Information processing device, method, and program

Abstract

To provide an information processing device, information processing system, method, and program capable of generating confirmation information for confirmation by verifiers related to a given event, tailored to the characteristics of multiple individuals, from text information about the event created by each of the multiple individuals related to that event. [Solution] The information processing device according to this disclosure comprises a text information acquisition unit that acquires text information about a target event created by each of several persons related to the target event, and a first trained model which is a program that uses the text information acquired by the text information acquisition unit and characteristic information indicating the characteristics of the several persons as input information and outputs confirmation information for confirmation by a verifier related to the target event. The first trained model generates confirmation information from the text information according to the characteristics of the several persons.

Description

Technical Field

[0001] This disclosure relates to an information processing apparatus, an information processing system, a method, and a program for processing text information related to a target event involving a plurality of persons.

Background Art

[0002] Conventionally, techniques for extracting specific information from various text information have been proposed. As an example of such a technique, Patent Document 1 discloses obtaining a text meeting record, dividing the text meeting record to form a plurality of meeting paragraphs, generating a meeting paragraph summary for each meeting paragraph, generating a meeting record summary based on the meeting paragraph summary for each meeting paragraph, extracting meeting instructions based on the text meeting record, and generating a minutes of meeting including generating a minutes of meeting based on the meeting record summary and the meeting instructions.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the method for creating minutes of meeting disclosed in Patent Document 1 has a problem that it is impossible to generate confirmation information for a confirmator related to the event according to the characteristics of a plurality of persons from text information related to the event created by each of the plurality of persons related to a predetermined event.

[0005] In view of the above problems, one object of this disclosure is to provide an information processing apparatus, an information processing system, a method, and a program capable of generating confirmation information for a confirmator related to an event according to the characteristics of a plurality of persons from text information related to the event created by each of the plurality of persons related to a predetermined event. [Means for solving the problem]

[0006] The information processing device related to this disclosure is A text information acquisition unit that acquires text information about the target event created by each of several people related to the target event, The system comprises a first trained model, which is a program that uses text information acquired by a text information acquisition unit and characteristic information indicating the characteristics of multiple individuals as input information, and outputs confirmation information for confirmation by a verifier related to the target event. The first pre-trained model generates verification information from text data, tailored to the characteristics of multiple individuals.

[0007] The information processing system related to this disclosure is A text information acquisition unit that acquires text information about the target event created by each of several people related to the target event, The system includes a first trained model, which is a program that uses text information acquired by a text information acquisition unit and characteristic information indicating the characteristics of multiple individuals as input information, and outputs confirmation information for confirmation by a verifier related to the target event. The first pre-trained model generates verification information from text data, tailored to the characteristics of multiple individuals.

[0008] The method relating to this disclosure is a computer, Obtain text information about the target event created by each of several individuals involved in the target event. Using a first pre-trained model, which is a program that takes acquired text information and characteristic information indicating the characteristics of multiple individuals as input information and outputs confirmation information for confirmation by verifiers related to the target event, confirmation information tailored to the characteristics of multiple individuals is generated from the text information.

[0009] The program relating to this disclosure is intended for computers, Obtain text information about the target event created by each of several individuals involved in the target event. Using the first pre-trained model, which takes acquired text information and characteristic information indicating the characteristics of multiple individuals as input information and outputs confirmation information for confirmation by verifiers related to the target event, confirmation information tailored to the characteristics of multiple individuals is generated from the text information. [Effects of the Invention]

[0010] This disclosure provides an information processing device, information processing system, method, and program that can generate confirmation information for confirmation by verifiers related to a given event, tailored to the characteristics of multiple individuals, from text information about the event created by each of the multiple individuals related to the event. [Brief explanation of the drawing]

[0011] [Figure 1] This figure shows an example of the configuration of the information processing device related to this disclosure. [Figure 2] This is a conceptual diagram illustrating the processing of the first trained model related to this disclosure. [Figure 3] This figure shows an example of a problem management table, which is one type of problem information related to this disclosure. [Figure 4] This figure shows an example of the verifier information related to this disclosure. [Figure 5] This figure shows an example of the creator information related to this disclosure. [Figure 6] This figure shows an example of a display image containing warning information output by the first trained model relating to this disclosure. [Figure 7] This figure shows another example of a display image containing warning information output by the first trained model relating to this disclosure. [Figure 8] This figure illustrates an example of a learning method for updating and outputting task management information in the first pre-trained model related to this disclosure. [Figure 9] This figure illustrates an example of a training method for the first pre-trained model related to this disclosure, which is used to generate text that matches the level of understanding of the reviewer. [Figure 10]This is a diagram for explaining an example of a learning method for causing a first learned model according to the present disclosure to create a sentence according to the degree of understanding of a checker. [Figure 11] This is a diagram for explaining an example of a learning method for causing a first learned model according to the present disclosure to create a sentence according to the degree of understanding of a checker and the type of checker. [Figure 12] This is a diagram for explaining an example of a learning method for causing a first learned model according to the present disclosure to output attention information. [Figure 13] This is a diagram for explaining an example of a learning method for causing a first learned model according to the present disclosure to output attention information. [Figure 14] This is a diagram showing an example of a display image including confirmation information output by a first learned model according to the present disclosure. [Figure 15] This is a diagram for explaining another example of a learning method for causing a first learned model according to the present disclosure to output attention information. [Figure 16] This is a flowchart showing an example of a process executed by an information processing apparatus according to the present disclosure. [Figure 17] This is a diagram showing main components of an information processing apparatus according to the present disclosure.

Mode for Carrying Out the Invention

[0012] Hereinafter, an exemplary embodiment will be described with reference to the drawings. FIG. 1 is a schematic diagram showing the configuration of an information processing apparatus 10 according to the present disclosure. The information processing apparatus 10 is an apparatus that processes text information regarding a target event involving a plurality of persons. Specific examples of the information processing apparatus 10 include various information processing apparatuses such as a server and a PC (Personal Computer). Specific examples of the target event include, for example, a project in business. The information processing apparatus 10 includes an arithmetic unit 11, a communication interface (I / F) 12, and a storage device 13.

[0013] The arithmetic unit 11 is a device that performs overall control of the information processing device 10. Specific examples of the arithmetic unit 11 include processors such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), and MPU (Micro Processing Unit). The arithmetic unit 11 executes the methods defined by the program by executing instructions contained in the program stored in the storage device 13. In other embodiments, the functions performed by the arithmetic unit 11 may also be performed by integrated circuits such as FPGAs (Field-Programmable Gate Arrays) and ASICs (Application Specific Integrated Circuits). These devices are equivalent to computers. Furthermore, the information processing device 10 can be equipped with multiple arithmetic units 11.

[0014] The communication interface 12 is an interface for data communication between the information processing device 10 and other devices. The storage device 13 is a storage device that stores various data, such as programs executed by the arithmetic unit 11 and data processed by the arithmetic unit 11. Specific examples of the storage device 13 include memory.

[0015] The arithmetic unit 11 includes a text information acquisition unit 110, a task information acquisition unit 111, a model control unit 112, a first trained model 113, a second trained model 114, and a display control unit 115. These functional units can be implemented by a program.

[0016] The text information acquisition unit 110 acquires text information about the target event created by each of several individuals involved in the target event. Specific examples of text information include communication information between the individuals involved in the target event and reports created by the individuals involved in the target event. Communication information includes, for example, the content of emails exchanged between the individuals involved in the target event, posts on communication apps, or chat content. Reports created by individuals involved in the target event include, for example, weekly reports related to the target event.

[0017] The issue information acquisition unit 111 acquires issue information that indicates issues related to the target event. Examples of issue information include an issue management table that summarizes issues related to a project.

[0018] The model control unit 112 causes the first trained model 113 to output verification information for verification by a verifier related to the target event. The verifier is a person who verifies text information and task information created by a person related to the target event.

[0019] The first pre-trained model 113 is a program that outputs confirmation information from the text information acquired by the text information acquisition unit. Specific examples of the first pre-trained model 113 include large language models (LLMs). The first pre-trained model 113 can be trained using machine learning methods such as deep learning.

[0020] Figure 2 is a conceptual diagram showing the processing of the first pre-trained model 113. The first pre-trained model 113 uses the following as input information: task management information, communication information, reports, verifier information regarding the verifier, creator information regarding the creator of the text information, and specification information to specify cautionary information that the verifier should pay attention to. The first pre-trained model 113 outputs the updated task management information and the cautionary information.

[0021] Task management information is information used to manage task information. Figure 3 shows an example of a task management table, which is one type of task management information. Task management information includes task identification information ("task number"), task information, task creation date, task category, desired response date, respondent's name, response date, and response content. Task identification information is unique identification information assigned to task information. Task creation date is the date the task information was entered. Task category is information used to classify task information. Desired response date is the date on which a response to the task information is desired, and corresponds to the date information indicating the deadline for the task information. Respondent's name is the name of the person who provided a response corresponding to the task indicated by the task information. Response date is the date the respondent provided the response. Response content is the content of the response corresponding to the task indicated by the task information.

[0022] Figure 4 shows an example of verifier information. Verifier information includes the type of verifier and the verifier's characteristics. Specific examples of verifier types include, for example, if the target event is a project to build an information system, the president of the company developing the information system, the person in charge of the cooperating company developing the information system, the information system reviewer, and the information system customer.

[0023] The verifier's characteristic information is information that indicates the verifier's characteristics. This characteristic information includes information indicating the verifier's level of understanding of the subject matter. Specific examples of this level of understanding include, for example, if the subject matter is a project to build an information system, the verifier's level of understanding of the project plan and the system itself. Furthermore, the verifier's characteristic information includes the verifier's IT knowledge, such as IT-related qualifications. Additionally, the verifier's characteristic information includes the verifier's years of experience in their role within the project.

[0024] Figure 5 shows an example of creator information. Creator information includes creator characteristic information. Creator characteristic information includes thought information that shows the creator's thinking about the subject matter. Thinking about the subject matter includes the creator's thinking about deadlines related to the subject matter, thinking about quality related to the subject matter, and thinking about costs related to the subject matter.

[0025] Furthermore, the creator's characteristic information includes information indicating the creator's level of understanding of the subject matter. Specific examples of this level of understanding include, for instance, the creator's understanding of the system if the subject matter is a project to build an information system. Additionally, the creator's characteristic information includes the creator's years of experience in their role within that project. Years of experience can be used as an indicator of the creator's knowledge level regarding the subject matter.

[0026] The designated information is information used to specify cautionary information that the reviewer should pay attention to. This cautionary information includes (1) cautionary information regarding communication between multiple individuals (persons in charge) related to the subject event, (2) cautionary information regarding individual individuals related to the subject event, and (3) cautionary information regarding specific keywords. Cautionary information regarding communication between multiple individuals related to the subject event includes, for example, information indicating disagreements among multiple individuals, or information indicating poor communication among multiple individuals. Cautionary information regarding individual individuals related to the subject event includes, for example, information indicating a decrease in the amount of information reported by each individual, information regarding the system development deadline, information regarding the system development cost, or information regarding the system development quality. Cautionary information regarding specific keywords includes, for example, words related to customer complaints, words related to increased system development costs, words related to increased system development scale, or negative words that could negatively impact system development. The designated information specifies such cautionary information.

[0027] Figure 6 shows an example of a display image containing warning information output by the first trained model 113. In the example shown in Figure 6, each piece of warning information is displayed for each person in charge.

[0028] Figure 7 shows another example of a display image containing warning information output by the first trained model 113. In the example shown in Figure 7, each warning is displayed for each task and person in charge.

[0029] Figure 8 illustrates an example of a learning method for updating and outputting task management information in the first trained model 113. In the example shown in Figure 8, a chat thread 20 from a communication application, a task management table 21, and an updated task management table 22 are used as training data. In this learning method, the model is trained to output the updated task management table 22 when the thread 20 and the task management table 21 are input to the model. Preferably, the thread 20 used as input information includes task information identification information ("task number"). This allows the model to appropriately extract text information indicating the answer corresponding to the task indicated by each task information from the various text information that is input information.

[0030] Furthermore, the model can be trained not only on text information such as task information contained in the task management table, but also on structural information of the data table of the task management table, such as text information indicating the number of rows and columns in the data table. In this case, the first trained model 113 can output the updated task management table and the structural information of the data table of the updated task management table. Moreover, the model can be trained to extract multiple answers for a single task information.

[0031] The first trained model 113 obtained through this learning method can output an updated issue management table 22 when thread 20 and issue management table 21 are input. More specifically, the first trained model 113 extracts information from thread 20 that indicates the answer corresponding to the issue indicated by the issue information. Specifically, the first trained model 113 extracts D's comment dated 8 / 10 and E's comment dated 8 / 30 from thread 20. Then, using the extracted information, the first trained model 113 creates a document that corresponds to the level of understanding of the verifier regarding the target event such as a project, updates the issue management table 21 using the created document, and outputs an updated issue management table 22.

[0032] Figures 9 and 10 illustrate an example of a training method for a first trained model 113 to generate text corresponding to the verifier's level of understanding. In the examples shown in Figures 9 and 10, training data is used in which text information 30 that may be included in communication information or reports and verifier characteristic information 31 or characteristic information 33 are input information, and answer information 32 or answer information 34 related to the answer corresponding to the task indicated by the task information is output information. In the training method shown in Figure 9, the model to be trained is trained to output answer information 32 when text information 30 and characteristic information 31 are input to the model. In the training method shown in Figure 10, the model to be trained is trained to output answer information 34 when text information 30 and characteristic information 33 are input to the model. In these training methods, the model to be trained is instructed to extract answer information related to the answer corresponding to the task indicated by the task information from the text information 30, to use the extracted answer information to generate text corresponding to the verifier's level of understanding of the target event, and to output verification information including the generated text.

[0033] The characteristic information 31 of the verifier shown in Figure 9 indicates that the verifier has a high level of understanding of the project plan for constructing the information system, which is the target event. The response information 32 is the response information when the verifier has a high level of understanding. The characteristic information 33 of the verifier shown in Figure 10 indicates that the verifier has a moderate level of understanding of the project plan for constructing the information system, which is the target event. The response information 34 is the response information when the verifier has a moderate level of understanding.

[0034] The amount of information in the response should be reduced as the reviewer's level of understanding increases, and conversely, increased as the reviewer's level of understanding decreases. Specifically, if the reviewer's level of understanding is high, supplementary explanations of terms included in the response can be omitted, while if the reviewer's level of understanding is moderate or lower, the response can include supplementary explanations of terms included in the response. For example, in response information 32 shown in Figure 9, the supplementary explanation for "related system A" is omitted, while in response information 34 shown in Figure 10, a sentence including the supplementary explanation for "related system A" can be used.

[0035] Figure 11 illustrates an example of a training method for a first trained model 113 to generate text corresponding to the verifier's level of understanding and type of verifier. In the example shown in Figure 11, training data is used in which text information 35 that may be included in communication information or reports, verifier characteristic information 31, and type information 36 indicating the type of verifier are input information, and response information 37 relating to the answer corresponding to the task indicated by the task information and the type of verifier is output information. In the training method shown in Figure 11, the model to be trained is trained to output response information 37 when text information 35, characteristic information 31, and type information 36 are input to the model. In these training methods, the model to be trained is instructed to extract response information relating to the answer corresponding to the task indicated by the task information from the text information 35, to use the extracted response information to generate text corresponding to the verifier's level of understanding and type of verifier regarding the target event, and to output verification information including the generated text.

[0036] Text information 35 includes sentences corresponding to the type of reviewer. In the example shown in Figure 11, the sentence "There is ample leeway in the subsequent processes, so we judge that there is no problem assuming that we can make up for the delay." corresponds to the sentence corresponding to the type of reviewer.

[0037] The type information 36 shown in Figure 11 is an example where the type of verifier is the person in charge at the cooperating company. The response information 37 is response information that includes sentences appropriate to the verifier's level of understanding of the target event and the type of verifier. In the example shown in Figure 11, the sentence "Furthermore, this adjustment is the cooperating company's task, and the project manager expects it to be recovered," included in the response information 37, corresponds to sentences appropriate to the type of verifier. Sentences appropriate to the type of verifier are information related to the type of verifier. For example, if the type of verifier is the person in charge at the cooperating company, sentences appropriate to the type of verifier are information related to the person in charge at the cooperating company. In the example in Figure 11, the information indicating that the design item is the cooperating company's task corresponds to information related to the person in charge at the cooperating company. Note that the sentences shown in Figure 11 are examples, and any sentences can be used to train the model.

[0038] Figures 12 and 13 illustrate an example of a learning method for causing a first trained model 113 to output attention information. In the example shown in Figures 12 and 13, training data is used in which text information 40, 50 that may be included in communication information or reports, thought information 41, 51 which is characteristic information of the creator, and designation information 42, 52 are used as input information, and confirmation information 43, 53 are used as output information. In the learning method shown in Figure 12, the model to be trained is trained to output confirmation information 43 when text information 40, thought information 41 and designation information 42 are input to the model. In the learning method shown in Figure 13, the model to be trained is trained to output confirmation information 53 when text information 50, thought information 51 and designation information 52 are input to the model. In these learning methods, the model to be trained is instructed to extract attention information contained in information about the schedule, information about costs and / or quality from the text information 40, 50, and to output the extracted attention information along with information indicating the degree of attention.

[0039] In the examples shown in Figures 12 and 13, text information 40 and 50 includes information about the schedule ("The schedule is behind schedule."), information about costs ("There is no need to add personnel for the work that may be delayed."), and information about quality ("The quality is fine."). The designation information 42 and 52 shown in Figures 12 and 13 is designation information that specifies cautionary information about individual people related to the event in question. Note that since personnel costs are incurred when adding personnel, the text information regarding the increase in personnel falls under cost information.

[0040] The thinking information 41 shown in Figure 12 includes information indicating an optimistic view of deadlines, a cautious view of costs, and an optimistic view of quality. The thinking information 51 shown in Figure 13 includes information indicating a cautious view of deadlines, an optimistic view of costs, and a cautious view of quality.

[0041] The schedule information contained in text information 40 and 50 is information that the verifier should pay attention to, so the model being trained is instructed to output the schedule information contained in text information 40 and 50. Here, since thought information 41 contains information indicating that the viewer is optimistic regarding deadlines, the model is instructed to output information indicating a high level of attention. On the other hand, since thought information 51 contains information indicating that the viewer is cautious regarding deadlines, the model is instructed to output information indicating a moderate level of attention.

[0042] The cost information contained in text information 40 and 50 is not information that the verifier should pay attention to. Since thought information 41 contains information indicating a cautious approach to costs, the model being trained will not output the cost information contained in text information 40. On the other hand, since thought information 51 contains information indicating an optimistic approach to costs, the model will output the cost information contained in text information 50, as well as information indicating a moderate level of attention.

[0043] The quality information contained in text information 40 and 50 is not information that the reviewer should pay attention to. Since thought information 41 contains information indicating that the attitude towards quality is optimistic, the model to be trained will output the quality information contained in text information 40, as well as information indicating that the level of attention is moderate. On the other hand, since thought information 51 contains information indicating that the attitude towards quality is cautious, the model will not output the quality information contained in text information 50.

[0044] Figure 14 shows an example of a display image containing confirmation information output by the first trained model 113. In this example, it is an example of confirmation information output when specified information (not shown) that specifies cautionary information for individual personnel is input to the first trained model 113.

[0045] The example shown in Figure 14 illustrates a case where two authors, A and B, with different cognitive information, create identical reports (weekly reports) 60. The cognitive information 61 indicates that author A is optimistic regarding deadlines, while author B is cautious. Both authors are cautious regarding costs. Both authors are optimistic regarding quality.

[0046] In this case, if the report 60 created by creator A contains text information indicating that there is a delay in the schedule, the first trained model 113 outputs confirmation information 62 that includes text information indicating that there is a delay in the schedule and information indicating that the level of attention is high. On the other hand, if the report 60 created by creator B contains text information indicating that there is a delay in the schedule, the first trained model 113 outputs confirmation information 62 that includes text information indicating that there is a delay in the schedule and information indicating that the level of attention is moderate.

[0047] Furthermore, if the reports 60 created by creator A and creator B each contain text information indicating that there are no quality issues, the first trained model 113 outputs confirmation information 62 that includes text information indicating that there are no quality issues and information indicating that the level of attention is moderate.

[0048] Figure 15 illustrates another example of a training method for causing the first trained model 113 to output attention information. In the example shown in Figure 15, training data is used in which text information 70 that may be included in communication information or reports and designation information 71 that specifies attention information regarding communication between multiple people related to the target event are used as input information, and confirmation information 72 is used as output information. In the training method shown in Figure 15, the model to be trained is trained to output confirmation information 72 when the text information 70 and designation information 71 are input to the model.

[0049] In the example shown in Figure 15, there is a difference of opinion between person A and person B, and person A's comments included in text information 70 constitute cautionary information that the reviewer should pay attention to. In this case, the model being trained is trained to output confirmation information 72 that includes a summary of person A's comments that constitute cautionary information.

[0050] Furthermore, the first pre-trained model 113 can extract a specific keyword from the input text information if the specified information specifies cautionary information about that specific keyword. In this case, by training the model using training data in which text information containing the specific keyword and specified information specifying cautionary information about that specific keyword are input information, and confirmation information containing the specific keyword is output information, a pre-trained model that outputs confirmation information containing the specific keyword can be obtained.

[0051] The second pre-trained model 114 is a program that generates proposal information regarding the updating of characteristic information. Specific examples of the second pre-trained model 114 include large-scale language models.

[0052] The display control unit 115 causes the confirmation information output by the first trained model 113 to be displayed on the display device. In this embodiment, the display device can be a display device provided by the information processing device 10, a display device that the information processing device 10 can communicate with, a display device provided by a device that the information processing device 10 can communicate with, etc.

[0053] Figure 16 is a sequence diagram showing an example of a process performed by the information processing device 10. In step S1, the text information acquisition unit 110 acquires text information. In step S2, the task information acquisition unit 111 acquires task information. In step S3, the model control unit 112 causes the first trained model 113 to output confirmation information. In step S4, the display control unit 115 displays the confirmation information on the display device, and the process shown in Figure 16 is completed.

[0054] Figure 17 shows the main components of the information processing device 10. The information processing device 10 comprises a text information acquisition unit 110 and a first trained model 113. The text information acquisition unit 110 acquires text information about the target event created by each of several people related to the target event. The first trained model 113 uses the text information acquired by the text information acquisition unit and characteristic information indicating the characteristics of the several people as input information, and generates and outputs confirmation information corresponding to the characteristics of the several people from the text information.

[0055] By adopting this configuration, it is possible to generate confirmation information for verification by verifiers involved in a given event, tailored to the characteristics of each individual involved in that event, from text information about that event created by each of the individuals involved in that event.

[0056] Furthermore, the information processing device 10 includes a task information acquisition unit 111 that acquires task information indicating tasks related to the target event. The first trained model 113 extracts answer information from text information regarding answers corresponding to tasks indicated by the task information, uses the extracted answer information to create a document that corresponds to the verifier's level of understanding of the target event indicated by the characteristic information, and outputs verification information including the created document.

[0057] By adopting this structure, it becomes possible to provide answers to the challenges that are tailored to the level of understanding of the subject matter by the person reviewing the information.

[0058] Furthermore, the first pre-trained model 113 uses type information indicating the type of verifier as input information. The first pre-trained model 113 extracts answer information from the text information regarding the answers corresponding to the tasks indicated by the task information, and uses the extracted answer information to create sentences according to the verifier's level of understanding of the target event and the type of verifier, and outputs verification information including the created sentences.

[0059] By adopting this structure, it is possible to provide answers to the challenges that are tailored to the level of understanding of the subject matter by the reviewer and the type of reviewer.

[0060] Furthermore, the information processing device 10 includes a second trained model 114 that generates suggestion information regarding the updating of characteristic information. When generating suggestion information regarding the level of understanding or knowledge related to a target event for each of several individuals, the second trained model 114 uses the results of an evaluation test on the level of understanding or knowledge related to the target event for each of the several individuals as input information, and outputs information indicating the level of understanding or knowledge according to the results of the evaluation test as suggestion information.

[0061] By adopting this configuration, it is possible to provide suggested information that indicates the level of understanding of each individual regarding the target phenomenon. This allows the suggested content indicated by the suggested information to be used to update characteristic information.

[0062] Furthermore, the characteristic information includes thought information that shows the thoughts of multiple individuals regarding the target event. The first trained model 113 uses designation information that specifies attention information to be extracted from text information and thought information as further input information, extracts the attention information specified by the designation information from the text information, and outputs confirmation information that includes the extracted attention information and information indicating the degree of attention the confirmer paid to the confirmation information.

[0063] By adopting this configuration, cautionary information is output along with information indicating the degree of attention the verifier paid to the confirmation information, allowing the verifier to understand the degree of attention paid to that cautionary information along with the cautionary information itself.

[0064] Furthermore, the information processing device 10 includes a second trained model 114 that generates suggestion information regarding the updating of characteristic information. When generating suggestion information regarding the approach to deadlines, the second trained model 114 uses date information indicating the deadline and date information on which each of the multiple individuals created the text information as input information, and outputs information indicating the approach to deadlines as suggestion information, based on the difference between the date information indicating the deadline and the date information on which the text information was created.

[0065] By adopting this configuration, it is possible to provide proposal information that shows the respective approaches of multiple individuals regarding deadlines. This allows the content of the proposal information to be used to update characteristic information.

[0066] Furthermore, when generating proposal information regarding approaches to quality, the second trained model 114 uses information indicating the quality of deliverables related to the target event created by each of several individuals as input information, and outputs information indicating approaches to quality related to the target event, corresponding to the quality of the deliverables, as proposal information.

[0067] By adopting this configuration, it is possible to provide proposal information that shows the different perspectives on quality of multiple individuals. This allows the content of the proposal information to be used to update the characteristic information.

[0068] Furthermore, when generating proposed information regarding cost considerations, the second trained model 114 uses the planned and actual work times for each of multiple individuals related to the respective event as input information, and outputs information as proposed information that shows how to consider costs depending on whether the actual work time exceeds the planned work time.

[0069] By adopting this configuration, it is possible to provide proposal information that shows the different cost perspectives of multiple individuals. This allows the proposed content shown in the proposal information to be used to update characteristic information.

[0070] Furthermore, the characteristic information includes comprehension information indicating the degree of understanding of each of the multiple individuals regarding the aforementioned target event, or knowledge level information indicating knowledge related to the target event. The first trained model 113 uses multiple text information items related to the target event created by the same person, along with the comprehension information or knowledge level information, as input information, and outputs confirmation information including cautionary information when the amount of information in the multiple text information items is less than a predetermined amount of information corresponding to the comprehension information or knowledge level information. The predetermined amount of information can be any number of characters. The higher the level of comprehension, the more information can be included. Also, the higher the level of knowledge, the more information can be included.

[0071] The first pre-trained model 113 can be trained by machine learning using training data that consists of multiple text information created by the same person, where the text information is below a predetermined amount corresponding to the same person's comprehension or knowledge level information, and the comprehension or knowledge level information of the same person as input, with confirmation information including attention-grabbing information as output. Alternatively, the first pre-trained model 113 can be trained by machine learning using training data that consists of multiple text information created by the same person, where the text information is above a predetermined amount corresponding to the same person's comprehension or knowledge level information, and the confirmation information does not include attention-grabbing information as output.

[0072] By adopting this configuration, if the text information created by a person related to the target event falls below the amount of information corresponding to that person's knowledge level, confirmation information including cautionary information can be output. This allows the reviewer to understand, for example, a decline in the person's performance due to poor health.

[0073] Furthermore, the information processing device 10 includes a display control unit 115 that displays the attention information contained in the confirmation information output by the first trained model 113 on a display device in a predetermined display manner according to the degree of attention the confirmer is paying to the confirmation information. The display manner includes the color and size of the characters. The display control unit 115 can display the attention information in different character colors and sizes according to the degree of attention. For example, the higher the degree of attention, the more the information can be displayed in a color that attracts the confirmer's attention (e.g., red). Also, the higher the degree of attention, the larger the characters can be displayed.

[0074] By adopting this configuration, verification information is displayed in a manner that corresponds to the level of attention the reviewer is paying. This allows the reviewer to understand the displayed verification information according to their level of attention.

[0075] Furthermore, if the confirmation information contains a specific keyword, the display control unit 115 can display that specific keyword in a different color and size than other words.

[0076] In other embodiments, the information processing device 10 may include a work information acquisition unit instead of the task information acquisition unit 111. In this case, the first trained model 113 extracts information related to work information from text information, uses the extracted information to create a document corresponding to the verifier's level of understanding of the target event indicated by the characteristic information, and outputs verification information including the created document.

[0077] By adopting this configuration, it is possible to provide confirmation information that includes text based on information related to the work, tailored to the level of understanding of the person reviewing the target event.

[0078] In other embodiments, the first trained model 113 may use multiple text information items related to a target event created by the same person, along with knowledge level information, as input information, and output confirmation information including warning information when the amount of information in the multiple text information items decreases by a predetermined amount or more. The predetermined amount of information can be any amount.

[0079] By adopting this configuration, if the amount of information in the text created by a person related to the target event decreases, confirmation information including warnings can be output. This allows the person reviewing the information to understand, for example, a decline in the person's performance due to poor health.

[0080] In the examples described above, the program includes a set of instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more of the functions described in the embodiments. The program may be stored on a non-temporary computer-readable medium or a physical storage medium. Examples, but not limited to, include random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technologies, CD-ROM, digital versatile disk (DVD), Blu-ray® disc or other optical disc storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices. The program may be transmitted over a temporary computer-readable medium or a communication medium. Examples, but not limited to, include temporary computer-readable medium or a communication medium that includes electrically, optically, acoustically, or otherwise propagating signals.

[0081] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure can be made as can be understood by those skilled in the art within the scope of the present disclosure. Furthermore, each embodiment can be combined with other embodiments as appropriate.

[0082] Each drawing is merely illustrative to illustrate one or more embodiments. Each drawing may be associated with one or more other embodiments rather than with only one specific embodiment. As those skilled in the art will understand, various features or steps described with reference to any one drawing can be combined with features or steps shown in one or more other drawings, for example, to create embodiments not explicitly shown or described. Not all features or steps shown in any one drawing to illustrate an exemplary embodiment are necessarily required, and some features or steps may be omitted. The order of steps shown in any of the drawings may be changed as appropriate.

[0083] Some or all of the above embodiments may also be described as follows, but are not limited to the following: (Note 1) A text information acquisition unit that acquires text information about the target event created by each of several people related to the target event, A first trained model is a program that uses the text information acquired by the text information acquisition unit and characteristic information indicating the characteristics of the multiple persons as input information, and outputs confirmation information for confirmation by the verifier related to the target event. Equipped with, The first trained model is an information processing device that generates the confirmation information corresponding to the characteristics of the multiple persons from the text information. (Note 2) The system includes a problem information acquisition unit that acquires problem information indicating issues related to the aforementioned target event, The characteristic information includes information indicating the degree of understanding of the verifier regarding the target event, The first trained model extracts answer information from the text information relating to the answer to the task indicated by the task information, uses the extracted answer information to create a sentence corresponding to the verifier's level of understanding of the target event, and outputs the verification information including the created sentence, as described in Appendix 1. (Note 3) The first trained model further uses type information indicating the type of the verifier as input information, The first trained model extracts answer information from the text information relating to the answer to the task indicated by the task information, uses the extracted answer information to create a document corresponding to the verifier's level of understanding and type of verifier regarding the target event, and outputs the verification information including the created document, as described in Appendix 2. (Note 4) The system includes a work information acquisition unit that acquires work information indicating tasks related to the aforementioned target event, The characteristic information includes information indicating the degree of understanding of the verifier regarding the target event, The first trained model extracts information related to the work information from the text information, uses the extracted information to create a document corresponding to the verifier's level of understanding of the target event, and outputs the verification information including the created document, as described in Appendix 1. (Note 5) The system comprises a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, In generating suggestion information regarding the degree of understanding of each of the multiple persons regarding the target event or the knowledge related to the target event, the second trained model uses the results of an evaluation test on the degree of understanding of each of the multiple persons regarding the target event or the knowledge related to the target event as input information, and outputs information indicating the degree of understanding or knowledge level according to the results of the evaluation test as the suggestion information, as described in any one of Appendix 2 to 4. (Note 6) The characteristic information includes thought information that shows the thoughts of the multiple persons regarding the target event, The information processing device described in Appendix 1 further uses the first trained model as input information, specifying information that specifies attention information to be extracted from the text information and the thought information, extracts the attention information specified by the specifying information from the text information, and outputs the extracted attention information and confirmation information that includes information indicating the degree of attention of the confirmer to the confirmation information. (Note 7) The information processing device described in Appendix 6, wherein the thinking regarding the aforementioned target event includes an approach to deadlines related to the aforementioned target event, an approach to quality related to the aforementioned target event, and an approach to costs related to the aforementioned target event. (Note 8) The system comprises a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, The information processing device described in Appendix 7, which generates proposed information regarding the approach to the deadline, uses the second trained model as input information, the date information indicating the deadline and the date information on which each of the multiple persons created the text information, and outputs information indicating the approach to the deadline, corresponding to the difference between the date information indicating the deadline and the date information on which the text information was created, as the proposed information. (Note 9) The system comprises a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, The information processing device described in Appendix 7, which generates proposed information regarding the aforementioned approach to quality, wherein the second trained model uses information indicating the quality of deliverables related to the target event created by each of the multiple persons as input information, and outputs information indicating the approach to quality related to the target event in accordance with the quality of the deliverables as proposed information. (Note 10) The system comprises a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, The information processing device described in Appendix 7, which generates proposed information regarding the approach to the aforementioned costs, uses the second trained model as input information for the planned work time and actual work time related to the aforementioned event for each of the multiple persons, and outputs information as proposed information indicating the approach to the aforementioned costs depending on whether or not the actual work time exceeds the planned work time. (Note 11) The characteristic information includes understanding level information indicating the degree of understanding of each of the multiple individuals regarding the subject event, or knowledge level information indicating knowledge related to the subject event. The information processing device described in Appendix 1, wherein the first trained model uses multiple text information relating to the target event created by the same person and the comprehension level information or the knowledge level information as input information, and outputs the confirmation information including information to draw attention when the amount of information in the multiple text information is less than a predetermined amount of information corresponding to the comprehension level information or the knowledge level information. (Note 12) An information processing apparatus according to any one of the appendices 1 to 4, 6 to 11, comprising a display control unit that displays the attention information contained in the confirmation information output by the first trained model on a display device in a predetermined display manner according to the degree of attention of the confirmer to the confirmation information. (Note 13) A text information acquisition unit that acquires text information about the target event created by each of several people related to the target event, A first trained model is a program that uses the text information acquired by the text information acquisition unit and characteristic information indicating the characteristics of the multiple persons as input information, and outputs confirmation information for confirmation by the verifier related to the target event. Includes, The first trained model is an information processing system that generates the confirmation information from the text information according to the characteristics of the multiple individuals. (Note 14) Computers Obtain text information about the target event created by each of several individuals involved in the target event, Using a first trained model, which is a program that uses acquired text information and characteristic information indicating the characteristics of the multiple persons as input information and outputs confirmation information for confirmation by a verifier related to the target event, the program generates the confirmation information from the text information according to the characteristics of the multiple persons. method. (Note 15) For computers, The system retrieves text information about the target event created by each of several individuals involved in the target event. Using a first trained model, which is a program that uses acquired text information and characteristic information indicating the characteristics of the multiple persons as input information and outputs confirmation information for confirmation by a verifier related to the target event, the confirmation information corresponding to the characteristics of the multiple persons is generated from the text information. program.

[0084] Some or all of the elements described in Appendices 2 to 12, which are dependent on Appendice 1, may also be dependent on Appendices 13, 14, and 15 in the same manner as those described in Appendices 2 to 12. Some or all of the elements described in any appendice may be applicable to various hardware, software, recording means, systems, and methods for recording software. [Explanation of symbols]

[0085] 10: Information Processing Device 11: Arithmetic device 12: Communication Interface 13:Storage device 110: Text information acquisition unit 111: Assignment information acquisition department 112: Model Control Unit 113: First pre-trained model 114: Second pre-trained model 115: Display Control Unit

Claims

1. A text information acquisition unit that acquires text information about the target event created by each of several people related to the target event, A first trained model is a program that uses the text information acquired by the text information acquisition unit and characteristic information indicating the characteristics of the multiple persons as input information, and outputs confirmation information for confirmation by the verifier related to the target event. Equipped with, The first trained model is an information processing device that generates the confirmation information corresponding to the characteristics of the multiple individuals from the text information.

2. The system includes a problem information acquisition unit that acquires problem information indicating issues related to the aforementioned target event, The characteristic information includes information indicating the degree of understanding of the verifier regarding the target event, The information processing device according to claim 1, wherein the first trained model extracts answer information from the text information relating to the answer to the task indicated by the task information, uses the extracted answer information to create a sentence according to the verifier's level of understanding of the target event, and outputs the verification information including the created sentence.

3. The first trained model further uses type information indicating the type of the verifier as input information, The information processing device according to claim 2, wherein the first trained model extracts answer information relating to the answer to the task indicated by the task information from the text information, uses the extracted answer information to create a document according to the level of understanding of the verifier and the type of verifier regarding the target event, and outputs the verification information including the created document.

4. The system includes a work information acquisition unit that acquires work information indicating tasks related to the aforementioned target event, The characteristic information includes information indicating the degree of understanding of the verifier regarding the target event, The information processing device according to claim 1, wherein the first trained model extracts information relating to the work information from the text information, uses the extracted information to create a document corresponding to the verifier's level of understanding of the target event, and outputs the verification information including the created document.

5. The system includes a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, In generating suggestion information regarding the degree of understanding of each of the multiple persons regarding the target event or the knowledge related to the target event, the second trained model uses the results of an evaluation test on the degree of understanding of each of the multiple persons regarding the target event or the knowledge related to the target event as input information, and outputs information indicating the degree of understanding or knowledge level according to the results of the evaluation test as the suggestion information, as described in any one of claims 2 to 4.

6. The characteristic information includes thought information that shows the thoughts of the multiple persons regarding the target event, The information processing device according to claim 1, wherein the first trained model further uses designation information that specifies attention information to be extracted from the text information and the thought information as input information, extracts the attention information specified by the designation information from the text information, and outputs the extracted attention information and confirmation information that includes information indicating the degree of attention of the confirmer to the confirmation information.

7. The information processing apparatus according to claim 6, wherein the thinking regarding the subject event includes a way of thinking regarding deadlines related to the subject event, a way of thinking regarding quality related to the subject event, and a way of thinking regarding costs related to the subject event.

8. The system includes a second trained model that generates proposed information regarding the updating of the aforementioned characteristic information, The information processing device according to claim 7, which generates proposed information regarding the approach to the deadline, wherein the second trained model uses date information indicating the deadline and date information on which each of the multiple persons created the text information as input information, and outputs information indicating the approach to the deadline according to the difference between the date information indicating the deadline and the date information on which the text information was created as proposed information.

9. Computers Obtain text information about the target event created by each of several individuals involved in the target event, Using a first trained model, which is a program that uses acquired text information and characteristic information indicating the characteristics of the multiple persons as input information and outputs confirmation information for confirmation by a verifier related to the target event, the program generates the confirmation information from the text information according to the characteristics of the multiple persons. method.

10. For computers, The system retrieves text information about the target event created by each of several individuals involved in the target event. Using a first trained model, which is a program that uses acquired text information and characteristic information indicating the characteristics of the multiple persons as input information and outputs confirmation information for confirmation by a verifier related to the target event, the confirmation information corresponding to the characteristics of the multiple persons is generated from the text information. program.

Images