Information processing device, information processing method, and information processing program

Abstract

An information processing device according to the present invention comprises a processor, wherein the processor acquires a plurality of document groups in which a plurality of documents are classified according to a prescribed criterion and which correspond to the classifications, acquires intermediate information groups including intermediate information extracted from each of the plurality of document groups, integrates the intermediate information groups on the basis of a prescribed rule, and uses a document generation model to generate a document as a final product from the integrated intermediate information groups.

Description

Information Processing Apparatus, Information Processing Method, and Information Processing Program

[0001] The present disclosure relates to an information processing apparatus, an information processing method, and an information processing program.

[0002] Techniques for generating a desired document from a certain document using a document generation model are known. As such a technique, for example, Japanese Patent Application Laid-Open No. 2023-137048 (Patent Document 1) describes a technique for generating a first document using a learned model from a second document related to the first document to be created.

[0003] In the above technique, when there are various multiple documents used for generation, it is necessary to handle these various documents integratively. However, if the types and characteristics of the documents are different, the important elements in the documents may be different, or the writing styles of the documents may be different. Therefore, when generating a final product document using a single document generation model from multiple document groups, appropriate information may not be extracted from each of the multiple document groups, and as a result, the final product document may not contain the desired information or may contain unnecessary information.

[0004] Thus, when generating a final product document from multiple document groups, the information contained in the final product document may be inappropriate.

[0005] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide an information processing apparatus, an information processing method, and an information processing program that can make the information contained in the final product document appropriate even when generating a final product document from multiple document groups.

[0006] To achieve the above object, the information processing apparatus of the present disclosure includes a processor. The processor acquires a plurality of document groups corresponding to classification in which a plurality of documents are classified according to a predetermined criterion, acquires an intermediate information group including intermediate information extracted from each of the plurality of document groups, integrates the intermediate information group based on a predetermined rule, and generates a final product document from the integrated intermediate information group using a document generation model.

[0007] Furthermore, in order to achieve the above objective, the information processing method disclosed herein acquires multiple document groups, each of which multiple documents are classified according to predetermined criteria; acquires an intermediate information group, which includes intermediate information extracted from each of the multiple document groups; integrates the intermediate information groups based on predetermined rules; and generates a final deliverable document from the integrated intermediate information group using a document generation model.

[0008] Furthermore, in order to achieve the above objectives, the information processing program disclosed herein obtains multiple document groups, each of which is classified according to predetermined criteria; obtains an intermediate information group, which includes intermediate information extracted from each of the multiple document groups; integrates the intermediate information group according to predetermined rules; and causes a computer to perform the process of generating a final output document from the integrated intermediate information group using a document generation model.

[0009] According to this disclosure, even when generating a final deliverable document from a set of multiple documents, the information contained in the final deliverable document can be made appropriate.

[0010] This is a diagram illustrating the overview of the information processing device of the embodiment. This is a block diagram showing an example of the configuration of the information processing device of the embodiment. This is a functional block diagram showing an example of the configuration of the information processing device of the embodiment. This is a flowchart showing an example of the information processing flow by the information processing device of the embodiment. This is a diagram illustrating the processing of the second acquisition unit and the integration unit. This is a diagram illustrating an example of the display method for intermediate information, integrated data, and medical summaries for each document group. This is a diagram illustrating the learning of the extraction model.

[0011] Embodiments of this disclosure will be described in detail below with reference to the drawings. These embodiments are not intended to limit the technology of this disclosure.

[0012] As shown in Figure 1, the information processing device 10 of this embodiment is a device that generates a final output document from multiple documents using a document generation model 33. In the example shown in Figure 1, a mode is shown in which one final output document is generated from three types of documents, documents A to C. Documents A to C are documents classified according to predetermined criteria based on at least one of the document type and document characteristics. In this embodiment, as a specific example, the case in which the information processing device 10 generates a summary as the final output document from multiple documents related to a patient's medical information will be described. Examples of documents A to C include document A being an electronic medical record, document B being a nursing record, and document C being a radiology report.

[0013] The information processing device 10 extracts intermediate information A from extraction model A_32A for document A, which is included in the extraction model group 32; intermediate information B from extraction model B_32B for document B; and intermediate information C from extraction model C_32C for document C. The information processing device 10 also integrates intermediate information A to C according to predetermined rules, and generates a summary, which is the final output document, from the integrated group of intermediate information using the document generation model 33.

[0014] Figure 2 shows a block diagram illustrating an example of the configuration of the information processing device 10 of this embodiment. As shown in Figure 2, the information processing device 10 of this embodiment comprises a control unit 20, a storage unit 22, a communication interface unit 24, an operation unit 26, and a display unit 28. The control unit 20, storage unit 22, communication interface unit 24, operation unit 26, and display unit 28 are connected to each other via a bus 29, such as a system bus or a control bus, enabling the exchange of various types of information.

[0015] The control unit 20 in this embodiment controls the overall operation of the information processing device 10. The control unit 20 is a processor and includes a CPU (Central Processing Unit) 20A. The control unit 20 is also connected to the storage unit 22, which will be described later. The control unit 20 may also include a GPU (Graphics Processing Unit).

[0016] The operation unit 26 is used for the user to input information related to document generation. The operation unit 26 is not particularly limited and may include, for example, various switches, a touch panel, a stylus, a mouse, and a microphone for voice input. The display unit 28 displays the final output document and various information. The operation unit 26 and the display unit 28 may be integrated to form a touch panel display.

[0017] The communication interface unit 24 communicates various types of information with external devices of the information processing device 10 via the network 9 using wireless or wired communication.

[0018] The memory unit 22 includes a ROM (Read Only Memory) 22A, a RAM (Random Access Memory) 22B, and a storage unit 22C. The ROM 22A is pre-stored with various programs executed by the CPU 20A. The RAM 22B temporarily stores various data. The storage unit 22C stores the information processing program 30 executed by the CPU 20A, the extraction model group 32, the document generation model 33, and other various information. The storage unit 22C is a non-volatile memory unit, such as an HDD (Hard Disk Drive) or SSD (Solid State Drive).

[0019] As described above, the extraction model group 32 includes multiple extraction models (extraction model A_32A to extraction model C_32C) for extracting intermediate information from each of multiple document groups (documents A to C) classified according to predetermined criteria. Extraction models A_32A to extraction model C_32C are extractors designed for each of the multiple document groups.

[0020] On the other hand, the document generation model 33, as described above, is a model that generates a final document from an integrated set of intermediate information. As an example, the document generation model 33 in this embodiment is a model that takes an integrated set of intermediate information as input and outputs a final document. Examples of such models include LLMs (Large Language Models) and generative AI (Artificial Intelligence).

[0021] Furthermore, Figure 3 shows a functional block diagram of an example configuration of the information processing device 10 of this embodiment. As shown in Figure 3, the information processing device 10 comprises a first acquisition unit 40, a second acquisition unit 42, an integration unit 44, a generation unit 46, and a display control unit 48. As an example, in the information processing device 10 of this embodiment, the CPU 20A of the control unit 20 executes an information processing program 30 stored in the storage 22C, so that the CPU 20A functions as the first acquisition unit 40, the second acquisition unit 42, the integration unit 44, the generation unit 46, and the display control unit 48.

[0022] As described above, the first acquisition unit 40 acquires multiple document groups (documents A to C) according to a classification, in which multiple documents are classified according to a predetermined standard. The first acquisition unit 40 outputs the acquired multiple document groups (documents A to C) to the second acquisition unit 42. In this embodiment, "document" includes at least one of words, sentences, texts, and images. For example, one document may include multiple sentences and images. Also, for example, one document may include only images. Furthermore, each document group may include one or more documents. For example, document A, which is an electronic medical record, may be an electronic medical record for multiple days, and document C, which is a radiographic interpretation report, may be a single radiographic interpretation report for an X-ray photograph. Furthermore, words, etc., include at least one of letters, numbers, and symbols. For example, a word may be numerical data representing an examination result.

[0023] As described above, in this embodiment, the first acquisition unit 40 acquires three document groups (documents A to C) in order to generate a summary, which is the final deliverable document. However, the number of document groups acquired by the first acquisition unit 40 is not limited to this embodiment. Furthermore, the acquisition source from which the first acquisition unit 40 acquires multiple document groups (documents A to C) is not particularly limited. For example, if multiple document groups are pre-stored in the storage 22C of the information processing device 10, the first acquisition unit 40 acquires the multiple document groups from the storage 22C. Alternatively, if multiple document groups are stored in a database of patient information at a hospital located outside the information processing device 10, the first acquisition unit 40 acquires the multiple document groups from the external database via a network using the communication I / F 24.

[0024] As a specific example, when the information processing device 10 generates a summary of a patient as a final output document, the first acquisition unit 40 acquires a group of documents (documents A to C) related to the patient based on information for identifying the patient.

[0025] The second acquisition unit 42 acquires an intermediate information group containing intermediate information extracted from each of the multiple document groups (documents A to C). In this embodiment, as described above, the second acquisition unit 42 inputs document A into extraction model A_32A and acquires intermediate information A extracted by extraction model A_32A. It also inputs document B into extraction model B_32B and acquires intermediate information B extracted by extraction model B_32B. Furthermore, it inputs document C into extraction model C_32C and acquires intermediate information C extracted by extraction model C_32C.

[0026] As described above, each of the extraction models A_32A to C_32C is individually designed according to a criterion based on the type of document and at least one of the characteristics of the document, in order to extract intermediate information (intermediate information A to C) related to the final output document, the summary, from their respective set of documents (documents A to C). For example, extraction model A_32A for document A, which is an electronic medical record, may be designed to primarily extract events. Similarly, extraction model C_32C for document C, which is a radiology report, may be designed to primarily extract findings. In this way, by individually designing each model according to a criterion based on the type of document and at least one of the characteristics of the document, it is possible to extract intermediate information that is appropriate for each document (documents A to C).

[0027] Regarding document types, examples include electronic medical records, nursing records, and radiology reports, as well as other types according to a prescribed classification. However, the specific content of the "types," or in other words, the method of classification, is not limited and may be determined according to the form of the final document, the desired information, and the multiple documents used in its creation. Furthermore, document characteristics include the nature of the information contained within the document. For example, differences in formatting, the proportion of important information included, and the location of important information within the document. Specific examples include domains such as language, images, and audio; document length; author attributes (e.g., doctor or nurse); target audience (e.g., doctor or nurse); document creation date and day of the week; and the device used.

[0028] Furthermore, examples of the form of intermediate information extracted from each of the extraction model group 32 (extraction model A_32A to extraction model C_32C) include words such as keywords and short sentences. It is preferable that the intermediate information be in a fine-grained unit.

[0029] The second acquisition unit 42 associates time information related to each document or element of each document with the extracted intermediate information and outputs it to the integration unit 44. Examples of such time information include the time the document was written, the time estimated from the context of the document, the time the event described in the document occurred, and the time the document was acquired from an external database. In addition, for multiple types of time, a priority order may be set for each of the multiple types of time, and the time information may be identified based on that priority order. For example, if the time corresponding to an element such as a word or sentence extracted as intermediate information can be estimated from the context, the time estimated from the context is used as the time information. If the time cannot be estimated from the context of the document, the time the document was written is used as the time information. If the time cannot be estimated from the context of the document, and the time the document was written is also unknown, the first acquisition unit 40 uses the time the event described in the document occurred as the time information. For example, if a radiology report created on May 18th contains the results of an examination on May 16th, but there is no record that the creation date was May 18th, then May 16th should be used as the time information. The method for identifying the time the document was written, the time estimated from the context of the document, the time the event described in the document occurred, or the time the document was acquired is not particularly limited and publicly known methods may be used. While this explanation describes a configuration where time information is associated with extracted intermediate information, the time information may also be referred to as intermediate information.

[0030] The integration unit 44 integrates the intermediate information group (intermediate information A to C) acquired by the second acquisition unit 42 based on predetermined rules. The predetermined rules include the chronological order of the intermediate information, based on the time information associated with the intermediate information, as described above. The predetermined rules may also include rules other than chronological order. Examples of rules other than chronological order include classification results based on predetermined criteria, rules based on information underlying the document, etc. Examples of predetermined criteria include criteria corresponding to the data source. Examples of underlying information include information related to the background of the document, such as diseases or hospitalization / outpatient treatment. Multiple predetermined rules may be used in combination.

[0031] In other words, the integration unit 44 integrates the intermediate information group by rearranging the intermediate information according to predetermined rules. The integration unit 44 outputs the integrated intermediate information group to the generation unit 46.

[0032] As described above, the generation unit 46 uses the document generation model 33 to generate a final output document (summary) from the integrated intermediate information group (intermediate information A to C). Specifically, the generation unit 46 inputs the integrated intermediate information group obtained from the integration unit 44 into the document generation model 33 and obtains the summary output from the document generation model 33 to generate the final output document. The generation unit 46 outputs the generated final output document (summary) to the display control unit 48.

[0033] In this embodiment, the generation unit 46 generates a final output document from an integrated group of intermediate information using the document generation model 33, but it is not limited to this. For example, it may generate the final output from the group of intermediate information according to predetermined rules. In this case, predetermined rules include, for example, expression methods such as nominalization, and formats indicating the order of entries such as date:item:numerical value.

[0034] The display control unit 48 controls the display of the final output document (summary) on the display unit 28.

[0035] Next, the operation of the information processing device 10 of this embodiment will be described with reference to the drawings. Figure 4 shows a flowchart illustrating an example of the flow of information processing performed in the information processing device 10 of this embodiment. In this embodiment, as an example, the information processing device 10 executes the information processing shown as an example in Figure 4 by having the CPU 20A of the control unit 20 execute the information processing program 30 stored in the storage 22C based on a user's start instruction made by the operation unit 26.

[0036] In step S100 of Figure 4, the acquisition unit 40 acquires a group of documents according to a classification based on predetermined criteria, as described above. As shown in Figure 1, etc., in this embodiment, as a specific example, the first acquisition unit 40 acquires documents A to C.

[0037] In the next step S102, the second acquisition unit 42 acquires an intermediate information group (intermediate information A to C) containing intermediate information (intermediate information A to C) extracted from each of the multiple document groups using the extraction model group 32 (extraction model A_32A to extraction model C_32C) as described above. As shown in Figure 5, in this embodiment, as a specific example, the second acquisition unit 42 acquires intermediate information A from document A, which includes "CT scan: 5 / 7", "CRP (C-Reactive Protein) 3.0: 5 / 8", and "CRP 3.3: 5 / 9". The second acquisition unit 42 also acquires intermediate information B from document B, which includes "SpO2 (Saturation of percutaneous Oxygen) 95%: 5 / 6" and "SpO2 93%: 5 / 8". The second acquisition unit 42 also acquires intermediate information C from document C, which includes "right lung ground-glass opacity: 5 / 8".

[0038] In the next step S104, the integration unit 44 integrates the intermediate information group based on predetermined rules, as described above. As a result, intermediate information A to C are integrated, for example, in chronological order. The example shown in Figure 5 is a specific example of integration in chronological order. The integration unit 44 integrates intermediate information A to C by rearranging each of the intermediate information based on the time information from 5 / 6 to 5 / 9.

[0039] In the next step S106, the generation unit 46 generates a final output document (summary) from the integrated intermediate information group using the document generation model 33, as described above. As shown in Figure 1, etc., in this embodiment, as a specific example, the generation unit 46 inputs the integrated intermediate information group from step S104 into the document generation model 33 and obtains a summary, which is the final output document, from the document generation model 33.

[0040] In the next step S108, the display control unit 48 controls the display unit 28 to display the generated final output document (summary), as described above. When the processing in step S108 is completed, the information processing shown in Figure 4 is completed.

[0041] As described above, the information processing device 10 of this embodiment comprises a first acquisition unit 40, a second acquisition unit 42, an integration unit 44, and a generation unit 46. The first acquisition unit 40 acquires multiple document groups, each of which multiple documents have been classified according to predetermined criteria. The second acquisition unit 42 acquires an intermediate information group, which includes intermediate information extracted from each of the multiple document groups. The integration unit 44 integrates the intermediate information groups based on predetermined rules. The generation unit 46 generates a final output document from the integrated intermediate information group using a document generation model 33.

[0042] As described above, the information processing device 10 of this embodiment is provided with an extraction model for each of the multiple document groups used to generate the final deliverable document. Therefore, even when generating the final deliverable document from multiple document groups, the information contained in the final deliverable document can be made into appropriate information.

[0043] The information processing device 10 may also display the group of intermediate information integrated by the integration unit 44 on the display unit 28 and accept modifications to the integrated group of intermediate information instructed by the user via the operation unit 26. In this case, the generation unit 46 generates the final output document from the modified group of intermediate information. The information processing device 10 may also display the group of intermediate information acquired by the second acquisition unit 42 on the display unit 28 for at least one of the source documents or groups of documents, for example, and accept modifications to the intermediate information instructed by the user via the operation unit 26. In this case, the integration unit 44 integrates the modified group of intermediate information. The information processing device 10 may also have the generation unit 46 convert the style of the integrated group of intermediate information to an output style before generating the final output document. Alternatively, the generation unit 46 may convert the style of the final output document to an output style. Known style converters or language models may be used for such style conversion.

[0044] Figure 6 shows an example of a display configuration when intermediate information 50A to 50C for multiple document groups, integrated data 52 of the intermediate information groups, and the final output document (generated text), the medical summary 54, are displayed side by side on the display unit 28. In the example shown in Figure 6, three pieces of intermediate information 50A extracted from document group A (electronic medical record), two pieces of intermediate information 50B extracted from document group B (nursing record), and one piece of intermediate information 50C extracted from document group C (radiography report) are presented for each source document group. In addition, each piece of intermediate information includes date information, showing both the date the event described in the document occurred and the date the document was written. For example, in the example shown in Figure A, for the intermediate information "CT scan" from document group A (electronic medical record), the date the CT scan was performed, "5 / 7", the date the doctor recorded the CT scan in the electronic medical record, "5 / 9", and the document name are displayed. Furthermore, in the example shown in Figure 6, the integrated data 52, which combines these intermediate information sets, is presented in a way that allows for the identification of the source document group from which each intermediate information 50A to 50C was extracted. In the example shown in Figure 6, intermediate information 50A extracted from document group A (electronic medical record) is enclosed in a solid line, intermediate information 50B extracted from document group B (nursing record) is enclosed in a dotted line, and intermediate information 50C extracted from document group C (radiology report) is enclosed in a dashed line. In the example shown in Figure 6, the source can be identified by using different types of lines to enclose the intermediate information 50A to 50C, but other methods may be used to identify the source. For example, the color of the text, the size of the text, and the type of underline may be different depending on the source. In addition, in the example shown in Figure 6, the medical summary 54 of the final output document (generated text) is presented in a way that allows for the identification of the source document from which the information in the document was originally written, similar to the integrated data. Furthermore, in the integrated data 52, if the source document group can be identified for each piece of intermediate information, it is not necessary to present the intermediate information 50A to 50C for each document group. For example, the integrated data 52 and the portion of the medical summary 54 shown in Figure 6 may be displayed on the display unit 28. In addition, each piece of intermediate information 50A to 50C presents both the date on which the event described in the document occurred and the date on which the document was written as date information, but is not limited to this.For example, only one of the date when the event described in the document occurred and the date when the document was described may be displayed.

[0045] In the above embodiment, the form in which the extraction model group 32 is stored in the storage 22C in advance has been described. However, the control unit 20 of the information processing apparatus 10 may generate the extraction model group 32, that is, learn the extraction model group 32.

[0046] For example, when learning the extraction model A_32A provided for the document A, as shown in FIG. 7, the control unit 20 extracts the intermediate information to be extracted from the document A (electronic medical record) as the correct answer data for the document A from the correct answer document (correct summary) corresponding to the final product document, and generates a combination of the document A (electronic medical record) and the correct answer data for the document A as learning data. Further, the control unit 20 learns the extractor with the generated learning data to learn the extraction model A_32A for the document A. Also, the control unit 20 extracts the intermediate information to be extracted from the document B (nursing record) as the correct answer data for the document B from the correct answer document (correct summary) corresponding to the final product document, and generates a combination of the document B (nursing record) and the correct answer data for the document B as learning data. Further, the control unit 20 learns the extractor with the generated learning data to learn the extraction model B_32B for the document B. Also, the control unit 20 extracts the intermediate information to be extracted from the document B (fluoroscopy data) as the correct answer data for the document C from the correct answer document (correct summary) corresponding to the final product document, and generates a combination of the document C (fluoroscopy data) and the correct answer data for the document C as learning data. Further, the control unit 20 learns the extractor with the generated learning data to learn the extraction model C_32C for the document C. Note that the control unit 20 may learn each of the extraction models A to C using learning data in which labels are assigned to the intermediate information (correct answer data) to be extracted by each of the extraction models A to C with respect to the correct answer document (correct summary). In this case, the learning data for the correct answer documents (correct summaries) with different labels assigned thereto is used.

[0047] Further, for each extraction model provided for each of a plurality of document groups, the control unit 20 may generate learning data from a document based on the degree of coincidence in a predetermined element between the document included in the document group and the document of the final product that is the correct answer, and learn each of the extraction models with the generated learning data. Further, the control unit 20 may generate, as learning data, a document in which a portion having a degree of coincidence with the correct final product greater than a threshold value is labeled. Further, in this case, for each of the plurality of document groups, a plurality of extraction models classified based on at least one of a predetermined criterion, a noun, an event, a sentence, a line, a paragraph, and a predetermined delimiter expression may be provided.

[0048] For example, as the threshold value, a value obtained by dividing the extraction rate by the number of extractions (extraction rate / number of extractions) may be used. A specific example using such a threshold value will be described. The degree of coincidence at the character unit between the sentence "Regarding variable-time DNAR" and the sentence "Sudden-change-time DNAR" is 7 / 11. On the other hand, the degree of coincidence at the character unit between the sentence "Regarding sudden-change-time DNAR" and the sentence "Regarding having a pain in the body" is 4 / 11. When the threshold value is set to 0.5, it is labeled. On the other hand, in the case of the word unit or the character unit, the degree of coincidence at the word unit between the sentence "CRP was as high as 3.0" and the sentence "CRP: 3.0" becomes 2 / 7. Since both of these sentences mean almost the same thing, they have a high degree of coincidence in terms of content, but when looking at the degree of coincidence at the word unit like this, it becomes lower than the threshold value. Therefore, important words and numerical values may be weighted before determining the degree of coincidence.

[0049] Furthermore, the method for generating the extraction model group 32, i.e., the learning method, may be varied according to the characteristics of each document. For example, it may be varied according to the importance threshold as described above. Also, for example, it may be an extractive summary or an abstractive summary, and an extractive summary is preferred when dealing with documents that have a lot of overlap. Also, missing information in the data may be added and used. Also, if the amount of input is large, the input data may be divided into multiple groups and processed in multiple stages, such as processing each group separately. Also, it may be possible to combine extractive summarization and abstract summarization and process in two stages, such as roughly extracting with an extractive summary and then carefully summarizing including expressions with an abstractive summary. Also, for example, electronic medical records and nursing records have similar content, and the extraction models corresponding to each may also be similar. In this way, when the extraction models for different documents are similar, they may be integrated into a single extraction model.

[0050] Furthermore, in this embodiment, each process is executed on any computer. Alternatively, any computer may execute these processes using a processor as hardware, a program as software, or a combination thereof. In that case, the processor is configured to work in cooperation with the program to execute the various processes in this embodiment, and can function as a unit or means in this embodiment. The execution order of the processes by the processor is not limited to the order described and may be changed as appropriate. Any computer may be a general-purpose computer, a computer designed for a specific purpose, a workstation, or any other system capable of executing each process.

[0051] A processor may consist of one or more hardware components, and the type of hardware is not limited. For example, a processor may consist of programmable logic devices such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), FPGA (Field Programmable Gate Array), dedicated circuits for performing specific processing such as an ASIC (Application Specific Integrated Circuit), a GPU (Graphic Processing Unit), or an NPU (Neural Processing Unit). Furthermore, the type of hardware may be a combination of different types of hardware. When multiple hardware components are configured to execute one or more processes of a processor, these components may reside in physically separate devices or in the same device. Furthermore, in any embodiment, the order of the processes performed by the processor is not limited to the order described above and may be changed as appropriate. The hardware components are composed of electrical circuits (circuits) and the like, which are combinations of circuit elements such as semiconductor elements.

[0052] Furthermore, the program may be firmware or software such as microcode. Alternatively, the program may be, for example, a set of program modules, each function of which may be implemented by a processor configured to perform its respective function. The program may be program code or multiple code segments stored on one or more non-temporary computer-readable media (e.g., storage media or other storage). The program may be divided and stored on multiple non-temporary computer-readable media located on physically separate devices. Program code or code segments may represent any combination of procedures, functions, subprograms, routines, subroutines, modules, software packages, classes, or instructions, data structures, or program statements. Program code or code segments may be connected to other code segments or hardware circuits by sending and receiving information, data, arguments, parameters, or memory contents.

[0053] Furthermore, although the above embodiment describes an embodiment in which the information processing program 30 is pre-stored (installed) in the storage 22C of the memory unit 22, the invention is not limited to this. The information processing program 30 may be provided in the form of a recording medium such as a CD-ROM (Compact Disc Read Only Memory), DVD-ROM (Digital Versatile Disc Read Only Memory), or USB (Universal Serial Bus) memory. Alternatively, the information processing program 30 may be provided in the form of a download from an external device via a network.

[0054] Furthermore, the technology disclosed herein extends to all program products. A program product includes all forms of products for providing programs. For example, a program product includes programs provided via a network such as the Internet, and non-temporary computer-readable recording media such as CD-ROMs, DVDs, and USB memory sticks on which programs are stored. The present invention is also applicable to programs and program products.

[0055] Furthermore, the configuration and operation of the information processing device 10, etc., described in each of the above embodiments are merely examples and can be modified as needed without departing from the spirit of the present invention. It also goes without saying that the above embodiments may be combined as appropriate.

[0056] The following additional information is disclosed regarding the above-described embodiments.

[0057] (Note 1) An information processing device comprising a processor, the processor acquires a plurality of document groups according to a predetermined standard, the plurality of documents being classified according to the classification, the processor acquires an intermediate information group including intermediate information extracted from each of the plurality of document groups, the intermediate information group is integrated according to a predetermined rule, and the information processing device generates a final output document from the integrated intermediate information group using a document generation model.

[0058] (Note 2) The information processing device described in Note 1, wherein the prescribed criteria are based on at least one of the type of document and the characteristics of the document.

[0059] (Note 3) The information processing device described in Note 1 or Note 2, wherein the prescribed rules are in chronological order of the intermediate information based on time information relating to each document or element of each document.

[0060] (Note 4) The information processing device described in Note 3, wherein the time information is one of the time the document was written, the time estimated from the context of the document, or the time the document was acquired.

[0061] (Appendix 5) The information processing device described in Appendix 3 or Appendix 4, wherein the prescribed rules further include rules other than the chronological order.

[0062] (Note 6) The information processing device described in Note 5, which has rules other than the chronological order, is a rule based on either the classification result based on a prescribed standard or the information that forms the basis of the document.

[0063] (Note 7) The information processing device according to any one of Notes 1 to 6, wherein the processor displays the integrated intermediate information group and, when a modification of the integrated intermediate information group is accepted, generates the final deliverable document from the modified intermediate information group.

[0064] (Note 8) The information processing device according to any one of Notes 1 to 7, wherein the processor converts the style of the integrated intermediate information group into an output style before generating the final output document, or converts the style of the final output document into an output style.

[0065] (Note 9) The information processing apparatus according to any one of Notes 1 to 8, wherein the processor uses an extractor designed for each of the plurality of document groups to obtain the intermediate information extracted from each of the plurality of document groups.

[0066] (Note 10) The information processing device according to any one of Notes 1 to 9, wherein the processor generates training data from the documents based on the degree of agreement between the documents included in the document group and the final output document which is the correct answer, for each extractor provided for each of the plurality of document groups, the extracted extractors are trained with the generated training data, and the intermediate information extracted for each of the plurality of document groups is obtained by each of the trained extractors.

[0067] (Note 11) The information processing device according to Note 10, wherein the processor generates the document as training data, with the portion of the document whose degree of agreement with the correct final output is greater than a threshold as a label.

[0068] (Note 12) The information processing device according to Note 10 or Note 11, wherein each of the plurality of document groups is provided with a plurality of extractors classified based on at least one of the predetermined criteria, nouns, events, sentences, lines, paragraphs, and predetermined delimiter expressions.

[0069] (Note 13) An information processing method which involves obtaining multiple document groups in which multiple documents are classified according to predetermined criteria, obtaining an intermediate information group which includes intermediate information extracted from each of the multiple document groups, integrating the intermediate information group according to predetermined rules, and generating a final output document from the integrated intermediate information group using a document generation model.

[0070] (Note 14) An information processing program that causes a computer to perform the following processes: obtaining multiple document groups according to a predetermined standard in which multiple documents are classified; obtaining an intermediate information group including intermediate information extracted from each of the multiple document groups; integrating the intermediate information group according to a predetermined rule; and generating a final output document from the integrated intermediate information group using a document generation model.

[0071] (Note 15) A computer program product that includes an information processing program for causing a processor to perform the following processes: acquire multiple document groups according to a predetermined standard, in which multiple documents are classified according to the said classification; acquire an intermediate information group including intermediate information extracted from each of the multiple document groups; integrate the intermediate information group according to a predetermined rule; and generate a final output document from the integrated intermediate information group using a document generation model.

[0072] (Note 16) A computer-readable storage medium storing an information processing program for a processor that causes the processor to acquire multiple document groups according to a predetermined standard, which are classified according to the classification; acquire an intermediate information group which includes intermediate information extracted from each of the multiple document groups; integrate the intermediate information group according to a predetermined rule; and use a document generation model to generate a final output document from the integrated intermediate information group.

[0073] The disclosure of Japanese Patent Application No. 2024-160428 is incorporated herein by reference in its entirety.

[0074] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

Claims

1. An information processing device comprising a processor, the processor acquires a plurality of document groups according to a predetermined standard, the plurality of documents being classified according to the classification, the plurality of documents being classified according to the classification, the plurality of document groups being acquired as an intermediate information group including intermediate information extracted from each of the plurality of document groups, the intermediate information group being integrated according to a predetermined rule, and the document generation model being used to generate a final output document from the integrated intermediate information group.

2. The information processing apparatus according to claim 1, wherein the predetermined criterion is a criterion based on at least one of the type of document and the characteristics of the document.

3. The information processing apparatus according to claim 1, wherein the prescribed rules are in chronological order of the intermediate information based on time information relating to each document or element of each document.

4. The information processing apparatus according to claim 3, wherein the time information is one of the time the document was written, the time estimated from the context of the document, and the time the document was acquired.

5. The information processing apparatus according to claim 3, wherein the prescribed rules further include rules other than the chronological order.

6. The information processing device according to claim 5, wherein the rules other than the chronological order are rules based on either the classification result based on a predetermined standard or the information underlying the document.

7. The information processing apparatus according to claim 1, wherein the processor displays the integrated intermediate information group, and when a modification of the integrated intermediate information group is accepted, generates the final output document from the modified intermediate information group.

8. The information processing apparatus according to claim 1, wherein the processor converts the style of the integrated intermediate information group into an output style before generating the final output document, or converts the style of the final output document into an output style.

9. The information processing apparatus according to claim 1, wherein the processor obtains the intermediate information extracted from each of the plurality of document groups using an extractor designed for each of the plurality of document groups.

10. The information processing apparatus according to claim 1, wherein the processor generates training data from the documents based on the degree of agreement between the documents included in the document group and the final output document which is the correct answer, for each extractor provided for each of the plurality of document groups, trains each of the extractors using the generated training data, and obtains the intermediate information extracted for each of the plurality of document groups by each of the trained extractors.

11. The information processing apparatus according to claim 10, wherein the processor generates the document as training data, with the portion of the document whose degree of agreement with the correct final output is greater than a threshold as a label.

12. The information processing apparatus according to claim 10, wherein each of the plurality of document groups is provided with a plurality of extractors classified based on at least one of the predetermined criteria, nouns, events, sentences, lines, paragraphs, and predetermined delimiter expressions.

13. An information processing method that obtains multiple document groups, each of which multiple documents are classified according to a predetermined standard; obtains an intermediate information group, which includes intermediate information extracted from each of the multiple document groups; integrates the intermediate information group according to a predetermined rule; and generates a final output document from the integrated intermediate information group using a document generation model.

14. An information processing program that causes a computer to perform the following processes: obtaining multiple document groups according to predetermined criteria, where multiple documents are classified according to said classification; obtaining an intermediate information group containing intermediate information extracted from each of the multiple document groups; integrating the intermediate information group according to predetermined rules; and generating a final output document from the integrated intermediate information group using a document generation model.

Images