Information processing system, program, and information processing method

The system addresses the challenge of generating user-specific summaries from unstructured text data by allowing interactive editing and incorporating subtopics, ensuring accurate and informative summaries.

JP7583495B1Active Publication Date: 2024-11-14QUANTUM NEXUS INC

Patent Information

Application Number
JP2024143420
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-08-23
Publication Date
2024-11-14
Estimated Expiration
2044-08-23

AI Technical Summary

Technical Problem

Existing technologies struggle to generate summaries from unstructured text data, particularly from voice recognition of conference audio, that are suitable for the user's needs and may contain errors or lack necessary information.

Method used

An information processing system that includes an interactive part to obtain multiple topics from text data using a generative model, allows user interaction for editing instructions, and generates summaries by inputting edited data to the model, incorporating subtopics and additional information, and provides a structure diagram.

Benefits of technology

Enables the generation of summaries tailored to user preferences, correcting errors, and ensuring inclusion of necessary information, facilitating accurate and language-independent information sharing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007583495000001_ABST
    Figure 0007583495000001_ABST
Patent Text Reader

Abstract

An information processing system, a program, and an information processing method are provided that are capable of acquiring a summary suited to a user from text data. [Solution] The system includes a dialogue unit 102 that acquires multiple topics by inputting text data into a generative model 210, a presentation unit 103 that presents the topics to a user, and an editing instruction acquisition unit 104 that acquires editing instructions for the topics obtained in response to user operations on the topics. The dialogue unit 102 generates input data to be input into the generative model 210 in accordance with the editing instructions, and acquires summaries corresponding to the topics related to the editing instructions by inputting the input data into the generative model.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an information processing system, a program, and an information processing method. [Background technology]

[0002] Conventionally, there are known techniques for generating summaries from text data. Patent Document 1 discloses a technique for generating candidate data of summary data constrained according to a predetermined constraint condition received from a client user, and further generating final data of the summary data based on the constraint condition updated based on the candidate data. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2022-165786 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, there is a problem in that it is difficult to generate a summary suitable for the user from unstructured text data, such as text data obtained by speech recognition of conference audio data.

[0005] The present invention has been made in consideration of the above points, and has an object to provide an information processing system, a program, and an information processing method that are capable of acquiring a summary suited to a user from text data. [Means for solving the problem]

[0006] The information processing system of the present invention comprises: A dialogue part that acquires multiple topics by inputting text data into a generative model; a presentation unit that presents the topic to a user; an editing instruction acquisition unit that acquires editing instructions for the topic obtained in response to a user operation on the topic; Equipped with The dialogue unit is characterized in that it generates input data to be input to the generative model in accordance with the editing instructions, and obtains a summary corresponding to the topic related to the editing instructions by inputting the input data to the generative model.

[0007] The information processing system of the present invention comprises: the dialogue unit further acquires the subtopic by inputting an instruction to generate a subtopic for the topic into the generative model; The dialogue unit may obtain the summary by inputting the text data, the topic, and the subtopic into the generative model.

[0008] The information processing system of the present invention comprises: The dialogue unit may obtain the summary by inputting the text data, the topic, and a subtopic obtained in response to a user operation into the generative model.

[0009] The information processing system of the present invention comprises: The editing instructions include instructions for modifying the subtopic; The dialogue unit may obtain the summary by inputting the text data, the topic, and the modified subtopic into the generative model.

[0010] The information processing system of the present invention comprises: the editing instruction includes an instruction to add additional information to the text data, The dialogue unit may obtain the summary by inputting the text data, the topic, and the added additional information into the generative model.

[0011] The information processing system of the present invention comprises: the editing instruction includes an instruction to modify the additional information, The dialogue unit may obtain the summary by inputting the text data, the topic, and the modified additional information into the generative model.

[0012] The information processing system of the present invention comprises: The dialogue unit may generate an overall summary for the text data by inputting the text data into a generative model, and obtain a revised summary corresponding to the topic by inputting the overall summary and a summary corresponding to the topic into the generative model.

[0013] The information processing system of the present invention comprises: The dialogue unit may obtain a structure diagram showing a document structure of the text data by inputting the summary into a generative model.

[0014] The information processing system of the present invention comprises: The apparatus may further include a translation unit that translates the summary into a language different from that of the text data.

[0015] The information processing system of the present invention comprises: The system may further include a recognition unit that acquires the text data by voice recognition or character recognition.

[0016] The program of the present invention comprises: A program for causing a computer to function as a dialogue unit, a presentation unit, and an editing instruction acquisition unit, The dialogue unit inputs text data into a generative model to obtain a plurality of topics; The presentation unit presents the topic to a user; the editing instruction acquisition unit acquires an editing instruction for the topic obtained in response to a user operation on the topic; The dialogue unit is characterized in that it generates input data to be input to the generative model in accordance with the editing instructions, and obtains a summary corresponding to the topic related to the editing instructions by inputting the input data to the generative model.

[0017] The information processing method of the present invention comprises the steps of: An information processing method performed by a computer having a control unit, The control unit inputs text data into a generative model to obtain a plurality of topics; the control unit presenting the topic to a user; a pre-control unit acquiring an editing instruction for the topic obtained in response to a user operation on the topic; a step of, by the control unit, generating input data to be input to the generative model in accordance with the editing instruction, and inputting the input data to the generative model, thereby obtaining a summary corresponding to the topic related to the editing instruction; The present invention is characterized by comprising: Effect of the Invention

[0018] According to the information processing system, the program, and the information processing method of the present invention, a summary suited to a user can be obtained from text data. [Brief description of the drawings]

[0019] [Figure 1] 1 is an overall configuration diagram of a summary generation system. [Diagram 2] 13 is a flowchart showing a summary generation process. [Diagram 3] FIG. 2 is a diagram illustrating an example of text data. [Figure 4] FIG. 13 is a diagram illustrating an example of a topic. [Diagram 5] FIG. 13 is a diagram illustrating an example of a summary. [Figure 6] FIG. 13 is a diagram showing an example of a displayed topic list after correction. [Figure 7] FIG. 13 is a diagram illustrating an example of a correspondence table. [Figure 8] FIG. 13 is a diagram showing an example of a revised summary. [Figure 9] FIG. 11 is a diagram illustrating an example of relationship information. [Figure 10] FIG. 1 is a diagram showing an example of a structural diagram. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Figures 1 to 10 are diagrams showing a program, an information processing system, an information processing method, etc. according to the present embodiment.

[0021] FIG. 1 is an overall configuration diagram of a summary generation system 1 according to this embodiment. The summary generation system 1 generates a summary for input text data using a generative model. The summary generation system 1 includes a summary generation device 10, a generative model server device 20, and a user device 30. The summary generation device 10 is an example of an information processing system. The summary generation device 10, the generative model server device 20, and the user device 30 are connected to each other so as to be able to communicate with each other via a communication network 2 such as the Internet.

[0022] The summary generation device 10 is a computer or the like, and generates a summary from text data by utilizing interaction with a generative model 210 stored in a generative model server device 20. The summary generation device 10 is configured with a computer or the like, and includes a control unit 100, a storage unit 110, and a communication unit 120.

[0023] The control unit 100 includes a processor such as a CPU (Central Processing Unit) and controls the operation of the summary generation device 10. The storage unit 110 includes, for example, a hard disk drive (HDD), a random access memory (RAM), a read only memory (ROM), and a solid state drive (SSD). The storage unit 110 is not limited to being built into the summary generation device 10, and may be a storage medium (for example, a USB memory) that is detachably attached to the summary generation device 10. In this embodiment, the storage unit 110 stores various programs executed by the control unit 100 and various data.

[0024] The communication unit 120 includes a communication interface for communicating with an external device wirelessly or via communication. The control unit 100 transmits and receives data to and from the generation model server device 20 and the user device 30 via the communication unit 120.

[0025] The generation model server device 20 is a computer or the like. The generation model server device 20 stores a generation model 210 in which a question sentence is input and an answer sentence is output. The generation model 210 is a learning model generated by machine learning, and generates an answer sentence in response to a question sentence (prompt) from a user who accesses via the communication network 2. The generation model may be a Generative Pre-trained Transformer or a language model based thereon. Examples of the generation model 210 include ChatGPT, GPT-3, GPT-4 by OpenAI, and GPT-J by EleutherAI.

[0026] The generative model server device 20 is used for large-scale natural language processing tasks, for example, automatically performing machine translation, sentence generation, document classification, question answering, etc. The generative model 210 learns a huge amount of data and then makes predictions for new inputs. The generative model server device 20 hosts the generative model 210, invokes the generative model 210 in response to a request, and returns the result. When the generative model server device 20 according to this embodiment receives a prompt from the summary generation device 10, it performs calculation processing to generate an answer sentence. The generated answer sentence is then transmitted to the summary generation device 10.

[0027] The user device 30 is an information processing device. The user device 30 may also be a portable device such as a smartphone or a PC tablet. The user device 30 includes a control unit 310, a storage unit 320, a communication unit 330, an operation unit 340, and a display unit 350. The control unit 310, the storage unit 320, and the communication unit 330 are similar to the control unit 100, the storage unit 110, and the communication unit 120 of the summary generating device 10, respectively. The communication unit 330 includes a communication interface that communicates with an external device wirelessly or via a wire. The operation unit 340 accepts user operations. Examples of the operation unit 340 include a keyboard and a mouse. The display unit 350 is, for example, a monitor, and displays various screens. The operation unit 340 and the display unit 350 may be provided integrally as a touch panel.

[0028] In this embodiment, it is assumed that an application for generating a summary is installed in the user device 30. It is assumed that information is transmitted and received to and from the summary generation device 10 by this application.

[0029] Next, a configuration of the control unit 100 of the summary generating device 10 will be described. The control unit 100 executes a program stored in the storage unit 110 to function as a recognition unit 101, a dialogue unit 102, a presentation unit 103, an editing instruction acquisition unit 104, an error registration unit 105, a structure diagram acquisition unit 106, and a translation unit 107. Note that, in the following, the processes described as being performed by the recognition unit 101, the dialogue unit 102, the presentation unit 103, the editing instruction acquisition unit 104, the error registration unit 105, the structure diagram acquisition unit 106, and the translation unit 107 are processes performed by the control unit 100 executing the program.

[0030] The recognition unit 101 acquires speech data to be used for generating a summary from the user device 30 via the communication unit 120. Examples of the speech data include data related to discussions on a research topic, research presentations, and corporate meetings. The recognition unit 101 acquires text data from the speech data through speech recognition processing. The recognition unit 101 may also acquire video data including speech, convert the data into speech data, and then perform speech recognition.

[0031] The dialogue unit 102 transmits a prompt to the generative model server device 20 via the communication unit 120, thereby acquiring an answer sentence obtained by the generative model 210 from the generative model server device 20. The dialogue unit 102 generates and transmits, for example, together with text data, a prompt requesting the generation of multiple topics and summaries corresponding to each topic for the content indicated in the text data. In this way, the dialogue unit 102 acquires multiple topics and summaries corresponding to the topics.

[0032] The presentation unit 103 presents to the user the topics and summaries obtained by the generative model 210. Specifically, the presentation unit 103 transmits the topics and summaries to the user device 30 via the communication unit 120, and causes the display unit 350 to display them.

[0033] The editing instruction acquisition unit 104 acquires editing instructions from the user device 30 via the communication unit 120. Here, the editing instructions are instructions such as to correct typos included in a topic.

[0034] The error registration unit 105 registers errors contained in the text data and the correct spellings corresponding to the errors in a substitution table stored in the storage unit 110. The structure diagram acquisition unit 106 acquires a structure diagram showing the structure of the document represented by the text data based on a summary of the text data. The translation unit 107 translates the topic and summary. The processing of each unit will be described in detail later.

[0035] FIG. 2 is a flowchart showing the summary generation process by the summary generation device 10. In step S100, the recognition unit 101 acquires voice data from the user device 30. Next, in step S102, the recognition unit 101 acquires text data by performing voice recognition on the acquired voice data. FIG. 3 is a diagram showing an example of text data. The text data shown in FIG. 3 is text data acquired by voice recognition on the voice data, and corresponds to voice data in which an explanation of the "Suzuki-Miyaura coupling reaction" is recorded. This data can be seen at the following website: https: / / www.youtube.com / watch?v=MiEqmtm3yRs Text data obtained by speech recognition may contain errors such as conversion errors into kanji. For example, in the example shown in Figure 3, the correct word is "Suzuki Miyaura coupling," but it is written in katakana as "Suzuki Miyaura coupling." Also, the correct word is "Suzuki Akira Sensei," but it is written as "Suzuki Akira Sensei."

[0036] Furthermore, a substitution table for changing a typographical error into a correct notation is stored in the storage unit 110, and the recognition unit 101 corrects the typographical error in the text data by utilizing the substitution table. The substitution table is generated by a user, transmitted from the user device 30 to the summary generation device 10, and stored in the storage unit 110, for example.

[0037] Next, in step S104, the dialogue unit 102 generates a prompt requesting the generation of multiple topics and the generation of a summary for each topic together with the text data obtained in step S100. Then, the dialogue unit 102 transmits this prompt to the generative model server device 20 via the communication unit 120. The dialogue unit 102 acquires multiple topics and summaries corresponding to each topic from the generative model server device 20 via the communication unit 120. Hereinafter, these topics and summaries are referred to as primary topics and primary summaries, respectively. The dialogue unit 102 requests the generation of, for example, 10 topics. Furthermore, the dialogue unit 102 may limit the number of characters obtained as topics in the prompt. This makes it possible to obtain a topic with the user's desired number of characters.

[0038] From the text data shown in Figure 3, for example, ten first-order topics as shown in Figure 4 can be obtained. Furthermore, as shown in Figure 5, first-order summaries corresponding to each first-order topic can be obtained. The first-order summary shown in Figure 5 contains typos such as "Professor Akira Suzuki." Also, it is written that "The original paper was published in a little-known journal," and information about the journal is insufficient. In this way, the initially generated summary, i.e., the first-order summary, contains typos and is insufficient in information.

[0039] Returning to FIG. 2 for the explanation. After the process of step S104, in step S106, the presentation unit 103 presents the plurality of first topics and the first summaries corresponding to each of the first topics to the user. Specifically, the presentation unit 103 transmits the plurality of first topics acquired in step S104 and the first summaries corresponding to each of the first topics to the user device 30 via the communication unit 120. In the user device 30, when the control unit 310 receives the plurality of first topics and the first summaries corresponding to each of the first topics via the communication unit 330, it causes the display unit 350 to display them. For example, the topic list shown in FIG. 4 and the summary list shown in FIG. 5 are displayed on the display unit 350. This allows the user to check the plurality of first topics as well as the first summaries corresponding to each of the first topics.

[0040] When the user wishes to modify a displayed primary topic, the user can modify the displayed primary topic by operating the operation unit 340. For example, on the topic list screen shown in Fig. 4, each primary topic is displayed in an editable manner, and the user can add or modify character strings of the primary topic on the topic list screen.

[0041] Figure 6 shows an example of a list of topics that have been modified according to user operations. In the example in Figure 6, for example, "Suzuki-Miyaura coupling" has been modified to "Suzuki-Miyaura coupling." In addition, "4. High functional group tolerance," "5. Reaction possible even in water," and "6. Application to chemical biology" shown in Figure 4 have been consolidated into "4. Reaction scope."

[0042] Furthermore, subtopics may be added as additional information to a topic. In the example in Figure 6, subtopics are shown in parentheses. For example, if you want to specifically mention "Changes in the valence of palladium catalysts" in the summary corresponding to the topic "Reaction mechanism of Suzuki-Miyaura coupling", you can add "Changes in the valence of palladium catalysts" as a subtopic, as shown in Figure 6.

[0043] Furthermore, the user can also provide additional information to the text data to be summarized. For example, additional information such as "The Suzuki-Miyaura coupling, discovered by Japanese chemists Professor Akira Suzuki and Professor Norio Miyaura, is introduced on YouTube (registered trademark)" can be provided. The additional information is information that is added to the entire text data. Examples of the additional information include the contents indicated by the text data, speaker information in the voice data that is the source of the text data, and the time of generation of the voice data corresponding to the text data.

[0044] As described above, the user can modify (including addition and deletion) topics, add subtopics, and add additional information as necessary. Furthermore, as described later, it is possible to repeatedly edit topics in response to user operations, and the user can also modify added subtopics and additional information in response to this. In this way, the user appropriately modifies (including addition and deletion) each of the topics, subtopics, and additional information, and then operates the operation unit 340 to input an execution instruction. As a result, editing instructions including the addition of subtopics, modification instructions, addition of additional information, modification instructions, etc. are transmitted from the user device 30 to the summary generation device 10 in response to user operations.

[0045] Returning to FIG. 2 for the explanation, after the process of step S106, when the control unit 100 of the summary generation device 10 receives an editing instruction from the user device 30 (Y in step S108), the process proceeds to step S110. In step S110, the dialogue unit 102 generates input data to be input to the generation model 210 in accordance with the editing instruction. For example, when the editing instruction is an instruction to modify a topic, the dialogue unit 102 generates a topic modified in accordance with the editing instruction as input data. When the editing instruction is an instruction to add a subtopic, the dialogue unit 102 generates a topic and the subtopic to be added as input data. When the editing instruction is an instruction to add additional information, the dialogue unit 102 generates a topic and the additional information to be added as input data.

[0046] Fig. 6 is an explanatory diagram of topic modification and subtopic addition. For example, the 10 primary topics shown in Fig. 4 are modified to 9 topics shown in Fig. 6. Hereinafter, the topic after modification of the primary topic is referred to as a secondary topic. In this manner, the number of topics may be changed. Furthermore, in the example shown in Fig. 6, subtopics have been added to topics 1, 6, and 9.

[0047] Next, in step S112, when the editing instruction includes a correction of a typo, the typo registration unit 105 registers the typo and the correct spelling in the substitution table in association with each other. As shown in Fig. 6, when "Suzuki Miyaura" is corrected to "Suzuki·Miyaura", "Suzuki Miyaura" and "Suzuki·Miyaura" are registered in the substitution table as the typo and the correct spelling, respectively. Fig. 7 is an example of the substitution table 400. In the substitution table 400, the typo and the correct spelling are stored in association with each other.

[0048] 2, after the processing of step S112, in step S114, the dialogue unit 102 transmits the text data, the input data, and a prompt requesting summary generation to the generative model server device 20 via the communication unit 120. Then, the dialogue unit 102 acquires summaries corresponding to each secondary topic from the generative model server device 20 via the communication unit 120. Note that, hereinafter, the topics included in the input data transmitted in step S114 are referred to as secondary topics, and the summaries obtained corresponding to the secondary topics are referred to as secondary summaries.

[0049] Next, in step S116, the presentation unit 103 presents the user with a plurality of secondary topics and secondary summaries corresponding to each secondary topic. This process is similar to the process in step S106. This allows the user to check the revised topics (secondary topics), subtopics, and summaries (secondary summaries) revised according to the additional information from the display contents of the display unit 350.

[0050] At this time, the display unit 350 displays the secondary topic as shown in Fig. 6, for example. Furthermore, the display unit 350 displays the secondary summary corresponding to the secondary topic shown in Fig. 6, as shown in Fig. 8. As shown in Fig. 8, in the secondary summary, for example, "Suzuki-Miyaura coupling" has been corrected to "Suzuki-Miyaura coupling". Also, details about the first paper, such as "Tetrahedron Letters", have been added.

[0051] After that, control unit 100 advances the process to step S108. The user can again modify the topic if desired. In this manner, the processes of steps S110 to S116 can be repeated. If steps S110 to S116 are subsequently repeated, a third-order topic and a third-order summary are generated, followed by a fourth-order topic and a fourth-order summary, and so on, resulting in topics and summaries with sequentially increasing degrees.

[0052] As another example, if the topic and summary are modified, the speech recognition may be performed again.

[0053] Furthermore, in step S108, if the user does not wish to edit, i.e., if an editing instruction is not received (N in step S108), the control unit 100 advances the process to step S118. In step S118, the dialogue unit 102 sets the finally obtained topics and summaries (e.g., secondary topics and secondary summaries) as final topics and final summaries. Then, the dialogue unit 102 transmits a prompt requesting generation of a document structure to the generative model server device 20 together with the final summaries, thereby acquiring relationship information between topics corresponding to the document structure. Next, in step S120, the structure diagram acquisition unit 106 acquires a structure diagram showing the document structure of the text data based on the relationship information.

[0054] FIG. 9 is a diagram showing an example of relationship information. In this way, the relationship information may be data expressed in Markdown format. Furthermore, FIG. 10 is an example of a structure diagram. In this way, the structure diagram may be a Mermaid diagram. In step S122, the presenting unit 103 presents this structure diagram to the user. Specifically, the presenting unit 103 transmits the structure diagram to the user device 30 and causes the display unit 350 to display it. This allows the user to visually grasp the relationship between each topic.

[0055] As another example, the dialogue unit 102 may obtain relationship information based on the text and the determined topic by sending them to the generative model server device 20 along with a prompt requesting the generation of a determined summary and document structure.

[0056] Furthermore, in step S124, translation unit 107 translates the corrected topic and summary into another language as appropriate in accordance with instructions from the user's operation, and presents the translated topic and summary to the user.

[0057] As described above, the summary generation system 1 of the present embodiment can generate topics from text data, edit the topics in response to user operations, and obtain summaries for each topic. This makes it possible to obtain summaries suited to the user.

[0058] In addition, when topics are extracted from text data and a summary is generated, it is not possible to obtain a summary that is suitable for the user or that contains necessary and sufficient information for the target text data. In addition, the summary may contain spelling variations or summary errors. In contrast, as described above, the summary generation system 1 of this embodiment can obtain a summary that is suitable for the user. Furthermore, a summary using appropriate technical terms can be generated even for text data that indicates technical content.

[0059] Furthermore, since a summary that is appropriate for the user and that contains necessary and sufficient information can be obtained, this summary can be used to obtain a highly accurate structure diagram. This summary can also be translated into other languages, making it easy to share information between users of different languages.

[0060] The above embodiment is one example for carrying out the present invention, and various other embodiments can be adopted. For example, various modifications and changes are possible within the scope of the gist of the present invention described in the claims, such as applying one modified example to another modified example. For example, some of the configurations of the above embodiment may be omitted, and the order of processing may be changed or omitted.

[0061] Such a first modified example will be described. The summary generation device of this embodiment is an example of an information processing system that generates a summary of text data, but as another example, the information processing system may be realized by a plurality of devices. That is, some functions of the summary generation device may be realized by a first device, and other functions of the summary processing device may be realized by a second device. Furthermore, the functions of the summary generation device may be realized by a user device 30. In this case, the user device 30 acquires the summary by directly interacting with the generative model server device 20.

[0062] As a second modified example, the contents indicated by the text data are not limited to those in the embodiment, and may be minutes data of a meeting attended by multiple people, data indicating the contents of a conversation, etc. Furthermore, the text data is not limited to data converted from voice data, and may be text data obtained by OCR (Optical Character Recognition) from characters printed on a recording medium such as paper, for example. In this case, the summary generating device 10 may be equipped with an OCR. Furthermore, as another example, the text data may be text data obtained based on a generation model from data including characters and images. Furthermore, as another example, the text data may be generated as text data, and may be, for example, a novel, a story, an essay, an impression or opinion on a certain subject, etc. In this way, the text data may be document data indicating some content.

[0063] Furthermore, the text data may correspond to a presentation. In this case, the dialogue unit 102 can obtain a summary of the presentation through dialogue with the generative model 210. In addition, the dialogue unit 102 can obtain advice on points that should be emphasized in the presentation, points that should be revised, etc. This can contribute to improving the quality of the presentation.

[0064] As a third modified example, the substitution table may be generated and managed for each user. In this case, in the voice recognition process for voice data provided by the same user, the substitution table corresponding to the user is referred to. This can improve the accuracy of voice recognition. As another example, the substitution table may be generated and managed for each attribute of the user, such as the department or company to which the user belongs. In this case, in the voice recognition process provided by a user with the same attribute, the correspondence table corresponding to the attribute is referred to. This is not limited to the voice recognition process, but is also the same for OCR.

[0065] As a fourth modified example, the dialogue unit 102 of the summary generation device 10 may acquire topics and summaries by using multiple different generative models. In this case, the generative model server device 20 may be provided with multiple generative models, and the summary generation device 10 may use these. As another example, the generative model server device 20 may be capable of communicating with multiple generative model server devices, and may use the generative models provided in each of these multiple generative model server devices.

[0066] As a fifth variant, the summary generation device 10 may first obtain a first number of topics, and then use the first number of topics to obtain a second number of topics that is less than the first number.

[0067] As a sixth modified example, the summary generation device 10 may generate primary topics using a generation model, and may also generate one or more subtopics corresponding to each primary topic, and present them to the user together with the primary topic. In this case, not only the primary topic but also the subtopics may be appropriately modified in response to user operations. This makes it possible to generate a summary that takes the subtopics into consideration. Furthermore, the generated subtopics may be added, modified, or deleted in response to user operations.

[0068] As a seventh modified example, the presentation unit 103 may present the relationship information shown in FIG. 9 or the structure diagram shown in FIG. 10 to the user. Then, the control unit 100 may accept an instruction to modify the relationship information or the structure diagram in response to a user operation. In this case, a Mermaid diagram reflecting the modification instruction can be generated. Furthermore, in this case, the dialogue unit 102 creates the modified relationship information and the topic as input data. Then, the dialogue unit 102 transmits a prompt for generating a summary together with the text data and the input data to the generative model server device 20, thereby obtaining a summary reflecting the modification instruction.

[0069] According to the information processing system, program, and information processing method of the present embodiment configured as described above, the information processing system includes a dialogue unit 102 that acquires multiple topics by inputting text data to a generative model 210, a presentation unit 103 that presents the topics to a user, and an editing instruction acquisition unit 104 that acquires editing instructions for the topics obtained in response to user operations on the topics. The dialogue unit 102 generates input data to be input to the generative model 210 in accordance with the editing instructions, and inputs the input data to the generative model 210 to acquire a summary corresponding to the pick related to the editing instructions. This makes it possible to acquire a summary suitable for the user from the text data.

[0070] In the information processing system, program, and information processing method of the present embodiment, the dialogue unit 102 may further acquire subtopics by inputting an instruction to generate subtopics for a topic to the generative model, and the dialogue unit 102 may acquire the summary by inputting the text data, the topic, and the subtopic to the generative model. This makes it possible to acquire a summary in which the subtopics are taken into consideration.

[0071] In the information processing system, the program, and the information processing method of the present embodiment, the dialogue unit 102 may obtain a summary by inputting text data, a topic, and a subtopic obtained in response to a user operation into a generative model. This makes it possible to obtain a summary suitable for the user.

[0072] In the information processing system, program, and information processing method of the present embodiment, the editing instruction may include an instruction to modify a subtopic of a topic, and the dialogue unit 102 may obtain a summary by inputting the text data, the topic, and the modified subtopic into a generative model. This makes it possible to obtain a summary that takes the subtopic into consideration.

[0073] In the information processing system, program, and information processing method of the present embodiment, the editing instruction may include an instruction to add additional information to the text data, and the dialogue unit 102 may obtain a summary by inputting the text data, the topic, and the added additional information into a generative model. This makes it possible to obtain a summary that takes the additional information into consideration.

[0074] In the information processing system, the program, and the information processing method of the present embodiment, the editing instruction may include an instruction to modify the additional information, and the dialogue unit 102 may generate a summary by inputting the text data, the topic, and the modified additional information into a pre-generated model. This makes it possible to obtain a summary that is more suitable for the user.

[0075] In the information processing system, program, and information processing method of the present embodiment, the dialogue unit 102 may generate an overall summary for the text data by inputting the text data to a generative model, and obtain a revised summary corresponding to the topic by inputting the overall summary and a summary corresponding to the topic to the generative model. This makes it possible to obtain a summary that is more suitable for the user.

[0076] In the information processing system, the program, and the information processing method of the present embodiment, the dialogue unit 102 may obtain a structure diagram showing the document structure of the text data by inputting the summary into the generation model. This makes it possible to obtain a structure diagram of a document corresponding to the text data.

[0077] In the information processing system, the program, and the information processing method of the present embodiment, the information processing system may further include a translation unit 107 that translates the summary into a language different from that of the text data, thereby making it possible to obtain a summary in another language.

[0078] In the information processing system, the program, and the information processing method of the present embodiment, the information processing system may further include a recognition unit 101 that acquires text data by voice recognition or character recognition. This makes it possible to acquire a summary from the text data acquired by the voice recognition and the character recognition. [Explanation of symbols]

[0079] 1 Summary Generation System 10 Summary Generator 20 Generative model server device 30 User device 100 Control section 101 Recognition part 102 Dialogue Section 103 Presentation section 104 Editing Instruction Acquisition Unit 105 Error Registration Section 106 Structural Drawing Acquisition Section 107 Translation Department 110 Storage section 120 Communications Department 210 Generative Model 310 Control section 320 Storage section 330 Communications Department 340 Operation section 350 Display section

Claims

1. a dialogue unit that acquires multiple topics by inputting a prompt requesting generation of multiple topics to a generative model together with text data; a presentation unit that presents the topic to a user; an editing instruction acquisition unit that acquires editing instructions for the topic obtained in response to a user operation on the topic; Equipped with the dialogue unit generates input data to be input to the generative model in accordance with the editing instruction, and inputs a prompt requesting summary generation together with the input data and the text data to the generative model, thereby obtaining a summary corresponding to the topic related to the editing instruction; An information processing system, wherein the input data includes an edited topic in accordance with the editing instructions.

2. the dialogue unit further acquires the subtopic by inputting a generation instruction for a subtopic for the topic to the generation model; the dialogue unit obtains the summary by inputting the topic and the subtopic as input data, the input data, the text data, and a prompt requesting the generation of the summary into the generation model; The information processing system according to claim 1 .

3. the dialogue unit receives as input data the topic and a subtopic obtained in response to a user operation, and obtains the summary by inputting the input data, the text data, and a prompt requesting the generation of the summary to the generation model; The information processing system according to claim 1 .

4. The editing instructions include instructions for modifying the subtopic, the dialogue unit receives the topic and the modified subtopic as input data, and obtains the summary by inputting the input data, the text data, and a prompt requesting the generation of the summary to the generation model; 4. The information processing system according to claim 2 or 3.

5. the editing instruction includes an instruction to add additional information to the text data, the dialogue unit receives the topic and the added additional information as input data, and obtains the summary by inputting the input data, the text data, and a prompt requesting the generation of the summary to the generation model; The information processing system according to claim 1 .

6. the editing instruction includes an instruction to modify the additional information, the dialogue unit receives the topic and the modified additional information as input data, and inputs the input data, the text data, and a prompt requesting generation of the summary to the generation model, thereby obtaining the summary.

6. The information processing system according to claim 5.

7. 2. The information processing system of claim 1, wherein the dialogue unit generates an overall summary for the text data by inputting a prompt requesting summary generation together with the text data to a generative model, and obtains a revised summary corresponding to the topic by inputting the input data, the text data, the overall summary, a summary corresponding to the topic, and the prompt requesting summary generation to a generative model.

8. The information processing system according to claim 1 , wherein the dialogue unit obtains a structure diagram showing a document structure of the text data by inputting, to a generative model, a prompt requesting generation of a document structure together with the summary.

9. the editing instruction acquisition unit acquires a correction instruction for the document structure, the dialogue unit generates input data to be input to a generative model in accordance with the correction instruction, and inputs a prompt requesting generation of a document structure to the generative model together with the input data and the text data, thereby obtaining a summary corresponding to the topic; The input data includes a document structure corrected in accordance with the correction instruction and the topic. The information processing system according to claim 8.

10. The information processing system according to claim 1 , further comprising a translation unit that translates the summary into a language different from that of the text data.

11. A recognition unit that acquires the text data by voice recognition or character recognition is further provided. The information processing system according to claim 1 .

12. A program for causing a computer to function as a dialogue unit, a presentation unit, and an editing instruction acquisition unit, the dialogue unit inputs a prompt requesting generation of a plurality of topics together with the text data into a generative model, thereby acquiring a plurality of topics; The presentation unit presents the topic to a user; the editing instruction acquisition unit acquires an editing instruction for the topic obtained in response to a user operation on the topic; the dialogue unit generates input data to be input to the generative model in accordance with the editing instruction, and inputs a prompt requesting summary generation together with the input data and the text data to the generative model, thereby obtaining a summary corresponding to the topic related to the editing instruction; The input data is a program including an edited topic in accordance with the editing instructions.

13. An information processing method performed by a computer having a control unit, a step of acquiring a plurality of topics by inputting a prompt requesting generation of a plurality of topics to a generative model together with the text data by the control unit; the control unit presenting the topic to a user; a pre-control unit acquiring an editing instruction for the topic obtained in response to a user operation on the topic; the control unit generates input data to be input to the generative model in accordance with the editing instruction, and inputs the input data and the text data together with a prompt requesting summary generation to the generative model, thereby obtaining a summary corresponding to the topic related to the editing instruction; Including, The input data includes an edited topic in accordance with the editing instruction. Information processing methods.

Citation Information

Patent Citations

  • Computer system and sentence data management method

    JP2019164504A

  • Document summarization device, document summarization system, document summarization method, and program

    JP2020181387A

  • Information processing device and information processing program

    JP7471026B1

  • Information processor, information processing method and information processing program

    JP2022165786A

  • JPP7471026B

Cited By

  • system

    JP2026155513A