prompt to extract patent documents that meet the characteristic value conditions. The patent analysis server 11 then sends the patent documents belonging to the cluster of the specified characteristics, along with the created prompt, to the artificial intelligence 14. If the user specifies characteristic values ​​in the patent map 30A for each usage mode, the patent analysis server extracts patent documents that meet the characteristic value conditions from the patent documents belonging to the cluster corresponding to that usage mode.

[0095] The prompt includes instructions to extract patent documents that meet the criteria for the characteristic value. The prompt may include instructions for determining whether the characteristics of individual raw materials or components, rather than a composite, meet the conditions for those characteristics.

[0096] In patent documents, characteristic values ​​such as melt index are sometimes expressed numerically, but in other cases, sensory evaluations are performed using evaluation levels other than numerical values, such as "○", "×", "Good", "Poor", and "A" to "C". For this reason, instructions may be given to exclude patent documents that express sensory evaluations from the extraction target. The prompts may include instructions to unify the order of units, and instructions to convert values ​​of the same type of characteristic but measured under different conditions so that they correspond to values ​​measured under the same conditions.

[0097] An example of a prompt is shown below. The example below is a summary of the prompt and may include more detailed extraction conditions, output format specifications, and notes. In this example, the melt index of polyethylene is specified as "10 g / min or higher" as a characteristic condition. Please extract only the patent documents related to polyethylene where the "melt index" is 10 g / min or higher. (subject) • The target material must be "polyethylene" alone. Only the Melt Index, which is listed as a numerical value, is included. (Extraction conditions) • The melt index of polyethylene must be 10 g / min or higher. • Any patent documents containing the following characteristic values ​​will be included in the extraction. Example: 10~30g / min. Example: 10g / min or more, exceeding 10g / min. For example, there are examples where polyethylene levels of 10 g / min or more are listed. • Patent documents containing the following characteristic values ​​will not be extracted. The upper limit is less than 10g / min. (Exclusion conditions) • Do not extract entries where the Melt Index is indicated by something other than a numerical value, such as "○", "×", "Good", "Poor", or "A-C". • Items that cannot be identified as polyethylene alone, such as "PE / PP," "PE+PP," and "composite materials of PE and other resins," will not be extracted. (Interpretation of numerical representations) Expressions such as "approximately," "about," "greater than," and "exceeding" should be interpreted as numerical values ​​while retaining their meaning. (output format) Output in the following format: "Publication number, relevant excerpt, melt index (g / min) value, remarks"

[0098] The patent analysis server 11 receives patent document data that matches the characteristic value conditions from the artificial intelligence 14. Then, the patent analysis server 11 updates the patent map 30 based on the received data and transmits the patent map 30 or map data to the user terminal 12. The user terminal 12 outputs the patent map 30 (step S22).

[0099] Figure 10 shows an example of a patent map 30D displayed on the user terminal 12 in the second embodiment. Bubbles 35 corresponding to the number of extracted patent documents are superimposed on the bubbles 31 of the base patent map 30A. The patent map may also display only the bubbles 35. In research and development, there are sometimes conditions regarding the characteristic values ​​of products that can be developed. Therefore, by showing the quantitative distribution of patent documents that satisfy the characteristic value conditions, it is possible to confirm the quantitative distribution of patent documents that meet the assumptions of the research and development.

[0100] When the user selects bubble 35, a process similar to steps S10 to S12 of the first embodiment is performed, and the patent list 22 of patent documents belonging to the cluster corresponding to bubble 35 is displayed.

[0101] [Effects of the second embodiment] According to the second embodiment, in addition to the effects described in (1-1) to (1-5) above, the following effects can be obtained.

[0102] (2-1) According to the second embodiment, the patent documents narrowed down by the form of use of the substance are used as the population, and the quantitative distribution of patent documents that meet the characteristic value conditions specified by the user is displayed. Therefore, it is possible to check the quantitative distribution of patent documents that are in line with the conditions of raw materials or materials that are prerequisites for research and development, based on each indicator. The user can efficiently read patent documents by checking the patent map 30D and reading only the narrowed down patent documents.

[0103] (2-2) A bubble 35, which represents the quantity of patent documents that satisfy the characteristic value conditions, is superimposed on each of the bubbles 31, which is one of the elements of the patent map 30D. Therefore, by comparing the sizes of bubbles 31 and 35, it is possible to easily compare the quantity of patent documents that satisfy the characteristic value conditions with the quantity of patent documents that do not satisfy the characteristic value conditions.

[0104] [Example of changes] Each of the above embodiments can be implemented with the following modifications. These embodiments and the following modifications can be combined with each other to the extent that they do not contradict each other technically. [Example of change 1: Retrieving search result list 21] The search results list 21 may be a pre-prepared search results list 21 transmitted from the user terminal 12. Alternatively, the search results list 21 may be generated from the user's past searches and registered with the patent analysis server 11. These search results lists 21 do not necessarily have to be linked to the search query 20 used to generate the search results list 21. The patent analysis server 11 may be configured to perform the extraction of usage patterns, issues, and characteristics using the search results list 21 received from the user terminal 12 or a previously used search results list 21.

[0105] [Example of change 2: Patent document search] The patent analysis server 11 may, when a relatively small number of patent documents are specified by the user, extract from the patent document database 13 patent documents in which at least one of the usage forms, characteristics, and problems described in those patent documents is similar. For example, the patent analysis server 11 sends a prompt and the specified patent documents to the artificial intelligence 14, and vectorizes the usage forms, characteristics, and problems of the specified patent documents. The artificial intelligence 14 then uses a clustering method such as the K-means method to create clusters of patent documents with similar vectors based on the vectors of the specified patent documents and the vectors of patent documents stored in the patent document database 13. In this embodiment, the effort of the user to enter appropriate search conditions is eliminated.

[0106] [Example of change 3: Extraction of characteristics] The properties extracted by artificial intelligence 14 may be those of a substance specified by the user at a time separate from the input of the search query 20. For example, if the user searches for patent documents using "polyolefin" as a keyword and specifies "polyethylene" as the material (or raw material) from which to extract properties, artificial intelligence 14 will extract the properties of polyethylene.

[0107] [Example Change 4: Search Extension] The artificial intelligence 14 may have a search extension function that searches external documents to refer to definitions of usage patterns, characteristics and problems, common technical knowledge, etc. The artificial intelligence 14 may autonomously refer to external websites, data collected by crawlers (engines that automatically visit and retrieve web pages), databases, etc. The artificial intelligence 14 may also refer to the URL of a website specified in the prompt.

[0108] [Example of change 5: Patent map 30] In each of the above embodiments, the patent analysis server 11 created a bubble chart as the patent map 30. Alternatively, or in addition to this, a scatter plot or a heat map may be displayed as the patent map 30. In a scatter plot, each patent document is represented by a single point, and the set of points shows the quantitative distribution of patent documents. In a heat map, the quantitative distribution of patent documents is represented not by the size of bubbles or the set of points, but by the intensity of the colors. Also, in each of the above embodiments, the patent map 30 was a two-dimensional map, but it may also be a three-dimensional map with three axes.

[0109] [Example of change 6: Setting conditions for characteristic values] In the second embodiment, after displaying a patent map 30 narrowed down by the form of use of the substance, the conditions for characteristic values ​​were accepted. Then, patent documents corresponding to the conditions for characteristic values ​​were extracted and patent map 30D was displayed. Alternatively, before extracting the form of use of the substance, a step may be performed to obtain a list of patent documents that satisfy the characteristics of the raw materials or components related to the substance. As shown in Figure 11, after obtaining a search result list 21 (step S1), the patent analysis server 11 accepts the characteristics and characteristic value conditions specified by the user (step SA). Then, after extracting the forms of use of the material, the patent analysis server 11 may create a patent map 30 by extracting characteristics and problems for each form of use of the material for patent documents that satisfy the characteristic value conditions (step S2). According to this embodiment, the characteristic values ​​that are the basis for product development can be narrowed down in advance.

[0110] [Example of change 7: User operations before outputting patent map 30] The user operation required from obtaining the search results list 21 to outputting the patent map 30 is not limited. The patent analysis server 11 has the artificial intelligence 14 perform at least the tasks of extracting usage patterns, extracting characteristics, and extracting problems. The patent analysis server 11 may send a prompt containing instructions for one task to the artificial intelligence 14 to obtain the extraction result, and then send another prompt containing instructions for the next task, or it may send a prompt containing multiple of the above tasks. For example, the patent analysis server 11 may have the artificial intelligence 14 perform all tasks in response to receiving a search query 20 from the user terminal 12, and then use the extraction results of the artificial intelligence to create the patent map 30 and send it to the user terminal 12. The patent analysis server 11 may also output the usage patterns of the patent documents to the user terminal 12 before displaying the patent map 30. Then, it may accept the user's specification of usage patterns and create a patent map 30 for those usage patterns. In addition to usage patterns, the patent analysis server 11 may also display characteristics and problems to the user terminal 12 whenever they are extracted. In each of the above embodiments, the patent analysis server 11 is said to create the patent map 30, but the artificial intelligence 14 may also be made to create the patent map 30. In this case as well, when the patent analysis server 11 receives the patent map 30, or in response to an operation by a user who has viewed the patent map 30, it obtains the patent list 22 corresponding to the patent map 30.

[0111] [Example of change 8: Extraction by raw material and ingredient] The patent analysis server 11 may extract characteristics by separating them into raw materials and materials using those raw materials, and create a patent map 30 for each raw material and each material. For example, the patent map 30 may be created by separating it into "polyethylene" as a raw material and "packaging materials" using "polyethylene".

[0112] [Example of change 9: Patent analysis server 11] In each of the above embodiments, the patent analysis server 11 performs tasks such as extracting usage patterns, characteristics, and problems, and generating a patent map 30, according to the patent analysis program. Each process performed by the patent analysis server 11 may be distributed across multiple servers. For example, the process of searching for patent documents using the patent document database 13 and the process of extracting usage patterns, etc., using artificial intelligence 14 may be performed on physically different servers.

[0113] [Example of modification 10: Patent analysis system 10] The user terminal 12 may store part or all of the patent analysis program in memory 102. For example, the user terminal 12 may execute part of each process performed by the patent analysis server 11. For example, the user terminal 12 may execute at least one of the following: searching for patent documents using the patent document DB 13, or extracting each indicator using artificial intelligence 14. The user terminal 12 may generate the patent map 30 in cooperation with the patent analysis server 11, or the user terminal 12 may execute each process independently.

[0114] [Example of modification 11: Patent analysis system 10] The patent analysis server 11, the patent document database 13, and the artificial intelligence 14 may be composed of the same server. Alternatively, any two of the patent analysis server 11, the patent document database 13, and the artificial intelligence 14 may be composed of the same server. In each of the above embodiments, the patent analysis system 10 is provided with a patent analysis server 11 and a user terminal 12, but it may also consist of only the patent analysis server 11, the patent analysis server 11 and the patent document database 13, the patent analysis server 11 and the artificial intelligence 14, the patent analysis server 11, the patent document database 13 and the artificial intelligence 14, and so on.

[0115] The embodiments and modifications described above are listed below. [1] A patent analysis program for analyzing patent documents, At least one processor, A step of obtaining a characteristic value condition which is a value of the characteristic, which is a condition specified by the user for the characteristics included in the data of the patent document. A step of obtaining a list of patent documents from the aforementioned patent documents that satisfy the conditions for the characteristic values, A step of obtaining a list of patents from artificial intelligence that includes the forms of use of substances extracted from patent documents that satisfy the aforementioned characteristic value conditions, A step of creating a patent map that visualizes the quantitative distribution of the patent documents in the patent list using the aforementioned usage and other indicators included in the patent documents, A patent analysis program that performs the process of outputting the aforementioned patent map to a user terminal. [Explanation of Symbols]

[0116] 10…Patent Analysis System 11…Patent analysis server 12…User terminal 13…Patent document DB 14… Artificial Intelligence

Claims

1. A patent analysis program for analyzing patent documents, At least one processor, A process of obtaining a patent list including the forms of use of substances extracted from patent documents using artificial intelligence, A step of creating a patent map that visualizes the quantitative distribution of the patent documents in the patent list using the aforementioned usage and other indicators included in the patent documents, A patent analysis program that performs the process of outputting the aforementioned patent map to a user terminal.

2. The aforementioned at least one processor further includes, The patent analysis program according to claim 1, which transmits a prompt to the artificial intelligence to perform the step of obtaining from the artificial intelligence the properties of raw materials or materials relating to the substance contained in the patent documents of the patent list as other indicators.

3. The aforementioned at least one processor further includes, A step of obtaining a characteristic value condition that is a value of the characteristic, which is a condition specified by the user. A patent analysis program according to claim 2, which performs the step of outputting to the user terminal the quantitative distribution of patent documents that satisfy the characteristic value conditions among the patent documents included in the patent list.

4. The patent analysis program according to claim 3, wherein in the step of outputting the quantity distribution of patent documents that satisfy the conditions of the characteristic value, an element indicating the quantity of patent documents that satisfy the conditions of the characteristic value is displayed in each element of the patent map.

5. The aforementioned at least one processor further includes, The patent analysis program according to claim 1, which causes the artificial intelligence to perform the step of obtaining, as other indicators, terms of the problem of the invention contained in the patent documents of the patent list.

6. The aforementioned at least one processor further includes, A step of accepting the user's selection of elements of the patent map, A patent analysis program according to claim 1, which performs the steps of creating a list of patent documents associated with the aforementioned elements and outputting the list to the user terminal.

7. The aforementioned at least one processor further includes, The system executes the process of extracting patent documents that satisfy the search criteria received from the user terminal from a database of stored patent documents, and creating a search result list using the extracted patent documents. The patent analysis program according to claim 1, wherein in the step of obtaining the patent list, the patent documents included in the search results list are used as the population to obtain the patent list including the forms of use of the substance included in the search conditions.

8. The aforementioned at least one processor further includes, Prior to the step of obtaining the patent list including the forms of use of the substance, The patent analysis program according to claim 1, which performs the step of obtaining a list of patent documents that satisfy the characteristics of raw materials or materials related to the aforementioned substance.

9. A patent analysis method for analyzing patent documents, At least one processor, A patented analysis method that performs the steps described in any one of claims 1 to 8.

10. A patent analysis system comprising at least one processor and memory, The at least one processor, A patented analysis system that performs the steps described in any one of claims 1 to 8.

Images