Information processing device, program, and information processing system

Abstract

This information processing device comprises: a summary sentence data generation unit that generates a plurality of summary sentence data from processing target sentence data in accordance with processing instruction data; an extraction unit that extracts key information included in the plurality of summary sentence data for each category constituting the processing target sentence data; and a summary sentence data selection unit that selects summary sentence data to be output from the plurality of summary sentence data using the key information for each category.

Description

Information processing device, program, and information processing system

[0001] The present disclosure relates to an information processing device, a program, and an information processing system.

[0002] While large-scale language models (LLMs) generate high-quality summaries, the content can vary from one summary to the next, potentially resulting in the failure to extract relevant information. Large-scale language models also tend to generate information not contained in the input document, known as hallucination. A two-stage method has been proposed that first extracts text from the input document to be used in the summary and then synthesizes it to generate a highly explainable summary (see, for example, Non-Patent Document 1).

[0003] Masafumi Enomoto et al., "Analysis of hallucination in two-stage summarization using large-scale language models," Proceedings of the 30th Annual Conference of the Association for Natural Language Processing (March 2024), pp.2641-2646

[0004] However, in two-stage summarization using large-scale language models, if appropriate information is not extracted in the first stage of summarization, appropriate information will not be extracted in the final result. Also, because text extracted from different parts of the input document is synthesized, the language model that performs the synthesis may not be able to grasp the original context, resulting in the generation of incorrect information.

[0005] The present disclosure aims to provide an information processing device, program, and information processing system that can prevent information deficiencies and prevent the generation of erroneous summary sentence data when generating summary sentence data from sentence data to be processed.

[0006] The present disclosure has the following configuration.

[0007] [1] An information processing device comprising: a summary sentence data generation unit that generates a plurality of summary sentence data from sentence data to be processed in accordance with processing instruction data; an extraction unit that extracts key information included in the plurality of summary sentence data for each category constituting the sentence data to be processed; and a summary sentence data selection unit that selects the summary sentence data to be output from the plurality of summary sentence data using the key information for each category.

[0008] [2] The information processing device according to [1], wherein the summary sentence data generation unit generates a plurality of summary sentence data for each category from the processing target sentence data in accordance with the processing instruction data.

[0009] [3] The information processing device according to [1] or [2], further comprising an importance calculation unit that calculates the importance of the key information, and wherein the summary sentence data selection unit uses the importance to select the summary sentence data to be output from the plurality of summary sentence data.

[0010] [4] The information processing device according to [3], wherein the summary sentence data selection unit: calculates the importance of each of the plurality of summary sentence data using the importance of the key information; selects a first summary sentence data having the highest importance from the plurality of summary sentence data; calculates the importance of each of the remaining summary sentence data excluding the first summary sentence data using the importance of the remaining key information excluding the key information included in the first summary sentence data; and selects a second summary sentence data having the highest importance from the remaining summary sentence data.

[0011] [5] The information processing device according to any one of [1] to [4], wherein the summary sentence data generation unit generates a plurality of summary sentence data from the same processing target sentence data in accordance with different processing instruction data.

[0012] [6] The information processing device according to any one of [1] to [4], wherein the summary sentence data generation unit generates a plurality of summary sentence data from different processing target sentence data in accordance with the same processing instruction data.

[0013] [7] The information processing device according to any one of [1] to [4], wherein the summary sentence data generation unit generates a plurality of summary sentence data from different processing target sentence data in accordance with different processing instruction data.

[0014] [8] The information processing device according to any one of [1] to [7], further comprising a detection unit that detects spelling variations of the key information, wherein the extraction unit treats the key information of the detected spelling variations as the same key information.

[0015] [9] The information processing device according to [3] or [4], wherein the importance calculation unit calculates the importance of the key information using one or more of the number of occurrences of the key information, TF-IDF (Term Frequency - Inverse Document Frequency), or co-occurrence relationships between words.

[0016]

[10] A program causing an information processing device to execute: a summary sentence data generation step of generating a plurality of summary sentence data from sentence data to be processed in accordance with processing instruction data; an extraction step of extracting key information included in the plurality of summary sentence data for each category constituting the sentence data to be processed; and a summary sentence data selection step of selecting the summary sentence data to be output from the plurality of summary sentence data using the key information for each category.

[0017]

[11] An information processing system comprising: an input receiving unit that receives input of processing instruction data and processing target sentence data; a summary sentence data generation unit that generates a plurality of summary sentence data from the processing target sentence data in accordance with the processing instruction data; an extraction unit that extracts key information included in the plurality of summary sentence data for each category constituting the processing target sentence data; a summary sentence data selection unit that selects the summary sentence data to be output from the plurality of summary sentence data using the key information for each category; and a display control unit that displays the selected summary sentence data on a display device.

[0018] According to the present disclosure, when generating summary sentence data from sentence data to be processed, it is possible to prevent information deficiencies and to prevent generation of erroneous summary sentence data.

[0019] FIG. 1 is a configuration diagram of an example of an information processing system according to the present embodiment. FIG. 1 is a hardware configuration diagram of an example of a computer according to the present embodiment. FIG. 2 is an explanatory diagram of an example of a process for generating summary sentence data from sentence data to be processed in accordance with a prompt. FIG. 3 is an explanatory diagram of an example of a process for an information processing system according to the present embodiment for generating summary sentence data from sentence data to be processed. FIG. 4 is a functional configuration diagram of an example of an information processing system according to the present embodiment. FIG. 5 is a flowchart of an example of a process for an information processing system according to the present embodiment. FIG. 6 is a flowchart of an example of a process for an information processing system according to the present embodiment. FIG. 7 is a flowchart of an example of a process for step S16. FIG. 8 is a flowchart of an example of a process for step S16. FIG. 9 is an explanatory diagram of an example of a process in the flowchart of FIG. 9. FIG. 10 is an explanatory diagram of an example of a process for step S12. FIG. 11 is an explanatory diagram of an example of a process for step S12. FIG. 12 is an explanatory diagram of an example of a process for step S12. FIG. 13 is a specific example of a process for the information processing system according to the present embodiment.

[0020] Next, an embodiment of the present invention will be described in detail, but the present invention is not limited to the following embodiment.

[0021] <System Configuration> Fig. 1 is a configuration diagram of an example of an information processing system 1 according to this embodiment. The information processing system 1 of Fig. 1 provides a summarization function using a large-scale language model (hereinafter referred to as LLM). The LLM generates summary sentence data from sentence data to be processed in accordance with a prompt. The prompt is an example of processing instruction data, and is data such as a character string that instructs the LLM on the content of the summary sentence data to be generated. Furthermore, the information processing system 1 selects summary sentence data to be output from the multiple summary sentence data generated by the LLM, as described below. In this way, the information processing system 1 performs a process of selecting and outputting summary sentence data selected from the multiple summary sentence data generated by the summarization function using the LLM.

[0022] The information processing system 1 includes an information processing device 10 and a user terminal 12, which are connected to each other via a network 18 so as to enable data communication. The network 18 is, for example, a local area network (LAN) or the Internet. The information processing device 10 can be realized by a workstation or a personal computer (PC). The user terminal 12 can be realized by an information processing terminal operated by a user, such as a PC, tablet terminal, or smartphone.

[0023] A user can operate the user terminal 12 and use the summarization function provided by the information processing device 10. The user operates the user terminal 12 and inputs processing target sentence data and a prompt to the information processing device 10. The user terminal 12 transmits the input processing target sentence data and prompt to the information processing device 10, thereby causing the information processing device 10 to output summary sentence data of the processing target sentence data, as will be described later.

[0024] The user terminal 12 receives the summary sentence data of the sentence data to be processed output from the information processing device 10 and displays it on a display device, etc. In this way, the user can check the summary sentence data of the sentence data to be processed displayed on a display device, etc.

[0025] The information processing device 10 also receives processing target sentence data and a prompt from the user terminal 12. The information processing device 10 performs processing to generate multiple pieces of summary sentence data from the received processing target sentence data in accordance with the received prompt. The information processing device 10 selects summary sentence data to be output from the generated multiple pieces of summary sentence data as described below, and transmits the summary sentence data to the user terminal 12 to display on a display device or the like.

[0026] The functions of the information processing device 10 may be provided as a cloud service. The information processing device 10 may use an LLM that it has, or an LLM that another device has. The LLM is a natural language processing model trained using a large amount of text data (a trained large-scale language model). A fine-tuned trained model can be used as the LLM. A GPT (Generative Pre-trained Transformer) or the like can be used as the LLM.

[0027] The configuration of the information processing system 1 shown in Fig. 1 is an example. The configuration of the information processing system 1 varies depending on the application and purpose. For example, the functions of the information processing device 10 and the user terminal 12 may be integrated and realized by a single computer. Furthermore, the information processing device 10 may be realized by multiple computers.

[0028] <Hardware Configuration> For example, the information processing apparatus 10 and the user terminal 12 in FIG. 1 can be realized by a computer 500 having the hardware configuration shown in FIG.

[0029] 2 is a diagram showing an example of the hardware configuration of a computer 500 according to this embodiment. The computer 500 includes, for example, an input device 501, an output device 502, an external I / F 503, a RAM (Random Access Memory) 504, a ROM (Read Only Memory) 505, a CPU (Central Processing Unit) 506, a communication I / F 507, and an auxiliary storage device 508, all of which are interconnected by a bus B. Note that the input device 501 and the output device 502 may be connected to the computer 500 via the external I / F 503 for use.

[0030] The input device 501 is a device that accepts user operations, and is, for example, a touch panel, operation keys, buttons, a keyboard, or a mouse. The output device 502 has a device that displays a screen and a device that outputs sound. The device that displays a screen is, for example, a display (display device) such as a liquid crystal display. The device that outputs sound is, for example, a speaker. The communication I / F 507 is an interface that allows the computer 500 to perform data communication.

[0031] The auxiliary storage device 508 is an example of a non-volatile storage device that stores programs and data. The auxiliary storage device 508 is, for example, a hard disk drive (HDD) or a solid state drive (SSD). The programs are, for example, an operating system (OS), which is basic software, and applications that provide various functions on the OS.

[0032] The external I / F 503 is an interface with an external device. The external device is a recording medium 503a or the like. The computer 500 can read programs and data from the recording medium 503a via the external I / F 503. The recording medium 503a is, for example, a flexible disk, a CD, a DVD, an SD memory card, or a USB memory.

[0033] The ROM 505 stores the BIOS, OS settings, network settings, etc., which are executed when the computer 500 starts up. The RAM 504 is a working memory that temporarily stores programs and data. The CPU 506 can realize various functions described below by reading programs and data from the ROM 505 or the auxiliary storage device 508 into the RAM 504 and executing the processes.

[0034] <Processing Overview> Fig. 3 is an explanatory diagram of an example of processing for generating summary sentence data 1004-1 to 1004-3 from sentence data to be processed 1000 in accordance with a prompt 1002. In Fig. 3, the sentence data to be processed 1000 shows an example of a patent document.

[0035] Summary sentence data 1004-1 is an example of summary sentence data generated by the LLM from the sentence data to be processed the first time in accordance with the prompt 1002. Summary sentence data 1004-2 is an example of summary sentence data generated by the LLM from the sentence data to be processed the second time in accordance with the prompt 1002. Summary sentence data 1004-3 is an example of summary sentence data generated by the LLM from the sentence data to be processed the third time in accordance with the prompt 1002.

[0036] The summary sentence data 1004-1 to 1004-3 show the output results multiple times using the same processing target sentence data 1000 and prompt 1002. The summary sentence data 1004-1 to 1004-3 generated by the LLM produce different output results even when using the same processing target sentence data 1000 and prompt 1002. Therefore, the summary sentence data 1004-1 to 1004-3 may lack information that provides an overview of the processing target sentence data 1000 as a whole.

[0037] Therefore, the information processing system 1 according to this embodiment generates summary sentence data 1008 from the sentence data 1000 to be processed, as shown in Fig. 4. Fig. 4 is an explanatory diagram of an example of a process in which the information processing system 1 according to this embodiment generates summary sentence data 1008 from the sentence data 1000 to be processed. Fig. 4 shows an example in which the sentence data 1000 to be processed is a patent document.

[0038] Summary sentence data 1004-1 is an example of summary sentence data generated by the LLM the first time from the processing target sentence data in accordance with the prompt 1002. The summary sentence data 1004-1 includes summary sentences of the problem, solution, and effect, which are examples of categories that make up the processing target sentence data 1000.

[0039] Summary sentence data 1004-2 is an example of summary sentence data generated by the LLM from the processing target sentence data for the second time in accordance with the prompt 1002. The summary sentence data 1004-2 includes summary sentences of the problem, the solution, and the effect, which are examples of categories that make up the processing target sentence data 1000.

[0040] Summary sentence data 1004-3 is an example of summary sentence data generated by the LLM from the processing target sentence data for the third time in accordance with the prompt 1002. The summary sentence data 1004-3 includes summary sentences of the problem, solution, and effect, which are examples of categories that make up the processing target sentence data 1000.

[0041] Summary sentence data 1006-1 is an example of summary sentence data that collects, from the summary sentence data 1004-1 to 1004-3, summary sentences on problems, which are an example of a category that makes up the sentence data to be processed 1000. Summary sentence data 1006-2 is an example of summary sentence data that collects, from the summary sentence data 1004-1 to 1004-3, summary sentences on solutions, which are an example of a category that makes up the sentence data to be processed 1000. Summary sentence data 1006-3 is an example of summary sentence data that collects, from the summary sentence data 1004-1 to 1004-3, summary sentences on effects, which is an example of a category that makes up the sentence data to be processed 1000.

[0042] In the information processing system 1 according to this embodiment, key information extraction processing using LLM is performed from the summary sentence data 1006-1. The key information extraction processing using LLM extracts one or more pieces of key information (key information list) of tasks, which are examples of categories that make up the sentence data 1000 to be processed. The key information may be, for example, a keyword or a key phrase. Furthermore, the key information may be, for example, a keyword and a key phrase.

[0043] In the information processing system 1 according to this embodiment, key information is extracted from the summary sentence data 1006-2 using the LLM. The key information is extracted from the summary sentence data 1006-2 using the LLM. The key information is extracted from the solution means (key information list), which is an example of a category that makes up the sentence data 1000 to be processed. In addition, in the information processing system 1 according to this embodiment, key information is extracted from the summary sentence data 1006-3 using the LLM. The key information is extracted from the effect (key information list), which is an example of a category that makes up the sentence data 1000 to be processed.

[0044] In the information processing system 1 according to this embodiment, key information of the assignment extracted by LLM from the summary sentence data 1006-1 is used to select the summary sentence data of the assignment to be output from the summary sentence data 1006-1, and the summary sentence data is added as the summary sentence of the assignment in the summary sentence data 1008 to be output.

[0045] For example, the information processing system 1 according to this embodiment selects summary data (text) of the assignment that broadly covers the key information of the assignment extracted by the LLM from the summary data 1006-1.

[0046] In the information processing system 1 according to this embodiment, the key information of the solution extracted by the LLM from the summary sentence data 1006-2 is used to select the summary sentence data of the solution to be output from the summary sentence data 1006-2, and the summary sentence data is added as the summary sentence of the solution to the summary sentence data 1008 to be output.

[0047] For example, the information processing system 1 according to this embodiment selects summary sentence data (sentences) of the solution means that broadly covers the key information of the solution means extracted by the LLM from the summary sentence data 1006-2.

[0048] In the information processing system 1 according to this embodiment, key information on effects extracted by LLM from the summary sentence data 1006-3 is used to select summary sentence data of effects to be output from the summary sentence data 1006-3, and the selected summary sentence data is added as a summary sentence of effects in the summary sentence data 1008 to be output.

[0049] For example, the information processing system 1 according to this embodiment selects summary data (sentences) of effects that broadly cover the key information of effects extracted by LLM from the summary data 1006-3.

[0050] <Functional Configuration> The functional configuration of the information processing system 1 according to this embodiment will be described.

[0051] Fig. 5 is a functional configuration diagram of an example of an information processing system 1 according to this embodiment. Note that the configuration diagram of Fig. 10 omits portions unnecessary for explaining this embodiment as appropriate. The information processing device 10 in Fig. 5 includes an input accepting unit 30, a summary sentence data generating unit 32, an extracting unit 34, a summary sentence data selecting unit 36, a display control unit 38, an importance calculating unit 40, a detecting unit 42, a communicating unit 44, and an LLM storage unit 50. The user terminal 12 in Fig. 5 includes an operation accepting unit 60, a display unit 62, and a communicating unit 64.

[0052] The operation reception unit 60 receives various operations from the user. The display unit 62 displays a screen to be viewed by the user. For example, the display unit 62 displays a screen for receiving a request from the user to cause the information processing device 10 to execute a process, and a screen for presenting the results of the process performed by the information processing device 10 to the user.

[0053] The communication unit 64 performs data communication with the information processing device 10. For example, the communication unit 64 transmits a request to the information processing device 10 and receives a response from the information processing device 10.

[0054] The input receiving unit 30 of the information processing device 10 receives input from the user terminal 12 operated by the user. For example, the input receiving unit 30 receives input of processing target sentence data 1000 and a prompt 1002.

[0055] The summary sentence data generation unit 32 uses the LLM to generate summary sentence data 1004-1 to 1004-3 from the sentence data 1000 to be processed in accordance with the prompt 1002. Alternatively, the summary sentence data generation unit 32 may use the LLM to generate summary sentence data 1004-1 to 1004-3 for each category (problem, solution, and effect) from the sentence data 1000 to be processed in accordance with the prompt 1002. The summary sentence data generation unit 32 generates summary sentence data 1006-1 to 1006-3 by collecting summary sentences for each category that make up the sentence data 1000 from the summary sentence data 1004-1 to 1004-3.

[0056] The extraction unit 34 extracts key information contained in the summary sentence data 1006-1 to 1006-3 for each category constituting the processing target sentence data 1000. The key information is extracted using LLM. Note that the key information may also be extracted using existing techniques for extracting key words.

[0057] The summary sentence data selection unit 36 ​​uses the key information extracted for each category to select summary sentence data of the category to be output from the summary sentence data 1006-1 to 1006-3, and adds it as a summary sentence to the summary sentence data to be output 1008. Details of the processing by the summary sentence data selection unit 36 ​​will be described later. The display control unit 38 performs display control to display the summary sentence data 1008 to be output on the display unit 62 of the user terminal 12.

[0058] The importance calculation unit 40 calculates the importance of the key information. The importance calculation unit 40 may calculate the importance of the key information extracted from the summary sentence data 1006-1 to 1006-3 using the number of occurrences of the key information.

[0059] The importance calculation unit 40 may also use TF-IDF (Term Frequency - Inverse Document Frequency) to calculate the importance of key information extracted from the summary sentence data 1006-1 to 1006-3. TF-IDF is an effective method for extracting important words that characterize a certain sentence. TF-IDF is calculated by multiplying tf and idf.

[0060] tf is the frequency of occurrence of a certain word in a certain sentence, and is calculated using the following formula (1). The higher the frequency of occurrence, the higher the importance.

[0061] tf = number of times word i appears in sentence d / sum of number of times all words appear in sentence d... (1) idf indicates how many sentences a word is used in and is calculated using the following formula (2). The lower the frequency of appearance, the higher the importance.

[0062] idf = log (total number of sentences / number of sentences containing a certain word i)...(2) The summary sentence data selection unit 36 ​​may use the importance of the key information calculated by the importance calculation unit 40 to select summary sentence data of a category to be output from the summary sentence data 1006-1 to 1006-3.

[0063] The importance calculation unit 40 may also calculate the importance of key information extracted from the summary sentence data 1006-1 to 1006-3 using co-occurrence relationships between words. A co-occurrence relationship between words refers to the simultaneous appearance of two words in a sentence, and indicates the strength of the connection between the words. More specifically, a co-occurrence matrix of words in the summary sentence data may be created, and the importance of key information may be calculated based on the co-occurrence matrix. Methods for calculating the importance of key information using co-occurrence relationships between words include TextRank and TopicRank. The key information and importance scores calculated using these methods may be used to select summary sentence data of a category to be output from the summary sentence data 1006-1 to 1006-3.

[0064] The detection unit 42 detects spelling variations of the key information. For example, the detection unit 42 detects spelling variations of the key information by determining synonyms using a synonym dictionary or embedding (embedded expressions). Note that the extraction unit 34 may absorb spelling variations of the key information by treating the spelling variations of the key information included in the summary sentence data 1006-1 to 1006-3 as the same key information.

[0065] The communication unit 44 performs data communication with the user terminal 12. For example, the communication unit 44 receives a request from the user terminal 12 and transmits a response to the user terminal 12. The LLM storage unit 50 stores the LLM.

[0066] <Processing> The information processing system 1 according to this embodiment executes the following processing: Fig. 6 is a flowchart illustrating an example of the processing of the information processing system 1 according to this embodiment.

[0067] In step S10, the input receiving unit 30 of the information processing device 10 receives input from the user of the sentence data to be processed 1000 and the prompt 1002. The sentence data to be processed 1000 and the prompt 1002 to be received from the user may be one or more.

[0068] In step S12, the summary sentence data generation unit 32 of the information processing device 10 performs summarization multiple times on the processing target sentence data 1000 input from the user in accordance with the prompt 1002 input from the user. For example, the summary sentence data generation unit 32 uses the LLM to generate summary sentence data 1004-1 to 1004-3 from the processing target sentence data 1000 in accordance with the prompt 1002. The summary sentence data 1004-1 to 1004-3 include summary sentences of the problem, the solution, and the effect, which are examples of categories constituting the processing target sentence data 1000.

[0069] The summary sentence data generating unit 32 also generates, for example, summary sentence data 1006-1 to 1006-3 by collecting the summary sentences for each category that make up the processing target sentence data 1000 from the summary sentence data 1004-1 to 1004-3.

[0070] In step S14, the extraction unit 34 extracts key information contained in the summary sentence data 1006-1 to 1006-3 for each summary sentence, such as the problem, solution, and effect, which are examples of categories that make up the sentence data to be processed 1000. Note that the extraction of key information is not limited to the extraction of key words; key information may also be extracted from multiple perspectives, such as technology, use, material, method, problem, or effect.

[0071] In step S16, the summary sentence data selection unit 36 ​​uses key information extracted for each category, such as the problem, solution, and effect, as described below, to select summary sentence data for each category to be output from the summary sentence data 1006-1 to 1006-3, and adds the selected summary sentence data as a summary sentence to the summary sentence data 1008 to be output.

[0072] In step S18, the display control unit 38 performs display control to display the summary sentence data 1008 to be output on the display unit 62 of the user terminal 12, thereby presenting the summary sentence data selected for each category in step S16 to the user.

[0073] The information processing system 1 according to this embodiment may add a process for absorbing variations in the notation of key information shown in step S15 of Fig. 7 in addition to the process shown in Fig. 6. Fig. 7 is a flowchart showing an example of the process of the information processing system 1 according to this embodiment. Note that the flowchart shown in Fig. 7 adds the process of step S15 to the flowchart of Fig. 6.

[0074] In step S15, the detection unit 42 detects spelling variations of the key information detected in step S14. The spelling variations of the key information detected in step S15 are absorbed by treating the spelling variations of the key information as the same key information. The processes other than step S15 are the same as those in the flowchart of Figure 6, so their explanation will be omitted.

[0075] The process of step S16 in Figures 6 and 7 is executed, for example, as shown in the flowchart of Figure 8. Figure 8 is a flowchart showing an example of the process of step S16.

[0076] In step S30, the importance calculation unit 40 calculates the importance of the key information extracted in step S14 of Fig. 6 or Fig. 7. For example, the importance calculation unit 40 calculates the importance of the key information extracted from the summary sentence data 1006-1 to 1006-3.

[0077] The importance calculation unit 40 calculates the importance of the key information extracted in step S14 using the TF-IDF of the key information. Alternatively, the importance calculation unit 40 may calculate the importance of the key information using the number of times the key information appears. When the number of times the key information is applied is used, the importance may be calculated so that, for example, the more the number of times the key information appears, the higher the importance, and the fewer the number of times the key information appears, the lower the importance.

[0078] In step S32, the importance calculation unit 40 calculates the importance of the summary sentence data generated in step S12 of Fig. 6 or Fig. 7. For example, the importance calculation unit 40 calculates the importance of each sentence included in the summary sentence data 1006-1 to 1006-3.

[0079] The importance calculation unit 40 calculates the importance of the summary sentence data generated in step S12 of Figure 6 or Figure 7 using the importance of each piece of key information calculated in step S30. The importance of summary sentence data is the sum of the importance of the key information included in the summary sentence data. For example, the importance calculation unit 40 calculates the importance of each sentence included in summary sentence data 1006-1 to 1006-3 (the sentences generated by the first to third summaries shown in Figure 4).

[0080] In step S34, the summary data selection unit 36 ​​selects summary data with a high degree of importance from the summary data generated in step S12 of FIG. 6 or FIG.

[0081] For example, the summary sentence data selection unit 36 ​​uses the importance of each sentence included in the summary sentence data 1006-1 to 1006-3 to select, for each of the summary sentence data 1006-1 to 1006-3, a sentence of summary sentence data with a high importance from the sentences included in the summary sentence data 1006-1 to 1006-3. The number of sentences of summary sentence data selected for each of the summary sentence data 1006-1 to 1006-3 may be one or more. For example, out of the sentences "[Assignment (first time)].... [Assignment (second time)].... [Assignment (third time)]... included in the summary sentence data 1006-1 in Figure 4, the summary sentence data selection unit 36 ​​selects "[Assignment (first time)]..." if the importance of "[Assignment (first time)]..." is higher than "[Assignment (second time)].... [Assignment (third time)]...".

[0082] The process of step S16 in Figures 6 and 7 may be executed as shown in, for example, the flowchart of Figure 9. Figure 9 is a flowchart showing an example of the process of step S16.

[0083] In step S50, the importance calculation unit 40 calculates the importance of the key information extracted in step S14 of Fig. 6 or Fig. 7. For example, the importance calculation unit 40 calculates the importance of the key information extracted from the summary sentence data 1006-1 to 1006-3.

[0084] The importance calculation unit 40 calculates the importance of the key information extracted in step S14 using the TF-IDF of the key information. Alternatively, the importance calculation unit 40 may calculate the importance of the key information using the number of times the key information appears.

[0085] In step S52, the importance calculation unit 40 calculates the importance of the summary sentence data generated in step S12 of Fig. 6 or Fig. 7. For example, the importance calculation unit 40 calculates the importance of each sentence included in the summary sentence data 1006-1 to 1006-3.

[0086] The importance calculation unit 40 calculates the importance of the summary data generated in step S12 of Fig. 6 or 7 using the importance of each piece of key information calculated in step S30. For example, the importance calculation unit 40 calculates the importance of each sentence included in the summary data 1006-1 to 1006-3.

[0087] In step S54, the summary data selection unit 36 ​​selects the first summary data with the highest importance from the summary data generated in step S12 of FIG. 6 or FIG.

[0088] For example, the summary sentence data selection unit 36 ​​uses the importance of each sentence contained in the summary sentence data 1006-1 to 1006-3 to select the sentence of the first summary sentence data with the highest importance from the sentences contained in the summary sentence data 1006-1 to 1006-3 for each of the summary sentence data 1006-1 to 1006-3.

[0089] In step S56, the summary sentence data selection unit 36 ​​calculates the importance of each sentence of the remaining summary sentence data, excluding the sentences of the first summary sentence data, using the importance of the remaining key information, excluding the key information contained in the first summary sentence data selected in step S54.

[0090] In step S58, the summary data selection unit 36 ​​selects the second summary data sentence with the highest importance from the remaining summary data sentences. The second summary data sentence has the least overlap with the first summary data sentence.

[0091] The processing of the flowchart in Fig. 9 will be further explained using Fig. 10. Fig. 10 is an explanatory diagram of an example of the processing of the flowchart in Fig. 9. Note that Fig. 10 shows an example in which the key information is a keyword. Fig. 10 also explains an example in which the importance of a keyword is a TF-IDF value.

[0092] For example, "Summary Sentence-1" in Fig. 10 is the sentence "[Assignment (1st time)] ..." included in the summary sentence data 1006-1 shown in Fig. 4. "Summary Sentence-2" in Fig. 10 is the sentence "[Assignment (2nd time)] ..." included in the summary sentence data 1006-1 shown in Fig. 4. "Summary Sentence-3" in Fig. 10 is the sentence "[Assignment (3rd time)] ..." included in the summary sentence data 1006-1 shown in Fig. 4.

[0093] In Fig. 10, KW-A to KW-F represent keywords A to F. The table shown in Fig. 10 is filled with check marks indicating the keywords contained in "Summary-1" to "Summary-3".

[0094] For example, the sentence "Summary-1" contains keywords A, B, and D. The sentence "Summary-2" contains keywords A, C, and E. The sentence "Summary-3" contains keywords D and F.

[0095] In step S50 of FIG. 9, the importance calculation unit 40 calculates the TF-IDF values ​​of the keywords A to F contained in "Summary-1" to "Summary-3" of FIG.

[0096] In step S52, the importance calculation unit 40 calculates the importance of "Summary Sentence-1" to "Summary Sentence-3" in Fig. 10 using the TF-IDF values ​​of the keywords A to F. The importance of "Summary Sentence-1" to "Summary Sentence-3" in Fig. 10 is the sum of the TF-IDF values ​​of the keywords A to F contained in "Summary Sentence-1" to "Summary Sentence-3" in Fig. 10, respectively.

[0097] In this example, the description will continue assuming that "Summary-1" has the highest importance. In step S54, the summary data selection unit 36 ​​selects "Summary-1" as the first summary data with the highest importance from "Summary-1" to "Summary-3" in FIG. 10.

[0098] In step S56, the summary data selection unit 36 ​​excludes keywords A, B, and D contained in "summary sentence-1" selected as the first summary data. The lower table shown in Fig. 10 shows "summary sentence-1" selected as the first summary data and the state after excluding keywords A, B, and D contained in "summary sentence-1" selected as the first summary data.

[0099] The importance calculation unit 40 calculates the importance of "Summary Sentence-2" to "Summary Sentence-3" in Figure 10 using the TF-IDF values ​​of the remaining keywords C, E, and F, excluding keywords A, B, and D, contained in "Summary Sentence-1" selected as the first summary data.

[0100] In this example, the description will continue assuming that "Summary-2" has the highest importance. In step S58, the summary data selection unit 36 ​​selects "Summary-2" as the second summary data with the highest importance from "Summary-2" to "Summary-3" in Figure 10.

[0101] "Summary-2," selected as the second summary data, contains different keywords from "Summary-1," selected as the first summary data. Therefore, by combining "Summary-1," selected as the first summary data, and "Summary-2," selected as the second summary data, it is possible to select sentences in the summary data that cover a wide range of keywords.

[0102] The process of step S12 in Figures 6 and 7 may be executed as shown in Figures 11 to 14. Figures 11 to 14 are explanatory diagrams showing an example of the process of step S12.

[0103] 11 shows an example of a process for generating multiple pieces of summary sentence data from the same sentence data to be processed in accordance with different prompts. The summary sentence data generation unit 32 in FIG. 11 receives as input two different prompts A and B and one piece of sentence data to be processed.

[0104] For example, prompt A might be "Please list three key points in 20 to 40 characters regarding the problem described in the [Problem] section of the patent document abstract." Prompt B might be "Please list three key points in 20 to 40 characters regarding the problem described in the [Problem] section of the patent document abstract and the [Problem to be Solved by the Invention] section of the specification."

[0105] The summary sentence data generating unit 32 executes summarization a plurality of times for the input target sentence data in accordance with different prompts A and B, thereby generating a plurality of summary sentence data for the target sentence data.

[0106] In FIG. 11, a plurality of summary sentence data can be generated from the sentence data to be processed in accordance with prompt A and a plurality of summary sentence data can be generated from the sentence data to be processed in accordance with prompt B.

[0107] Figure 12 shows an example of a process for generating multiple summary sentence data from different target sentence data in response to the same prompt. The summary sentence data generation unit 32 in Figure 12 receives as input one prompt and different target sentence data A and B. For example, in the case of the patent document in Figure 4, target sentence data A may be a sentence described in the "Problem" section of the abstract. In the case of the patent document in Figure 4, target sentence data B may be a sentence described in the "Problem" section of the abstract and the "Problem to be Solved by the Invention" section of the specification.

[0108] The summary sentence data generating unit 32 can generate multiple pieces of summary sentence data for the processing target sentence data A and B by executing summarization multiple times in accordance with the same prompt for the processing target sentence data A and B that have been received as input.

[0109] In FIG. 12, a plurality of summary sentence data can be generated from the sentence data A to be processed in accordance with the prompt, and a plurality of summary sentence data can be generated from the sentence data B to be processed in accordance with the prompt.

[0110] 13 shows an example of a process for generating a plurality of summary sentence data from different target sentence data in accordance with different prompts. The summary sentence data generation unit 32 in FIG. 13 receives as input two different prompts A and B and two different target sentence data A and B.

[0111] The summary sentence data generation unit 32 can generate multiple summary sentence data of the sentence data to be processed by executing summarization multiple times on the input sentence data A and B in accordance with prompts A and B. In Fig. 13, multiple summary sentence data can be generated: summary sentence data generated from the sentence data to be processed A in accordance with prompt A, summary sentence data generated from the sentence data to be processed B in accordance with prompt A, summary sentence data generated from the sentence data to be processed A in accordance with prompt B, and summary sentence data generated from the sentence data to be processed B in accordance with prompt B.

[0112] 14 shows an example of a process for generating a plurality of summary sentence data from different target sentence data in accordance with different prompts. The summary sentence data generation unit 32 in FIG. 14 receives as input different prompts A to C and different target sentence data A and B.

[0113] For example, prompt A might be "Please list the main points regarding the problem described in each of the two patent documents in 20 to 40 characters." Prompt B might be "Please list the similarities regarding the problem described in each of the two patent documents in 20 to 40 characters." Prompt C might be "Please list the differences regarding the problem described in each of the two patent documents in 20 to 40 characters."

[0114] The summary sentence data generating unit 32 executes summarization a plurality of times for the input target sentence data A and B in accordance with prompts A to C, thereby generating a plurality of summary sentence data of the target sentence data.

[0115] 14, for example, a plurality of summary sentence data can be generated from the processing target sentence data A or B in accordance with prompt A. The summary sentence data generated from the processing target sentence data A or B in accordance with prompt A is, for example, a sentence that expresses the main points of the processing target sentence data A or B.

[0116] 14, a plurality of summary sentence data can be generated from the target sentence data A and B in accordance with prompt B. The summary sentence data generated from the target sentence data A and B in accordance with prompt B is, for example, a sentence that shows the similarities between the target sentence data A or B.

[0117] 14, a plurality of summary sentence data can be generated from the target sentence data A and B in accordance with prompt C. The summary sentence data generated from the target sentence data A and B in accordance with prompt C is, for example, a sentence that indicates the differences between the target sentence data A and B.

[0118] 15 is a specific example of processing by the information processing system 1 according to this embodiment. As shown in Fig. 15, the information processing device 10 generates multiple summary sentence data of key points related to the "problem" by LLM from patent documents accepted as sentence data to be processed, in accordance with prompts.

[0119] The information processing device 10 extracts key information such as keywords and key phrases from the multiple generated summary sentence data, calculates the TF-IDF value of the extracted key information, and calculates the TF-IDF value for each sentence included in the multiple generated summary sentence data.

[0120] The information processing device 10 selects the sentence with the highest TF-IDF value from among the sentences included in the multiple pieces of summary data generated. In Figure 15, the sentence "Resolves variations in the reliability of organizational data due to measurement, observation, and analysis skills and subjectivity" included in the summary data generated by the second summarization is selected.

[0121] The information processing device 10 selects the sentence with the highest TF-IDF value from among the sentences containing key information that is not included in the selected sentence with the highest TF-IDF value. The sentence "To provide a device that can easily acquire data and perform highly accurate estimation" included in the summary sentence data generated by the second summarization is selected.

[0122] Other Embodiments The information processing system 1 according to this embodiment selects a summary result from the generated plurality of summary sentence data or the plurality of sentences included in the generated summary sentence data according to importance, but the selected summary result may be regenerated using LLM. The regeneration of the summary result may be performed, for example, by using key information included in the sentences of the summary result so as not to omit key information of high importance.

[0123] For example, the information processing system 1 according to this embodiment may regenerate summary sentence data of the entire sentence to be processed using the summary results for each selected category, using the LLM.

[0124] In this embodiment, summary results are selected from the generated plurality of summary sentence data or the plurality of sentences contained in the generated summary sentence data according to importance, thereby preventing the generation of erroneous (low-accuracy) summary sentence data from the sentence data to be processed. Furthermore, in this embodiment, a summary result that includes a wide range of key information and provides an overview of the entire sentence data to be processed can be output.

[0125] Although the present embodiment has been described above, it will be understood that various modifications in form and details are possible without departing from the spirit and scope of the claims. While the present invention has been described above based on examples, the present invention is not limited to the above examples and various modifications are possible within the scope of the claims. This application claims priority from basic application No. 2024-085146, filed with the Japan Patent Office on May 24, 2024, the entire contents of which are incorporated herein by reference.

[0126] REFERENCE SIGNS LIST 1 Information processing system 10 Information processing device 12 User terminal 18 Network 30 Input reception unit 32 Summary sentence data generation unit 34 Extraction unit 36 ​​Summary sentence data selection unit 38 Display control unit 40 Importance calculation unit 42 Detection unit 44 Communication unit 50 LLM storage unit 60 Operation reception unit 62 Display unit 64 Communication unit 500 Computer

Claims

1. A summary sentence data generation unit generates multiple summary sentence data from the sentence data to be processed according to the processing instruction data, An extraction unit that extracts key information contained in the plurality of summary sentence data for each category that constitutes the sentence data to be processed, A summary sentence data selection unit selects the summary sentence data to be output from the plurality of summary sentence data using the key information for each category, An information processing device having

2. The summary sentence data generation unit generates a plurality of summary sentence data for each category from the sentence data to be processed according to the processing instruction data. The information processing apparatus according to claim 1, characterized in that

3. The system further includes an importance calculation unit that calculates the importance of the aforementioned key information, The summary data selection unit selects the summary data to be output from the plurality of summary data using the importance level. An information processing apparatus according to claim 1 or 2, characterized by the above.

4. The summary data selection unit said above, Using the importance of the aforementioned key information, the importance of each of the multiple summary sentence data is calculated. From the aforementioned plurality of summary data, select the first summary data with the highest importance. Using the importance of the remaining key information after removing the key information contained in the first summary data, the importance of each of the remaining summary data after removing the first summary data is calculated. Select the second summary data with the highest importance from the remaining summary data. The information processing apparatus according to claim 3, characterized by the following:

5. The summary sentence data generation unit generates multiple summary sentence data from the same processed sentence data according to different processing instruction data. An information processing apparatus according to claim 1 or 2, characterized by the above.

6. The summary sentence data generation unit generates multiple summary sentence data from different processed sentence data according to the same processing instruction data. An information processing apparatus according to claim 1 or 2, characterized by the above.

7. The summary sentence data generation unit generates multiple summary sentence data from different processed sentence data according to different processing instruction data. An information processing apparatus according to claim 1 or 2, characterized by the above.

8. The system further includes a detection unit for detecting variations in the notation of the aforementioned key information, The extraction unit treats the key information of the detected variations in notation as the same key information. An information processing apparatus according to claim 1 or 2, characterized by the above.

9. The importance calculation unit calculates the importance of the key information using one or more of the following: the number of occurrences of the key information, TF-IDF (Term Frequency - Inverse Document Frequency), or the co-occurrence relationship between words. The information processing apparatus according to claim 3, characterized by the following:

10. In an information processing device, Summary sentence data generation step, which generates multiple summary sentence data from the sentence data to be processed according to the processing instruction data, Extraction step of extracting key information contained in the plurality of summary sentence data for each category that constitutes the sentence data to be processed, A summary sentence data selection step in which, using the key information for each category, a summary sentence data is selected from the plurality of summary sentence data to be output, A program that executes the command.

11. An input receiving unit that receives processing instruction data and data to be processed, A summary sentence data generation unit that generates a plurality of summary sentence data from the sentence data to be processed according to the processing instruction data, An extraction unit that extracts key information contained in the plurality of summary sentence data for each category that constitutes the sentence data to be processed, A summary sentence data selection unit selects the summary sentence data to be output from the plurality of summary sentence data using the key information for each category, A display control unit that displays the selected summary data on a display device, An information processing system having