Information processing apparatus, information processing method, and recording medium

Abstract

To provide an information processing apparatus capable of modifying a content by reflecting user information.SOLUTION: In the information processing apparatus, acquiring means acquires reader information and content information. Prompt generating means generates a prompt based on the reader information and the content information. Content modifying means inputs the prompt into a large-scale language model and acquires output from the large-scale language model as a modified content. The information processing apparatus can optimize the content.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present disclosure relates to techniques for modifying content. [Background technology]

[0002] There are known filtering techniques that prevent the display of content that includes inappropriate content, such as violent content. For example, Patent Document 1 describes a technique for filtering video and / or audio media that includes undesirable content. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2021-175187 A Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional filtering techniques mainly perform filtering according to predefined rules, making it difficult to perform filtering flexibly in response to user requests.

[0005] One object of the present disclosure is to provide an information processing device that makes it possible to modify content by reflecting user information. [Means for solving the problem]

[0006] According to one aspect of the present disclosure, there is provided an information processing device, An acquisition means for acquiring reader information and content information; a prompt generating means for generating a prompt based on the reader information and the content information; content modification means for inputting the prompt into a large scale language model and obtaining an output from the large scale language model as modified content; Equipped with.

[0007] In another aspect of the present disclosure, an information processing method includes: Acquire reader information and content information, generating a prompt based on the reader information and the content information; The prompt is input to a large language model, and output from the large language model is obtained as modified content.

[0008] According to yet another aspect of the present disclosure, a recording medium includes: Acquire reader information and content information, generating a prompt based on the reader information and the content information; A program is recorded that causes a computer to input the prompt into a large scale language model and obtain output from the large scale language model as modified content. Effect of the Invention

[0009] According to the present disclosure, it becomes possible to modify content to reflect user information. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram showing an overall configuration of a content modification system. [Diagram 2] FIG. 2 is a block diagram showing a hardware configuration of the information processing device. [Diagram 3] FIG. 2 is a block diagram showing a hardware configuration of the terminal device. [Figure 4] FIG. 2 is a block diagram showing a functional configuration of the information processing device. [Diagram 5] An example of reader information is shown below. [Figure 6] Here is an example prompt: [Figure 7] The content before and after the revision is shown. [Figure 8] 13 is a flowchart of a content modification process. [Figure 9]1 shows the overall configuration of a content modification system 1a according to a modified example. [Figure 10] FIG. 11 is a block diagram showing a functional configuration of an information processing device according to a second embodiment. [Figure 11] Here is an example of a collection of documents: [Figure 12] 13 is a flowchart of a content modification process according to the second embodiment. [Figure 13] FIG. 11 is a block diagram showing a functional configuration of an information processing device according to a third embodiment. [Figure 14] FIG. 13 is a block diagram showing a functional configuration of an information processing apparatus according to a fourth embodiment. [Figure 15] 13 is a flowchart of a process performed by an information processing device according to a fourth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, preferred embodiments of the present disclosure will be described with reference to the drawings. First Embodiment [Overall configuration] 1 shows the overall configuration of a content modification system to which an information processing device according to the present disclosure is applied. The content modification system 1 includes an information processing device 10 and a terminal device 20. The information processing device 10 and the terminal device 20 can communicate with each other via wired or wireless communication. It is assumed that there are a plurality of terminal devices 20. It is also assumed that a user of the terminal device 20 has registered in advance to use the content modification system.

[0012] The terminal device 20 is operated by a user. The user can view web content, mobile content, and the like via the terminal device 20. In this embodiment, it is assumed that the user viewing the content is a minor (child). When the user accesses content, the terminal device 20 transmits information about the user and information about the content to the information processing device 10. The information about the content is, for example, the URL (Uniform Resource Locator) of the content.

[0013] The information processing device 10 modifies the content accessed by the user and transmits the modified content (hereinafter also referred to as "modified content") to the terminal device 20. Specifically, the information processing device 10 generates an instruction sentence (hereinafter also referred to as "prompt") for a generation AI (Artificial Intelligence) based on information about the user and information about the content. The generated prompt includes an instruction to modify the content and the content to be modified. The information processing device 10 inputs the generated prompt to the generation AI and obtains an answer to the prompt from the generation AI. In this embodiment, for example, a large-scale language model is used as the generation AI.

[0014] The information processing device 10 transmits the answer from the generation AI as modified content to the terminal device 20. The terminal device 20 displays the modified content received from the information processing device 10 on the display instead of the content accessed by the user.

[0015] In this way, the content modification system 1 modifies the content using information about the user, and therefore it is possible to provide modified content suited to each individual user.

[0016] [About large-scale language models] A large-scale language model on which this embodiment is based will be described. A large-scale language model (also simply referred to as a "language model") is a model that learns the relationships between words in a sentence and generates related strings related to a target string from the target string. By using a language model that has learned sentences and texts in various contexts, it is possible to generate related strings with appropriate content related to the target string. For example, a case in which a language model is used in question answering will be described. The language model accepts an input of a question such as "What kind of country is Japan?" as a target string. The language model generates a string such as "Japan is an island country in the northern hemisphere..." as an answer to the question.

[0017] The learning method of the language model is not particularly limited, but as an example, the language model may be one that is learned to output at least one sentence including an input string. As a specific example, the language model may be a GPT (Generative Pre-Training) that outputs a sentence including an input string by predicting a string that is likely to follow the input string. In addition, for example, T5 (Text-to-Text Transfer Transformer), BERT (Bidirectional Encoder Representations from Transformers), RoBERTa (Robustly optimized BERT approach), ELECTRA (Efficiently Learning an Encoder that Classifies Token Replacements Accurately), etc. are also language models.

[0018] Furthermore, the character string generated by the language model is not limited to a natural language. For example, the language model may output an artificial language (such as a program source code) for a character string input in a natural language. For example, the language model accepts an input question such as "How can I retrieve data containing a specific character string from a database?" as a target character string. The language model may output a program source code for performing database processing. Alternatively, the language model may output a natural language corresponding to a character string input in an artificial language.

[0019] Furthermore, the content generated by the language model is not limited to character strings. For example, the language model may generate image data, video data, audio data, or other data formats corresponding to the input character string. Furthermore, the data input to the language model is not limited to character strings. For example, the language model may receive audio data as input and output a character string based on the content of the audio data.

[0020] [Hardware configuration] (Information processing device) 2 is a block diagram showing a hardware configuration of the information processing device 10. As shown in the figure, the information processing device 10 includes a processor 11, an interface (IF) 12, a read only memory (ROM) 13, a random access memory (RAM) 14, and a storage device 15. Each component is connected to each other via a bus 18, for example.

[0021] The processor 11 is a computer such as a CPU (Central Processing Unit), and executes a prepared program to control the entire information processing device 10. Specifically, the processor 11 may be a CPU, a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), an MPU (Micro Processing Unit), an FPU (Floating Point number Processing Unit), a PPU (Physics Processing Unit), a TPU (Tensor Processing Unit), a quantum processor, a microcontroller, or a combination of these.

[0022] Moreover, the processor 11 loads a program stored in the ROM 13, the storage device 15, or the like, and executes each process coded in the program. The processor 11 also functions as a part or the whole of the information processing device 10. The processor 11 then executes a content modification process, which will be described later.

[0023] The IF 12 transmits and receives data to and from an external device. Specifically, the information processing device 10 receives information about a user and information about content from the terminal device 20 through the IF 12. The information processing device 10 also transmits modified content to the terminal device 20 through the IF 12.

[0024] The ROM 13 stores various programs executed by the processor 11. In addition, the RAM 14 is used as a working memory while the processor 11 is executing various processes.

[0025] The storage device 15 is a non-volatile, non-transitory storage device such as a disk-shaped recording medium or a semiconductor memory. The storage device 15 may be configured to be detachable from the information processing device 10. The storage device 15 records various programs executed by the processor 11. The storage device 15 may include a database for recording reader information, a database for recording revised content, and a document collection, which will be described later. The storage device 15 may also store a machine learning model, etc.

[0026] In addition to the above, the information processing device 10 may include a display device such as a liquid crystal display or a projector, and an input device such as a keyboard or a mouse. These display devices and input devices are used by, for example, an administrator of the information processing device 10 to perform necessary management.

[0027] (Terminal Device) 3 is a block diagram showing a hardware configuration of the terminal device 20. The terminal device 20 is, for example, a smartphone. As shown in the figure, the terminal device 20 includes a processor 21, an IF 22, a ROM 23, a RAM 24, an input unit 26, and a display unit 27. Each component is connected to each other via, for example, a bus 28.

[0028] The processor 21 is a computer such as a CPU, and executes a prepared program to control the entire terminal device 20. The processor 21 may be a GPU, an FPGA, a DSP, an ASIC, or the like.

[0029] The IF 22 transmits and receives data to and from an external device. Specifically, the terminal device 20 transmits information about a user and information about content to the information processing device 10 through the IF 22. In addition, the terminal device 20 receives modified content from the information processing device 10 through the IF 22.

[0030] The ROM 23 stores various programs executed by the processor 21. In addition, the RAM 24 is used as a working memory while the processor 21 is executing various processes.

[0031] The input unit 26 is, for example, an input device such as a touch panel, a microphone, etc. The display unit 27 is, for example, a display that performs display based on the control of the processor 21, a speaker that performs audio output, etc.

[0032] [Function configuration] 4 is a block diagram showing a functional configuration of the information processing device 10 according to the first embodiment. The information processing device 10 functionally includes a reader information acquisition unit 111, a content information acquisition unit 112, a prompt generation unit 113, a content correction unit 114, and a corrected content output unit 115.

[0033] The reader information acquisition unit 111 acquires information about the user from the terminal device 20. Then, based on the information about the user, the reader information acquisition unit 111 estimates information such as the user's "age group," "favorite characters and themes," "preferred content format," "past browsing history," "feedback and ratings," "reader's region and cultural background," and "requests from parents" (hereinafter also referred to as "reader information"). The reader information acquisition unit 111 outputs the reader information to the prompt generation unit 113.

[0034] In the above description, the information processing device 10 estimates the reader information, but instead, the terminal device 20 may estimate the reader information. In this case, the terminal device 20 transmits the reader information to the information processing device 10 instead of information related to the user.

[0035] The content information acquisition unit 112 acquires information on the content accessed by the user from the terminal device 20. The content information acquisition unit 112 acquires the relevant content based on the content information, and outputs the acquired content to the prompt generation unit 113. For example, when the content information is a URL, the content information acquisition unit 112 acquires the content corresponding to the linked URL, and outputs the acquired content to the prompt generation unit 113. The content information acquisition unit 112 outputs the content to the prompt generation unit 113.

[0036] The prompt generation unit 113 generates a prompt based on the reader information input from the reader information acquisition unit 111 and the content input from the content information acquisition unit 112. The prompt includes, for example, an "instruction unit" that provides the model with a task to be executed or an instruction, and a "context unit" that provides the model with information required to generate an answer. The "instruction unit" is generated based on the reader information. The prompt generation unit 113 generates an instruction such as "Please create a story for 8-year-olds based on the following content" based on the reader information. The "context unit" is generated so as to include the content. The prompt generation unit 113 outputs the generated prompt to the content modification unit 114.

[0037] The content correction unit 114 obtains a prompt from the prompt generation unit 113. The content correction unit 114 inputs the prompt to the generation AI and obtains an answer to the prompt from the generation AI. The generation AI corrects the content included in the input prompt in accordance with the instructions included in the prompt, and outputs the corrected content as an answer. The content correction unit 114 outputs the corrected content to the corrected content output unit 115. The corrected content output unit 115 outputs the corrected content to the terminal device 20.

[0038] In the above configuration, the reader information acquisition unit 111 and the content information acquisition unit 112 are examples of an acquisition means, the prompt generation unit 113 is an example of a prompt generation means, and the content correction unit 114 and the corrected content output unit 115 are examples of a content correction means.

[0039] [Example of reader information] Next, a description will be given of reader information estimated by the reader information acquisition unit 111. Fig. 5 shows examples of reader information corresponding to each of the items of "age group," "favorite characters and themes," "preferred content format," "past browsing history," "feedback and ratings," "regional and cultural background of the reader," and "requests from parents."

[0040] "Age group" refers to the age or age range of the user. In FIG. 5, age ranges such as "6-10 years old" and "11-15 years old" are given as examples of age groups. The reader information acquisition unit 111 can estimate the age group from the registration information at the time of user registration. If the age is not entered at the time of user registration, the reader information acquisition unit 111 can detect content in which the user is interested based on the application usage history, website browsing history, search history, etc., and estimate the age group from the target demographic of the content.

[0041] "Favorite characters and themes" are characters and content themes in which the user is particularly interested. In FIG. 5, "dragons, magic, and friendship" are given as examples of favorite characters and themes. The reader information acquisition unit 111 can estimate the user's favorite characters and themes based on, for example, the browsing history and search history of websites, favorite data, bookmark data, and the like.

[0042] "Preferred content format" is the format of content that the user prefers. In Fig. 5, examples of preferred content formats are "manga," "novels," and "animation." The reader information acquisition unit 111 can estimate the content format that the user prefers based on access data such as the time spent browsing a website, the number of times the website is viewed, and the number of content downloads.

[0043] "Past browsing history" is a list of content that the user has viewed in the past. In FIG. 5, examples of past browsing history include story titles such as "Magical Adventure" and "A Story of Friendship." The reader information acquisition unit 111 can create a list of content that the user has viewed in the past based on, for example, browser cookies, session data, and app usage history.

[0044] "Feedback and evaluation" is a user's opinion or evaluation of the content. In FIG. 5, examples of feedback and evaluation include "I want more adventure scenes" and "I like character A." The reader information acquisition unit 111 can estimate the feedback and evaluation based on, for example, comments, reviews, and evaluation scores input by users regarding the content.

[0045] "Region and cultural background" is information on the user's place of residence and cultural background. In FIG. 5, country names such as "Japan," "America," and "India" are given as examples of the reader's region and cultural background. The reader information acquisition unit 111 can identify the region in which the user lives from, for example, the IP address and GPS information of the terminal device 20. The reader information acquisition unit 111 can also estimate the user's cultural background from the language setting of the terminal device 20.

[0046] "Parental requests" are requests regarding content that parents want to allow or restrict for their children. In FIG. 5, examples of parental requests include "I'm allowed to read manga" and "Please avoid violent scenes." The reader information acquisition unit 111 can estimate parental requests based on, for example, setting information such as filtering and functional restrictions set in the terminal device 20, and survey responses and feedback from parents. The reader information acquisition unit 111 may also estimate parental requests by analyzing settings of access restrictions for specific content.

[0047] [Prompt Generation Method] Next, a method for generating a prompt by the prompt generation unit 113 will be described. The prompt generation unit 113 can generate a prompt by using a machine learning model. This machine learning model is a model generated by machine learning using, as learning data, a pair of reader information (age, favorite characters, themes, etc.) and a prompt created based on the reader information. The prompt generation unit 113 can generate an optimal prompt by inputting new reader information into the machine learning model.

[0048] Furthermore, the prompt generator 113 may generate the prompt according to predetermined rules, such as age-specific rules, such as "if the user is 10 years old or younger, add the phrase 'Children's story' to the prompt."

[0049] Furthermore, the prompt generation unit 113 may generate a prompt using an ontology, a knowledge graph, or the like. For example, the prompt generation unit 113 uses external knowledge to hierarchically organize knowledge and information related to the reader information and generate a knowledge graph. Then, the prompt generation unit 113 extracts information related to the reader information from the knowledge graph, and generates a prompt based on the reader information and the extracted information. For example, if the user's "favorite characters and themes" include "dragon," the prompt generation unit 113 extracts "fantasy," which is a superordinate concept of "dragon," from the knowledge graph, and generates a prompt such as "Please create a dragon adventure story in a fantasy world."

[0050] Furthermore, the prompt generation unit 113 may present the generated prompt to the user and modify the prompt based on the user's reaction or feedback. For example, the prompt generation unit 113 generates a prompt such as "Please create an adventure story" and presents it to the user. Then, if the user provides feedback such as "I would like to see more adventure stories that use magic," the prompt generation unit 113 modifies the prompt to "Please create an adventure story that uses magic."

[0051] [Example of generated prompt] Next, a description will be given of the prompts generated by the prompt generating unit 113. Fig. 6 shows examples of prompts generated based on information on each of the items "age group," "favorite characters and themes," "preferred content format," "past browsing history," "feedback and ratings," "regional and cultural background of the reader," and "requests from parents."

[0052] The "age group" is used to provide age-appropriate content, such as filtering out violent expressions. In the example of Fig. 6, the user's age is 8 years old, and the prompt generation unit 113 generates a prompt based on the user's age, saying, "Based on the content below, please create a story suitable for 8-year-olds. Please avoid violent scenes and difficult language."

[0053] "Favorite character or theme" is used to customize content according to the user's preferences. In the example of Fig. 6, the user's favorite character is a dragon, and the prompt generation unit 113 generates a prompt based on the user's favorite character, saying "Based on the following content, please create an adventure story with a dragon as the main character."

[0054] "Preferred content format" is used to change the content format to suit the user's preference. In the example of Fig. 6, the user's preferred content format is manga, and prompt generator 113 generates a prompt such as "Please rewrite this story in the format of a manga scenario" based on the user's preferred content format.

[0055] The "past browsing history" is used to analyze the user's interests and preferences and to recommend new content. In the example of Fig. 6, the past browsing history includes a story called "Magical Adventure", and the prompt generating unit 113 generates a prompt such as "Please create a sequel to the magical adventure you read last time" based on the past browsing history.

[0056] The "feedback and ratings" are used to improve the quality of the content. The "feedback and ratings" are also used to generate content that meets the needs of the user. In the example of FIG. 6, the user provides feedback that "I like character A," and the prompt generation unit 113 generates a prompt based on the feedback, saying, "Based on the feedback from readers, please create a story that puts character A in a more active role."

[0057] The "region and cultural background" is used to provide content that matches the country, region, and culture that the user resides in. In the example of Fig. 6, the user resides in Japan, and the prompt generation unit 113 generates a prompt, "Please create a story that incorporates Japanese culture and scenery, based on the content below," based on the country in which the user resides.

[0058] "Parental requests" are used to provide content that meets the parent's wishes. In the example of Fig. 6, the parental requests are "You can let me read manga" and "Please avoid violent scenes", and the prompt generation unit 113 generates a prompt, "Please create a story that does not include violent scenes in accordance with the parental requests," based on the parental requests.

[0059] In the above, prompt generator 113 generates a prompt based on one item, but it may also generate a prompt based on a combination of multiple items.

[0060] [Example of modified content] Next, an example of the modified content will be described. It is assumed that a user in the younger age group accesses, via the terminal device 20, content that includes violent expressions.

[0061] FIG. 7 shows content before and after correction. FIG. 7(A) shows the content before correction. The content before correction includes a story about "murder." FIG. 7(B) shows the content after correction by the information processing device 10. In the content after correction, the story about "murder" has been changed to a story about "treasure hunting." In this way, the information processing device 10 replaces undesirable expressions with other expressions and transmits them to the terminal device 20. This allows the user of the terminal device 20 to avoid coming into contact with harmful information.

[0062] [Content Correction Processing] Next, the content modification process will be described. Fig. 8 is a flowchart of the content modification process by the information processing device 10. This process is realized by the processor 11 shown in Fig. 2 executing a program prepared in advance and operating as each element shown in Fig. 4.

[0063] First, the reader information acquisition unit 111 acquires information about the user from the terminal device 20. Then, the reader information acquisition unit 111 estimates reader information based on the information about the user (step S11). The reader information acquisition unit 111 outputs the reader information to the prompt generation unit 113. Next, the content information acquisition unit 112 acquires information about the content accessed by the user from the terminal device 20. Then, the content information acquisition unit 112 acquires the corresponding content based on the content information (step S12). The content information acquisition unit 112 outputs the content to the prompt generation unit 113.

[0064] Next, the prompt generation unit 113 generates a prompt based on the reader information input from the reader information acquisition unit 111 and the content input from the content information acquisition unit 112 (step S13). The prompt generation unit 113 outputs the generated prompt to the content correction unit 114. Next, the content correction unit 114 inputs the prompt to the generation AI and obtains an answer to the prompt from the generation AI (step S14). The generation AI corrects the content included in the input prompt in accordance with the instructions included in the prompt, and outputs the corrected content as an answer. The content correction unit 114 outputs the corrected content to the corrected content output unit 115. Next, the corrected content output unit 115 outputs the corrected content to the terminal device 20 (step S15), and the process ends.

[0065] [Variations] Next, a modified example of the first embodiment will be described. The information processing device 10 of the present embodiment can be applied to adaptive parental control, etc. Adaptive parental control is a control function that appropriately controls the browsing of content on the Internet by children (minors) based on standards set by guardians.

[0066] Fig. 9 shows the overall configuration of a content modification system 1a in a modified example. Fig. 9 includes a content server that provides content, a terminal device 20a used by a subject (user), and a terminal device 20b used by a guardian of the subject. The content server includes an information processing device 10.

[0067] The guardian of the subject sets attributes of the subject and specific instructions via terminal device 20b and transmits the set contents to the content server. The subject connects to the content server via terminal device 20a and browses the content. At this time, the information processing device 10 included in the content server dynamically modifies the content browsed by the subject according to the set contents received from terminal device 20b.

[0068] For example, if a parent restricts access to content related to a specific keyword or theme, content containing that keyword or theme will not be displayed on the target user's terminal device 20a. In FIG. 9, based on the parent's restriction of "do not use violent expressions," the content server modifies violent expressions to other expressions and transmits them to the terminal device 20a. The content server may also change kanji to hiragana or generate illustrations. In FIG. 9, based on the parent's instruction of "write difficult kanji to hiragana," the content server adds furigana in parentheses next to the kanji and transmits them to the terminal device 20a. This function allows for flexible control of Internet content browsing in accordance with the values ​​of each family and the parent's wishes.

[0069] <Second embodiment> Next, a second embodiment will be described. The information processing device 10a of the second embodiment can generate text content based on image content such as comics. Note that the overall configuration and hardware configuration of the information processing device 10a are similar to those of the information processing device 10 of the first embodiment, and therefore the description will be omitted.

[0070] [Function configuration] Fig. 10 is a block diagram showing a functional configuration of an information processing device 10a according to the second embodiment. The information processing device 10a functionally includes a reader information acquisition unit 111, a content information acquisition unit 112, a prompt generation unit 113a, a content correction unit 114, a corrected content output unit 115, and a document set 116. The document set 116 is realized by the storage device 15 shown in Fig. 2. The reader information acquisition unit 111, the content information acquisition unit 112, the prompt generation unit 113a, the content correction unit 114, and the corrected content output unit 115 are configured by the processor 11 shown in Fig. 2.

[0071] The reader information acquisition unit 111, the content information acquisition unit 112, the content correction unit 114, and the corrected content output unit 115 have the same configuration and operate in the same manner as the information processing device 10 of the first embodiment, so their explanations are omitted.

[0072] The prompt generating unit 113 a acquires reader information from the reader information acquiring unit 111 , and acquires image content from the content information acquiring unit 112 .

[0073] The prompt generation unit 113a generates sentences similar to the image content (hereinafter also referred to as "similar sentences") using a similar sentence generation method described below. Then, the prompt generation unit 113a generates a prompt based on the reader information and the similar sentences. Note that the "instruction portion" of the prompt is generated based on the reader information, as in the first embodiment. Also, the "context portion" is generated so as to include the similar sentences. The prompt generation unit 113a outputs the generated prompt to the content correction unit 114.

[0074] [How to generate similar sentences] Next, a method for generating similar sentences by the prompt generation unit 113a will be described. The prompt generation unit 113a extracts keywords and descriptions related to the image content from the image content. Then, the prompt generation unit 113a searches for and extracts sentences similar to the keywords and descriptions to generate similar sentences.

[0075] (How to extract keywords and descriptions) First, the prompt generation unit 113a recognizes an object included in the image content using an image recognition technique. For example, the prompt generation unit 113a identifies a specific object or scene using an existing semantic segmentation model. Examples of the existing semantic segmentation model include U-Net, DeepLabV3, and PSPNet. The prompt generation unit 113a extracts keywords and descriptions related to the image content based on the recognized object or scene. For example, the prompt generation unit 113a extracts keywords such as "tree" and "house" based on the recognized object.

[0076] (Method of extracting similar sentences) Next, the prompt generation unit 113a searches for and extracts sentences similar to the keywords and explanatory sentences from the document collection 116. Fig. 11 is an example of the document collection 116. The document collection 116 is a document collection made up of a plurality of narrative sentences, etc., and includes the sentences themselves and sentence IDs that uniquely identify the sentences.

[0077] Specifically, the prompt generation unit 113a calculates the importance of words included in each sentence in the document set 116 using TF-IDF (Term Frequency-Inverse Document Frequency). The prompt generation unit 113a then compares the similarity between the important words in each sentence and the keyword, and extracts the sentence containing the most similar important word as a similar sentence. The prompt generation unit 113a uses, for example, cosine similarity as a method for calculating the similarity. The prompt generation unit 113a uses a vector space model to calculate the angle between the vectors of the important words in each sentence and the keyword, and obtains the similarity.

[0078] Furthermore, the prompt generator 113a may use a word embedding technique to represent words included in each sentence in the document set 116 as vectors and calculate the similarity between the words and keywords. The prompt generator 113a extracts sentences that include the most similar words as similar sentences. Examples of word embedding techniques include Word2Vec and GloVe.

[0079] Furthermore, the prompt generation unit 113a may represent each sentence or keyword in the document set 116 as a vector using a Transformer-based model and calculate the similarity between the vectors. The prompt generation unit 113a extracts a sentence that is most similar to a keyword as a similar sentence. Examples of Transformer-based models include BERT and RoBERTa. Furthermore, the prompt generation unit 113a may use LSI (Latent Semantic Indexing) to extract sentences that are semantically related to a keyword from the document set 116 as similar sentences.

[0080] [Content Correction Processing] Next, the content modification process of the second embodiment will be described. Fig. 12 is a flowchart of the content modification process by the information processing device 10a. This process is realized by the processor 11 shown in Fig. 2 executing a program prepared in advance and operating as each element shown in Fig. 10. Note that the processes of steps S21 to S22 and S25 to S27 are similar to the processes of steps S11 to S15 of the first embodiment shown in Fig. 8, and therefore description thereof will be omitted.

[0081] The prompt generation unit 113a determines whether the content acquired from the content information acquisition unit 112 is image content (step S23). If the content is image content (step S23: Yes), the prompt generation unit 113a generates similar sentences from the image content (step S24). Specifically, the prompt generation unit 113a generates similar sentences by searching for and extracting sentences similar to keywords and descriptions related to the image content. On the other hand, if the content is not image content (step S23: No), the process proceeds to step S25.

[0082] As described above, the information processing device 10a of the second embodiment can generate text content based on image content such as comics, etc. This makes it possible to meet the needs of parents who say, "Comics are not good, but novels are good."

[0083] <Third embodiment> Next, a third embodiment will be described. The information processing device 10b of the third embodiment generates modified content based on the user's speech and outputs the modified content as audio. Note that the overall configuration and hardware configuration of the information processing device 10b are the same as those of the information processing device 10 of the first embodiment, and therefore the description will be omitted.

[0084] [Function configuration] 13 is a block diagram showing the functional configuration of an information processing device 10b according to the third embodiment. The information processing device 10b functionally comprises a reader information acquisition unit 111, a content information acquisition unit 112b, a prompt generation unit 113b, a content correction unit 114b, and a corrected content output unit 115b. The reader information acquisition unit 111 has the same configuration as the information processing device 10 of the first embodiment and operates in the same manner, so a description thereof will be omitted.

[0085] The terminal device 20 collects the user's speech and transmits it to the information processing device 10b. For example, the user speaks requests such as "play the XX video" or "read the XX story" to the terminal device 20. The terminal device 20 transmits the voice data of the voice spoken by the user to the information processing device 10b.

[0086] The user's voice data received from the terminal device 20 is input to the content information acquisition unit 112b. The content information acquisition unit 112b recognizes the voice of the voice data, for example, by using a voice recognition model. Then, the content information acquisition unit 112b searches for content from a website or the like based on the recognized voice, and acquires the corresponding content. The content information acquisition unit 112b outputs the acquired content to the prompt generation unit 113b.

[0087] The prompt generating unit 113b generates a prompt based on the reader information and the content. The "instruction portion" of the prompt is generated based on the reader information, as in the first embodiment. The "context portion" is generated to include the content. If the content is video content, the prompt generating unit 113b may include the audio file of the video as it is in the context. Alternatively, the prompt generating unit 113b may use a voice recognition model to transcribe the audio of the video and include it in the context of the prompt. The voice recognition model is generated, for example, by having a neural network learn a combination (dataset) of the speaker's voice and the corresponding text information. The prompt generating unit 113b outputs the generated prompt to the content modifying unit 114b.

[0088] The content correction unit 114b obtains a prompt from the prompt generation unit 113b. The content correction unit 114b inputs the prompt to the generation AI and obtains an answer to the prompt from the generation AI. The content correction unit 114b outputs the corrected content to the corrected content output unit 115b. The corrected content output unit 115b outputs the corrected content to the terminal device 20 by voice.

[0089] Furthermore, the content modification unit 114b can generate a response based on the user's reaction to the modified content. The terminal device 20 transmits voice data (hereinafter also referred to as "reaction information") indicating the user's reaction to the modified content, such as "What's the punch line?" or "What does XX mean?" to the information processing device 10b. The reaction information received from the terminal device 20 is input to the content modification unit 114b. The content modification unit 114b inputs the reaction information to the generation AI and obtains a response to the reaction information from the generation AI. Then, the content modification unit 114b outputs the response to the reaction information to the modified content output unit 115b. The modified content output unit 115b outputs the response to the reaction information by voice to the terminal device 20.

[0090] As described above, the information processing device 10b of the third embodiment can provide modified content in an interactive manner with the user. This allows the information processing device 10b to provide modified content to a young user who cannot operate a terminal device by hand. For example, the information processing device 10b can direct the user's attention to the content while the user's parents are doing housework. Furthermore, in the field of medical health care, the information processing device 10b can provide modified content to a patient who is required to rest.

[0091] [Variations] Next, modified examples of the present embodiment will be described. The following modified examples may be applied to the above embodiment in any combination.

[0092] (Variation 1) The information processing device 10b according to the third embodiment may add an instruction to the prompt to output clean language aloud. Clean language is, for example, language that does not include slang. The content correction unit 114b adds an instruction such as "Please write in clean Japanese" to the "instruction section" of the prompt. This allows the user to hear content that uses clean language, so that the user's parents can allow the user to use the content without feeling guilty.

[0093] (Variation 2) The information processing device 10b according to the third embodiment may make the edited content public to other users. The information processing device 10b may make all the edited content it has created public, or may make only good quality edited content public. Good quality edited content is edited content that the user is obsessed with. For example, the information processing device 10b accumulates edited content uploaded by the user's parent as edited content that the user is obsessed with. Then, the information processing device 10b makes the accumulated edited content public to other users. This allows the user to share the edited content with other users.

[0094] Furthermore, the information processing device 10b may generate new content using the edited content with the highest number of plays from among the published edited content. The information processing device 10b selects multiple edited contents based on the number of plays of the edited content. Then, the information processing device 10b generates new edited content by including a combination of the selected multiple edited contents in the context part of the prompt. This increases the possibility of generating better content.

[0095] In the above, the information processing device 10b generates new content based on the edited content that has been played many times, but instead, the information processing device 10b may generate new content based on edited content selected by the user or his / her parent. Specifically, the user or his / her parent selects multiple edited contents from among the edited contents that have been made public. Then, the information processing device 10b generates new edited content by including a combination of the selected multiple edited contents in the context part of the prompt. This increases the possibility that better content that matches the preferences of the user or his / her parent will be generated.

[0096] <Fourth embodiment> 14 is a block diagram showing the functional configuration of an information processing apparatus according to the fourth embodiment. The information processing apparatus 400 includes an acquisition unit 401, a prompt generation unit 402, and a content modification unit 403.

[0097] 15 is a flowchart of processing by the information processing device of the fourth embodiment. The acquisition means 401 acquires reader information and content information (step S401). The prompt generation means 402 generates a prompt based on the reader information and the content information (step S402). The content correction means 403 inputs the prompt into a large-scale language model and acquires the output from the large-scale language model as corrected content (step S403).

[0098] According to the information processing device 400 of the fourth embodiment, it is possible to modify the content by reflecting the user's information. Furthermore, the information processing device 400 can optimize the content.

[0099] A part or all of the above-described embodiments can be described as, but is not limited to, the following supplementary notes.

[0100] (Appendix 1) An acquisition means for acquiring reader information and content information; a prompt generating means for generating a prompt based on the reader information and the content information; content modification means for inputting the prompt into a large scale language model and obtaining an output from the large scale language model as modified content; An information processing device comprising:

[0101] (Appendix 2) The information processing device according to claim 1, wherein the content information is text content.

[0102] (Appendix 3) the content information is image content, The information processing device according to claim 1, wherein the prompt generation means generates a similar sentence that is similar to the image content based on the image content, and generates a prompt based on the reader information and the similar sentence.

[0103] (Appendix 4) A text storage means for storing a plurality of texts, The information processing device of claim 3, wherein the prompt generation means uses image recognition technology to extract keywords or explanatory sentences related to the image content, and generates similar sentences by extracting similar sentences to the keywords or explanatory sentences from the sentence storage means.

[0104] (Appendix 5) an output means for outputting the modified content to a user's terminal device; 3. The information processing device according to claim 2, wherein the output means outputs the edited content as audio.

[0105] (Appendix 6) A modified content storage means for storing the modified content is provided, The modified content storage means stores modified content uploaded by a user; 6. The information processing device according to claim 5, wherein the prompt generating means generates a prompt based on the reader information and any combination of the modified content stored in the modified content storage means.

[0106] (Appendix 7) have a publishing means for publishing the modified content to other users; 7. The information processing device according to claim 6, wherein the publishing means publishes the modified content stored in the modified content storage means.

[0107] (Appendix 8) The information processing device described in Appendix 7, wherein the prompt generation means generates a prompt based on the reader information and any combination of published revised content that has been played a predetermined threshold or more.

[0108] (Appendix 9) Acquire reader information and content information, generating a prompt based on the reader information and the content information; The method further comprises inputting the prompt into a large scale language model and obtaining output from the large scale language model as modified content.

[0109] (Appendix 10) Acquire reader information and content information, generating a prompt based on the reader information and the content information; A recording medium having a program recorded thereon for causing a computer to execute a process of inputting the prompt into a large-scale language model and obtaining an output from the large-scale language model as modified content.

[0110] Although the present disclosure has been described above with reference to the embodiments and examples, the present disclosure is not limited to the above-mentioned embodiments and examples. Various modifications that can be understood by a person skilled in the art can be made to the configuration and details of the present disclosure within the scope of the present disclosure. [Explanation of symbols]

[0111] 10. Information processing device 111 Reader Information Acquisition Department 112 Content information acquisition unit 113 Prompt Generation Unit 114 Content Modification Department 115 Modified Content Output Unit 116 Document Collection

Claims

1. An acquisition means for acquiring reader information and content information; a prompt generating means for generating a prompt based on the reader information and the content information; content modification means for inputting the prompt into a large scale language model and obtaining an output from the large scale language model as modified content; An information processing device comprising:

2. The information processing apparatus according to claim 1 , wherein the content information is text content.

3. the content information is image content, The information processing apparatus according to claim 1 , wherein the prompt generating means generates a similar sentence that is similar to the image content based on the image content, and generates a prompt based on the reader information and the similar sentence.

4. A text storage means for storing a plurality of texts, The information processing device according to claim 3, wherein the prompt generation means generates similar sentences by extracting keywords or explanatory sentences related to the image content using image recognition technology and extracting similar sentences similar to the keywords or explanatory sentences from the sentence storage means.

5. an output means for outputting the modified content to a user's terminal device; The information processing apparatus according to claim 2 , wherein the output means outputs the edited content as audio.

6. A modified content storage means for storing the modified content is provided, The modified content storage means stores modified content uploaded by a user; 6. The information processing apparatus according to claim 5, wherein the prompt generating means generates a prompt based on an arbitrary combination of the reader information and the modified content stored in the modified content storage means.

7. A publishing means for publishing the modified content to other users; The information processing apparatus according to claim 6 , wherein the publishing means publishes the revised content stored in the revised content storage means.

8. The information processing apparatus according to claim 7 , wherein the prompt generating means generates a prompt based on the reader information and an arbitrary combination of the revised content that has been published and has been played a predetermined threshold or more in number.

9. Acquire reader information and content information, generating a prompt based on the reader information and the content information; The method further comprises inputting the prompt into a large scale language model and obtaining output from the large scale language model as modified content.

10. Acquire reader information and content information, generating a prompt based on the reader information and the content information; A recording medium having recorded thereon a program for causing a computer to execute a process of inputting the prompt into a large-scale language model and obtaining an output from the large-scale language model as modified content.

Images