Information processing device and information processing method

Abstract

An information processing device (10S) is provided with: a search unit (16) for retrieving, from an already-added document which is to be referred to by an LLM (30) and which has had a caption added in the form of generation request information, a similar document which is an already-added document having had generation request information newly inputted and being similar or conforming to the generation request information; an acquisition unit (17) for acquiring a generation result from the LLM (30) by issuing, to the LLM (30), a generation instruction based on the similar document obtained by the search and a directive to generate a generation result for the generation request information on the basis of the similar document; and an output unit (18) for outputting the acquired generation result.

Description

Information processing device and information processing method

[0001] The present disclosure relates to an information processing device and an information processing method.

[0002] Conventionally, systems have been provided that perform natural language processing such as question answering using large language models (LLMs). LLMs are natural language processing models trained using large amounts of text data, and take sentences as input and output sentences. When LLMs are applied to a question-answering system that performs question answering, a question is input to the LLM, and an answer is output from the LLM (see Patent Literature 1). Furthermore, the LLM receives input of generation request information for various requests, such as a language translation request or a content generation request, in addition to a question, and outputs a generation result (e.g., a translated document or generated content) in response to the generation request information, in addition to an answer to the question.

[0003] Another known technique involves inputting documents related to the content of a generation request, such as a question, into the LLM along with that request, and then having the LLM refer to those documents, thereby obtaining more appropriate content from the LLM.

[0004] Patent No. 7396582

[0005] However, in the above method, it is known that if the style of the generation request information input to the LLM differs from the style of the document, the accuracy of information retrieval by the LLM decreases.

[0006] Therefore, an object of the present disclosure is to suppress the decrease in accuracy of information search using LLM as described above and to improve the quality of generated results.

[0007] The information processing device of the present disclosure includes a search unit that searches for similar documents, which are annotated documents that are documents to be referenced by a large-scale language model and are attached with captions in the format of generation request information, from annotated documents, which are annotated documents whose generation request information matches or is similar to newly input generation request information; an acquisition unit that acquires generation results from the large-scale language model by issuing a generation instruction to the large-scale language model based on the similar documents obtained by the search and an instruction statement to generate generation results for the generation request information based on the similar documents; and an output unit that outputs the generation results acquired by the acquisition unit.

[0008] According to the present disclosure, it is possible to suppress a decrease in the accuracy of information search by LLM and improve the quality of answers.

[0009] FIG. 1 is a system configuration diagram of a first embodiment. FIG. 1 is a diagram for explaining processing into a Q&A format. FIG. 2 is a flow diagram showing advance preparation processing of the first embodiment. FIG. 3 is a diagram for explaining the processing of FIG. 3. FIG. 4 is a flow diagram showing processing for responding to a question of the first embodiment. FIG. 1 is a system configuration diagram of a second embodiment. FIG. 1 is a diagram for explaining adding a caption. FIG. 2 is a flow diagram showing advance preparation processing of the second embodiment. FIG. 3 is a diagram for explaining the processing of FIG. 8. FIG. 4 is a flow diagram showing processing for responding to a question of the second embodiment. FIG. 5 is a diagram showing an example of the hardware configuration of an information processing device.

[0010] Various embodiments of an information processing device and an information processing method according to the present disclosure will be described below with reference to the drawings. A form will be described below in which a question is input to an LLM (large-scale language model) as an example of "generation request information," and an answer to the question is output from the LLM as an example of a "generation result." Furthermore, as a first embodiment, an embodiment will be described below in which a document referenced by the LLM is processed into a description format in which a question and an answer are associated (hereinafter referred to as a "Q&A format"), and as a second embodiment, an embodiment will be described below in which a question-style caption is added to a document referenced by the LLM.

[0011] First Embodiment FIG. 1 illustrates a configuration diagram of a system 1 including an information processing device 10 according to a first embodiment. As illustrated in FIG. 1 , the system 1 includes a terminal 20 operated by a user, an external server (hereinafter referred to as “LLM”) 30 running a large-scale language model (LLM), and the information processing device 10. The information processing device 10 is a device constituting a Retrieval-Augmented Generation (RAG) system. The RAG system is a type of prompt extension technology used for, for example, corporate information linkage of LLMs. Specifically, when issuing a generation request to an LLM based on an instruction statement (generation request information), the system searches for similar documents, etc. in a search system in advance, and outputs the obtained similar documents, etc., along with the instruction statement to the LLM to request generation. To realize the functions described herein, the information processing device 10 includes a question receiving unit 11, a question history storage unit 12, an FAQ acquisition unit 13, a raw document storage unit 14, a processing unit 15, a search unit 16, an answer acquisition unit 17, and an answer output unit 18. The functions of each part will be explained below.

[0012] The question receiving unit 11 is a functional unit that receives questions sent from the terminal 20 .

[0013] The question history storage unit 12 is a functional unit that stores question history information, which is history information relating to various questions received in the past and their answers.

[0014] The FAQ acquisition unit 13 is a functional unit that acquires FAQ (Frequently Asked Questions) information, which is a collection of frequently asked questions and their answers, based on conditions required for the FAQ (e.g., a predetermined question frequency that is used as a criterion for recognizing a question as an FAQ) and question history information, and stores the information in the built-in FAQ storage unit 13A. The FAQ acquisition unit 13 acquires the conditions required for the FAQ from an external device (such as the terminal 20) or a built-in memory (not shown), and acquires question history information from the question history storage unit 12. Regarding the acquisition of the FAQ, the FAQ acquisition unit 13 may generate the FAQ itself, or may instruct the LLM 30 to generate the FAQ and acquire the FAQ from the LLM 30; details will be described later.

[0015] The raw document storage unit 14 is a functional unit that stores unprocessed documents (raw documents) previously acquired from the outside (such as the terminal 20).

[0016] The processing unit 15 is a functional unit that processes documents referenced by the LLM 30 into a Q&A format and stores them in the built-in processed document storage unit 15A. Processing into a Q&A format means, for example, as shown in FIG. 2, converting the contents of the original document (raw document) into a processed document in Q&A format, as shown in the lower part of FIG. 2, based on anticipated questions about the original document (e.g., "I would like to know what ingredients are needed to make curry."). The processing unit 15 may perform this processing itself, or it may instruct the LLM 30 to perform the processing and obtain the processed document from the LLM 30. Details will be described later. The processing unit 15 may also narrow down the raw documents to be processed based on FAQs. For example, only raw documents related to questions and answers included in the FAQs may be processed.

[0017] The search unit 16 is a functional unit that searches processed documents for similar documents that match or are similar to a newly input question.

[0018] The answer acquisition unit 17 is a functional unit that acquires an answer from the LLM 30 by issuing a generation instruction to the LLM 30 based on similar documents obtained by the search and an instruction statement to generate an answer to the newly input question based on the similar documents. Regarding the generation instruction, the answer acquisition unit 17 issues the generation instruction to the LLM 30 by outputting the instruction statement and the similar documents to a prompt that is input information to the LLM 30.

[0019] The answer output unit 18 is a functional unit that outputs the answer acquired by the answer acquisition unit 17. In the configuration of the information processing device 10 described above, the functional units involved in acquiring and storing the processed document (i.e., the question history storage unit 12, the FAQ acquisition unit 13, the raw document storage unit 14, and the processing unit 15) may be provided in an external device rather than inside the information processing device 10.

[0020] The following describes the preparation process (FIG. 3) and the question response process (FIG. 5) as processes executed by the information processing device 10, in that order, with reference to the flow charts of FIGS.

[0021] In order to respond to various questions, the information processing device 10 executes a preparatory process (FIG. 3) for acquiring and storing processed documents to be referenced by the LLM 30. First, the raw document storage unit 14 stores unprocessed documents (raw documents) acquired in advance from an external device (such as the terminal 20) (step S1 in FIG. 3). The FAQ acquisition unit 13 acquires the conditions required for FAQs from an external device (such as the terminal 20) or an internal memory (not shown) and the question history information from the question history storage unit 12. Referring to the question history information, the device acquires a combination of a question that should be recognized as an FAQ in light of the above conditions (for example, a question received more frequently than a reference frequency) and its answer as an FAQ, and stores the combination in the FAQ storage unit 13A (step S2). In another embodiment of this step S2, as shown in FIG. 4, the FAQ acquisition unit 13 may cause the LLM 30 to generate an FAQ by sending (instructing the LLM 30 to generate) (1) question history information, (2) conditions required for the FAQ, and (3) an instruction to generate an FAQ based on (1) and (2) above, and then acquire the generated FAQ from the LLM 30.

[0022] Next, the processing unit 15 processes the raw documents stored in the raw document storage unit 14 into a Q&A format by narrowing down the target to questions that are frequently asked based on the FAQ (for example, by only processing raw documents related to questions and answers included in the FAQ), and stores the processed documents in the processed document storage unit 15A (step S3). In another embodiment, in step S3, as shown in Figure 4, the processing unit 15 may transmit (instruct the LLM 30 to generate) (1) the raw documents, (2) the FAQ, and (3) an instruction statement to process the raw documents by narrowing down the target based on (1) and (2) above, thereby causing the LLM 30 to process the raw documents and obtaining the processed documents from the LLM 30.

[0023] Through the above advance preparation process (FIG. 3), the processed document is acquired and stored in the processed document storage unit 15A.

[0024] The following describes the question response process ( FIG. 5 ). First, the question receiving unit 11 receives a question (step S11). The search unit 16 searches the processed documents stored in the processed document storage unit 15A for similar documents, which are processed documents that match or are similar to the question received in step S11 (step S12). The answer obtaining unit 17 then outputs the similar documents obtained by the search unit 16 and an instruction to generate an answer to the question based on the similar documents to a prompt that serves as input information to the LLM 30 (instructing the LLM 30 to generate an answer). This causes the LLM 30 to generate an answer by referencing the similar documents, and the generated answer is obtained from the LLM 30 (step S13). The answer output unit 18 then outputs the obtained answer to the terminal 20 (step S14). This presents the answer to the question to the user.

[0025] According to the first embodiment described above, documents to be referenced by the LLM 30 are extracted by searching from processed documents that have been processed in advance into a Q&A format, so the question input to the LLM 30 and the document share a common writing style. This eliminates the conventional problem (deterioration in the accuracy of information retrieval by the LLM 30) that arises from differences in writing style between the question input to the LLM 30 and the document, and can improve the quality of the results (answers) generated by the LLM 30.

[0026] Furthermore, the processing unit 15 processes raw documents by narrowing down the target to questions that are frequently asked based on the FAQ (for example, by processing only raw documents related to frequently asked questions and answers included in the FAQ), thereby enabling efficient processing by effectively narrowing down the target.

[0027] Furthermore, when the FAQ acquisition unit 13 instructs the LLM 30 to generate an FAQ and acquires the FAQ from the LLM 30, the LLM 30 can be effectively utilized to acquire the FAQ.

[0028] Furthermore, even in the case where the processing unit 15 instructs the LLM 30 to execute processing and acquires the processed document from the LLM 30, the LLM 30 can be effectively utilized to acquire the processed document.

[0029] [Second Embodiment] Hereinafter, as a second embodiment, an embodiment in which a question-style caption is assigned to a document referenced by an LLM will be described. FIG. 6 shows a configuration diagram of a system 1S including an information processing device 10S in the second embodiment. As shown in FIG. 6, the system 1S includes a terminal 20 operated by a user, an LLM 30, and an information processing device 10S. Like the information processing device 10 of the first embodiment, the information processing device 10S is also a device that constitutes the aforementioned RAG system. To realize the functions related to the present disclosure, the information processing device 10S includes a question receiving unit 11, a question history storage unit 12, an FAQ acquisition unit 13, an unassigned document storage unit 24, an assignment unit 25, a search unit 16, an answer acquisition unit 17, and an answer output unit 18. The functions of each unit will be described below.

[0030] The question receiving unit 11 is a functional unit that receives questions sent from the terminal 20 .

[0031] The question history storage unit 12 is a functional unit that stores question history information, which is history information relating to various questions received in the past and their answers.

[0032] The FAQ acquisition unit 13 is a functional unit that acquires FAQ (Frequently Asked Questions) information, which is a collection of frequently asked questions and their answers, based on conditions required for the FAQ (e.g., a predetermined question frequency that is used as a criterion for recognizing a question as an FAQ) and question history information, and stores the information in the built-in FAQ storage unit 13A. The FAQ acquisition unit 13 acquires the conditions required for the FAQ from an external device (such as the terminal 20) or a built-in memory (not shown), and acquires question history information from the question history storage unit 12. Regarding the acquisition of the FAQ, the FAQ acquisition unit 13 may generate the FAQ itself, or may instruct the LLM 30 to generate the FAQ and acquire the FAQ from the LLM 30; details will be described later.

[0033] The uncaptioned document storage unit 24 is a functional unit that stores documents (uncaptioned documents) that have not been captioned and that have been acquired in advance from the outside (such as the terminal 20).

[0034] The captioning unit 25 is a functional unit that captions documents referenced by the LLM 30 with question-type captions and stores them in the built-in captioned document storage unit 25A. Captioning question-type captions means, for example, as shown in FIG. 7, captioning the contents of an original document (raw document) with a question-type caption (as shown in the lower part of FIG. 7) based on anticipated questions about the contents (e.g., "I want to know what ingredients are needed to make curry.") to obtain a captioned document. The captioning unit 25 may perform this captioning itself, or may instruct the LLM 30 to perform the captioning and obtain the captioned document from the LLM 30. Details will be described later. The captioning unit 25 may also narrow down the list of uncaptioned documents to be captioned based on FAQs. For example, only uncaptioned documents related to questions and answers included in the FAQ may be subject to captioning.

[0035] The search unit 16 is a functional unit that searches for similar documents, which are documents with annotations that match or are similar to a newly input question, from among the documents with annotations.

[0036] The answer acquisition unit 17 is a functional unit that acquires an answer from the LLM 30 by issuing a generation instruction to the LLM 30 based on similar documents obtained by the search and an instruction statement to generate an answer to the newly input question based on the similar documents. Regarding the generation instruction, the answer acquisition unit 17 issues the generation instruction to the LLM 30 by outputting the instruction statement and the similar documents to a prompt that is input information to the LLM 30.

[0037] The answer output unit 18 is a functional unit that outputs the answer acquired by the answer acquisition unit 17. In the configuration of the information processing device 10S described above, the functional units involved in acquiring and storing the tagged documents (i.e., the question history storage unit 12, the FAQ acquisition unit 13, the untagged document storage unit 24, and the tagging unit 25) may be provided in an external device instead of inside the information processing device 10.

[0038] The following describes the preparation process (FIG. 8) and the question response process (FIG. 10) as processes executed by the information processing device 10, in that order, with reference to the flow charts of FIGS. 8 and 10.

[0039] In order to respond to various questions, the information processing device 10S executes a preparatory process (FIG. 8) for acquiring and storing in advance captioned documents to be referenced by the LLM 30. First, the uncaptioned document storage unit 24 stores uncaptioned documents (uncaptioned documents) acquired in advance from an external device (such as the terminal 20) (step S21 in FIG. 8). The FAQ acquisition unit 13 acquires conditions required for FAQs from an external device (such as the terminal 20) or an internal memory (not shown) and question history information from the question history storage unit 12. Referring to the question history information, the device acquires a combination of a question that should be recognized as an FAQ in light of the above conditions (for example, a question received more frequently than a reference frequency) and its answer as an FAQ, and stores the combination in the FAQ storage unit 13A (step S22). In another embodiment of this step S22, as shown in FIG. 9, the FAQ acquisition unit 13 may cause the LLM 30 to generate an FAQ by sending (instructing the LLM 30 to generate) (1) question history information, (2) conditions required for the FAQ, and (3) an instruction to generate an FAQ based on (1) and (2) above, and then acquire the generated FAQ from the LLM 30.

[0040] Next, the captioning unit 25 narrows down the uncaptioned documents stored in the uncaptioned document storage unit 24 to questions that are frequently asked based on the FAQ (for example, only uncaptioned documents related to questions and answers included in the FAQ), captions them in the question format, and stores the captioned documents in the captioned document storage unit 25A (step S23). In another embodiment, in step S23, as shown in FIG. 9 , the captioning unit 25 may transmit (instruct the LLM 30 to generate) (1) the uncaptioned documents, (2) the FAQ, and (3) an instruction statement to caption the uncaptioned documents by narrowing down the targets based on (1) and (2) above, thereby causing the LLM 30 to caption them and acquiring the captioned documents from the LLM 30.

[0041] By the above-described advance preparation process (FIG. 8), the attached document is acquired and stored in the attached document storage unit 25A.

[0042] The following describes the question response process ( FIG. 10 ). First, the question receiving unit 11 receives a question (step S31). The search unit 16 searches the annotated documents stored in the annotated document storage unit 25A for similar documents that match or are similar to the question received in step S31 (step S32). The answer acquisition unit 17 then outputs the similar documents found by the search unit 16 and an instruction to generate an answer to the question based on the similar documents to a prompt (instructing the LLM 30 to generate an answer) as input information to the LLM 30, thereby causing the LLM 30 to generate an answer by referencing the similar documents and acquiring the generated answer from the LLM 30 (step S33). The answer output unit 18 then outputs the acquired answer to the terminal 20 (step S34). The answer to the question is then presented to the user.

[0043] According to the second embodiment described above, documents to be referenced by the LLM 30 are extracted by search from documents to which question-style captions have been added in advance, so the question input to the LLM 30 and the document share a common writing style. This eliminates the conventional problem (deterioration in the accuracy of information retrieval by the LLM 30) caused by differences in writing style between the question input to the LLM 30 and the document, and can improve the quality of the results (answers) generated by the LLM 30.

[0044] In addition, the captioning unit 25 narrows down the target of uncaptioned documents to questions that are frequently asked based on the FAQ (for example, it targets only uncaptioned documents related to frequently asked questions and answers included in the FAQ), so that efficient captioning can be performed by effectively narrowing down the target.

[0045] Furthermore, when the FAQ acquisition unit 13 instructs the LLM 30 to generate an FAQ and acquires the FAQ from the LLM 30, the LLM 30 can be effectively utilized to acquire the FAQ.

[0046] Furthermore, even in the case where the adding unit 25 instructs the LLM 30 to add a caption and acquires the captioned document from the LLM 30, the LLM 30 can be effectively utilized to acquire the captioned document.

[0047] The gist of the present disclosure lies in the following [1] to [6]. [1] An information processing device comprising: a search unit that searches for similar documents, which are annotated documents that are documents to be referenced by a large-scale language model and have been annotated with captions in the format of generation request information, from the annotated documents, which are documents to be referenced by a large-scale language model and have the captions annotated in the format of generation request information, and the generation request information matches or is similar to newly input generation request information; an acquisition unit that acquires generation results from the large-scale language model by issuing a generation instruction to the large-scale language model based on the similar documents obtained by the search and an instruction statement to generate a generation result in response to the generation request information based on the similar documents; and an output unit that outputs the generation result acquired by the acquisition unit. [2] The information processing device according to [1], further comprising: an addition unit that adds the caption to the unannotated document, which is a document to which no caption has been added. [3] The information processing device according to [2], further comprising: an information acquisition unit that acquires frequently requested information based on a generation request history and conditions required for frequently requested information that is information collecting frequently requested generation request information and generation results for the requests, wherein the attachment unit narrows down targets to be attached based on the acquired frequently requested information. [4] The information processing device according to [3], further comprising: an information acquisition unit that acquires frequently requested information based on a generation request history and conditions required for frequently requested information that is information collecting frequently requested generation request information and generation results for the requests, wherein the attachment unit narrows down targets to be attached based on the acquired frequently requested information. [4] The information processing device according to [3], further comprising: an information acquisition unit that acquires frequently requested information based on a generation request history and conditions required for frequently requested information that is information collecting frequently requested generation request information and generation results for the requests, wherein the attachment unit narrows down targets to be attached based on the acquired frequently requested information. [5] The information processing device according to [3] or [4], wherein the information acquisition unit acquires the frequently requested information from the large-scale language model by issuing a generation instruction to the large-scale language model based on the generation request history, conditions required for the frequently requested information, and an instruction statement to generate the frequently requested information based on the generation request history and the conditions required for the frequently requested information.[6] An information processing method comprising: a step in which an information processing device searches for similar documents, which are annotated documents that are documents to be referenced by a large-scale language model and that have been annotated in the format of generation request information, from the annotated documents, which are documents whose generation request information matches or is similar to newly input generation request information; a step in which the information processing device acquires generation results from the large-scale language model by issuing a generation instruction to the large-scale language model based on the similar documents obtained by the search and an instruction statement to generate generation results for the generation request information based on the similar documents; and a step in which the information processing device outputs the acquired generation results.

[0048] [Explanation of Terms, Explanation of Hardware Configuration (FIG. 11), etc.] The block diagrams used in the description of the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using a single device that is physically or logically coupled, or may be realized using two or more physically or logically separated devices that are directly or indirectly connected (e.g., wired, wireless, etc.) and these multiple devices. The functional block may be realized by combining the single device or the multiple devices with software.

[0049] Functions include, but are not limited to, judgment, determination, assessment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating, mapping, and assignment. For example, a functional block (component) that performs transmission is called a transmitting unit or transmitter. As mentioned above, there are no particular limitations on how these functions are implemented.

[0050] For example, an information processing device according to an embodiment of the present disclosure may function as a computer that executes the processes of the present disclosure. Fig. 11 is a diagram illustrating an example of a hardware configuration of an information processing device 10 according to an embodiment of the present disclosure. The information processing device 10 described above may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, etc.

[0051] In the following description, the term "apparatus" can be interpreted as a circuit, a device, a unit, etc. The hardware configuration of the information processing device 10 may be configured to include one or more of the apparatuses shown in the drawings, or may be configured to exclude some of the apparatuses.

[0052] Each function of the information processing device 10 is realized by loading specified software (programs) onto hardware such as the processor 1001 and memory 1002, causing the processor 1001 to perform calculations, control communication via the communication device 1004, and control at least one of reading and writing data in the memory 1002 and storage 1003.

[0053] The processor 1001 controls the entire computer by running, for example, an operating system, and may be configured as a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc.

[0054] The processor 1001 also reads programs (program codes), software modules, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes in accordance with these. The programs used are those that cause a computer to execute at least some of the operations described in the above-described embodiments. While the various processes have been described as being executed by one processor 1001, they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The programs may be transmitted from a network via a telecommunications line.

[0055] The memory 1002 is a computer-readable recording medium and may be configured by, for example, at least one of a read-only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a random access memory (RAM), etc. The memory 1002 may also be called a register, a cache, a main memory (primary storage device), etc. The memory 1002 can store executable programs (program codes), software modules, etc. for implementing a wireless communication method according to an embodiment of the present disclosure.

[0056] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, etc. Storage 1003 may also be referred to as an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, a server, or other appropriate medium including at least one of memory 1002 and storage 1003.

[0057] The communication device 1004 is hardware (transmission / reception device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc. to realize at least one of frequency division duplex (FDD) and time division duplex (TDD).

[0058] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that receives input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that outputs to the outside. The input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).

[0059] Furthermore, each device, such as the processor 1001 and the memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.

[0060] The information processing device 10 may also be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.

[0061] The notification of information is not limited to the aspects / embodiments described in the present disclosure and may be performed using other methods. For example, the notification of information may be performed by physical layer signaling (e.g., Downlink Control Information (DCI) and Uplink Control Information (UCI)), higher layer signaling (e.g., Radio Resource Control (RRC) signaling, Medium Access Control (MAC) signaling, broadcast information (Master Information Block (MIB) and System Information Block (SIB))), other signals, or a combination thereof. Furthermore, the RRC signaling may be referred to as an RRC message, and may be, for example, an RRC Connection Setup message, an RRC Connection Reconfiguration message, or the like.

[0062] Each aspect / embodiment described in the present disclosure may be implemented using any of the following standards: LTE (Long Term Evolution), LTE-Advanced (LTE-A), SUPER 3G, IMT-Advanced, 4G (4th generation mobile communication system), 5G (5th generation mobile communication system), 6th generation mobile communication system (6G), xth generation mobile communication system (xG) (xG (x is, for example, an integer or a decimal number)), FRA (Future Radio Access), NR (new Radio), New radio access (NX), Future generation radio access (FX), W-CDMA (registered trademark), GSM (registered trademark), CDMA2000, UMB (Ultra Mobile Broadband), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.17 (WiMAX (registered trademark)), IEEE 802.19 (WiMAX (registered trademark)), IEEE 802.20 (WiMAX (registered trademark)), IEEE 802.21 (Wi-Fi (registered trademark)), IEEE 802.22 (WiMAX (registered trademark)), IEEE 802.23 (WiMAX (registered trademark)), IEEE 802.24 (WiMAX (registered trademark)), IEEE 802.25 (WiMAX (registered trademark)), IEEE 802.26 (WiMAX (registered trademark)), IEEE 802.27 (WiMAX (registered trademark)), IEEE 802.28 (WiMAX (registered trademark)), IEEE 802.29 (WiMAX (registered trademark)), IEEE 802.30 (WiMAX (registered trademark)), IEEE 802.31 (Wi-Fi (registered trademark)), IEEE 802.32 (WiMAX (registered trademark)), IEEE 802.33 (WiMAX (registered trademark)), IEEE 802.34 ( The present invention may be applied to at least one of systems using 802.20, UWB (Ultra-Wide Band), Bluetooth (registered trademark), or other suitable systems, and next-generation systems that are extended, modified, created, or defined based on these systems. The present invention may also be applied to a combination of multiple systems (e.g., a combination of LTE and / or LTE-A with 5G).

[0063] The order of the procedures, sequences, flowcharts, etc. of each aspect / embodiment described in this disclosure may be changed unless it is consistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0064] Input and output information may be stored in a specific location (for example, memory) or may be managed using a management table. Input and output information may be overwritten, updated, or added to. Output information may be deleted. Input information may be transmitted to another device.

[0065] The determination may be made based on a value represented by one bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).

[0066] The aspects / embodiments described in this disclosure may be used alone, in combination, or switched depending on the implementation. Notification of predetermined information (e.g., notification that "X is true") is not limited to explicit notification, but may be implicit (e.g., not notifying the predetermined information).

[0067] Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure.

[0068] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0069] Software, instructions, information, etc. may also be transmitted or received over a transmission medium. For example, if software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave), then these wired and / or wireless technologies are included within the definition of transmission media.

[0070] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0071] Note that terms described in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of a channel and a symbol may be a signal (signaling). Furthermore, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, a cell, a frequency carrier, etc.

[0072] As used in this disclosure, the terms "system" and "network" are used interchangeably.

[0073] Furthermore, the information, parameters, etc. described in the present disclosure may be expressed using absolute values, may be expressed using relative values ​​from a predetermined value, or may be expressed using other corresponding information. For example, a radio resource may be indicated by an index.

[0074] The names used for the above-described parameters are not intended to be limiting in any way. Furthermore, the mathematical expressions using these parameters may differ from those explicitly disclosed in this disclosure. The various channels (e.g., PUCCH, PDCCH, etc.) and information elements may be identified by any suitable names, and therefore the various names assigned to these various channels and information elements are not intended to be limiting in any way.

[0075] As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, inquiring (e.g., searching in a table, database, or other data structure), ascertaining, and the like. "Determining" and "determining" may also include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in memory), and the like. Furthermore, "judgment" and "decision" can include regarding resolving, selecting, choosing, establishing, comparing, etc. as having been "judged" or "decided." In other words, "judgment" and "decision" can include regarding some action as having been "judged" or "decided." Furthermore, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.

[0076] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0077] As used in this disclosure, any reference to an element using a designation such as "first," "second," etc. does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must in some way precede the second element.

[0078] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, when the term "or" is used in this disclosure, it is not intended to be an exclusive or.

[0079] In this disclosure, where articles are added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0080] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different."

[0081] 1, 1S...system, 10, 10S...information processing device, 11...question receiving unit, 12...question history storage unit, 13...FAQ acquisition unit, 13A...FAQ storage unit, 14...raw document storage unit, 15...processing unit, 15A...processed document storage unit, 16...search unit, 17...answer acquisition unit (acquisition unit), 18...answer output unit (output unit), 20...terminal, 24...unassigned document storage unit, 25...assignment unit, 25A...assigned document storage unit, 30...LLM, 1001...processor, 1002...memory, 1003...storage, 1004...communication device, 1005...input device, 1006...output device, 1007...bus.

Claims

1. An information processing device comprising: a search unit that searches for similar documents, which are annotated documents that are documents to be referenced by a large-scale language model and that have been annotated with captions in the format of generation request information, from annotated documents that are annotated documents whose generation request information matches or is similar to newly input generation request information; an acquisition unit that acquires generation results from the large-scale language model by issuing a generation instruction to the large-scale language model based on the similar documents obtained by the search and an instruction statement to generate generation results for the generation request information based on the similar documents; and an output unit that outputs the generation results acquired by the acquisition unit.

2. The information processing device according to claim 1, further comprising an adding unit that adds the caption to an uncaptioned document, which is a document to which the caption has not been added.

3. The information processing device according to claim 2, further comprising an information acquisition unit that acquires frequently requested information based on a generation request history and conditions required for frequently requested information, which is information that collects frequently requested generation request information and generation results for those requests, and the assignment unit narrows down the targets to be assigned based on the acquired frequently requested information.

4. The information processing device described in claim 3, wherein the annotation unit obtains from the large-scale language model the annotated document for which annotation has been performed by narrowing down the target based on the frequently requested information, by instructing the large-scale language model to generate the document based on the frequently requested information, the unannotated document, and an instruction statement to narrow down the target based on the frequently requested information.

5. The information processing device described in claim 3, wherein the information acquisition unit acquires the frequently requested information from the large-scale language model by issuing a generation instruction to the large-scale language model based on the generation request history, the conditions required for the frequently requested information, and an instruction statement to generate the frequently requested information based on the generation request history and the conditions required for the frequently requested information.

6. An information processing method comprising: a step in which an information processing device searches for similar documents, which are annotated documents that are documents to be referenced by a large-scale language model and that have been given captions in the format of generation request information, from annotated documents whose generation request information matches or is similar to newly input generation request information; a step in which the information processing device obtains generation results from the large-scale language model by issuing a generation instruction to the large-scale language model based on the similar documents obtained by the search and an instruction statement to generate generation results for the generation request information based on the similar documents; and a step in which the information processing device outputs the obtained generation results.