Computer program, information processing method, and information processing device

Abstract

To provide a computer program, an information processing method, and an information processing device that can be expected to assist in adding information such as a tag or a label to textual information including a patent document.SOLUTION: A computer program according to the present embodiment causes a computer to execute processing for: acquiring textual information pertaining to a patent; making a first determination as to whether the acquired textual information belongs to a predetermined technical field; making a second determination as to which of a plurality of detailed fields included in the predetermined technical field the textual information that is determined to belong to the predetermined technical field belongs to; and adding one or more labels corresponding to the detailed field to the textual information, on the basis of a result of the second determination. The second determination is performed using a classification model which has been machine-trained in advance on the basis of learning data in which the textual information and detailed field are associated.SELECTED DRAWING: Figure 9

Description

[Technical Field]

[0001] The present disclosure relates to a computer program, an information processing method, and an information processing device. [Background technology]

[0002] Patent Document 1 proposes a learning module generation device that selects a program that matches learning source information having multiple patent classification information that has patent information and classification information from among multiple learning module generation programs that generate learning modules using different algorithms, and stores the predictive learning modules generated by the selected program in association with the learning source information or the corresponding correspondence information. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2020-077018 Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the present disclosure is to provide a computer program, an information processing method, and an information processing device that are expected to assist in adding information such as tags or labels to text information such as patent documents. [Means for solving the problem]

[0005] A computer program according to one embodiment causes a computer to acquire text information relating to a patent, make a first determination as to whether the acquired text information belongs to a specified technical field, make a second determination as to which of a plurality of detailed fields included in the specified technical field the text information determined to belong to belongs to, and, based on the result of the second determination, execute a process of assigning one or more labels corresponding to the detailed field to the text information, and perform the second determination using a classification model that has been machine-learned in advance based on learning data in which the text information and the detailed field are associated.

[0006] A computer program in one embodiment acquires patent classification information associated with the text information, and in the first judgment, determines whether the text information belongs to the specified technical field based on the acquired patent classification information.

[0007] A computer program according to one embodiment makes the first determination using a learning model that has undergone machine learning in advance based on learning data in which text information and information indicating whether or not the text information belongs to the specified technical field are associated.

[0008] In one embodiment, a computer program classifies, in the second determination, which of a plurality of first detailed fields the text information belongs to using a first classification model that has been machine-learned in advance based on learning data in which the text information and the first detailed field are associated, and classifies which of a plurality of second detailed fields different from the first detailed field the text information belongs to using a second classification model that has been machine-learned in advance based on learning data in which the text information and the second detailed field are associated, and assigns to the text information one or more first labels corresponding to the first detailed field and one or more second labels corresponding to the second detailed field based on the classification results of the first classification model and the classification results of the second classification model.

[0009] A computer program according to one embodiment changes the first label or the second label when the first label and the second label assigned to the text information based on the classification results of the first classification model and the classification results of the second classification model are a predetermined combination.

[0010] In one embodiment of the computer program, the classification model is a model that outputs multiple predicted probabilities corresponding to multiple detailed fields, and a label corresponding to the detailed field with a high predicted probability output by the classification model is assigned to the text information.

[0011] A computer program according to one embodiment accepts corrections to labels assigned to text information and retrains the classification model based on training data in which the text information and detailed fields corresponding to the corrected labels are associated.

[0012] An information processing method according to one embodiment includes an acquisition step in which an information processing device acquires text information relating to a patent, a first determination step in which the acquired text information belongs to a specified technical field, a second determination step in which the text information determined to belong to the specified technical field belongs to one of a plurality of detailed fields included in the specified technical field, and a label assignment step in which the text information is assigned one or more labels corresponding to the detailed field based on the determination result of the second determination step, and in the second determination step, classification is performed using a classification model that has been machine-learned in advance based on learning data in which the text information and the detailed field are associated.

[0013] An information processing device according to one embodiment includes a processing unit, which acquires text information relating to a patent, makes a first determination as to whether the acquired text information belongs to a specified technical field, and makes a second determination as to which of a plurality of detailed fields included in the specified technical field the text information determined to belong to belongs to, and makes the second determination using a classification model that has been machine-learned in advance based on learning data in which the text information and the detailed fields are associated, and assigns one or more labels corresponding to the detailed fields to the text information based on the results of the second determination.

[0014] In one embodiment, it is expected that the addition of information such as tags or labels to text information such as patent documents will be supported. [Brief explanation of the drawings]

[0015] [Figure 1] 1 is a schematic diagram illustrating an overview of an information processing system according to an embodiment of the present invention. [Figure 2] 1 is a block diagram showing an example of a configuration of an information processing device according to an embodiment of the present invention; [Figure 3] FIG. 2 is a block diagram showing the configuration of a terminal device according to the present embodiment. [Figure 4] FIG. 1 is a schematic diagram illustrating an example of the configuration of a learning model for making a first determination regarding a chemistry-related field. [Figure 5] FIG. 10 is a schematic diagram illustrating an example of the configuration of a learning model for making a second determination regarding technology classification labels in the air conditioning-related field. [Figure 6] FIG. 10 is a schematic diagram illustrating an example of the configuration of a learning model for making a second determination regarding structural labels in the air conditioning-related field. [Figure 7] 10A and 10B are schematic diagrams for explaining an example of a determination condition and a correction method for a combination of a technology classification label and a structure classification label. [Figure 8] FIG. 10 is a schematic diagram illustrating an example of the configuration of a learning model for making a second determination regarding classification labels in a chemistry-related field. [Figure 9]10 is a flowchart showing an example of a procedure of a label assignment process performed by the information processing device according to the present embodiment. [Figure 10] 10 is a flowchart illustrating an example of a procedure of a process performed by a terminal device according to the present embodiment. [Figure 11] 10 is a flowchart illustrating an example of a procedure for a re-learning process of a learning model performed by the information processing device according to the present embodiment. [Figure 12] FIG. 10 is a schematic diagram illustrating a configuration of an information processing system according to a modified example. DETAILED DESCRIPTION OF THE INVENTION

[0016] Specific examples of the information processing system according to the present embodiment will be described below with reference to the drawings. The present technology is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope of the claims.

[0017] <System Overview> FIG. 1 is a schematic diagram for explaining an overview of an information processing system according to this embodiment. The information processing system according to this embodiment classifies text information related to patents, such as patent documents (hereinafter referred to as patent text), by classifying the technical field to which the content of the patent text belongs, and assigns additional information (hereinafter referred to as classification labels), such as tags or labels, that represent the classification results. In this embodiment, tags or labels are information such as words, word combinations, or sentences (shorter than the original text) that succinctly express the content of the text information, and are, for example, information on character strings of several to several tens of characters. In this embodiment, classification labels are information such as words, word combinations, or sentences that succinctly express the technical field of an invention described in text information such as patent documents. In this embodiment, patent text is text information included in documents that publish the content of patent applications, utility model registration applications, etc., and may include text information such as published patent gazettes, patent gazettes, and utility model gazettes.

[0018] In this example, the information processing system is configured to include an information processing device 1, a terminal device 3, and a patent management server device 5. The terminal device 3 according to this embodiment is a device used by a user, and may be configured using a general-purpose information processing device such as a personal computer, a smartphone, or a tablet terminal. The user uses the terminal device 3 to upload data (files) of patent documents to which the user wishes to assign a classification label to the information processing device 1. Alternatively, the user may notify the information processing device 1 of identification information (such as application number, publication number, publication number, announcement number, or patent number) of the patent documents to which the user wishes to assign a classification label.

[0019] The information processing device 1 according to this embodiment can be configured by installing a computer program according to this embodiment in a general-purpose information processing device such as a server computer or a personal computer. The information processing device 1 acquires patent document data from a terminal device 3, classifies the technical field based on the content of the acquired patent document, and assigns a classification label to the patent document. If the terminal device 3 acquires patent document identification information rather than patent document data, the information processing device 1 acquires the corresponding patent document data from a patent management server device 5 managed and operated by the Japan Patent Office or the like. The information processing device 1 requests the patent management server device 5 to transmit the patent document by specifying the identification information acquired from the terminal device 3. In response to this document request, the patent management server device 5 reads the patent document data stored in a patent document database (DB) 6 and transmits it to the information processing device 1. The information processing device 1 receives the patent document data from the patent management server device 5, classifies the technical field based on the patent document information, and assigns a classification label to the patent document.

[0020] In this embodiment, the information processing device 1 classifies acquired patent documents as belonging to an "air conditioning-related field," a "chemistry-related field," or other fields. The information processing device 1 first determines whether the patent documents belong to an air conditioning-related field, and then determines whether the patent documents determined not to belong to the air conditioning-related field belong to a chemistry-related field, thereby classifying the patent documents as belonging to the air conditioning-related field, chemistry-related field, or other fields. The information processing device 1 further classifies the patent documents classified as the air conditioning-related field or the chemistry-related field in more detail by field and assigns classification labels to them. In this way, the information processing device 1 according to this embodiment performs two-stage judgments on the patent documents: a first judgment to determine whether the acquired patent documents belong to a desired technical field, and a second judgment to perform a more detailed classification of the patent documents belonging to the desired technical field, thereby improving the accuracy of assigning classification labels to the patent documents.

[0021] The first determination of whether a patent document belongs to a desired technical field can be a rule-based determination based on information contained in the patent document or information previously attached to the patent document, or a determination using a learning model previously generated by machine learning. The information processing device 1 according to this embodiment performs a rule-based determination of whether a patent document obtained from a user or the patent management server device 5 belongs to an air conditioning-related field. For patent documents determined not to belong to the air conditioning-related field, the information processing device 1 determines whether the patent document belongs to a chemistry-related field by using a learning model previously subjected to machine learning to perform a binary classification of whether the patent document belongs to a chemistry-related field. In this embodiment, the information processing device 1 assigns a classification label of "unclassified" to patent documents determined not to belong to either the air conditioning-related field or the chemistry-related field.

[0022] The information processing device 1 may first determine whether an acquired patent document belongs to a chemistry-related field, and then determine whether a patent document determined not to be a chemistry-related field belongs to an air conditioning-related field. The order of the technical fields in which the first determination is made is determined appropriately by the designer of the information processing system, etc. In this embodiment, the air conditioning-related field and the chemistry-related field are listed as two technical fields in which the first determination is made. However, this is merely an example and is not limiting. The information processing device 1 may make a determination for a technical field other than the air conditioning-related field and the chemistry-related field. The information processing device 1 may be configured to make a determination for one technical field, or may be configured to make a determination for three or more technical fields. The information processing device 1 may make a determination for each technical field using either a rule-based determination or a determination using a learning model.

[0023] The second determination, which performs detailed classification of patent documents determined to belong to a predetermined technical field, is performed using a learning model that has undergone machine learning in advance. The number of classification labels assigned to one patent document may be one or more. In this embodiment, the information processing device 1 assigns two classification labels, one each of a technology classification label and a structure classification label, to patent documents determined to belong to the air conditioning-related field. The information processing device 1 selects one of 16 predetermined labels as a technology classification label and one of four predetermined labels as a structure classification label, thereby assigning two classification labels to one patent document. For patent documents determined to belong to the chemistry-related field, the information processing device 1 selects three of 24 predetermined labels related to the chemistry-related field, thereby assigning three classification labels to one patent document. The types and numbers of classification labels assigned to these patent documents are merely examples and are not limited thereto.

[0024] That is, the information processing device 1 according to this embodiment assigns two classification labels in the air conditioning-related field, three classification labels in the chemistry-related field, or one unclassified classification label to the acquired patent text. The information processing device 1 transmits label information relating to the classification labels assigned to the patent text to the terminal device 3. The terminal device 3 receives the label information from the information processing device 1 and displays the received label information.

[0025] <Device configuration> 2 is a block diagram showing an example of the configuration of an information processing device 1 according to this embodiment. The information processing device 1 according to this embodiment can be realized by installing a predetermined application program or the like in a general-purpose information processing device such as a personal computer or a server computer. The information processing device 1 according to this embodiment is configured to include a processing unit (processor) 11, a memory unit (storage) 12, and a communication unit (transceiver) 13. In this embodiment, the processing will be described as being performed by one information processing device 1, but the processing of the information processing device 1 may be distributed among a plurality of devices.

[0026] The processing unit 11 is configured using an arithmetic processing device such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit) or a quantum processor, a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The processing unit 11 reads and executes a program 12a stored in the storage unit 12 to perform various processes such as a process of acquiring patent texts and a process of assigning classification labels to patent texts.

[0027] The storage unit 12 is configured using a large-capacity storage device such as a hard disk or an SSD (Solid State Drive). The storage unit 12 stores various programs executed by the processing unit 11 and various data required for the processing of the processing unit 11. In this embodiment, the storage unit 12 stores a program 12a executed by the processing unit 11. The storage unit 12 is provided with a model information storage unit 12b that stores information related to a trained learning model used in the processing performed by the information processing device 1.

[0028] In this embodiment, the program (computer program, program product) 12a is provided in a form recorded on a recording medium 99 such as a memory card or an optical disc, and the information processing device 1 reads the program 12a from the recording medium 99 and stores it in the storage unit 12. However, the program 12a may also be written to the storage unit 12 during the manufacturing stage of the information processing device 1. The program 12a may be distributed by a remote server device or the like and acquired by the information processing device 1 via communication. The program 12a may be recorded on the recording medium 99 and read by a writing device and written to the storage unit 12 of the information processing device 1. The program 12a may be provided in a form distributed via a network, or may be provided in a form recorded on the recording medium 99.

[0029] The model information storage unit 12b stores information about learning models that have been previously subjected to machine learning. The information about the learning models may include information indicating the configuration of the learning models, and information such as the values ​​of internal parameters determined by machine learning. In this embodiment, the model information storage unit 12b stores information about a learning model that classifies whether patent documents belong to a chemistry-related field, a learning model that classifies patent documents in the air conditioning-related field using 16 technology classification labels, a learning model that classifies patent documents in the air conditioning-related field using four structural classification labels, a learning model that classifies patent documents in the chemistry-related field using 24 classification labels, and a learning model that translates patent documents.

[0030] In this embodiment, information about the learning model is stored in the information processing device 1, and processing using the learning model is performed by the information processing device 1, but this is not limited to this. Information about the learning model may be stored in a device different from the information processing device 1, and this device may perform processing using the learning model, and the information processing device 1 may acquire the processing results from this device. Machine learning processing for generating the learning model may be performed by the information processing device 1, or may be performed by a device different from the information processing device 1.

[0031] The communication unit 13 transmits and receives data to and from devices such as the terminal device 3 and the patent management server device 5 via a wired or wireless network N. In this embodiment, the information processing device 1 can acquire patent texts or patent text identification information, etc. from the terminal device 3 and transmit information regarding classification labels assigned to the patent texts to the terminal device 3 by the communication unit 13 communicating with one or more terminal devices 3. The information processing device 1 can request the patent management server device 5 to transmit patent text data corresponding to the patent text identification information acquired from the terminal device 3 and receive the patent text data transmitted from the patent management server device 5 in response to the request by the communication unit 13 communicating with the patent management server device 5. The communication unit 13 transmits data provided by the processing unit 11 to other devices, receives data from other devices, and provides the received data to the processing unit 11.

[0032] The storage unit 12 may be an external storage device connected to the information processing device 1. The information processing device 1 may be a multi-computer including multiple computers, or may be a virtual machine virtually constructed by software. The information processing device 1 is not limited to the above configuration, and may include a reading unit that reads information stored in a portable storage medium, an input unit that accepts operation input, or a display unit that displays images.

[0033] In the information processing device 1 according to this embodiment, the processing unit 11 reads and executes the program 12a stored in the storage unit 12, thereby realizing a patent text acquisition unit 11a, a technical field determination unit 11b, an air conditioning label assignment unit 11c, a chemical label assignment unit 11d, a label information transmission processing unit 11e, and the like as software functional units in the processing unit 11. In the figure, functional units related to the process of assigning classification labels to patent texts are shown as functional units of the processing unit 11, and functional units related to other processes are not shown.

[0034] The patent text acquisition unit 11a performs a process of acquiring data of a patent text to be assigned a classification label. The patent text acquisition unit 11a communicates with the terminal device 3 via the communication unit 13 and acquires the patent text data by receiving the patent text data transmitted from the terminal device 3. When the patent text acquisition unit 11a receives identification information of a patent text from the terminal device 3, it communicates with the patent management server device 5 via the communication unit 13 and requests the patent management server device 5 to transmit the patent text data corresponding to the identification information acquired from the terminal device 3. The patent text acquisition unit 11a acquires the patent text data by receiving via the communication unit 13 the data transmitted by the patent management server device 5 in response to this request. The patent text acquisition unit 11a stores the acquired patent text data in the memory unit 12.

[0035] In this embodiment, the patent text data acquired by the patent text acquisition unit 11a may be data in a format that includes text information and image information in one file, such as PDF (Portable Document Format), or may be data divided into multiple files, such as text data and image data such as JPEG (Joint Photographic Experts Group) or GIF (Graphics Interchange Format). In this embodiment, the patent text acquisition unit 11a acquires at least information on patent text, and does not necessarily have to acquire images related to the patent.

[0036] The technical field determination unit 11b performs a process of determining whether or not the patent text acquired by the patent text acquisition unit 11a belongs to a predetermined technical field. In this embodiment, the technical field determination unit 11b first determines whether or not the patent text belongs to an air conditioning-related field using a rule-based method, and then determines whether or not the patent text determined not to belong to the air conditioning-related field belongs to a chemistry-related field using a learning model that has been previously subjected to machine learning.

[0037] Patent documents such as patent documents published by the Patent Office and other organizations are pre-assigned classification information such as the International Patent Classification (IPC), the Cooperative Patent Classification (CPC), and the File Index (FI). The information processing device 1 acquires the classification information contained in the patent document and determines whether it matches classification information predetermined for the air conditioning field, thereby determining whether the patent document belongs to the air conditioning field.

[0038] In this embodiment, the information processing device 1 receives a patent document as an input and stores information about a learning model that has been machine-learned in advance to perform a binary classification of whether or not the patent document belongs to a chemistry-related field in the model information storage unit 12b. The technical field determination unit 11b inputs a patent document that has been determined not to belong to the air conditioning-related field into this learning model and obtains the value output by the learning model, thereby determining whether or not the patent document belongs to a chemistry-related field.

[0039] The air conditioning label assignment unit 11c assigns two types of classification labels, a technology classification label and a structural classification label, to patent documents determined by the technology field determination unit 11b to belong to the air conditioning-related field. In this embodiment, the information processing device 1 receives patent documents as input and stores in the model information storage unit 12b information about a learning model that has been machine-learned in advance to perform a multi-value classification of the patent documents as to which of 16 predetermined technology classification labels the patent documents belong to. The air conditioning label assignment unit 11c inputs patent documents determined to belong to the air conditioning-related field into the learning model and obtains the value output by the learning model, thereby determining which technology classification label in the air conditioning-related field the patent documents should be assigned. In this embodiment, the information processing device 1 receives patent documents as input and stores in the model information storage unit 12b information about a learning model that has been machine-learned in advance to perform a multi-value classification of the patent documents as to which of four predetermined structural classification labels the patent documents belong to. The air conditioning label assignment unit 11c inputs a patent document that has been determined to belong to the air conditioning-related field into this learning model and obtains the value output by the learning model, thereby determining which structural classification label in the air conditioning-related field should be assigned to this patent document.

[0040] The air conditioning label assignment unit 11c determines the technology classification label and the structure classification label to be assigned to the patent document using the learning model, and then determines whether the combination of the technology classification label and the structure classification label is an appropriate combination, and if it is not an appropriate combination, corrects the classification label. In this embodiment, inappropriate combinations of technology classification labels and structure classification labels and correction contents of the structure classification label for the inappropriate combination are predetermined. The air conditioning label assignment unit 11c determines whether the combination corresponds to a predetermined inappropriate combination, and if it corresponds to an inappropriate combination, corrects the structure classification label to the predetermined content.

[0041] The chemical labeling unit 11d performs a process of assigning three classification labels to patent documents determined by the technical field determination unit 11b to belong to a chemistry-related field. In this embodiment, the information processing device 1 receives a patent document as input and stores, in the model information storage unit 12b, information on a learning model that has been pre-machined to perform a multi-value classification of the patent document to one of 24 classification labels related to a predetermined chemistry-related field. This learning model outputs 24 values ​​corresponding to the 24 classification labels, with larger values ​​indicating a higher probability of the document belonging to the corresponding classification label. The chemical labeling unit 11d inputs the patent document determined to belong to a chemistry-related field into this learning model, obtains and compares the 24 values ​​output by the learning model, and obtains the top three classification labels in descending order of value, thereby determining the classification label in a chemistry-related field to assign to the patent document.

[0042] The above-described multiple learning models used to assign classification labels in this embodiment accept English patent text as input. Therefore, the information processing device 1 determines whether the acquired patent text is in English, translates patent text in a language other than English, such as Japanese, into English, and inputs the English patent text into the learning model. The translation of patent documents into English may be performed by the information processing device 1 or by a device different from the information processing device 1. When the information processing device 1 performs the translation, information about the learning model that will perform the translation is stored in advance in the model information storage unit 12b. When another device performs the translation, the information processing device 1 transmits the original text to the other device to request a translation, and obtains the translated text transmitted as a response from the other device.

[0043] The label information transmission processing unit 11e performs processing to transmit information about the classification labels assigned to the patent text to the terminal device 3 that requested the generation of the classification labels. The label information transmission processing unit 11e acquires the technical classification labels and structural classification labels in the air conditioning-related field assigned to the patent text by the air conditioning label assignment unit 11c, the classification labels in the chemistry-related field assigned by the chemical label assignment unit 11d, or the unclassified labels assigned to patent texts in fields other than the air conditioning-related field and the chemistry-related field by the technical field determination unit 11b, and transmits information associating the identification information of the patent text with the classification labels assigned thereto to the terminal device 3 as the result of assigning the classification labels.

[0044] 3 is a block diagram showing the configuration of a terminal device 3 according to this embodiment. The terminal device 3 according to this embodiment is configured to include a processing unit (processor) 31, a memory unit (storage) 32, a communication unit (transceiver) 33, a display unit (display) 34, and an operation unit 35. The terminal device 3 is used by a user who performs work involving patent documents, and can be configured using an information processing device such as a personal computer, a smartphone, or a tablet terminal device.

[0045] The processing unit 31 is configured using an arithmetic processing unit such as a CPU or an MPU, a ROM, etc. The processing unit 31 reads and executes a program 32a stored in the storage unit 32, thereby performing various processes such as a process of accepting input of information related to a patent document for which a classification label is to be generated, and a process of displaying information related to the classification label assigned to the patent document.

[0046] The storage unit 32 is configured using a non-volatile memory element such as a flash memory or a storage device such as a hard disk. The storage unit 32 stores various programs executed by the processing unit 31 and various data required for processing by the processing unit 31. In this embodiment, the storage unit 32 stores the program 32a executed by the processing unit 31. In this embodiment, the program 32a is distributed by a remote server device or the like, and the terminal device 3 acquires the program 32a via communication and stores it in the storage unit 32. However, the program 32a may also be written to the storage unit 32 during the manufacturing stage of the terminal device 3. The program 32a may be read by the terminal device 3 from a recording medium 98 such as a memory card or an optical disk and stored in the storage unit 32. The program 32a may also be read from the recording medium 98 by a writing device and written to the storage unit 32 of the terminal device 3. The program 32a may be provided in the form of distribution via a network or in the form of being recorded on the recording medium 98.

[0047] The communication unit 33 communicates with various devices via a network N including a mobile phone communication network, a wireless LAN, the Internet, etc. In this embodiment, the communication unit 33 communicates with the information processing device 1 via the network N. The communication unit 33 transmits data provided by the processing unit 31 to other devices, and provides data received from other devices to the processing unit 31.

[0048] The display unit 34 is configured using a liquid crystal display or the like, and displays various images, characters, etc. based on processing by the processing unit 31. The operation unit 35 accepts user operations and notifies the processing unit 31 of the accepted operations. The operation unit 35 accepts user operations via input devices such as mechanical buttons or a touch panel provided on the surface of the display unit 34. The operation unit 35 may be input devices such as a mouse and a keyboard, and these input devices may be configured to be detachable from the terminal device 3.

[0049] In the terminal device 3 according to this embodiment, the processing unit 31 reads and executes the program 32a stored in the storage unit 32, thereby realizing the patent document acquisition unit 31a, the display processing unit 31b, etc. as software functional units in the processing unit 31. The program 32a may be a program dedicated to the information processing system according to this embodiment, or may be a general-purpose program such as an internet browser or a web browser.

[0050] The patent text acquisition unit 31a performs processing to acquire information about one or more patent texts to which a classification label is to be assigned. If the user already has the data (file) of the patent text, the patent text acquisition unit 31a acquires the patent text by accepting a data selection operation from the user, and transmits the acquired patent text data to the information processing device 1 to request the assignment of a classification label. If the user does not have the data of the patent text, the patent text acquisition unit 31a acquires information about the patent text by accepting input of identification information of the patent text (application number, publication number, publication number, announcement number, patent number, etc.) from the user, and transmits the acquired identification information of the patent text to the information processing device 1 to request the assignment of a classification label.

[0051] The display processing unit 31b performs processing to display various characters, images, etc. on the display unit 34. In this embodiment, the display processing unit 31b displays information regarding classification labels assigned to patent texts on the display unit 34. The display processing unit 31b receives, via the communication unit 13, information regarding classification labels sent by the information processing device 1 in response to a request from the patent text acquisition unit 31a, and based on the received information, displays a list of identification information for the patent text specified by the user in association with one or more classification labels assigned to the patent text by the information processing device 1.

[0052] <First judgment> The information processing device 1 according to this embodiment makes a first determination as to whether or not a patent text acquired from the terminal device 3 or the patent management server device 5 belongs to a predetermined technical field. In this embodiment, the information processing device 1 makes a first determination as to whether or not the patent text belongs to an air conditioning-related field, and a first determination as to whether or not the patent text belongs to a chemistry-related field.

[0053] In this embodiment, the information processing device 1 performs a first determination of whether a patent document belongs to an air conditioning-related field by a rule-based determination based on information contained in the patent document. The information processing device 1 acquires classification information such as IPC, CPC, or FI from the target patent document and determines whether the acquired classification information matches classification information previously defined as belonging to the air conditioning-related field, thereby determining whether the target patent document belongs to the air conditioning-related field. Because patent documents do not necessarily have IPC, CPC, and FI classification information, the information processing device 1 attempts to acquire the IPC from the patent document. If the IPC cannot be acquired, the information processing device 1 attempts to acquire the CPC. If the CPC cannot be acquired, the information processing device 1 attempts to acquire the FI. The information processing device 1 attempts to acquire the IPC, CPC, and FI in that order, and performs a first determination using the acquired information.

[0054] In this embodiment, the information processing device 1 determines that a patent document belongs to an air conditioning-related field when the classification information acquired from the patent document matches the classification information shown below. Examples of classification information for the air conditioning-related field are predetermined, such as "F24F, F25B, H02M, G06N, B60H, F04B, F04C, H02K, H02P, F28F, F28D, B65D, A61L, F28G, F24H, G10L." The classification information that is to be considered to belong to the air conditioning-related field is determined in advance by a designer of the system, etc., and is stored in advance as a determination condition in the information processing device 1. The classification information for the air conditioning-related field described above is merely an example and is not limited to this.

[0055] In the first determination, the information processing device 1 according to this embodiment first determines whether the acquired patent text belongs to an air conditioning-related field, and then determines whether the patent text determined not to belong to the air conditioning-related field belongs to a chemistry-related field. In this embodiment, the information processing device 1 determines whether the patent text belongs to a chemistry-related field by using a learning model that has been previously machine-learned.

[0056] FIG. 4 is a schematic diagram illustrating an example of the configuration of a learning model for making a first determination regarding a chemistry-related field. The learning model 71 used by the information processing device 1 according to this embodiment for making the first determination regarding a chemistry-related field is a learning model that has undergone machine learning in advance so as to accept a patent document as input and output a numerical value indicating whether or not the patent document belongs to a chemistry-related field. The learning model 71 may be a learning model such as a Transformer with an attention mechanism in a large-scale neural network, BERT (Bidirectional Encoder Representations from Transformers), or GPT (Generative Pre-trained Transformer). The learning model 71 is generated in advance by machine learning using training data (teacher data) that associates patent documents with flags (correct values) indicating whether or not the patent documents belong to a chemistry-related field.

[0057] The patent text input to the learning model 71 may be all of the text data contained in the patent publication, or may be only a portion of the text data, such as an abstract, contained in the patent publication. The numerical value output by the learning model 71 is, for example, a decimal value ranging from 0 to 1, and the larger the value, the more likely the patent text is to belong to a chemistry-related field. The output value of the learning model 71 is a value that may be called a predicted probability or confidence level. The information processing device 1 inputs the patent text to the learning model 71, obtains the numerical value, such as the predicted probability, output by the learning model 71, and can determine that the patent text belongs to a chemistry-related field if this numerical value exceeds a predetermined threshold (for example, 0.7).

[0058] <Second Judgment (Air Conditioning Related Field)> The information processing device 1 according to this embodiment performs a second determination process to assign two types of labels, a technology classification label and a structure classification label, to patent documents determined in the first determination process to belong to an air conditioning-related field. Sixteen types of technology classification labels according to this embodiment are predefined, for example, "structural design, refrigerant control, compression / expansion, power electronics, refrigerant circuit, air conditioning control, heat exchange, motor, humidity control, data processing, IAQ (Indoor Air Quality), ventilation, sensing, facility management, ventilation / gas management, and communication." Four types of structure classification labels according to this embodiment are predefined, for example, "control, material, structure, and manufacturing." The information processing device 1 determines which of these 16 technology classification labels the patent document belongs to and which of the four structure labels it belongs to. The information processing device 1 performs the determination of these technology classification labels and structure classification labels using a learning model that has been previously subjected to machine learning.

[0059] FIG. 5 is a schematic diagram illustrating an example of the configuration of a learning model for performing a second determination regarding technology classification labels in the air conditioning-related field. The technology classification model 75 used by the information processing device 1 according to this embodiment for the second determination of assigning technology classification labels is a learning model that has undergone machine learning in advance so as to accept a patent document as input and output 16 numerical values ​​indicating the predicted probabilities of whether or not the patent document corresponds to one of 16 types of technology classification labels. The technology classification model 75 may employ a learning model such as Transformer, BERT, or GPT. The technology classification model 75 is generated in advance by machine learning using learning data (teacher data) that associates a patent document with information indicating which technology classification label the patent document corresponds to (information of correct answer values ​​in which the corresponding technology classification label is set to 1 and other technology classification labels are set to 0).

[0060] The patent text input to the technology classification model 75 may be all text data contained in the patent publication, or may be only a portion of the text data, such as an abstract, contained in the patent publication. The 16 numerical values ​​output by the technology classification model 75 correspond to 16 types of technology classification labels. Each numerical value is a decimal value ranging from 0 to 1, and the larger the value, the more likely the patent text belongs to the corresponding technology classification label. The technology classification model 75 calculates and outputs 16 numerical values ​​so that the sum of these 16 numerical values ​​is 1. The 16 output values ​​of the technology classification model 75 are values ​​that may be called predicted probabilities or confidence levels. The information processing device 1 inputs the patent text to the technology classification model 75, obtains the 16 numerical values, such as predicted probabilities, output by the technology classification model 75, and selects the largest value from the 16 numerical values ​​to determine which technology classification label the patent text belongs to.

[0061] FIG. 6 is a schematic diagram illustrating an example of the configuration of a learning model for performing a second determination regarding structural labels in the air conditioning-related field. The structural classification model 76 used by the information processing device 1 according to this embodiment for the second determination of assigning structural classification labels is a learning model that has undergone machine learning in advance so as to accept a patent document as input and output four numerical values ​​indicating the predicted probabilities of whether or not the patent document corresponds to one of four structural classification labels. The structural classification model 76 may employ a learning model such as Transformer, BERT, or GPT. The structural classification model 76 is generated in advance by machine learning using learning data (teacher data) that associates a patent document with information indicating which structural classification label the patent document corresponds to (information of correct answer values ​​in which the corresponding structural classification label is set to 1 and other structural classification labels are set to 0).

[0062] The patent text input to the structural classification model 76 may be all text data contained in the patent publication, or may be only a portion of the text data, such as an abstract, contained in the patent publication. The information input to the structural classification model 76 and the information input to the technology classification model 75 may be the same information, or may be different information extracted from different parts of the patent text. The four numerical values ​​output by the structural classification model 76 correspond to four structural classification labels. Each numerical value is a decimal value ranging from 0 to 1, and the larger the value, the more likely the patent text belongs to the corresponding structural classification label. The structural classification model 76 calculates and outputs four numerical values ​​so that the sum of these four numerical values ​​is 1. The four output values ​​of the structural classification model 76 are values ​​that may be referred to as predicted probabilities or confidence levels. The information processing device 1 inputs the patent text to the structural classification model 76, obtains the four numerical values, such as predicted probabilities, output by the structural classification model 76, and selects the largest value from the four numerical values ​​to determine which structural classification label the patent text belongs to.

[0063] The information processing device 1 according to this embodiment determines whether or not there is a contradiction in the combination of a technology classification label obtained using the technology classification model 75 and a structure classification label obtained using the structure classification model 76, depending on whether or not the combination corresponds to a predetermined combination. If the information processing device 1 determines that there is a contradiction in the combination of a technology classification label and a structure classification label, it corrects the structure tag to resolve the contradiction. The designer or the like of this system determines in advance what combinations of technology classification labels and structure classification labels are considered to be contradictory, and how to correct the contradictory combinations, and information regarding the contradiction determination conditions, label correction methods, etc. is stored in the information processing device 1.

[0064] FIG. 7 is a schematic diagram illustrating an example of a determination condition and a correction method for a combination of a technology classification label and a structure classification label. The table shown in this figure defines structure classification labels that are inconsistent with a technology classification label as a combination, and the corrected structure classification label when the structure classification label for this combination is corrected. A combination of the technology classification label "refrigerant control" and the structure classification label "material," "structure," or "manufacturing" is inconsistent, and it is determined that the structure classification label should be corrected to "control" in this combination. Similarly, for each of the technology classification labels "air conditioning control," "hot and cold water control," "system control," and "flow control," there is a contradiction when combined with the structure classification label "material," "structure," or "manufacturing," and it is determined that the structure classification label should be corrected to "control" in this combination.

[0065] While this diagram shows five examples of contradictory combinations of technology classification labels and structure classification labels, the contradictory combinations are not limited to those shown. While this diagram shows an example in which the structure classification label is corrected to "control," this is not limited to this, and combinations in which the structure classification label is corrected to "material," "structure," or "manufacturing" can also be defined in the same way. The designers of this system can determine as appropriate which combinations of technology classification labels and structure classification labels are considered contradictory, and how to correct the structure classification label in the case of a contradictory combination.

[0066] If the combination of the technology classification label and the structural classification label obtained from the learning model for the patent text is not inconsistent, the information processing device 1 assigns the obtained technology classification label and the structural classification label to the patent text. If the combination of the technology classification label and the structural classification label is inconsistent, the information processing device 1 corrects the structural classification label and assigns the obtained technology classification label and the corrected structural classification label to the patent text.

[0067] <Second Judgment (Chemistry-related fields)> The information processing device 1 according to this embodiment performs a second determination process in which one or more classification labels are assigned to patent documents determined in the first determination process to belong to a chemistry-related field. In this embodiment, 24 classification labels for the chemistry-related field are predefined, for example, "PTFE (Poly Tetra Fluoro Ethylene), perfluororubber, fluororubber, paint, weather-resistant paint, optical material, liquid crystal, fuel cell, battery material, resist, fluorosilicone, laminate (fuel tube), polymer blend, functional / composite / other, repellent, surface treatment agent (functional material), etchant, fluorocarbon gas, gas / other, organic compound, inorganic compound, recycle, water treatment, melt summary." The information processing device 1 uses a learning model that has undergone machine learning in advance to determine which of these 24 classification labels the patent document belongs to.

[0068] FIG. 8 is a schematic diagram illustrating an example of the configuration of a learning model for making a second determination regarding classification labels in chemistry-related fields. The classification model 77 used by the information processing device 1 according to this embodiment for the second determination of assigning classification labels in chemistry-related fields is a learning model that has undergone machine learning in advance to accept a patent document as input and output 24 numerical values ​​indicating the predicted probabilities of whether or not the patent document corresponds to one of 24 classification labels. A learning model such as Transformer, BERT, or GPT may be employed as the classification model 77. The classification model 77 is generated in advance by machine learning using training data (teacher data) that associates a patent document with information indicating which classification label the patent document corresponds to (information of correct answer values ​​in which the classification label in the corresponding chemistry-related field is set to 1 and other classification labels are set to 0).

[0069] The patent text input to the classification model 77 may be all of the text data contained in the patent publication, or may be only a portion of the text data, such as the abstract, contained in the patent publication. The 24 numerical values ​​output by the classification model 77 correspond to 24 classification labels. Each numerical value is a decimal value between 0 and 1, and the larger the value, the more likely the patent text belongs to the corresponding classification label. The 24 output values ​​of the classification model 77 are values ​​that may be called predicted probabilities or confidence levels.

[0070] The information processing device 1 inputs the patent text into the classification model 77, obtains 24 numerical values ​​such as predicted probabilities output by the classification model 77, and selects three classification labels from the 24 numerical values ​​in descending order of value. At this time, the information processing device 1 may exclude values ​​output by the classification model 77 that do not exceed a predetermined threshold from the selection target, and may select up to three classification labels from the 24 classification labels. If none of the 24 numerical values ​​exceed the threshold, the information processing device 1 can select the classification label with the largest value from the 24 numerical values. The information processing device 1 assigns one to three classification labels selected based on the output values ​​of the classification model 77 to the patent text.

[0071] <Flowchart> 9 is a flowchart showing an example of the procedure of the labeling process performed by the information processing device 1 according to this embodiment. The patent text acquisition unit 11a of the processing unit 11 of the information processing device 1 according to this embodiment acquires a patent text to be assigned a classification label by communicating with the terminal device 3 or the patent management server device 5 via the communication unit 13 (step S1). At this time, if the patent text acquisition unit 11a receives data of the patent text from the terminal device 3, it acquires this data as the patent text, and if it receives identification information of the patent text from the terminal device 3, it acquires data of the patent text corresponding to this identification information from the patent management server device 5.

[0072] The technical field determination unit 11b of the processing unit 11 determines whether the patent text acquired in step S1 belongs to an air conditioning-related field (step S2). At this time, the technical field determination unit 11b acquires classification information such as IPC, CPC, or FI included in the patent text, and determines whether the patent text belongs to an air conditioning-related field based on whether the acquired classification information matches classification information predetermined for the air conditioning-related field.

[0073] If the patent text belongs to an air conditioning-related field (S2: YES), the air conditioning label assignment unit 11c of the processing unit 11 uses the technology classification model 75 stored in the model information storage unit 12b to determine which of 16 predetermined technology classification labels the patent text corresponds to (step S3). At this time, the air conditioning label assignment unit 11c inputs all or part of the patent text into the technology classification model 75, obtains 16 predicted probabilities output by the technology classification model 75, and selects the technology classification label corresponding to the highest predicted probability. The air conditioning label assignment unit 11c uses the structural classification model 76 stored in the model information storage unit 12b to determine which of four predetermined structural classification labels the patent text corresponds to (step S4). At this time, the air conditioning label assignment unit 11c inputs all or part of the patent text into the structural classification model 76, obtains the four predicted probabilities output by the structural classification model 76, and selects the structural classification label corresponding to the highest predicted probability.

[0074] Next, the air conditioning label assignment unit 11c determines whether or not there is a contradiction between the combination of the technology classification label obtained by the determination in step S3 and the structure classification label obtained by the determination in step S4 (step S5). At this time, the air conditioning label assignment unit 11c determines whether or not there is a contradiction by determining whether or not the combination of the technology classification label and the structure classification label obtained by the determination matches a contradictory combination stored in advance. If the combination of the technology classification label and the structure classification label is contradictory (S5: YES), the air conditioning label assignment unit 11c corrects the structure classification label included in the contradictory technology classification label and the structure classification label to a predetermined structure classification label (step S6), and proceeds to step S7. If the combination of the technology classification label and the structure classification label is not contradictory (S5: NO), the air conditioning label assignment unit 11c proceeds to step S7. The air conditioning label assignment unit 11c assigns the technical classification label obtained by the judgment in step S3 and the structural classification label obtained by the judgment in step S4 or the structural classification label corrected in step S6 to the target patent document as the classification result (step S7), and proceeds to step S12.

[0075] If the patent text acquired in step S1 does not belong to the air conditioning-related field (S2: NO), the technical field determination unit 11b determines whether or not this patent text belongs to the chemistry-related field (step S8). At this time, the technical field determination unit 11b inputs all or part of the patent text acquired in step S1 into the learning model 71, obtains the predicted probability output by the learning model 71, and determines whether or not the patent text belongs to the chemistry-related field based on whether or not this predicted probability exceeds a predetermined threshold.

[0076] If the patent text belongs to a chemistry-related field (S8: YES), the chemical labeling unit 11d of the processing unit 11 uses the classification model 77 stored in the model information storage unit 12b to determine which of 24 predetermined classification labels the patent text corresponds to (step S9). At this time, the chemical labeling unit 11d inputs all or part of the patent text into the classification model 77, obtains 24 predicted probabilities output by the classification model 77, and selects three classification labels in descending order of predicted probability. However, the chemical labeling unit 11d may exclude classification labels whose predicted probabilities do not exceed a predetermined threshold. The chemical labeling unit 11d assigns up to three classification labels obtained by the determination in step S9 to the target patent text as the classification result (step S10), and proceeds to step S12. If the patent text does not belong to a chemistry-related field (S8: NO), the chemical labeling unit 11d of the processing unit 11 assigns a label of "unclassified" to the target patent text (step S11), and the process proceeds to step S12.

[0077] The processing unit 11 stores information about the label assigned to the patent text acquired in step S1 in step S7, S10, or S11, together with information about the patent text or the patent text's identification information, etc., in the storage unit 12 (step S12). The label information transmission processing unit 11e of the processing unit 11 transmits information about the label assigned to the patent text in step S7, S10, or S11 to the terminal device 3 that requested the assignment of a label to the patent text (step S13), and the process ends.

[0078] 10 is a flowchart showing an example of a processing procedure performed by the terminal device 3 according to this embodiment. The patent text acquisition unit 31a of the processing unit 31 of the terminal device 3 according to this embodiment acquires data of the patent text to which a classification label is to be assigned or identification information of the patent text based on a user's operation on the operation unit 35 (step S31). At this time, the patent text acquisition unit 31a may display a file selection screen or the like to accept selection of a file of the patent text data, or may accept input of the identification information of the patent text. The patent text acquisition unit 31a communicates with the information processing device 1 via the communication unit 33, transmits the data or identification information of the patent text acquired in step S31 to the information processing device 1 (step S32), and requests assignment of a classification label to the patent text.

[0079] The display processing unit 31b of the processing unit 31 receives the label information sent by the information processing device 1 in response to the request for assigning a classification label to the patent text sent in step S32 (step S33). The label information received from the information processing device 1 includes identification information of the patent text and one or more classification labels assigned to the patent text. The display processing unit 31b displays the label information received in step S33 on the display unit 34 (step S34).

[0080] Next, the processing unit 31 determines whether a request to correct the classification label has been made based on whether a predetermined operation has been performed by the user on the label information displayed in step S34 (step S35). If a request to correct the classification label has not been made (S35: NO), the processing unit 31 ends the process.

[0081] If a request to correct a classification label is received (S35: YES), the processing unit 31 receives a user operation via the operation unit 35 to receive the content of the correction to the classification label assigned to the patent text (step S36). The content of the correction received may include deleting the assigned classification label or changing the assigned classification label to a different classification label. The processing unit 31 transmits correction information including information such as the identification information of the patent text to be corrected, the classification label before correction, and the classification label after correction to the information processing device 1 (step S37), and ends the processing.

[0082] 11 is a flowchart showing an example of the procedure of the re-learning process of the learning model performed by the information processing device 1 according to this embodiment. The processing unit 11 of the information processing device 1 according to this embodiment repeatedly performs the process shown in this flowchart. The processing unit 11 determines whether or not correction information regarding the classification label has been received from the terminal device 3 that transmitted information about the classification label assigned to the patent document (step S41). If correction information has been received from the terminal device 3 (S41: YES), the processing unit 11 stores the received correction information in the storage unit 12 (step S42) and ends the process.

[0083] If no correction information has been received from the terminal device 3 (S41: NO), the processing unit 11 determines whether it is time to re-learn the learning model based on whether a condition is met, such as whether the amount of correction information stored in the storage unit 12 exceeds a predetermined threshold or whether a predetermined re-learning period has elapsed (step S43). If it is not time to re-learn (S43: NO), the processing unit 11 ends the process.

[0084] If it is time to perform re-learning (S43: YES), the processing unit 11 acquires learning data based on the correction information stored in the storage unit 12, in which the patent text is used as input information and the corrected classification label for the patent text is associated as output information (correct value) (step S44). The processing unit 11 selects a learning model to be re-learned from the learning model 71, technology classification model 75, structural classification model 76, or classification model 77 stored in the model information storage unit 12b, reads out its information, and performs machine learning processing on the learning model using the learning data acquired in step S44 (step S45). The machine learning performed in step S45 is so-called supervised machine learning, and since it is an existing technology, detailed processing procedures will not be described here. The processing unit 11 stores information such as parameters of the learning model obtained by the machine learning in step S45 in the model information storage unit 12b (step S46), and ends the processing.

[0085] <Modification> 12 is a schematic diagram for explaining the configuration of an information processing system according to a modified example. In the information processing system according to the modified example, the information processing device 1 communicates with the patent management server device 5 at a predetermined cycle, such as once a day or once a week, to acquire patent documents stored in the patent document DB 6. At this time, the information processing device 1 acquires from the patent management server device 5 patent documents that meet conditions set by an administrator of the information processing system or the like and that have not been acquired before. Various conditions, such as the technical field or the name of the applicant, can be specified as the conditions for acquiring patent documents.

[0086] The information processing device 1 assigns classification labels to one or more patent documents acquired from the patent management server device 5 using the learning model 71, the technology classification model 75, the structural classification model 76, or the classification model 77. The information processing device 1 according to the modified example is provided with a patent document DB2, and stores in the patent document DB2 the patent documents (or the identification information of these patent documents) acquired from the patent management server device 5 in association with one or more classification labels assigned to these patent documents. This allows the information processing device 1 to collect necessary patent documents from the patent documents stored in the patent document DB6 of the patent management server device 5 and accumulate the information in its own patent document DB2.

[0087] A user of the information processing system according to the modified example can access the information processing device 1 using his / her own terminal device 3 and search, acquire, and view patent documents stored in the patent document DB2. The terminal device 3 according to the modified example accepts input of classification label wording as search conditions from the user and issues a search request for patent documents to the information processing device 1 using the classification label input by the user as the search condition. Upon receiving the search request from the terminal device 3, the information processing device 1 extracts patent documents to which classification labels that match or are similar to the classification labels specified as the search conditions are assigned from the patent document DB2. The information processing device 1 transmits the patent documents (or their identification information) extracted from the patent document DB2 to the terminal device 3 that issued the search request. The terminal device 3 receives information on the search results in response to the search request from the information processing device 1 and displays information on the patent documents corresponding to the classification labels input as search conditions on the display unit 34.

[0088] <Summary> In the information processing system according to the present embodiment, the information processing device 1 acquires information about patent texts related to patents from the terminal device 3, the patent management server device 5, or the like, and performs a first determination as to whether the acquired patent texts belong to a predetermined technical field (such as an air conditioning-related field or a chemistry-related field). The information processing device 1 then performs a second determination as to whether the patent texts determined in the first determination to belong to a predetermined technical field belong to one of multiple subfields represented by technical classification labels in the air conditioning-related field, structural classification labels, or classification labels in the chemistry-related field. The information processing device 1 then performs the second determination using a learning model that has undergone machine learning in advance based on training data (teacher data) in which text information and classification labels corresponding to the subfields are associated. Based on the results of the second determination, the information processing device 1 assigns one or more classification labels corresponding to the subfields to the patent texts. This allows the information processing system according to the present embodiment to be expected to assist in assigning label information to patent texts.

[0089] In the information processing system according to this embodiment, the information processing device 1 acquires classification information such as IPC, CPC, or FI associated with a patent document, and makes a first determination as to whether the patent document belongs to a predetermined technical field based on the acquired classification information. The information processing device 1 can make the first determination as to whether the patent document belongs to an air conditioning-related field by using the classification information such as IPC, CPC, or FI. As a result, the information processing system according to this embodiment is expected to easily determine whether the patent document belongs to a predetermined technical field by making a rule-based determination based on conditions, etc., predetermined by the designer of the system, etc.

[0090] In the information processing system according to this embodiment, the information processing device 1 makes a first determination using a learning model that has been machine-learned in advance based on learning data in which patent texts are associated with information on whether the text belongs to a predetermined technical field. The information processing device 1 can make a first determination on whether the patent text belongs to a chemistry-related field using this learning model. As a result, the information processing system according to this embodiment can be expected to accurately determine whether the patent text belongs to a predetermined technical field.

[0091] In the information processing system according to this embodiment, the information processing device 1 uses a first classification model (technical classification model 75) to classify a patent text into a plurality of first detailed fields (e.g., detailed fields classified by 16 technical classification labels), and uses a second classification model (structural classification model 76) to classify a patent text into a plurality of second detailed fields (e.g., detailed fields classified by four structural classification labels). The first classification model is generated in advance by machine learning using training data in which the patent text is associated with labels of the first detailed fields. The second classification model is generated in advance by machine learning using training data in which the patent text is associated with labels of the second detailed fields. Based on the classification results of the first and second classification models, the information processing device 1 assigns to the patent text a first label (technical classification label) corresponding to the first detailed field and a second label (structural classification label) corresponding to the second detailed field. This allows the information processing system according to this embodiment to assign two different types of labels to a single patent text.

[0092] In the information processing system according to this embodiment, when the first label and the second label to be assigned to the patent document based on the classification results of the first classification model and the second classification model are a predetermined combination such as a contradictory combination, the information processing device 1 corrects the first label or the second label. As a result, when there is a contradiction between the classification results of the first classification model and the classification results of the second classification model, the information processing system according to this embodiment is expected to correct the contradiction and assign an appropriate label to the patent document.

[0093] In the information processing system according to this embodiment, classification models such as the technology classification model 75 for the air conditioning-related field, the structural classification model 76, and the classification model 77 for the chemistry-related field are learning models that output multiple prediction probabilities corresponding to multiple detailed fields. The information processing device 1 assigns to the patent text a label corresponding to the detailed field with a high prediction probability output by the classification model. As a result, the information processing system according to this embodiment is expected to obtain a prediction probability for each detailed field using a classification model generated in advance by machine learning, and assign to the patent text a label for the detailed field with a higher prediction probability.

[0094] In the information processing system according to this embodiment, the information processing device 1 accepts corrections to classification labels assigned to patent documents, and re-trains the classification model using training data in which the patent documents and detailed fields corresponding to the corrected labels are associated. As a result, the information processing system according to this embodiment is expected to improve the accuracy of classification of detailed fields by the classification model.

[0095] Although the embodiments have been described above, it will be understood that various changes in form and details can be made without departing from the spirit and scope of the claims. [Explanation of symbols]

[0096] 1. Information processing equipment (computer) 3 Terminal Devices 5 Patent management server device 6 Patent document DB 11 Processing section 11a Patent Document Acquisition Department 11b Technical Field Determination Division 11c Air conditioning labeling section 11d Chemical Labeling Division 11e Label information transmission processing unit 12 Storage section 12a Program (computer program) 12b Model information storage section 13 Communications Department 31 Processing section 31a Patent Document Acquisition Department 31b Display processing section 32 Storage section 32a Program 33 Communications Department 34 Display section 35 Control section 71 Learning Model 75 Technology Classification Model (First Classification Model) 76 Structural Classification Model (Second Classification Model) 77 Classification Model 98,99 Recording media N Network

Claims

1. On the computer, Obtaining patent-related text information, making a first determination as to whether the acquired text information belongs to a predetermined technical field; making a second determination as to which of a plurality of sub-fields included in the predetermined technical field the text information determined to belong to the predetermined technical field belongs; Based on the result of the second determination, one or more labels corresponding to the detailed field are assigned to the text information. Execute the process, In the second determination, classifying the text information into one of a plurality of first detailed fields using a first classification model that has been machine-learned in advance based on learning data in which the text information and the first detailed fields are associated; classifying the text information into one of a plurality of second detailed fields different from the first detailed field using a second classification model that has been machine-learned in advance based on learning data in which the text information and the second detailed fields are associated; In assigning the label, assigning one or more first labels corresponding to the first detailed field and one or more second labels corresponding to the second detailed field to the text information based on the classification results of the first classification model and the classification results of the second classification model; changing the first label or the second label when the first label and the second label assigned to the text information based on the classification result of the first classification model and the classification result of the second classification model are a predetermined combination; Computer program.

2. obtaining patent classification information associated with the text information; In the first determination, it is determined whether the text information belongs to the predetermined technical field based on the acquired patent classification information.

2. The computer program of claim 1.

3. The first determination is made using a learning model that has been machine-learned in advance based on learning data in which text information and information indicating whether the text information belongs to the predetermined technical field are associated with each other.

2. The computer program of claim 1.

4. the first classification model and the second classification model are models that output a plurality of prediction probabilities corresponding to a plurality of detailed fields, assigning a label corresponding to a detailed field with a high prediction probability output by the first classification model to the text information; assigning a label corresponding to a detailed field with a high prediction probability output by the second classification model to the text information; A computer program according to any one of claims 1 to 3.

5. Accept corrections to the labels assigned to text information, re-learning the first classification model or the second classification model based on learning data in which the sentence information and the detailed fields corresponding to the corrected labels are associated with each other; A computer program according to any one of claims 1 to 3.

6. The information processing device an acquisition step of acquiring text information related to a patent; a first determination step of determining whether the acquired text information belongs to a predetermined technical field; a second determination step of determining to which of a plurality of sub-fields included in the predetermined technical field the text information determined to belong to the predetermined technical field belongs; a labeling step of assigning one or more labels corresponding to the detailed fields to the text information based on the determination result of the second determination step; Including, In the second determination step, classifying the text information into one of a plurality of first detailed fields using a first classification model that has been machine-learned in advance based on learning data in which the text information and the first detailed fields are associated; classifying the text information into one of a plurality of second detailed fields different from the first detailed field using a second classification model that has been machine-learned in advance based on learning data in which the text information and the second detailed fields are associated; In the labeling step, assigning one or more first labels corresponding to the first detailed field and one or more second labels corresponding to the second detailed field to the text information based on the classification results of the first classification model and the classification results of the second classification model; changing the first label or the second label when the first label and the second label assigned to the text information based on the classification result of the first classification model and the classification result of the second classification model are a predetermined combination; Information processing methods.

7. a processing unit; The processing unit Obtaining patent-related text information, making a first determination as to whether the acquired text information belongs to a predetermined technical field; making a second determination as to which of a plurality of sub-fields included in the predetermined technical field the text information determined to belong to the predetermined technical field belongs; assigning one or more labels corresponding to the detailed fields to the text information based on the result of the second determination; In the second determination, classifying the text information into one of a plurality of first detailed fields using a first classification model that has been machine-learned in advance based on learning data in which the text information and the first detailed fields are associated; classifying the text information into one of a plurality of second detailed fields different from the first detailed field using a second classification model that has been machine-learned in advance based on learning data in which the text information and the second detailed fields are associated; In assigning the label, assigning one or more first labels corresponding to the first detailed field and one or more second labels corresponding to the second detailed field to the text information based on the classification results of the first classification model and the classification results of the second classification model; changing the first label or the second label when the first label and the second label assigned to the text information based on the classification result of the first classification model and the classification result of the second classification model are a predetermined combination; Information processing device.

Images