Information processing system, information processing method, and program

Abstract

To enable information closely related to a question sentence to be transversely acquired even if a language of information to be a retrieval object is different from the question sentence, and to achieve an answer corresponding to the language of the question sentence on the basis of the acquired information.SOLUTION: The present invention includes question acquisition means for acquiring a question, estimation means for estimating a language related to the question acquired by the question acquisition means, retrieval means for retrieving retrieval object information by a retrieval query created on the basis of the question acquired by the question acquisition means, and output means for outputting an answer to the question in the language estimated by the estimation means by using the information retrieved by the retrieval means and the acquired information, and the retrieval means retrieves the retrieval object information by the retrieval query created by the language related to the retrieval object information on the basis of the question in the case that the language related to the question estimated by the estimation means is different from the language related to the retrieval object information.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] Increasingly, companies and public institutions are introducing full-text search systems, including enterprise search, to find the information they need from the vast amount of digital documents they have stored. Furthermore, the introduction of question-answering systems, which allow users to search for and answer questions through dialogue, has made it easier for them to find the information they need.

[0003] Furthermore, in recent years, the emergence of generative AI based on large-scale language models such as ChatGPT / GPT has made it possible to provide natural and highly accurate question-answering. Generative AI can also provide natural question-answering responses to inputs in a variety of languages, depending on the language of the question.

[0004] On the other hand, while generative AI has a high ability to respond to open-domain questions, it cannot accurately answer domain-specific questions that are not included in the training data, such as questions about specific documents within a company.

[0005] Non-Patent Document 1 reports on a mechanism called RAG (Retrieval-Augmented Generation) that links a generative AI with a search system to enable answers to unlearned matters. RAG uses a search system that registers domain knowledge to acquire domain knowledge that is closely related to the question, and by including the domain knowledge in the instructions (prompts) given to the generative AI, it enables the generative AI to answer unlearned matters.

[0006] Furthermore, Non-Patent Document 2 reports a method for improving the accuracy of answers in RAG by using both semantic vector search and full-text search as a search system in RAG.

[0007] In Patent Document 1, in a system for determining legal restrictions on chemical products, knowledge necessary for the determination is acquired by searching for keywords included in a query. The method disclosed involves extracting the name of the country that is subject to the law (target country) from the keywords during the search, determining whether other keywords are in the language of the target country, and, if the keywords are different from the language of the target country, translating the keywords into the language of the target country before performing the search. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Patent Publication No. 2021-68008 [Non-Patent Document 1] Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Meng Wang, and Haofen Wang, Retrieval-Augmented Generation for Large Language Models: A Survey, arXiv:2312.10997, 27 Mar 2024 (https: / / arxiv.org / abs / 2312.10997) [Non-patent document 2] Michael Glass, Gaetano Rossiello, Md Faisal Mahbub Chowdhury, Ankita Rajaram Naik, Pengshan Cai, Alfio Gliozzo, Re2G: Retrieve, Rerank, Generate, arXiv:2207.06300, 13 Jul 2022 (https: / / arxiv.org / abs / 2207.06300) DISCLOSURE OF THE INVENTION [Problem to be solved by the invention]

[0009] In Non-Patent Document 2, full-text search is used to acquire knowledge, so if the language of the domain knowledge is different from the language of the question, there is a problem that the keywords in the question do not match the keywords in the domain knowledge, making it impossible to search for the necessary knowledge.

[0010] In Patent Document 1, it is possible to search for knowledge in response to questions written in various languages ​​from a collection of documents that serve as knowledge.

[0011] However, the target language is determined by the target country name included as a keyword in the question, and the country name must be clearly stated in the question.

[0012] Furthermore, because documents for multiple countries are assumed to be written in the respective languages ​​of those countries, there is a problem in that knowledge cannot be searched across the entire document set.Furthermore, even if the document set is composed of a single language, multilingual search cannot be realized.

[0013] Therefore, even if the question and domain knowledge are written in different languages, it is desirable to be able to search for domain knowledge that is closely related to the question and to provide an answer according to the language of the question.

[0014] Therefore, the present invention aims to be able to acquire information that is closely related to the question across the board, even if the question and the information to be searched are written in different languages, and to provide an answer that corresponds to the language of the question based on the acquired information. [Means for solving the problem]

[0015] The present invention comprises a question acquisition means for acquiring a question, an estimation means for estimating the language related to the question acquired by the question acquisition means, a search means for searching for search target information using a search query created based on the question acquired by the question acquisition means, and an output means for outputting an answer to the question in the language estimated by the estimation means using the information searched by the search means and the acquired question, wherein when the language related to the question estimated by the estimation means is different from the language related to the search target information, the search means searches for the search target information using the search query created in the language related to the search target information based on the question. [Effects of the Invention]

[0016] According to the present invention, it is possible to provide an accurate answer that is appropriate to the language of a question sentence in a variety of languages, regardless of the language of the information to be searched for. [Brief explanation of the drawings]

[0017] [Figure 1] 1 is a diagram illustrating an example of a system configuration of a question answering device according to an embodiment of the present invention. [Figure 2] 1 is a block diagram showing an example of the hardware configuration of a question answering device and a user terminal according to an embodiment of the present invention. [Figure 3] FIG. 2 is a diagram illustrating an example of a functional configuration of a question answering device according to an embodiment of the present invention. [Figure 4] 10 is a flowchart illustrating an example of a question answering process according to an embodiment of the present invention. [Figure 5] FIG. 2 is a diagram illustrating an example of a configuration of a domain knowledge storage area in the embodiment of the present invention. [Figure 6] FIG. 10 is a diagram illustrating an example of a domain language management table according to an embodiment of the present invention. [Figure 7] FIG. 3 is a diagram illustrating an example of a domain knowledge management table according to an embodiment of the present invention. [Figure 8]FIG. 10 is a diagram showing an example of a prompt created in response to a question sentence in an embodiment of the present invention. [Figure 9] 10 is a flowchart showing an example of a question answering process in the second embodiment of the present invention. [Figure 10] FIG. 10 is a diagram illustrating an example of a chatbot screen displayed on a user terminal. [Figure 11] FIG. 1 is a conceptual diagram of the process of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0018] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[0019] FIG. 1 is a diagram showing an example of a system configuration of a question answering device according to an embodiment of the present invention.

[0020] The question answering device 100 is configured to be connected to a user terminal 110 via a network 120 .

[0021] The question answering device 100 presents an answer to a question received from the user terminal 110.

[0022] The user terminal 110 sends a question entered by a user to the question answering apparatus 100 and displays an answer returned by the question answering apparatus 100. Specific examples of the user terminal 110 include, but are not limited to, a personal computer (such as a notebook PC or desktop PC), a tablet terminal, or a smartphone. The configuration of the various terminals or devices connected to the network 120 in FIG. 1 is merely an example, and it goes without saying that there are various configuration examples depending on the application and purpose.

[0023] FIG. 2 is a block diagram showing an example of the hardware configuration of the question answering apparatus 100 and the user terminal 110 according to the embodiment of the present invention.

[0024] As shown in FIG. 2, the information processing device includes a CPU (Central Processing Unit) 201, a RAM (Random Access Memory) 202, a ROM (Read Only Memory) 203, an input controller 205, a video controller 206, a memory controller 207, and a communication I / F controller 208 connected via a system bus 204.

[0025] The CPU 201 comprehensively controls each device and controller connected to the system bus 204 .

[0026] The RAM 202 functions as a main memory, a work area, etc. for the CPU 201. The CPU 201 loads programs and the like required for executing processing from the ROM 203 or the external memory 211 into the RAM 202, and executes the loaded programs to realize various operations.

[0027] ROM 203 or external memory 211 stores the BIOS (Basic Input / Output System) and OS (Operating System), which are control programs executed by CPU 201, computer-readable and executable programs for realizing this information processing method, and various necessary data (including data tables).

[0028] The input controller 205 controls input from an input device such as a keyboard 209 or a pointing device such as a mouse (not shown). If the input device is a touch panel, the user can issue various instructions by pressing (touching with a finger or the like) icons, cursors, or buttons displayed on the touch panel.

[0029] The touch panel may also be a touch panel capable of detecting positions touched by multiple fingers, such as a multi-touch screen.

[0030] The video controller 206 controls the display on an external output device such as a display 210. The display also includes the display of a notebook computer integrated into the main body. Note that the external output device is not limited to a display, and may be, for example, a projector. In addition, for devices capable of receiving the above-mentioned touch operation, an input device is also provided.

[0031] The video controller 206 can control a video memory (VRAM) for display control, and can use part of the RAM 202 as a video memory area, or can provide a separate dedicated video memory.

[0032] The memory controller 207 controls access to the external memory 211. The external memory may be an external storage device (hard disk) that stores a boot program, various applications, font data, user files, edited files, and various data, a flexible disk (FD), or a CompactFlash (registered trademark) memory connected to a PCMCIA card slot via an adapter.

[0033] The communication I / F controller 208 connects and communicates with external devices via a network, and executes communication control processing on the network. For example, communication using TCP / IP, telephone lines such as ISDN, and 3G lines for mobile phones are possible.

[0034] The CPU 201 enables display on the display 210 by, for example, executing a process of expanding (rasterizing) an outline font into a display information area in the RAM 202. The CPU 201 also enables user instructions using a mouse cursor (not shown) or the like on the display 210.

[0035] 3 is a diagram showing an example of the functional configuration of the question answering apparatus 100 and the user terminal 110. The functions of each functional unit will be explained in the description of the flowcharts in FIG.

[0036] First, an overview of the present invention will be described using FIG. 11 . In this embodiment, the present invention will be described assuming a scenario in which, for example, an employee on an employee portal site asks a chatbot a question about information (documents) on the site and receives an answer. In this example, it is assumed that the information on the site is written in Japanese. For example, a user enters a question into a question input form on the site, such as "Tell me the address of the Osaka office." In this case, a search query is created from the question to search for information on the site. However, if the language of the question (English in this case) does not match the language of the information on the site (Japanese in this case), the search will not yield information suitable for generating an answer. Therefore, the present invention performs the following processes: (1) The language of the question is estimated and it is determined whether the estimated language of the question matches the language of the information on the site. (2) If they do not match, the question is translated into the language of the information on the site. (3) Domain knowledge is searched for using the Japanese search query created from the translated question. (4) A prompt is generated using the retrieved domain knowledge and the question, and the AI ​​is instructed to respond in the language of the user's question. This allows the user to obtain an accurate answer in the language of their choice, regardless of the language the information on the site is written in. In this embodiment, for simplicity, an example has been described in which large language models (LLMs) are used as the language estimation processing unit 303, translation processing unit 304, and answer generation processing unit 307, but this is not limiting, and generation AI, machine learning models, and other methods capable of performing language estimation, translation, and answer generation, respectively, may be used.

[0037] (First embodiment) Next, the process executed by the question and answer processing unit 302 in the first embodiment of the present invention will be described with reference to the flowchart of FIG.

[0038] (Question-answering processing) The flowchart in Figure 4 is a process in which the CPU 201 of the question answering device 100 reads and executes a predetermined control program, and is a flowchart showing the process in which the question answering processing unit 302 generates an answer sentence to a question sentence in response to a request from the dialogue management processing unit 301.

[0039] The dialogue management processing unit 301 realizes a dialogue (for example, a question and answer session) with the user by repeating the processing of the flowchart in FIG.

[0040] The domain knowledge necessary for question answering processing is registered in the domain knowledge storage area 306 shown in FIG. 5 for each piece of domain knowledge to be processed.

[0041] In this invention, domain knowledge refers to information that serves as the basis for answers to questions from users. For example, if the present invention is applied as a search system for a company's employee website (including a dialogue (chat) style question-answering system), the information posted on the employee website (such as information on web pages and posted document files) is domain knowledge. Domain knowledge is registered and managed by, for example, registering site information obtained by searching for a prefix of a URL and updating it periodically, or by having an administrator register data in advance. Domain knowledge is an example of search target information.

[0042] In an embodiment of the present invention, the main language of the domain knowledge (the language in which the domain knowledge is written) preset by the administrator is stored in the domain language management table 501, and the domain knowledge extracted from the registered documents is stored in the domain knowledge management table 502 and is available for search.

[0043] In step S401, the question and answer processing unit 302 receives the question sentence input by the user from the dialogue management processing unit 301.

[0044] In step S402, the question answering processor 302 uses the language estimation processor 303 to estimate the language of the question sentence.

[0045] In the embodiment of the present invention, the method for estimating the language is not particularly limited. The frequency of appearance of characters or basic vocabulary may be used, or estimation may be performed using a machine learning model or a large-scale language model.

[0046] In step S403, the question and answering processor 302 determines whether the language estimated in step S402 matches the main language of the domain knowledge. If they match, the process proceeds to step S404. If they do not match, the process proceeds to step S405.

[0047] In step S404, the question and answering processor 302 creates a search query from the question. A search query is a sentence, phrase, or combination thereof for searching for information suitable for answering the question. There are no particular limitations on the method for creating a search query from a question, but it may be determined, for example, by processing such as extracting only independent words from the results of morphological analysis of the question.

[0048] In step S405, the question and answering processor 302 translates the question into the main language of the domain knowledge using the translation processor 304. In the embodiment of the present invention, the translation method is not particularly limited, and machine translation or a large-scale language model may be used.

[0049] In step S406, the question answering processor 302 creates a search query from the question translated in step S405. In this embodiment, the question is translated in step S405, and then the search query is created in step S406. However, these processes may be performed in reverse. That is, the process may be performed in the following order: first, a search query is created from the question, and if the language of the search query does not match the primary language of the domain knowledge, the search query is translated into the language of the question, and the domain knowledge is searched for using the translated search query.

[0050] In step S407, the question and answering processor 302 uses the search processor 305 to acquire a specified number of pieces of domain knowledge that are highly relevant to the search query from the domain knowledge stored in the domain management table 502.

[0051] In the embodiment of the present invention, the search method is not particularly limited, and a full-text search, a search using semantic vectors, or other search methods may be used. Also, different search methods may be used in combination.

[0052] In step S408, the question and answering processor 302 creates a prompt (instruction sentence) that instructs an answer to the question from the question sentence acquired in step S401, the language estimated in step S402, and the domain knowledge acquired in step S407. Details of the prompt will be described later with reference to FIG.

[0053] In step S409, the question and answering processor 302 uses the answer generation processor 307 to obtain an answer to the prompt created in step S408.

[0054] In the embodiment of the present invention, answer generation using a large-scale language model or generative AI is assumed, but is not particularly limited to this. A general-purpose large-scale language model that has been pre-trained in an unsupervised manner using a large amount of text data to learn grammar, word meanings, etc. may also be used. Furthermore, a fine-tuned large-scale language model that has been trained in a supervised manner to learn specific domain knowledge may also be used. Furthermore, in this embodiment, the language estimation processing unit 303, translation processing unit 304, and answer generation processing unit 307 are configured to be located inside the question and answering processing device 100, but they may also be located externally, and answers may be obtained via an API, for example.

[0055] (Question-answering processing example) Next, as a specific example of the question-answering process shown in the flowchart of Figure 4, we will explain the case where question-answering process is performed for a domain with a domain ID of 1 (where the main language of the domain knowledge is "Japanese") in the domain language management table 501 of Figure 6 and the domain knowledge management table 502 of Figure 7.

[0056] In step S401, the question and answering processor 302 receives the question "Tell me the address of the Osaka office." input by the user from the dialogue management processor 301. In this embodiment, it is assumed that the user inputs the text using the input controller 205 or the like, but the question may also be input by voice recognition.

[0057] In step S402, the question and answering processor 302 uses the language estimation processor 303 to estimate "English" as the language of the question sentence.

[0058] In step S403, the question and answering processor 302 determines that the language estimated in step S402 is "English", which does not match the main language 601 of the domain knowledge, "Japanese", and therefore moves the process to step S405.

[0059] In step S405, the question and answering processor 302 uses the translation processor 304 to translate the question into Japanese, the main language of the domain knowledge, and obtains the Japanese translation of the question, "Please tell me the address of the Osaka office."

[0060] In step S406, the question and answering processor 302 creates a search query from the question translated in step S405.

[0061] In step S407, the question and answering processor 302 uses the search processor 305 to acquire domain knowledge 701 to 703 that is highly relevant to the search query from the domain knowledge management table 502. At this time, the relevance between the search query and the domain knowledge may be evaluated using a score or the like.

[0062] In step S408, the question answering processor 302 creates a prompt 801 shown in Fig. 8. In this embodiment of the present invention, a predetermined template is used to create a prompt 804 by applying the question sentence (802) acquired in step S401, the language "English" (803) estimated in step S402, and the domain knowledge (804) acquired in step S407. Note that in this embodiment, the prompt instructs the user to answer in the language in which the question is written, but it is also possible to have the user answer in the primary language of the domain knowledge and then translate it into the language of the question.

[0063] Furthermore, the domain knowledge input in the prompt may be the most relevant domain knowledge between the search query and the domain knowledge.

[0064] In step S409, the question and answering processor 302 uses the answer generation processor 307 to obtain the answer "The address of the Osaka office is Tosabori Dai Building, Tosabori XXX, Nishi-ku, Osaka, Osaka 550-0001." in response to the prompt 801 created in step S408.

[0065] FIG. 10 shows an example of a screen displaying an answer to a user. A user's question, "Tell me the address of the Osaka office," is entered in a question form 1001. The answer generated in step S409 is displayed in an answer form 1002. The domain knowledge acquired in step S407, which served as the basis for generating the answer, is displayed together with the answer as basis information 1003. The basis information 1003 is a link, allowing the content of the basis information to be displayed by clicking or otherwise instructing the user. The link may display the contents of the domain knowledge management table 502 or may refer to external information. A function may also be provided that allows the user to evaluate the answer generated from the question. A follow-up question form 1004 allows the user to continuously enter questions they wish to ask. In addition to simply displaying the basis information 1003, the basis information 1003 may be translated and displayed based on the language of the question estimated in step S401.

[0066] In the embodiment of the present invention, for the sake of simplicity, it is assumed that the domain knowledge extracted from a document is written in a single language and is registered as is. However, the language in which the domain knowledge is written may be estimated at the time of registration, and if it differs from the primary language of the domain, the domain knowledge may be translated and registered. This allows, for example, even if a document written primarily in Japanese contains some English text, to perform a cross-sectional document search by translating the relevant parts into Japanese and registering them as domain knowledge.

[0067] In this way, by inferring the language of the question and translating the question if it is in a different language than the main language of the domain knowledge, it is possible to realize question-answering in various languages ​​even if only a set of documents in a single language is available.

[0068] In addition, in this embodiment, it is assumed that the language of the question and the primary language of the domain knowledge are different languages, such as English and Japanese, respectively, but they may also be dialects. For example, if the primary language of the domain knowledge is written in standard Japanese and the question is input (by voice) in Niigata dialect, such as "Can we exchange the shared PC, Bokkoretasuke?", the question may be translated into standard Japanese as "The shared PC is broken, can we exchange it?", and then a search query may be created, and an answer may be created in Niigata dialect using the obtained information.

[0069] The present invention can also be applied to the translation of programming languages. For example, if source code written in C is registered as search target information and a question received from a user contains source code written in Java, the Java source code included in the question can be translated into C to search for the search target information.

[0070] (Second embodiment) Next, a second embodiment of the present invention will be described.

[0071] FIG. 9 is a flowchart showing a question and answer process in the second embodiment of the present invention.

[0072] The second embodiment of the present invention differs from the first embodiment in that it repeats the processes from step S403 to step S405 for each of the main languages ​​of the domain knowledge (S901, S903), it temporarily stores the domain knowledge obtained by searching for each language from the domain knowledge management table 502 (S902), and it selects a predetermined number of domain knowledge from the temporarily stored domain knowledge (S904).

[0073] In the second embodiment of the present invention, the process of selecting the predetermined number from the domain knowledge acquired by searching multiple languages ​​is not particularly limited. The so-called search scores may be compared as they are, or normalization or weighting by language may be performed. Furthermore, a ranking learning technique may be used.

[0074] In the second embodiment of the present invention, the main language of the domain knowledge may be explicitly determined at the time of registration, or may be acquired by performing language estimation for each document to be registered.

[0075] (Specific example of question answering process in the second embodiment) Next, as a specific example of the question-answering process shown in the flowchart of FIG. 9, we will explain the case where question-answering process is performed for domains with domain IDs of 1 and 2 (where the main languages ​​of the domain knowledge are "Japanese" and "English", respectively) in the domain language management table 501 of FIG. 6 and the domain knowledge management table 502 of FIG. 7.

[0076] In step S401, the question and answer processing unit 302 receives the question "Tell me the address of the Osaka office." input by the user from the dialogue management processing unit 301.

[0077] In step S402, the question and answering processor 302 uses the language estimation processor 303 to estimate "English" as the language of the question sentence.

[0078] In step S901, in the second embodiment, the processes from step S403 to step S406 in the first embodiment are repeated for each of the main languages ​​of the multiple domain knowledge ("Japanese" with id 1, "English" with id 2).

[0079] First, if id is 1, in step S403, the question and answering processor 302 determines that the language estimated in step S402 is "English" and does not match the main language 601 of the domain knowledge, "Japanese," and moves the process to step S405.

[0080] In step S405, the question and answering processor 302 uses the translation processor 304 to translate the question into Japanese, which is the main language of the domain knowledge, and obtains the Japanese translation of the question, "Please tell me the address of the Osaka office."

[0081] In step S406, the question and answering processor 302 creates a Japanese search query from the question translated in step S405.

[0082] In step S902, the question and answering processor 302 uses the search processor 305 to obtain domain knowledge that is highly relevant to the search query from the domain knowledge management table 502 and temporarily store it.

[0083] Next, when id is 2, in step S403, the question and answering processor 302 determines that the language estimated in step S402 is "English", which matches the main language 602 of the domain knowledge, "English", and therefore moves the process to step S404.

[0084] In step S404, the question and answer processing unit 302 creates an English search query from the question sentence.

[0085] In step S902, the question and answering processor 302 uses the search processor 305 to obtain domain knowledge that is highly relevant to the search query from the domain knowledge management table 502 and temporarily store it.

[0086] When the search for all domain knowledge with ids 1 to 2 is completed, a process of selecting a prescribed number from the temporarily stored domain knowledge is performed in step S904. In the second embodiment of the present invention, the process of selecting a prescribed number from the domain knowledge acquired by searching in multiple languages ​​is not particularly limited.

[0087] The processing in steps S408 to S409 is the same as in the first embodiment.

[0088] As described above, in the second embodiment, even when the main language of the domain knowledge is multiple, it is possible to search for the domain knowledge by determining each time whether it matches the language of the question, and if it does not match, translating it into the main language. This makes it possible to acquire information that is closely related to the question across the board, regardless of the languages ​​of the question and domain knowledge, and to provide an answer that corresponds to the language of the question based on the acquired information.

[0089] Although the embodiments have been described above, the present invention can be embodied as, for example, a system, an apparatus, a method, a program, a recording medium, etc. Specifically, the present invention may be applied to a system made up of multiple devices, or may be applied to an apparatus made up of a single device.

[0090] Furthermore, the program of the present invention is a program that enables a computer to execute the processing method of the flowchart shown in Fig. 3, and the storage medium of the present invention stores a program that enables a computer to execute the processing method of Fig. 3. Note that the program of the present invention may be a program for each processing method of each device in Fig. 3.

[0091] As described above, it goes without saying that the object of the present invention can also be achieved by supplying a recording medium on which a program that realizes the functions of the above-mentioned embodiments is recorded to a system or device, and having the computer (or CPU or MPU) of that system or device read and execute the program stored on the recording medium.

[0092] In this case, the program itself read from the recording medium will realize the novel functions of the present invention, and the recording medium on which the program is recorded will constitute the present invention.

[0093] Examples of recording media for supplying the program include flexible disks, hard disks, optical disks, magneto-optical disks, CD-ROMs, CD-Rs, DVD-ROMs, magnetic tapes, non-volatile memory cards, ROMs, EEPROMs, and silicon disks.

[0094] Furthermore, it goes without saying that not only are the functions of the above-mentioned embodiments realized by the computer executing a program it has read, but also cases are included in which an OS (operating system) running on the computer performs some or all of the actual processing based on the instructions of the program, and the functions of the above-mentioned embodiments are realized through that processing.

[0095] Furthermore, it goes without saying that this also includes cases where a program read from a recording medium is written into a memory provided on a function expansion board inserted into a computer or a function expansion unit connected to the computer, and then a CPU or the like provided on the function expansion board or function expansion unit performs some or all of the actual processing based on the instructions of the program code, thereby realizing the functions of the above-mentioned embodiments.

[0096] Furthermore, the present invention may be applied to a system consisting of multiple devices, or to a device consisting of a single device. It goes without saying that the present invention can also be applied to a case where the present invention is achieved by supplying a program to a system or device. In this case, the system or device can enjoy the effects of the present invention by reading a recording medium containing a program for achieving the present invention into the system or device.

[0097] Furthermore, by downloading and reading a program for achieving the present invention from a server, database, etc. on a network using a communication program, the system or device can enjoy the effects of the present invention. Note that the present invention also includes configurations that combine the above-mentioned embodiments and their modified examples. [Explanation of symbols]

[0098] 100 Question Answering Device 110 User terminal 120 Network

Claims

1. a question acquisition means for acquiring a question; an estimation means for estimating a language related to the question acquired by the question acquisition means; a search means for searching for search target information using a search query created based on the question acquired by the question acquisition means; an output means for outputting an answer to the question in the language estimated by the estimation means, using the information searched by the search means and the acquired question; Equipped with The information processing device is characterized in that, when the language related to the question estimated by the estimation means is different from the language related to the search target information, the search means searches for the search target information using a search query created in the language related to the search target information based on the question.

2. a translation means for translating the query into a language related to the search target information, The information processing device according to claim 1, characterized in that, if the language of the question estimated by the estimation means matches the language of the search target information, the search means searches for the search target information using a search query created based on the question without translating the question using the translation means, and if they do not match, the search means searches for the search target information using a search query created based on the question translated by the translation means into the language of the search target information.

3. 3. The information processing apparatus according to claim 2, wherein the information searched by said search means is information highly related to the question acquired by said acquisition means.

4. 4. The information processing apparatus according to claim 3, wherein the output means outputs, in an identifiable manner, information used to generate a response from among the information retrieved by the retrieval means.

5. 5. The information processing apparatus according to claim 4, wherein the output means outputs information used to generate an answer from the information retrieved by the retrieval means in a manner that allows the information to be identified in a language related to the question.

6. 6. The information processing apparatus according to claim 5, wherein the question acquisition means is capable of acquiring a question by voice.

7. 7. The information processing apparatus according to claim 6, wherein said output means is capable of outputting answers by voice.

8. 2. The information processing apparatus according to claim 1, wherein the answer output by said output means is generated using a large-scale language model.

9. 9. The information processing apparatus according to claim 8, wherein the output means outputs an answer generated by inputting the question acquired by the question acquisition means and the information searched by the search means into the large-scale language model.

10. 2. The information processing apparatus according to claim 1, wherein the search target information and a language related to the search target information are stored in association with each other.

11. 11. The information processing apparatus according to claim 10, wherein the search target information is related to a plurality of languages.

12. The information processing device according to claim 11, characterized in that, when the search target information is in multiple languages, the search target information is searched for using search queries in multiple languages ​​created in the languages ​​related to the multiple search target information based on the question.

13. 2. The information processing apparatus according to claim 1, wherein the search target information is information specified by a URL that satisfies a predetermined condition.

14. a question acquisition step of acquiring a question; an estimation step of estimating a language related to the question acquired by the question acquisition means; a search step of searching for search target information using a search query created based on the question acquired by the question acquisition means; an output step of outputting an answer to the question in the language estimated by the estimation means, using the information searched by the search means and the acquired question; Equipped with A control method for an information processing device, characterized in that the search step, when the language related to the question estimated by the estimation means is different from the language related to the search target information, searches for the search target information using a search query created in the language related to the search target information based on the question.

15. A program for causing at least one computer to function as each of the means of the information processing system according to any one of claims 1 to 13.

Images