Information processing device, information processing method

Abstract

To provide technology for presenting appropriate answers to questions. [Solution] Based on the input question, textual information relating to the answer to the question is obtained using a large-scale language model, and based on this textual information, it is decided whether or not to present non-textual information. If it is decided to present non-textual information, both the textual information and the non-textual information are transmitted, and if it is decided not to present non-textual information, only the textual information is transmitted.

Description

Technical Field

[0001] The present invention relates to a technique for presenting an answer to a question.

Background Art

[0002] A conventional technique is known in which a prompt created based on a user's question and documents related to the question is input into a large language model to generate an answer text and present it to the user (for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Non-Patent Documents

[0004]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, in the conventional technique, when information important as an answer cannot be represented only by a character string, the user cannot determine whether the answer is appropriate based solely on the presented answer text, which may take time and effort to investigate additional judgment materials. The present invention provides a technique for presenting an appropriate answer to a question.

Means for Solving the Problems

[0006] One aspect of the present invention is characterized by comprising: acquisition means for acquiring character information relating to an answer corresponding to an input question; determination means for determining whether or not to present non-character information based on the question and / or the character information; and transmission means for transmitting the character information and the non-character information if it is determined that the non-character information should be presented, and transmitting the character information if it is determined that the non-character information should not be presented. [Effects of the Invention]

[0007] According to the present invention, it is possible to provide appropriate answers to questions. [Brief explanation of the drawing]

[0008] [Figure 1] A diagram illustrating the system's usage scenarios. [Figure 2] (a) is a block diagram showing an example of the hardware configuration of tablet 3, and (b) is a block diagram showing an example of the hardware configuration of question answering system 4. [Figure 3] A block diagram showing example functional configurations for both Tablet 3 and Question Answering System 4. [Figure 4] A flowchart illustrating the processes performed by tablet 3 and question answering system 4 to create and present answers to questions entered by the user. [Figure 5] A diagram showing an example of the screen display. [Figure 6] A flowchart showing the details of the process in step S404. [Figure 7] A diagram showing an example of the screen display. [Figure 8] A diagram showing an example of the screen display. [Figure 9] A diagram showing an example of the screen display. [Figure 10] A diagram showing an example of the screen display. [Modes for carrying out the invention]

[0009] The embodiments will be described in detail below with reference to the attached drawings. Note that the following embodiments do not limit the invention to the claims. While the embodiments describe multiple features, not all of these features are essential to the invention, and the features may be combined in any way. Furthermore, in the attached drawings, the same or similar configurations are given the same reference numerals, and redundant descriptions are omitted.

[0010] [First Embodiment] First, we will explain the usage scenario of the system according to this embodiment using Figure 1. In the usage scenario in Figure 1, a problem has occurred with printer 1. The problem with printer 1 is, for example, an image defect such as smudges on the leading edge of the printed image. Service technician 2 (user), who has been called to repair printer 1, uses tablet 3, an example of an information processing device, to investigate how to solve the problem. First, service technician 2 operates tablet 3 to input a question, and after inputting the question, instructs it to send the question. When instructed to send the question, tablet 3 transmits the input question to question answering system 4 via a wireless network. Question answering system 4 creates an answer corresponding to the question received from tablet 3 and transmits the created answer to tablet 3. When tablet 3 receives the answer transmitted from question answering system 4, it presents the received answer to service technician 2. Based on the presented answer, service technician 2 identifies how to solve the problem with printer 1.

[0011] Next, an example of the hardware configuration of Tablet 3 will be explained using the block diagram in Figure 2(a). Note that the hardware configuration example shown in Figure 2(a) is just one example of a hardware configuration applicable to Tablet 3, and can be modified or changed as appropriate.

[0012] The CPU 101 executes various types of processing using computer programs and data stored in the RAM 105. Thereby, the CPU 101 controls the overall operation of the tablet 3, and also executes or controls various types of processing described as the processing performed by the tablet 3.

[0013] The ROM 102 stores setting data of the tablet 3, computer programs and data related to startup of the tablet 3, computer programs and data related to basic operations of the tablet 3, and the like.

[0014] The input device 103 is a user interface such as a keyboard, a mouse, or a touch panel, and can input various instructions and information to the tablet 3 through user operations.

[0015] The presentation device 104 is an interface for presenting the processing results of the tablet 3 to the user. For example, the presentation device 104 has a liquid crystal screen or a touch panel screen, and can display the processing results by the CPU 101 as images or text. Also for example, the presentation device 104 has a speaker, and can output the processing results by the CPU 101 as sound.

[0016] The RAM 105 has an area for storing computer programs and data loaded from the ROM 102 or the storage device 106, and an area for storing computer programs and data received from an external device via the communication interface 107. The RAM 105 also has a work area used when the CPU 101 executes various types of processing. In this way, the RAM 105 can appropriately provide various types of areas.

[0017] The storage device 106 is a non-volatile memory device such as an SSD (Solid State Drive) or HDD (Hard Disk Drive). The storage device 106 stores the OS, computer programs and data for the CPU 101 to execute or control various processes described as processes performed by the tablet 3.

[0018] The communication interface 107 is an interface that allows the tablet 3 to communicate data with external devices via a network such as a LAN or the internet. The network may be a wired network or a wireless network. Furthermore, the type of network is not limited to a specific type such as USB or serial communication.

[0019] The CPU 101, ROM 102, input device 103, display device 104, RAM 105, storage device 106, and communication interface 107 are all connected to the system bus 108.

[0020] Next, an example of the hardware configuration of the question answering system 4 will be explained using the block diagram in Figure 2(b). Note that the following description will focus on the case where the question answering system 4 is implemented on a single computer device; however, it is not limited to this, and the question answering system 4 may also be implemented using multiple computer devices. Furthermore, the hardware configuration example shown in Figure 2(b) is just one example of a hardware configuration applicable to the question answering system 4, and can be modified or changed as appropriate.

[0021] The CPU 201 executes various processes using computer programs and data stored in the RAM 205. In doing so, the CPU 201 controls the overall operation of the question answering system 4, and also executes or controls the various processes described as those performed by the question answering system 4.

[0022] ROM202 stores configuration data for the question answering system 4, computer programs and data related to the startup of the question answering system 4, computer programs and data related to the basic operation of the question answering system 4, and so on.

[0023] RAM 205 has areas for storing computer programs and data loaded from ROM 202 and storage device 206, and areas for storing computer programs and data received from external devices via the communication interface 207. RAM 205 also has a work area used by the CPU 201 when executing various processes. In this way, RAM 205 can provide various areas as appropriate.

[0024] The storage device 206 is a non-volatile memory device such as an SSD (Solid State Drive) or HDD (Hard Disk Drive). The storage device 206 stores computer programs and data that cause the CPU 201 to execute or control various processes described as being performed by the OS and the question answering system 4.

[0025] The communication interface 207 is an interface for the question answering system 4 to communicate data with external devices via a network such as a LAN or the Internet. The CPU 201, ROM 202, RAM 205, storage device 206, and communication interface 207 are all connected to the system bus 208.

[0026] Examples of the functional configurations in the tablet 3 and the question answering system 4 are shown in the block diagrams of Figure 3. In this embodiment, we will describe a case where each functional part of the tablet 3 shown in Figure 3 is implemented by software (computer program). In the following, each functional part of the tablet 3 in Figure 3 may be described as the main processing unit, but in reality, the function of the functional part is realized by the CPU 101 executing the computer program corresponding to the functional part. Similarly, in this embodiment, we will describe a case where each functional part of the question answering system 4 shown in Figure 3 is implemented by software (computer program). In the following, each functional part of the question answering system 4 in Figure 3 may be described as the main processing unit, but in reality, the function of the functional part is realized by the CPU 201 executing the computer program corresponding to the functional part.

[0027] The processes performed by the tablet 3 and the question answering system 4 to create and present answers to questions entered by the user will be explained according to the flowchart in Figure 4.

[0028] In step S401, the question acquisition unit 301 acquires the question entered in response to user operation. For example, the question acquisition unit 301 displays the screen 500 exemplified in Figure 5(a) on the display device 104. The user enters a question into the input box 501 using the input device 103, and once the question has been entered, the user uses the input device 103 to indicate button 502. When button 502 is indicated, the question acquisition unit 301 acquires the entered question. The question acquisition unit 301 then displays the acquired question as a user-side chat bubble 510a on the display device 104, as exemplified in Figure 5(b). The question acquisition unit 301 also transmits the acquired question to the question answering system 4 via the communication interface 107.

[0029] In step S402, the generation unit 302 acquires a question transmitted from the tablet 3. The generation unit 302 then uses the large-scale language model 307 stored in the storage device 206 to generate character information relating to the answer to the acquired question. For example, the method described in Non-Patent Document 1 above can be applied to generate the character information relating to the answer to the acquired question.

[0030] For example, the generation unit 302 uses the acquired question as a query to search the "database related to answers 306" stored in the storage device 206 for the address (source location) of the source file (e.g., an HTML file) corresponding to the question. The generation unit 302 then inputs the acquired question and the source file found in the database 306 into the large-scale language model 307 and performs calculations in the large-scale language model 307 to generate character information related to the answer corresponding to the question. The source file does not necessarily have to be input into the large-scale language model in its entirety; it is also possible to input only a portion of the source file (source location), such as sentences, paragraphs, or chapters that match the query through a string search. By reducing the amount of information to what is necessary for the large-scale language model to answer, it is possible to reduce wasted resources. The large-scale language model 307 is a model that has been pre-trained to output character information related to the answer corresponding to the question and the set of source files corresponding to that question, and various models such as machine learning models can be applied to such a model.

[0031] In step S403, the acquisition unit 303 uses a dictionary registered in the database 306 to obtain representative non-textual information associated with the text, paragraphs, chapters, etc., of the source location searched in step S402. In this embodiment, the case in which the non-textual information is an image will be described.

[0032] The dictionary represents the relationship between a source location in database 306 and its corresponding representative non-textual information. For example, the dictionary is a JSON file containing the addresses of source locations and their corresponding non-textual information in database 306 in key-value format, and is stored within database 306. The "representative non-textual information" corresponding to a source location is, for example, non-textual information associated with that source location (e.g., linked within the source file (html file)) that has a high semantic similarity to the title of the source file at that location. Such dictionaries are pre-created and registered in database 306. The database can also register product-related knowledge information (information unlikely to be learned when generating a large-scale language model), such as printer product manuals, as dictionaries.

[0033] Dictionary creation may be performed manually in response to user input, or it may be performed by the device. When dictionary creation is performed by the device, for example, the device uses a modal cospace model of natural language and non-textual information to obtain embedding representations of title texts and non-textual information, respectively. The device then calculates the cosine similarity between the obtained embedding representations as the similarity score mentioned above.

[0034] In step S404, the determination unit 304 determines the manner in which the answer to the question will be presented. In this embodiment, the determination unit 304 determines the manner in which the answer to the user's question will be presented, including the manner in which non-textual information will be presented, with the view that the content of the answer to be presented in the subsequent processing will be easy for the user to understand. Details of the processing in step S404 will be described later.

[0035] In step S405, the presentation unit 305 causes the presentation device 104 to present the answer to the question according to the presentation mode determined by the decision unit 304. Details of the process in step S405 will be described later.

[0036] In step S406, the CPU 101 determines whether the processing termination conditions have been met. Various conditions can be applied to the processing termination conditions. For example, the CPU 101 determines that the processing termination conditions have been met when the user inputs a processing termination instruction using the input device 103.

[0037] As a result of this judgment, if the termination conditions for the process are met, the process according to the flowchart in Figure 4 will end. If the termination conditions for the process are not met, the process will proceed to step S401.

[0038] Next, the details of the process in step S404 described above will be explained according to the flowchart in Figure 6. In step S601, the decision unit 304 creates a text prompt. Specifically, the decision unit 304 generates a text prompt by concatenating a pre-set template string, the question sent from tablet 3, and the "character information related to the answer to the question" generated by the generation unit 302. Below, as an example, we will explain the case where the template string is "In the following question and answer, please answer Yes or No whether the focus is on the appearance of the object."

[0039] Note that the template string is not limited to this. For example, the determination unit 304 may obtain a string representing non-textual information acquired by the acquisition unit 303 and update the template string by inserting the acquired string into the template string. For example, the template string could be set to "In the following question and answer, please answer Yes or No whether the focus is on the appearance of {non-textual information}," and the part corresponding to "{non-textual information}" could be replaced with the "string representing non-textual information" acquired by the acquisition unit 303. Note that methods for obtaining the string representing non-textual information include using the alternative text of the non-textual information pasted into the source HTML file, or using a model that captions images, which are non-textual information, into strings.

[0040] In step S602, the determination unit 304 inputs the text prompt generated in step S601 into the large-scale language model 307 and performs calculations on the large-scale language model 307 to obtain "Yes" or "No" as the answer corresponding to the text prompt. The determination unit 304 may use a language model different from the large-scale language model 307 to obtain the answer corresponding to the text prompt.

[0041] In step S603, the decision unit 304 determines whether the answer corresponding to the text prompt is "Yes" or "No". As a result of this determination, if the answer corresponding to the text prompt is "Yes", the process proceeds to step S604; if the answer corresponding to the text prompt is "No", the process proceeds to step S605.

[0042] In step S604, the decision unit 304 determines the mode of presentation of non-textual information to "present non-textual information". In step S605, the decision unit 304 determines the mode of presentation of non-textual information to "do not present non-textual information".

[0043] In step S606, the determination unit 304 determines the "presentation mode of the response" according to the presentation mode of the non-textual information determined in step S604 or step S605.

[0044] For example, if the decision unit 304 determines that the mode of presentation of non-textual information is "present non-textual information," it will determine that the "mode of presentation of the answer is "present an answer that includes non-textual information." On the other hand, if the decision unit 304 determines that the mode of presentation of non-textual information is "not to present non-textual information," it will determine that the "mode of presentation of the answer is "present an answer that does not include non-textual information."

[0045] Then, if the decision unit 304 determines that the "mode of presenting the answer" is "to present an answer that includes non-textual information," it transmits the textual information related to the answer generated in step S402 and the non-textual information acquired in step S403 to the tablet 3 via the communication interface 207.

[0046] On the other hand, if the decision unit 304 determines that the "mode of presenting the answer" is "to present an answer that does not contain non-textual information," it transmits the textual information related to the answer generated in step S402 to the tablet 3 via the communication interface 207.

[0047] Next, the response presented according to the presentation mode determined by the determination unit 304 (the response presented by the presentation unit 305 using the presentation device 104 in step S405) will be explained with reference to Figure 7.

[0048] Figure 7(a) shows an example of the display of the response screen shown on the presentation device 104 by the presentation unit 305 when the decision unit 304 decides that the "response presentation mode" is "present a response that includes non-textual information".

[0049] In this case, the display unit 305 receives the "textual information related to the answer" and "non-textual information" transmitted from the question answering system 4, generates an answer screen including the received "textual information related to the answer" and "non-textual information," and displays it on the display device 104.

[0050] The response screen in Figure 7(a) displays a user-side chat bubble 510a representing a question entered by the user, and a system-side chat bubble 520a representing the answer to that question. The system-side chat bubble 520a includes an introductory section 521a and an individual case section 522a. Each individual case included in the individual case section 522a includes text related to the source content described in the source file of the source location searched by the generation unit 302 from the database 306, and corresponding non-textual information. The textual information and images in the system-side chat bubble 520a correspond to "textual information related to the answer" and "non-textual information," respectively.

[0051] Each individual case in section 522a has a string of characters such as "[1]" or "[2]" at the end of the sentence, which is a link that takes you to the corresponding source in database 306. The user can use input device 103 to select the link and learn the details of each individual case.

[0052] Figure 7(b) shows an example of the display of the response screen shown on the presentation device 104 by the presentation unit 305 when the decision unit 304 decides that the "response presentation mode" is "present a response that does not contain non-textual information".

[0053] In this case, the display unit 305 receives the "textual information related to the answer" transmitted from the question answering system 4, generates an answer screen including the received "textual information related to the answer," and displays it on the display device 104.

[0054] The response screen in Figure 7(b) displays a user-side chat bubble 510b representing the question entered by the user, and a system-side chat bubble 520b representing the answer to that question. The system-side chat bubble 520b includes an introductory section 521b and an individual case section 522b. Each individual case included in the individual case section 522b contains text related to the source content described in the source file of the source location searched by the generation unit 302 from the database 306, but unlike Figure 7(a), it does not include corresponding non-textual information. The textual information in the system-side chat bubble 520b corresponds to "textual information related to the answer".

[0055] As shown in Figure 7(a), when a user's question and its answer focus on the appearance of an object, and there is a risk that important information cannot be expressed solely through text, the answer is presented with relevant non-textual information, such as an image. This allows the user to compare the appearance of the object important to the problem they are facing (in the example in Figure 7(a), a printed image with a dirty tip) with the images of individual cases in the answer, and determine which individual case matches their problem. Therefore, the decision is easier to make compared to when the answer is textual only. In other words, the user does not have to struggle to make a decision or research additional information, thus reducing the effort required from the user.

[0056] On the other hand, as shown in Figure 7(b), when the user's question and the answer do not focus on the appearance of an object, it is assumed that the important information in the answer can be sufficiently represented by text alone, and images, which are non-textual information related to the answer, are not included. This ensures that the answer provides sufficient information for the user to determine which individual case matches their problem, while improving the readability of the answer compared to when images are included as part of the answer, and ultimately reducing the effort required from the user.

[0057] Thus, according to this embodiment, even when important information for the answer cannot be represented solely by text, by presenting non-textual information that is important for the answer, the user can more easily understand the content of the answer, and consequently, the user's effort can be reduced.

[0058] (Modification 1 of the first embodiment) In the first embodiment, the determination unit 304 determined the format of the answer presentation based on the user's question and the textual information relating to the answer, but the method for determining the format of the answer presentation is not limited to a specific method.

[0059] For example, the decision unit 304 may determine the format of the answer presentation based on the user's question and a pre-configured template string. In this case, for example, in step S601, the decision unit 304 generates a text prompt by concatenating the pre-configured template string and the question sent from the tablet 3. Here, the template string is, for example, "In the following question, please answer Yes or No whether the focus is on the appearance of the object."

[0060] Furthermore, the determination unit 304 may determine the format of the response presentation based on the "textual information related to the response" generated by the generation unit 302 and a pre-set template string. In this case, for example, in step S601, a string is generated as a text prompt by concatenating the pre-set template string and the "textual information related to the response" generated by the generation unit 302. Here, the template string is, for example, "In the following response, please answer Yes or No whether the focus is on the appearance of the object."

[0061] If either the user's question or answer contains lengthy text, the format of the answer can be determined by using the shorter of the two (the question or the answer) and a pre-configured template string, as described above. This reduces the computational complexity of the processing performed by the decision unit 304 using the large-scale language model.

[0062] Thus, regardless of which method is used, it is possible to determine whether the important information in the answer can be represented solely by text, and to appropriately decide whether or not to include non-textual information in the way the answer is presented.

[0063] (Modification 2 of the first embodiment) In the first embodiment, the decision unit 304 used a large-scale language model to determine the manner in which the response is presented, but the method for determining the manner in which the response is presented is not limited to a specific method.

[0064] For example, instead of a large-scale language model, a pre-trained machine learning model could be used to determine the presentation style of the answer. Possible algorithms for the machine learning model to use include LSTM and BERT. In this case, the expected questions and answers are prepared as input, and the results of the presentation style determination are prepared as output, which are used to train the machine learning model.

[0065] Furthermore, it is also possible to determine whether the input text contains predetermined keywords. These predetermined keywords may include, for example, "image," "picture," or "photograph." In addition, the determination methods described above may be combined to perform the determination. By using these methods, the processing time can be reduced compared to processing using a large-scale language model.

[0066] In any of the above methods, the decision unit 304 can determine whether important information for the answer can be represented solely by text characters, and appropriately decide whether to include non-textual information in the format of the answer presentation.

[0067] (Modification 3 of the first embodiment) In the first embodiment, the decision unit 304 determined whether to present an image, which is non-textual information, as the answer based on whether the user's question and the answer focused on the appearance of an object, but it is not limited to this. In this modified example, the decision unit 304 determines whether the user's question and the answer focused on the fine details of an object, and determines the presentation mode by changing the resolution (display resolution) when displaying the image, which is non-textual information, as the answer. This makes it easier for the user to understand the content of the answer presented.

[0068] The determination unit 304 determines whether each of the text information relating to the user's question and answer contains predetermined keywords related to fine details. The predetermined keywords are, for example, "dot," "streak," and "fine." If at least one of the text information relating to the user's question and answer contains the predetermined keywords related to fine details, the determination unit 304 determines the presentation mode of the non-text information (the image presented as an answer) so that its display resolution is higher (for example, higher than the default display resolution). On the other hand, if neither the text information relating to the user's question nor the answer contains the predetermined keywords related to fine details, the determination unit 304 determines the presentation mode of the non-text information (the image presented as an answer) so that its display resolution is lower (for example, lower than the default display resolution).

[0069] In this manner, if at least one of the user's questions or the textual information related to the answers contains predetermined keywords related to fine details, then those fine details are considered important to the user's judgment. Accordingly, tablet 3 displays images accompanying each individual case presented as an answer at a higher resolution. This allows the user to focus on the fine details of the presented images, making it easier for the user to determine which individual case should be focused on, thus reducing the effort required.

[0070] On the other hand, if neither the user's question nor the textual information related to the answer contains specific keywords related to fine details, then it can be assumed that fine details are not very important to the user's judgment. Therefore, tablet 3 displays images accompanying each individual example presented in the answer at a relatively low resolution. This reduces the amount of data in the images presented as answers compared to when the images are high resolution, and consequently enables the presentation of answers with low latency, thereby reducing the user's waiting time for an answer.

[0071] (Modification 4 of the first embodiment) In the first embodiment, the acquisition unit 303 obtained representative non-textual information associated with the source location using a dictionary registered in the database 306, but the method for obtaining representative non-textual information is not limited to a specific acquisition method.

[0072] For example, the acquisition unit 303 may acquire the first non-textual information that appears from the beginning among the non-textual information associated with the source (for example, linked within the source file (html file)).

[0073] Furthermore, if the source location represents a portion of the document data in the database 306, the acquisition unit 303 may acquire the associated non-textual information within that source location, or it may acquire the non-textual information that is closest to the source location in terms of the number of characters or lines.

[0074] Furthermore, the acquisition unit 303 may calculate the similarity between the source location and the associated textual information and non-textual information, and acquire the non-textual information with the highest similarity. The textual information associated with the source location may include, for example, the textual information of the entire source file or the string of characters in the title of the document data containing the source location.

[0075] Furthermore, the acquisition unit 303 may calculate the similarity between the textual information and non-textual information related to the response, and obtain the non-textual information with the highest similarity. These similarities may be calculated using, for example, the cosine similarity between embedded representations converted using a multimodal model of language and non-textual information modal (e.g., images).

[0076] Furthermore, if there are headings for non-textual information, the acquisition unit 303 may calculate the similarity between the textual information related to the answer and the headings for each of the non-textual pieces of information, and acquire the non-textual information with the highest similarity. For example, this similarity may be calculated using the cosine similarity of the embedding representations obtained by transforming each piece of textual information using a language model.

[0077] Regardless of which method is used, the acquisition unit 303 can appropriately acquire non-textual information to be presented as an answer to a question, and obtain the same effects as described in the first embodiment.

[0078] (Modification 5 of the first embodiment) In the first embodiment, we described the case where non-textual information is an image, but non-textual information is not limited to images. Below, we will describe the case where non-textual information is audio. Below, we will describe the differences from the first embodiment.

[0079] For example, suppose in step S401, the question acquisition unit 301 acquires the question "An unusual noise is coming from the ABC unit. What is the cause and how can it be dealt with?". In this case, in step S402, the generation unit 302 acquires textual information related to the answer using the question and the source file corresponding to the question, in the same manner as in the first embodiment. Then, in step S403, the acquisition unit 303 acquires the audio (audio file) associated with the source location in the source file as non-textual information.

[0080] Then, the determination unit 304 generates a text prompt by performing the same processing as in step S601 above, using a template string related to audio (for example, "In the following question and answer, please answer Yes or No whether the focus is on audio.") as the template string.

[0081] In step S602, the decision unit 304 inputs the text prompt generated in step S601 into the large-scale language model 307 and performs calculations on the large-scale language model 307 to obtain "Yes" or "No" as the answer corresponding to the text prompt. In other words, the decision unit 304 determines from the text prompt whether the question and answer are focused on speech or not.

[0082] In step S405, the presentation unit 305 presents the answer to the question to the user via the presentation device 104 according to the presentation mode determined by the decision unit 304. Figure 8 shows an example of the answer screen displayed on the presentation device 104 by the presentation unit 305 when the decision unit 304 determines the "answer presentation mode" to be "present an answer that includes non-textual information".

[0083] In this case, the display unit 305 receives the "text information related to the answer" and "non-text information" transmitted from the question answering system 4, generates an answer screen including the received "text information related to the answer" and an interface for playing back the received "non-text information," and displays it on the display device 104.

[0084] The response screen in Figure 8 displays a user-side chat bubble 510c representing a question entered by the user, and a system-side chat bubble 520c representing the answer to that question. The system-side chat bubble 520c includes an introductory section 521c and an individual case section 522c. Each individual case included in the individual case section 522c includes text related to the source content described in the source file of the source location retrieved by the generation unit 302 from the database 306, and an interface for reproducing the corresponding non-textual information. The textual information and interface in the system-side chat bubble 520c correspond to "textual information related to the answer" and "interface for reproducing non-textual information," respectively.

[0085] When a user operates this interface using the input device 103 to input a command to play non-textual information, the presentation unit 305 plays the non-textual information and outputs the corresponding audio through an audio output device such as a speaker or headphones on the presentation device 104.

[0086] Thus, when the user's question and the answer are audio-focused, and there is a risk that important information in the answer cannot be expressed solely through text, the answer will be presented including the relevant audio.

[0087] Through the methods described above, when audio is important information in an answer, users can listen to the audio related to the answer, thus avoiding the need to struggle to make a decision or to research additional information, and ultimately reducing the effort required from the user.

[0088] (Modification 6 of the first embodiment) The following describes the case where non-textual information is a video. The following describes the differences from the first embodiment. For example, suppose in step S401, the question acquisition unit 301 acquires the question "How do you replace the ABC unit?". In this case, in step S402, the generation unit 302 acquires textual information related to the answer using the question and the source file corresponding to the question, in the same manner as in the first embodiment. Then, in step S403, the acquisition unit 303 acquires a video (video file) associated with the source location in the source file as non-textual information.

[0089] Then, the determination unit 304 uses a template string related to video (for example, "In the following question and answer, please answer Yes or No whether it would be better to have a video to support the answer.") as a template string to generate a text prompt in the same way as in step S601 above.

[0090] In step S602, the decision unit 304 inputs the text prompt generated in step S601 into the large-scale language model 307 and performs calculations on the large-scale language model 307 to obtain "Yes" or "No" as the answer corresponding to the text prompt. In other words, the decision unit 304 determines from the text prompt whether the question and answer are focused on the video.

[0091] In step S405, the presentation unit 305 presents the answer to the question to the user via the presentation device 104 according to the presentation mode determined by the decision unit 304. If the decision unit 304 determines that the "presentation mode of the answer" is "present an answer that includes non-textual information," the answer screen displayed on the presentation device 104 by the presentation unit 305 is the same as the answer screen in Figure 7(a), except that instead of an image, it displays a moving image and an interface for playing the moving image.

[0092] When a user uses the input device 103 to operate this interface and input a command to play a video, the presentation unit 305 plays the video and displays it on the answer screen of the presentation device 104.

[0093] In this way, when a user's question and its answer are focused on video, and there is a risk that important information cannot be expressed solely through text, the answer will include the relevant video. This reduces the user's effort by preventing them from having to struggle to make a decision or research additional information.

[0094] Furthermore, if the non-textual information is a video, the determination unit 304 may use the template string "In the following question and answer, please answer Yes or No if it is sufficient to have only the thumbnail image in the video that supports the answer." as the template string used to generate the text prompt. In this case, the acquisition unit 303 may acquire a thumbnail image of any one frame of the acquired video as non-textual information. By doing so, the answer can be conveyed concisely by displaying only the image, similar to the first embodiment.

[0095] Alternatively, the template string "In the following question and answer, please answer Yes or No if it is sufficient to have only the audio information in the video that supports the answer." may be used as a template string to create a text prompt. In this case, the acquisition unit 303 may acquire the audio in the video as non-text information. By doing so, similar to the modification 5 of the first embodiment, it is possible to present an answer that includes audio and convey the answer to the user concisely.

[0096] (Modification 7 of the first embodiment) In the first embodiment, the question acquisition unit 301 acquires questions entered by the user into the input box 501 on the screen 500 using the input device 103, but the method of acquiring questions is not limited to a specific method. For example, the question acquisition unit 301 may acquire the voice spoken by the user through the microphone provided in the input device 103, transcribe the acquired voice into text, and acquire the resulting string as the question. Even with this method, the same effects as in the first embodiment can be obtained.

[0097] (Modification 8 of the first embodiment) In the first embodiment, the generation unit 302 used the method described in Non-Patent Document 1 to generate textual information relating to the answer, but the method for generating textual information relating to the answer is not limited to a specific generation method.

[0098] For example, the generation unit 302 generates a text prompt using the user's question acquired by the question acquisition unit 301 (without using the character information in the database 306). The generation unit 302 then inputs the generated text prompt into a large-scale language model and performs calculations on the large-scale language model to generate the output corresponding to the text prompt as character information related to the answer.

[0099] In this case, the generation unit 302 has not acquired the source location information in the database 306 necessary for the acquisition unit 303 to acquire non-textual information in the first embodiment. Therefore, the acquisition unit 303 needs to separately search the database 306 based on the textual information related to the user's question and the answer generated by the generation unit 302, and acquire non-textual information related to the question and answer.

[0100] When textual information from database 306 is not required to answer user questions, using the above method can reduce the processing time for retrieving data from database 306 and suppress the decrease in answer accuracy caused by inputting unnecessary data from database 306 into the large-scale language model. This method can also achieve the same effects as the first embodiment.

[0101] [Second Embodiment] In the following embodiments and modifications, including this embodiment, the differences from the first embodiment will be described, and unless otherwise specified, they will be the same as the first embodiment.

[0102] In the first embodiment, the determination unit 304 determined the manner in which the answer would be presented based on the importance of non-textual information in the content of the answer to be presented. In this embodiment, the determination unit 304 determines the manner in which the answer would be presented, including the manner in which non-textual information would be presented, from the perspective of whether the answer to be presented is easy to read.

[0103] In this embodiment, the processes performed by the tablet 3 and the question answering system 4 to create and present answers to questions entered by the user are the same as in the first embodiment, except for the process in step S404 in the flowchart of Figure 4. The process in step S404 according to this embodiment will be described below.

[0104] The determination unit 304 counts the number of characters in the text information related to the answer generated by the generation unit 302. If the counted number of characters is less than a threshold, the determination unit 304 determines the presentation mode of the non-text information, which is the image presented as the answer, to be displayed at a larger display size (for example, a display size larger than the default display size). On the other hand, if the counted number of characters is equal to or greater than the threshold, the determination unit 304 determines the presentation mode of the non-text information, which is the image presented as the answer, to be displayed at a smaller display size (for example, a display size smaller than the default display size).

[0105] As a result, in step S405, the tablet 3 displays non-textual information in a display size appropriate to the presentation mode of the non-textual information. The response presented according to the presentation mode determined by the determination unit 304 (the response presented by the presentation unit 305 in step S405) will be explained with reference to Figure 9.

[0106] Figure 9(a) shows an example of the display of the response screen shown on the presentation device 104 by the presentation unit 305 when the decision unit 304 decides to "present an image, which is non-textual information, in a larger display size."

[0107] The response screen in Figure 9(a) displays a user-side chat bubble 510e representing a question entered by the user, and a system-side chat bubble 520e representing the answer to that question. The system-side chat bubble 520e includes an introductory section 521e and an individual case section 522e. Each individual case included in the individual case section 522e includes text related to the source content described in the source file of the source location retrieved by the generation unit 302 from the database 306, and corresponding non-textual information. The textual information and images in the system-side chat bubble 520e correspond to "textual information related to the answer" and "non-textual information," respectively.

[0108] In Figure 9(a), the presentation method for non-textual information is determined to be "presenting images, which are non-textual information, at a larger display size." Therefore, the presentation device 104 displays images, which are non-textual information, at a larger display size. As shown in Figure 9(a), when the number of characters in the answer is small, increasing the display size of the image makes both the text and the image of the answer easier to see.

[0109] Figure 9(b) shows an example of the display of the response screen shown on the presentation device 104 by the presentation unit 305 when the decision unit 304 decides to "present an image, which is non-textual information, in a smaller size."

[0110] The response screen in Figure 9(b) displays a user-side chat bubble 510f representing a question entered by the user, and a system-side chat bubble 520f representing the answer to that question. The system-side chat bubble 520f includes an introductory section 521f and an individual case section 522f. Each individual case included in the individual case section 522f includes text related to the source content described in the source file of the source location searched by the generation unit 302 from the database 306, and corresponding non-textual information. The textual information and images in the system-side chat bubble 520f correspond to "textual information related to the answer" and "non-textual information," respectively.

[0111] In Figure 9(b), the presentation mode for non-textual information is determined to be "presenting images, which are non-textual information, at a smaller display size." Therefore, the presentation device 104 displays images, which are non-textual information, at a smaller display size. As shown in Figure 9(b), when the number of characters in the answer is large, reducing the display size of the image improves the overall readability of the answer and maintains a state that is easy to see. If the number of characters in the answer is large and the display size of the image is large, the entire answer will extend far beyond the screen, making it difficult to see the entire answer, and consequently, it will take more effort for the user to scroll the screen to view and judge the answer. In this embodiment, this effort can be reduced.

[0112] As described above, according to the method of this embodiment, by changing the display size of the non-textual information, such as the image, that is presented along with the answer to the question, it is possible to present the answer in a way that is easy for the user to read.

[0113] (Modification 1 of the second embodiment) In this modified example, the determination unit 304 determines whether or not to present non-textual information as an answer, according to the number of individual cases in the individual case section 522 of the textual information of the answer generated by the generation unit 302.

[0114] For example, the decision unit 304 determines the manner in which non-textual information is presented, such that if the number of individual cases is greater than or equal to a threshold (e.g., 10), non-textual information is not presented as an answer, and if the number of individual cases is less than the threshold, non-textual information is presented.

[0115] Thus, when there are many individual cases, listing non-textual information alongside the textual information makes it difficult for the user to read, so the non-textual information is omitted. By presenting only textual information, the user's readability is maintained, thereby reducing the effort required for the user to make a judgment. However, this is not the only way to do this; if the number of individual cases exceeds a threshold (for example, 10 cases), non-textual information may be presented up to that threshold number.

[0116] (Modification 2 of the second embodiment) In the second embodiment, the display size of the image, which is non-textual information in the answer, was changed according to the number of characters in the answer to the question. However, the display size of the image, which is non-textual information presented as an answer, may be changed according to the number of non-textual pieces of information obtained.

[0117] For example, if the number of non-textual information acquired by the decision unit 304 is greater than or equal to a threshold (e.g., 4 items), the decision unit 304 decides that the presentation mode for the non-textual information is to "display the image that is non-textual information at a smaller display size." On the other hand, if the number of non-textual information acquired by the decision unit 304 is less than the threshold, the decision unit 304 decides that the presentation mode for the non-textual information is to "display the image that is non-textual information at a larger display size."

[0118] This means that if a large amount of non-textual information is collected, displaying it in a large size makes it difficult for the user to read. However, by presenting the non-textual information in a small size, the overall readability of the user's response can be maintained, and the effort required for the user to make a judgment can be reduced.

[0119] [Third Embodiment] In this embodiment, when there are multiple examples presented as answers, the manner in which the answers are presented, including the manner in which non-textual information is presented, is determined with respect to whether it is easy to distinguish between individual examples.

[0120] In this embodiment, the processes performed by the tablet 3 and the question answering system 4 to create and present answers to questions entered by the user are the same as in the first embodiment, except for the process in step S404 in the flowchart of Figure 4. The process in step S404 according to this embodiment will be described below.

[0121] The determination unit 304 obtains character information related to individual cases from the character information of the responses generated by the generation unit 302, and calculates the similarity between the character information of each individual case obtained. The method for calculating the similarity between the character information of each individual case is not limited to a specific method. For example, the determination unit 304 obtains the cosine similarity between embedded representations obtained by inputting the character information of each individual case into a language model and performing calculations on the language model.

[0122] The determination unit 304 then determines the presentation mode to present an image, which is non-textual information, as the answer if the highest similarity among the calculated similarities is equal to or greater than a threshold (for example, 0.8). On the other hand, if the highest similarity is less than the threshold, the determination unit 304 determines the presentation mode not to present an image, which is non-textual information, as the answer.

[0123] The response presented according to the presentation mode determined by the decision unit 304 (the response presented by the presentation unit 305 in step S405) will be explained using Figure 10. Figure 10(a) shows an example of the response screen displayed on the presentation device 104 by the presentation unit 305 when the decision unit 304 determines the "response presentation mode" to be "present a response that includes non-textual information".

[0124] The response screen in Figure 10(a) displays a user-side chat bubble 510g representing a question entered by the user, and a system-side chat bubble 520g representing the answer to that question. The system-side chat bubble 520g includes an introductory section 521g and an individual case section 522g. Each individual case included in the individual case section 522g includes text related to the source content described in the source file of the source location retrieved by the generation unit 302 from the database 306, and corresponding non-textual information. The textual information and images in the system-side chat bubble 520g correspond to "textual information related to the answer" and "non-textual information," respectively.

[0125] Figure 10(b) shows an example of the display of the response screen shown on the presentation device 104 by the presentation unit 305 when the decision unit 304 decides that the "response presentation mode" is "present a response that does not contain non-textual information".

[0126] The response screen in Figure 10(b) displays a user-side chat bubble 510h representing the question entered by the user, and a system-side chat bubble 520h representing the answer to that question. The system-side chat bubble 520h includes an introductory section 521h and an individual case section 522h. Each individual case included in the individual case section 522h contains text related to the source content described in the source file of the source location searched by the generation unit 302 from the database 306, but unlike Figure 10(a), it does not include corresponding non-textual information. The textual information in the system-side chat bubble 520h corresponds to "textual information related to the answer".

[0127] As shown in Figure 10(a), when the similarity between the textual information of each individual case is high, presenting an image, which is non-textual information, allows the user to identify which individual case is desired, even though it would be difficult to distinguish from the textual information alone. On the other hand, as shown in Figure 10(b), when the similarity between the textual information of each individual case is low, the user can identify which individual case is desired using only the textual information, without presenting an image. In this case, the overall readability of the response can be maintained by deliberately not presenting an image.

[0128] As described above, according to this embodiment, by deciding whether or not to present non-textual information as an answer based on the similarity of the textual information of individual answer cases, it is possible to present answers while maintaining the ease of distinguishing between individual answer cases.

[0129] (Modification 1 of the third embodiment) In this modified example, the size of the image, which is the non-textual information presented as the answer, is changed according to the similarity between the non-textual information. For example, the determination unit 304 determines the similarity between the non-textual information acquired by the acquisition unit 303. Whether the non-textual information is an image or audio, methods for determining the similarity between images or between audio are well known, and in this embodiment, the method for determining the similarity between non-textual information is not limited to a specific method.

[0130] The determination unit 304 then determines the presentation mode of the non-textual information, such as presenting the image, which is the non-textual information to be presented as the answer, at a larger display size (for example, a larger display size than the default display size) if the highest similarity among the determined similarities is equal to or greater than a threshold (for example, 0.8). On the other hand, the determination unit 304 determines the presentation mode of the non-textual information, such as presenting the image, which is the non-textual information to be presented as the answer, at a smaller display size (for example, a smaller display size than the default display size) if the highest similarity among the determined similarities is less than the threshold. As a result, in step S405, the tablet 3 displays the non-textual information at a display size corresponding to the presentation mode of the non-textual information.

[0131] Therefore, if images, which are non-textual information, have a high degree of similarity, users may have difficulty distinguishing them if the displayed image size is small. By increasing the size of the displayed image, user identification can be maintained.

[0132] [Fourth Embodiment] In the first to third embodiments, the decision unit 304 determined the manner in which the answer would be presented from different perspectives. In this embodiment, the decision unit 304 determines the manner in which the answer would be presented based on all the perspectives of the decision unit 304 in the first to third embodiments. The decision unit 304 uses a machine learning model to determine a judgment score (0 to 1) for each of the perspectives of the decision unit 304 in the first to third embodiments.

[0133] For example, the decision unit 304 inputs the likelihood output from the large-scale language model 307 as the likelihood of the response obtained in step S602 of the first embodiment into a machine learning model and obtains a score obtained by performing calculations on the machine learning model as a first judgment score corresponding to the perspective of the decision unit 304 of the first embodiment.

[0134] Furthermore, the determination unit 304 inputs the number of characters in the character information related to the answer generated by the generation unit 302 into a machine learning model and obtains a score obtained by performing calculations on the machine learning model as a second judgment score corresponding to the viewpoint of the determination unit 304 in the second embodiment.

[0135] Furthermore, the determination unit 304 inputs the maximum similarity into the machine learning model and obtains a score obtained by performing calculations on the machine learning model as a third judgment score corresponding to the viewpoint of the determination unit 304 in the third embodiment.

[0136] The decision unit 304 then calculates the overall decision score as the sum or weighted linear sum of the first decision score, the second decision score, and the third decision score. If the overall decision score is equal to or greater than a threshold (for example, 0.8), the decision unit 304 decides to present non-textual information as the answer; if the overall decision score is less than the threshold, it decides not to present non-textual information as the answer.

[0137] Furthermore, the decision unit 304 may generate a text prompt by concatenating a question from the user, character information relating to the answer generated by the generation unit 302, and a string indicating that the decision will be made using the various perspectives of the decision unit 304 in the first to third embodiments. In this case, the decision unit 304 inputs the text prompt into a large-scale language model and performs calculations on the large-scale language model to obtain a presentation mode, which is whether or not to present non-character information.

[0138] By using the above method, it is possible to appropriately determine whether or not to present non-textual information from all perspectives, including whether the content of the answer presented is easy for the user to understand, whether the answer presented is easy for the user to see, and whether the individual cases presented as answers are easy to distinguish from each other. In turn, this reduces the effort required for the user to determine whether the answer is what they want.

[0139] [Fifth Embodiment] The system usage scenarios described in the above embodiments and modifications are just examples of troubleshooting, and the system can be applied to other troubleshooting situations as well. Furthermore, the system's usage scenarios are not limited to troubleshooting.

[0140] Furthermore, the tablet 3 and the question answering system 4 may be located in a different location from the malfunctioning printer 1, for example, at the maintenance company's location. In other words, various placements of the tablet 3 and the question answering system 4 are possible depending on the application scenario of the system.

[0141] Furthermore, although the above embodiments and modifications describe a case where the tablet 3 and the question answering system 4 are separate devices, the invention is not limited to this. For example, a single information processing device may be configured that integrates the tablet 3 and the question answering system 4. In this case, the information processing device executes or controls the various processes (excluding communication processes between devices) described as being performed by the tablet 3 and the question answering system 4, respectively.

[0142] Furthermore, the above embodiments and modifications describe a case where the tablet 3 has a question acquisition unit 301 and a presentation unit 305, and the question answering system 4 has a generation unit 302, an acquisition unit 303, a decision unit 304, a database 306, and a large-scale language model 307. However, it is not limited to this case. For example, the tablet 3 may have a decision unit 304. Also, the database 306 and the large-scale language model 307 may be provided by the tablet 3. In addition, one or more of the functional units shown in Figure 3 may be implemented in hardware.

[0143] The numerical values, processing timing, processing order, processing entity, data (information) structure / acquisition method / destination / source / storage location, etc., used in the above embodiment are given as examples for the purpose of providing a concrete explanation, and are not intended to limit the scope to such examples.

[0144] Furthermore, some or all of the embodiments described above may be used in appropriate combinations. Alternatively, some or all of the embodiments described above may be used selectively.

[0145] (Other embodiments) The present invention can also be realized by supplying a program that implements one or more of the functions of the above-described embodiments to a system or device via a network or storage medium, and by having one or more processors in the computer of that system or device read and execute the program. It can also be realized by a circuit (e.g., an ASIC) that implements one or more functions.

[0146] The inventions described herein include the following information processing devices, information processing methods, and computer programs. (Item 1) An acquisition means for acquiring textual information related to the answer to an input question using a large-scale language model, A determination means for determining whether or not to present non-textual information based on the aforementioned textual information, If it is decided to present the non-textual information, the transmission means transmits the textual information and the non-textual information, and if it is decided not to present the non-textual information, the transmission means transmits the textual information. An information processing device characterized by comprising: (Item 2) The information processing device according to item 1, characterized in that the acquisition means acquires character information relating to the answer corresponding to the question based on the question and a dictionary registered in a database corresponding to the question. (Item 3) The information processing device according to item 1 or 2, characterized in that the determination means acquires the non-character information based on the question and the character information. (Item 4) The information processing device according to any one of items 1 to 3, characterized in that the determination means generates a text prompt including the question and / or the character information, and determines whether or not to present the non-character information based on the text prompt. (Item 5) The information processing device according to any one of items 1 to 3, characterized in that the determination means generates a text prompt whichever of the question and the text information has fewer characters, and determines whether or not to present the non-text information based on the text prompt. (Item 6) The information processing device according to any one of items 1 to 3, characterized in that the determination means determines whether or not to present the non-character information based on the number of individual cases in the character information. (Item 7) The information processing device according to any one of items 1 to 3, characterized in that the determination means determines whether or not to present the non-character information based on the degree of similarity between the character information of each individual case in the character information. (Item 8) The information processing apparatus according to any one of items 1 to 7, characterized in that the determination means determines the display resolution of the non-character information depending on whether the question and / or the character information includes keywords relating to fine details. (Item 9) The information processing apparatus according to any one of items 1 to 8, characterized in that the determination means determines the display size of the non-character information according to the number of characters in the character information. (Item 10) The information processing apparatus according to any one of items 1 to 8, characterized in that the determination means determines the display size of the non-character information based on the number of non-character information items. (Item 11) The information processing apparatus according to any one of items 1 to 8, characterized in that the determination means determines the display size of the non-character information based on the similarity between the non-character information. (Item 12) The information processing device according to any one of items 1 to 11, characterized in that the non-textual information is an image, audio, video, a thumbnail image of a single frame in a video, or audio in a video. (Item 13) An acquisition means for acquiring textual information related to the answer to an input question using a large-scale language model, A determination means for determining whether or not to present non-textual information based on the aforementioned textual information, If it is decided to present the non-textual information, the textual information and the non-textual information will be presented, and if it is decided not to present the non-textual information, the presentation means for presenting the textual information will be presented. An information processing device characterized by comprising: (Item 14) An information processing method performed by an information processing device, The acquisition means of the information processing device includes an acquisition step of acquiring character information relating to the answer corresponding to the input question using a large-scale language model, The decision means of the information processing device includes a decision step of determining whether or not to present non-character information based on the character information, A transmission step in which, if the transmission means of the information processing device determines to present the non-character information, it transmits the character information and the non-character information, and if it determines not to present the non-character information, it transmits the character information. An information processing method characterized by comprising: (Item 15) An information processing method performed by an information processing device, The acquisition means of the information processing device includes an acquisition step of acquiring character information relating to the answer corresponding to the input question using a large-scale language model, The decision means of the information processing device includes a decision step of determining whether or not to present non-character information based on the character information, If the presentation means of the information processing device determines to present the non-textual information, it presents the textual information and the non-textual information; if it determines not to present the non-textual information, it presents the textual information. An information processing method characterized by comprising: (Item 16) A computer program that causes a computer to function as one of the means of an information processing device described in any one of items 1 through 13.

[0147] The invention is not limited to the embodiments described above, and various modifications and variations are possible without departing from the spirit and scope of the invention. Accordingly, claims are attached to disclose the scope of the invention. [Explanation of Symbols]

[0148] 301: Question acquisition unit 302: Generation unit 303: Acquisition unit 304: Decision unit 305: Presentation unit 306: Database 307: Large-scale language model

Claims

1. An acquisition means for acquiring textual information related to the answer to an input question using a large-scale language model, A determination means for determining whether or not to present non-textual information based on the aforementioned textual information, If it is decided to present the non-textual information, the transmission means transmits the textual information and the non-textual information, and if it is decided not to present the non-textual information, the transmission means transmits the textual information. An information processing device characterized by comprising:

2. The information processing apparatus according to claim 1, wherein the acquisition means acquires character information relating to the answer corresponding to the question based on the question and a dictionary registered in a database corresponding to the question.

3. The information processing apparatus according to claim 1, wherein the determination means acquires the non-character information based on the question and the character information.

4. The information processing apparatus according to claim 1, characterized in that the determination means generates a text prompt including the question and / or the character information, and determines whether or not to present the non-character information based on the text prompt.

5. The information processing apparatus according to claim 1, characterized in that the determination means generates a text prompt whichever of the question and the character information has fewer characters, and determines whether or not to present the non-character information based on the text prompt.

6. The information processing apparatus according to claim 1, characterized in that the determination means determines whether or not to present the non-character information based on the number of individual cases in the character information.

7. The information processing apparatus according to claim 1, characterized in that the determination means determines whether or not to present the non-character information based on the degree of similarity between the character information of each individual case in the character information.

8. The information processing apparatus according to claim 1, wherein the determination means determines the display resolution of the non-character information depending on whether the question and / or the character information includes keywords relating to fine details.

9. The information processing apparatus according to claim 1, characterized in that the determination means determines the display size of the non-character information according to the number of characters in the character information.

10. The information processing apparatus according to claim 1, wherein the determination means determines the display size of the non-character information based on the number of non-character information items.

11. The information processing apparatus according to claim 1, wherein the determination means determines the display size of the non-character information based on the similarity between the non-character information.

12. The information processing apparatus according to claim 1, characterized in that the non-textual information is any of the following: an image, audio, a video, a thumbnail image of a single frame in a video, or audio in a video.

13. An acquisition means for acquiring textual information related to the answer to an input question using a large-scale language model, A determination means for determining whether or not to present non-textual information based on the aforementioned textual information, If it is decided to present the non-textual information, the textual information and the non-textual information will be presented, and if it is decided not to present the non-textual information, the presentation means for presenting the textual information will be presented. An information processing device characterized by comprising:

14. An information processing method performed by an information processing device, The acquisition means of the information processing device includes an acquisition step of acquiring character information relating to the answer corresponding to the input question using a large-scale language model, The decision means of the information processing device includes a decision step of determining whether or not to present non-character information based on the character information, A transmission step in which, if the transmission means of the information processing device determines to present the non-character information, it transmits the character information and the non-character information, and if it determines not to present the non-character information, it transmits the character information. An information processing method characterized by comprising:

15. An information processing method performed by an information processing device, The acquisition means of the information processing device includes an acquisition step of acquiring character information relating to the answer corresponding to the input question using a large-scale language model, The decision means of the information processing device includes a decision step of determining whether or not to present non-character information based on the character information, If the presentation means of the information processing device determines to present the non-textual information, it presents the textual information and the non-textual information; if it determines not to present the non-textual information, it presents the textual information. An information processing method characterized by comprising:

16. A computer program for causing a computer to function as one of the means of an information processing apparatus described in any one of claims 1 to 13.

Images