System

Abstract

A system is provided.SOLUTION: A system comprising: means for loading a generative language model; means for acquiring learning data from an information provider; means for generating fiction news by the generative language model based on the acquired learning data; means for inserting an advertisement into the generated fiction news; and means for publishing the fiction news including the inserted advertisement.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional news delivery systems are required to provide only information whose veracity is guaranteed. However, due to "hallucinations," a characteristic of generative language models, they can output false information that lacks validity. Taking advantage of this drawback, there are challenges to overcome in automatically generating and disseminating fictional news that has a sense of authenticity, and utilizing it as a new form of entertainment and information delivery. [Means for solving the problem]

[0005] In order to solve the above problems, the present invention provides the following means. A system is configured that includes a means for loading a generative language model, a means for acquiring training data from an information provider, and a means for generating fictional news using the generative language model based on the acquired training data. The system also includes a means for inserting advertisements into the generated fictional news and a means for publishing the fictional news containing the inserted advertisements. This makes it possible to effectively utilize the "hallucination" function of the generative language model and realize a new business model that links fictional news with advertising revenue.

[0006] A "generative language model" is an artificial intelligence model that is trained to generate a particular form of text based on a large dataset.

[0007] An "information provider" is a data holder that holds news articles and various data, and is the entity that provides the data used to train the generative language model.

[0008] "Training data" is the underlying dataset on which the generative language model is trained to generate fictional news.

[0009] "Fictional news" is fictional news that is automatically generated by a generative language model and has a sense of authenticity but does not exist in reality.

[0010] "Advertising" refers to prepared marketing messages or promotional content that are inserted into fictional news as a means of generating advertising revenue.

[0011] "Publication" refers to the act of distributing the generated fictional news and inserted advertisements over the Internet or other media in a form accessible to users.

[0012] A "system" is the set of hardware and software required to execute a series of processes, such as loading a generative language model, acquiring training data, generating fictional news, inserting advertisements, and publishing it. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14]FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0014] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0015] First, the terms used in the following description will be explained.

[0016] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0017] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0018] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0019] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0021] [First embodiment]

[0022] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0023] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0026] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0029] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0033] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0034] To implement the present invention, the following system configuration and processing must be included.

[0035] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model trained to generate text in a specific format based on a large dataset. By loading a pre-trained generative language model onto the server, it becomes possible to generate fictional news.

[0036] Next, the server obtains training data from information providers. Information providers are data holders that hold news articles and various data, and this data will be used to train the generative AI model. The server sends an HTTP request from a specific API endpoint to obtain the news data. This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0037] The server then generates fictional news using a generative language model based on the acquired training data. Fictional news is automatically generated by the generative language model and is fictional news that has a sense of authenticity but does not exist in reality. The server provides the training data as input to the generative language model to generate the fictional news.

[0038] Next, the server inserts advertisements into the generated fictional news. Advertisements are prepared marketing messages or promotional content that are inserted into the fictional news as a means of earning advertising revenue. The server obtains advertising data from an advertising network and randomly combines advertisements for each generated fictional news article to create the final content.

[0039] Finally, the server publishes the fictional news with the inserted advertisements on the website. Publishing is the act of distributing the generated fictional news and the inserted advertisements on the Internet or other media in a form that is accessible to users. This allows the advertisements to be displayed along with the fictional news when users visit the website, making it possible to earn advertising revenue.

[0040] Specific examples

[0041] The server loads the generative AI model "some-ai-model." For example, a generative language model is loaded from a hard disk or cloud storage and deployed in memory.

[0042] The server retrieves news data from "http: / / example.com / news-data." The retrieved data includes categories such as politics, economics, and sports.

[0043] Based on the acquired news data, the server generates fictional news stories such as "A mysterious event happened in New York City last night" or "New technology may change the future."

[0044] The server retrieves advertisements such as "Buy this product now!" from the advertising network and randomly inserts them into the generated fictional news.

[0045] The server publishes the fictional news containing the inserted advertisements on a website, so that the fictional news and advertisements can be viewed when a user accesses the website.

[0046] This invention makes it possible to effectively utilize the "hallucination" function of generative language models, thereby enabling the creation of fictional news as a new form of entertainment and information provision while simultaneously generating advertising revenue.

[0047] The processing flow will be explained below.

[0048] Step 1:

[0049] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name. This operation deploys the model in memory and prepares it for generating fictional news.

[0050] Step 2:

[0051] The server obtains training data from the information provider. Specifically, it sends an HTTP request to obtain data from the specified URL. For example, it obtains news data from "http: / / example.com / news-data." This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0052] Step 3:

[0053] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the fictional news output by the model is obtained. This fictional news has a sense of verisimilitude, but is fictional news that does not exist in reality.

[0054] Step 4:

[0055] The server inserts advertisements into the generated fictional news. Specifically, it obtains advertisement data from an advertising network and randomly inserts advertisements into each fictional news article. For example, it obtains an advertisement "Buy this product now!" from the advertising network and inserts it into the generated news article.

[0056] Step 5:

[0057] The server publishes the fictional news articles, interspersed with advertisements, to a website by sending a POST request to a publishing API endpoint and uploading each article to the website, where it can be accessed by users over the internet.

[0058] Step 6:

[0059] Users access the website and view the published fiction news and advertisements, thereby generating advertising revenue along with the viewing of the fiction news.

[0060] Example 1

[0061] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0062] Providing information quickly and generating revenue are important challenges in today's world. However, while there are existing systems that combine real news with advertising, there are no systems that generate fictional articles, insert advertisements into them, and publish them, creating new forms of entertainment and revenue opportunities. Therefore, there is a need for a system that uses generative artificial intelligence models to automatically generate fictional articles and effectively insert advertisements into them for publication.

[0063] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0064] In this invention, the server includes means for loading a generative artificial intelligence model, means for acquiring training data from an information provider, means for generating fictional articles using the generative artificial intelligence model based on the acquired training data, means for inserting advertisements into the generated fictional articles, and means for publishing the fictional articles including the inserted advertisements, thereby enabling the automatic generation of fictional articles and the effective insertion and publication of advertisements.

[0065] A "generative artificial intelligence model" is an artificial intelligence model that learns from large datasets and enables language generation.

[0066] An "information provider" refers to a data provider that has training data and provides the data necessary to train and generate a generative model.

[0067] "Training data" refers to large sets of text and other data used to train generative models.

[0068] "Fictional articles" are fictional news articles or article-style content generated by a generative model based on training data.

[0069] "Advertising" means marketing messages or promotional content that are inserted into and displayed alongside fictional articles.

[0070] "Publication" refers to the act of distributing the generated fictional article and inserted advertisements over the Internet or other media in a manner that makes it accessible to users.

[0071] To implement the present invention, the following system configuration and processes are involved: The components of the server, generative AI model, information provider, advertising network, and website work together.

[0072] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model that enables language generation based on a large dataset and is specialized for generating news articles. The server downloads this model from cloud storage, expands it into memory, and makes it executable.

[0073] Next, the server obtains training data from information providers. Information providers are data providers that hold news articles and other text data, and the server obtains the data by sending an HTTP request to them. The obtained data is provided in JSON format, so the server parses it and converts it into a usable format.

[0074] The server then uses the acquired training data to generate fictional articles using a generative AI model. For example, it provides prompt sentences such as "A strange event happened in New York City last night" and "New technology may change the future" as input, and generates fictional news articles based on these prompts.

[0075] The server then inserts advertisements into the generated fictional stories. Advertisements are pre-prepared marketing messages or promotional content obtained from an advertising network. For example, an advertising message such as "Buy this product now!" is inserted randomly into the fictional stories by the server.

[0076] Finally, the server publishes the fictional story with the inserted advertisements on a website. Publishing is the act of distributing the generated fictional story and the inserted advertisements in a user-accessible form over the Internet or other media, so that users can view the fictional story and advertisements when they visit the website, and advertising revenue can be generated.

[0077] In this way, generative AI models can be used to generate fictional articles and effectively insert advertisements, providing new forms of entertainment and revenue.

[0078] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0079] Step 1:

[0080] The server loads the generative AI model from cloud storage. First, the server accesses cloud storage and downloads a model file called "some-ai-model". Then, it expands the downloaded model file into memory and makes it executable. The input is the cloud storage URL, and the output is the generative AI model expanded in memory.

[0081] Step 2:

[0082] The server obtains training data from information providers. This means accessing an API endpoint using an HTTP request to obtain the training data. Specifically, the server sends a GET request to "http: / / example.com / news-data" and receives news data in JSON format as a response. The input is the URL of the API endpoint, and the output is the parsed news data.

[0083] Step 3:

[0084] The server generates a fictional article using a generative AI model based on the acquired training data. First, the server creates a prompt sentence from the news data and inputs it into the generative AI model. For example, the server inputs the prompt sentence "A strange event happened in New York City last night" and obtains the article output by the model. The input is the prompt sentence, and the output is the generated fictional article.

[0085] Step 4:

[0086] The server inserts advertisements into the generated fictional article. The server obtains advertising data from an advertising network and randomly inserts it into the fictional article. For example, it inserts an advertising message such as "Buy this product now!" The input is the generated fictional article and advertising data, and the output is the fictional article with the advertisement inserted.

[0087] Step 5:

[0088] The server publishes the fictional article with the advertisements inserted on a website. Specifically, the server uploads the generated content, which is a combination of the fictional article and the advertisements, to a web server and distributes it in a form that can be accessed by users. The input is the fictional article with the advertisements inserted, and the output is the content published on the website.

[0089] (Application example 1)

[0090] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0091] Conventional advertising display systems were unable to blend in with the real world, limiting the entertainment value and advertising effectiveness for users. In particular, in situations where augmented reality technology such as smart glasses and head-mounted displays is utilized, it is necessary to display instantly generated content in a natural way, but the technological means to achieve this have not yet been established.

[0092] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0093] In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for publishing the fictional news containing the inserted advertisements, and means for displaying the generated fictional news and advertisements using augmented reality technology. This makes it possible to display the fictional news and advertisements in a manner that harmonizes with real space, thereby improving the entertainment value and advertising effectiveness for users.

[0094] A "generative language model" is an artificial intelligence model trained to generate natural-sounding sentences in the same way as humans do, based on input text data.

[0095] An "information provider" is a data holder that holds news articles and various data and from which the server obtains learning data.

[0096] "Training data" is a collection of news articles and other information used to train a generative language model.

[0097] "Fictional news" refers to fictional news that is automatically generated by a generative language model and has a sense of authenticity but does not exist in reality.

[0098] "Advertising" means marketing messages or promotional content used to promote a particular product or service.

[0099] "Publishing" refers to the act of distributing the generated fictional news and advertisements over the Internet or other media in a form accessible to users.

[0100] "Augmented reality technology" is a technology that overlays computer-generated information onto real space, and is often used with smart glasses or head-mounted displays.

[0101] A "server" is a computer system that has the functionality to load generative language models, obtain training data, generate fictional news, insert advertisements, and publish them.

[0102] Program Generation and Processing Description

[0103] 1. Load the generative language model:

[0104] The server loads a generative language model, which has been trained to generate natural-sounding sentences, using a deep learning library such as TensorFlow or PyTorch. The generative language model is loaded from cloud storage (e.g., AWS S3 or Google Cloud Storage) and deployed in memory.

[0105] 2. Obtaining training data:

[0106] The server obtains training data from the information provider (data holder). This is done using a communication library (e.g., requests) by sending an HTTP request from a specific API endpoint. The obtained data is then parsed from JSON format to extract the necessary information.

[0107] 3. Fictional News Generation:

[0108] The server provides input to a generative language model based on the training data to generate fictional news. The generated fictional news is fictional news that has a sense of authenticity but does not exist in reality. For example, a prompt sentence could be, "A mysterious event occurred in New York City last night that shocked many people. Here are the details about this event."

[0109] 4. Ad Insertion:

[0110] The server retrieves advertising data from the advertising network's API (e.g., Google Ads API) and inserts advertisements into the generated fictional news. The advertisements are inserted randomly, incorporating marketing messages and promotional content into the generated fictional news.

[0111] 5. Publication of generated fictional news and advertisements:

[0112] The server publishes fictional news stories with inserted advertisements using augmented reality technology. AR development kits such as Unity and Vuforia are used to display the stories superimposed on the real world, and the stories are visually presented to users through smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest).

[0113] Adding specific examples

[0114] The hardware used is smart glasses or a head-mounted display, which allows users to experience fictional news and advertisements generated in real space in real time.

[0115] Using the example of a specific prompt sentence, "Last night, a mysterious event occurred in New York City that shocked many people. Here are the details about the event," the generative language model generates natural-sounding fictional news.

[0116] In this way, through a series of processes, the server can provide fictional news and advertisements in a manner that is in harmony with the real world, thereby improving the entertainment value for users and the effectiveness of advertisements.

[0117] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0118] Step 1:

[0119] The server loads the generative AI model. It retrieves the generative AI model as input from cloud storage (e.g., AWS S3) and deploys it in memory using a deep learning library (e.g., TensorFlow or PyTorch). This process makes the generative AI model ready for use.

[0120] Step 2:

[0121] The server obtains training data from information providers. As input, it sends an HTTP request to a specific API endpoint to obtain news data in JSON format. The obtained data is processed using a data parsing library (e.g., json). The output of this step is structured training data.

[0122] Step 3:

[0123] The server inputs a prompt sentence into a generative language model based on the acquired training data to generate fictional news. Specifically, it inputs a prompt sentence (e.g., "A strange event happened in New York City last night") into a generative language model (e.g., GPT-3) and generates fictional news text. The output is the generated fictional news.

[0124] Step 4:

[0125] The server retrieves ad data from the ad network. It sends a request to the ad network's API (e.g., Google Ads API) as input to retrieve ad message data. The retrieved ad data is structured and converted into text format. The output of this step is the ad data.

[0126] Step 5:

[0127] The server inserts advertisements into the generated fictional news. It randomly combines the fictional news with advertisement data. Specifically, it inserts advertisement messages into appropriate positions in the fictional news. The output of this step is fictional news with advertisements inserted.

[0128] Step 6:

[0129] The server publishes the fictional news with the inserted advertisements. Specifically, the server distributes the generated content in a user-accessible form on a website or application. The output of this step is the published fictional news in a user-accessible form.

[0130] Step 7:

[0131] The server displays the generated fictional news and advertisements using augmented reality technology. Specifically, it uses an AR development kit such as Unity or Vuforia to overlay the content on smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest). The output of this step is the fictional news and advertisements visually presented to the user on the AR device.

[0132] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0133] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0134] First, the server loads a generative language model, which uses pre-trained models and is deployed in memory to enable the generation of fictional news.

[0135] Next, the server retrieves training data from information providers by sending an HTTP request to a specific API endpoint to retrieve news data from major data holders. This data is provided in JSON format, which the server parses and converts into a usable format.

[0136] The acquired training data is fed into a generative language model, and the server generates fictional news stories that appear realistic but are based on information that does not actually exist.

[0137] This is where the emotion engine comes in. Based on input data (e.g., text, voice, and images) provided by the device or user, the emotion engine recognizes the user's emotions. Based on this recognition result, the server adjusts the content of the generated fictional news. This adjustment includes the news title, body text, and even the overall tone and emotional nuances.

[0138] The server then inserts advertisements into the generated fictional news. The display of advertisements is optimized based on the recognition results of the emotion engine. For example, if the user has positive emotions, advertisements that stimulate purchasing desire are displayed, and if the user has negative emotions, advertisements for relaxation items are displayed.

[0139] Finally, the server publishes the fiction news with the advertisements inserted on the website, and users can access the website and view the tailored fiction news and the optimized advertisements.

[0140] Specific examples

[0141] The server loads the generative AI model "some-ai-model". The model is read from storage and loaded into memory.

[0142] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0143] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0144] The device analyzes the user's emotions through an emotion engine based on emotional input from the user (e.g., text message or voice). For example, if the user feels "fun," the device adjusts the news article accordingly.

[0145] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0146] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0147] The present invention enables the provision of fictional news and advertisements based on user emotions, improving the user experience and optimizing advertising revenue.

[0148] The processing flow will be explained below.

[0149] Step 1:

[0150] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name, "some-ai-model." This operation deploys the model in memory and prepares it to generate fictional news.

[0151] Step 2:

[0152] The server obtains training data from the information provider by sending an HTTP request to retrieve news data from the specified URL (e.g., "http: / / example.com / news-data"). This data is provided in JSON format, which the server parses and converts into a usable format.

[0153] Step 3:

[0154] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the model outputs fictional news, including news articles that appear believable but do not actually exist.

[0155] Step 4:

[0156] The device uses an emotion engine to recognize the user's emotions. Specifically, it analyzes emotions from text, voice, images, etc. input by the user. The analysis results in the user's emotional state (e.g., joy, sadness, surprise, etc.).

[0157] Step 5:

[0158] The server adjusts the content of the generated fictional news based on the user's emotions recognized by the emotion engine. For example, if the user is feeling "happy," the news will be tailored to a positive story that matches that emotion. On the other hand, if the user is feeling "sad," the news will be tailored to include comforting and encouraging content.

[0159] Step 6:

[0160] The server inserts advertisements into the generated fictional news. Appropriate advertisements are selected based on the recognition results of the emotion engine. For example, advertisements that stimulate purchasing desire are inserted for positive emotions, and advertisements for relaxation items are inserted for negative emotions. Advertisements are obtained from an advertising network and inserted randomly into articles.

[0161] Step 7:

[0162] The server publishes the fictional news articles, interspersed with advertisements, to the website by sending a POST request to the publishing API endpoint and uploading each article to the website, where it can then be accessed by users.

[0163] Step 8:

[0164] Users access the website and view published fiction news and advertisements. This allows the website to display fiction news and advertisements based on the user's emotions, providing a more personalized experience. Furthermore, advertising revenue is generated by maximizing the effectiveness of advertisements.

[0165] Example 2

[0166] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0167] In today's information-saturated society, the quantity and quality of content viewed by users are important. However, conventional systems have difficulty optimizing content and advertisements based on user emotions, making it difficult to improve user experience and maximize advertising revenue. Furthermore, randomly inserted advertisements may not attract user attention and may reduce advertising effectiveness. Therefore, there is a need for a system that can adjust content and advertisements based on user emotional data.

[0168] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for acquiring emotion data from a user, means for adjusting the content of the fictional news based on the emotion data, means for inserting advertisements into the generated fictional news, means for optimizing the advertisements based on the emotion data, and means for publishing the fictional news including the inserted advertisements. This makes it possible to provide fictional news and advertisements according to the user's emotions, thereby improving the user experience and optimizing advertising revenue.

[0169] A "generative language model" is a machine learning model that has been trained on a large amount of text data in advance and is used to generate new text data and complete text.

[0170] An "information provider" is an individual or organization that serves to provide training data to the server.

[0171] "Training data" refers to text data and related information used to train a generative language model.

[0172] "Fictional news" refers to fictional news articles that are based on events or phenomena that do not actually exist.

[0173] "Emotion data" is information that represents the user's emotional state and is provided in the form of text, audio, images, or the like.

[0174] An "emotion engine" is a system that has the functionality to analyze emotion data provided by a user and identify the user's emotional state.

[0175] "Advertising" means promotional content, such as messages, images, banners, etc., created to advertise a product or service.

[0176] "Advertising optimization" is the process of selecting and displaying the most effective advertisements based on a user's emotional state and preferences.

[0177] "Publishing" means making the generated fictional news and inserted advertisements available to users on a website or other platform.

[0178] A "server" is a computer system and associated software that processes, stores, and provides various types of data.

[0179] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0180] First, the server loads the generative AI model. The generative AI model to be used is "some-ai-model", which is read from the server's storage and deployed in memory to become operational. This allows new text data to be generated immediately.

[0181] Next, the server obtains training data from the information provider. In this case, the server sends an HTTP request to the API endpoint "http: / / example.com / news-data" to obtain news data from major data holders. The obtained data is provided in JSON format, which the server parses and converts into a usable format. The obtained news data includes a variety of categories, such as politics, economics, and sports.

[0182] The server inputs the acquired news data into a generative language model to generate fictional news stories, such as "A mysterious event occurred in New York City last night." Although the news stories generated are not real, they are detailed and realistically described.

[0183] Next, the emotion engine is introduced. The device or user provides emotion data such as text, voice, and images. For example, if the user feels "happy," the device sends that emotion to the emotion engine and receives the analysis results. Based on this, the emotion engine recognizes the user's emotional state.

[0184] The server receives the analysis results of the emotion engine and adjusts the content of the generated fictional news accordingly: for example, if positive emotion is detected, the news title, text, overall tone, and emotional nuances will be changed to be more cheerful.

[0185] The server then inserts advertisements into the generated fictional news. Based on the analysis results of the emotion engine, the content and display method of the advertisements are optimized. For example, if positive emotions are detected, a high-energy advertisement such as "Buy this product now!" is selected.

[0186] Finally, the server publishes the fiction news with advertisements inserted on the website, and users access the website through their browsers to view the tailored fiction news and optimized advertisements, thereby improving the user experience and increasing the effectiveness of the advertisements.

[0187] Specific examples

[0188] The server loads the generative AI model "some-ai-model", which is read from the server's storage and loaded into memory.

[0189] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0190] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0191] The device analyzes emotional input from the user (e.g., text message or voice) using an emotion engine. For example, if the user feels "fun," the device adjusts the news article based on that emotion.

[0192] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0193] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0194] Prompt Sentence Examples

[0195] "Generate a detailed fictional news story about an event. For example, about a mysterious event that happened in New York City last night."

[0196] "Tailor news articles based on users' positive emotions (e.g., 'I feel happy')."

[0197] "Generate ad copy to display when positive sentiment is detected."

[0198] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0199] Step 1:

[0200] The server reads the generative AI model "some-ai-model" from storage and expands it into memory. The input is the path to the model file in storage, and the output is the generative AI model loaded into memory. The server first retrieves the model file from a specific path in storage and then loads it into memory.

[0201] Step 2:

[0202] The server retrieves news data by sending an HTTP request to an API endpoint, such as "http: / / example.com / news-data." The input is the API endpoint URL, and the output is the news data in JSON format. The server generates an HTTP request, sends it to the API endpoint, and then parses the JSON data received as a response and converts it into a usable format.

[0203] Step 3:

[0204] The server inputs the acquired news data into a generative AI model to generate fictional news. The news data and prompt sentences are used as input, and fictional news is generated as output. Specifically, the server sends the news data and prompt sentences (e.g., "A mysterious event occurred in New York City last night") to the generative AI model and receives the fictional news generated by the model.

[0205] Step 4:

[0206] The terminal collects emotion data from the user. As input, it receives emotion data such as text messages and voice from the user, and prepares to send the emotion data to the emotion engine as output. The terminal provides an interface for, for example, speech recognition or text input, and acquires emotion data from the user.

[0207] Step 5:

[0208] The device sends the collected emotion data to the emotion engine and receives the analysis results. The device sends emotion data to the emotion engine as input and receives the emotion analysis results (positive, negative, neutral, etc.) as output. The device sends an HTTP request to the emotion engine and receives the emotion analysis results.

[0209] Step 6:

[0210] The server adjusts the content of the generated fictional news based on the results of the emotion engine. It uses the fictional news and the results of the emotion analysis as input, and obtains adjusted fictional news as output. For example, the server modifies the title, body, and tone of the news according to the emotion.

[0211] Step 7:

[0212] The server selects advertisements based on the analysis results of the emotion engine and inserts them into the fictional news. The inputs are the fictional news, the emotion analysis results, and the advertisement data obtained from the advertising network, and the output is the fictional news with the advertisement inserted. The server communicates with the advertising network to obtain the appropriate advertisements and insert them into the news article.

[0213] Step 8:

[0214] The server publishes the fictional news with advertisements inserted on a website. The server uses the fictional news with advertisements inserted as input and obtains a web page ready for user viewing as output. The server generates a web page, embeds the fictional news and advertisements, and publishes it.

[0215] Step 9:

[0216] A user accesses a website in a browser and views tailored fiction news and advertisements. The website URL is used as input, and tailored fiction news and optimized advertisements are displayed as output. The user opens the web page using a web browser and views its content.

[0217] (Application example 2)

[0218] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0219] Existing fiction news generation systems have a problem in that news content and advertisement display do not correspond to user emotions, limiting the improvement of user experience. In addition, since advertisement display is not optimized based on user emotions, it is difficult to maximize advertising revenue.

[0220] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for analyzing emotions from user input data, means for adjusting the content of the generated fictional news based on the analyzed emotions, means for inserting advertisements selected based on the analyzed emotions, and means for publishing the adjusted fictional news and the selected advertisements. This makes it possible to display news content that corresponds to the user's emotions and optimized advertisements.

[0221] A "generative language model" is an artificial intelligence model that has the ability to generate, translate, and summarize text based on a pre-trained dataset.

[0222] "Training data" is a collection of data, such as text, images, and audio, that is used to train and apply a generative language model.

[0223] "Fictional news" is fictional, artificially generated news stories that are not based on actual events.

[0224] "Advertising" means information or a message created to promote a particular product or service.

[0225] "User input data" refers to data such as text messages, voice, images, etc. that a user provides to the system.

[0226] "Analyzing emotions" refers to identifying the user's current emotional state based on input data.

[0227] "Adjusting the news content" means changing the title, text, and overall tone and emotional nuances of the generated fictional news based on the user's emotions.

[0228] "Inserting advertisements" means incorporating advertisements into the generated fictional news.

[0229] "Publishing" means making the generated fictional news and inserted advertisements accessible to users.

[0230] The present invention is a system that generates fictional news using a generative language model, analyzes user emotions using an emotion engine, and adjusts the news content and advertisement display based on the results. This system is composed of a server and a terminal.

[0231] First, the server loads the generative language model. This uses a high-performance natural language processing model such as Hugging Face's T5 model. The generative language model is loaded from storage and deployed in memory. Next, training data is obtained from information providers. This training data is often obtained using web APIs and includes a wide range of datasets from news providers, such as politics, economics, and sports. The obtained training data is provided in a format such as JSON, and the server parses it appropriately and converts it into a usable format.

[0232] The server then uses a generative language model to generate fictional news stories based on the acquired training data. The generated news stories appear realistic at first glance, but are based on events that do not actually exist. For example, if you provide a prompt such as "A mysterious event occurred in New York City last night," a detailed fictional news story will be generated based on that content.

[0233] After the fictional news is generated, a user sentiment analysis is performed. The sentiment engine analyzes sentiment based on input data (text, audio, images, etc.) provided by the user from their device (smartphone, PC, etc.). TextBlob and other sentiment analysis libraries are used for sentiment analysis. The analyzed sentiment information is classified as positive or negative.

[0234] The server adjusts the content of the generated fictional news based on the analyzed emotion information, for example, adjusting the tone of the news article to be lighthearted and fun if positive emotion is detected, or to be calmer and more soothing if negative emotion is detected.

[0235] Furthermore, based on the results of the sentiment analysis, appropriate advertisements are selected and inserted from the advertising network. For positive emotions, advertisements that stimulate purchasing desire are selected, and for negative emotions, advertisements for relaxation items are selected.

[0236] Finally, the server publishes the tailored fictional news and inserted advertisements on websites and applications, allowing users to view the news and advertisements optimized based on their emotions.

[0237] For example, if a user types the text message "I'm feeling happy!" into their device, the emotion engine will interpret this as a positive emotion. This will allow the server to adjust the tone of the news article generated based on the prompt "A strange event happened in New York City last night" to be lighthearted, and insert a high-energy advertisement saying "Buy this product now!" These will then be displayed when the user visits the site.

[0238] In this way, the present invention achieves improved user experience and optimized advertising revenue by individually optimizing news content and advertisements based on user sentiment.

[0239] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0240] Step 1:

[0241] The server loads the generative language model. This is done by extracting a high-performance natural language processing model, such as Hugging Face's T5 model, from storage and storing it in memory. The file path of the model is specified as input, and the generative language model extracted in memory is obtained as output.

[0242] Step 2:

[0243] The server obtains training data from information providers. In this step, the server sends an HTTP request to the specified API endpoint to obtain news data from the data holder. The URL of the API endpoint is given as input, and the obtained news data in JSON format is obtained as output.

[0244] Step 3:

[0245] The server generates fictional news using a generative language model based on the acquired training data. It receives a prompt sentence provided by the user (e.g., "A strange event happened in New York City last night") as input and generates fictional news using the generative language model. The prompt sentence and model input are given as input, and the generated fictional news is obtained as output.

[0246] Step 4:

[0247] The device receives user input data (e.g., text, voice, image) and analyzes the emotion using an emotion engine. The input is the user input, and the output is the analyzed emotion information (positive, negative, etc.). In this step, an emotion analysis library such as TextBlob is used.

[0248] Step 5:

[0249] The server adjusts the content of the generated fictional news based on the analyzed emotional information. For example, if positive emotions are detected, the tone of the news article is adjusted to be lighthearted and fun. The input is given as emotional information and the generated news, and the output is the adjusted news article.

[0250] Step 6:

[0251] The server selects advertisements based on the analyzed emotional information and inserts them into the generated fictional news. Advertisements that stimulate purchasing motivation are selected for positive emotions, and advertisements for relaxation items are selected for negative emotions. The input is given as emotional information and a list of available advertisements, and the output is the selected advertisements and the inserted news article.

[0252] Step 7:

[0253] The server publishes the tailored fictional news and selected advertisements on a website or application, which is then displayed when a user accesses the website or application. The input is the tailored news article and advertisements, and the output is the published news article and advertisements.

[0254] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0255] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0256] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0257] [Second embodiment]

[0258] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0259] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0260] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0261] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0262] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0263] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0264] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0265] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0266] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0267] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0268] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0269] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0270] To implement the present invention, the following system configuration and processing must be included.

[0271] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model trained to generate text in a specific format based on a large dataset. By loading a pre-trained generative language model onto the server, it becomes possible to generate fictional news.

[0272] Next, the server obtains training data from information providers. Information providers are data holders that hold news articles and various data, and this data will be used to train the generative AI model. The server sends an HTTP request from a specific API endpoint to obtain the news data. This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0273] The server then generates fictional news using a generative language model based on the acquired training data. Fictional news is automatically generated by the generative language model and is fictional news that has a sense of authenticity but does not exist in reality. The server provides the training data as input to the generative language model to generate the fictional news.

[0274] Next, the server inserts advertisements into the generated fictional news. Advertisements are prepared marketing messages or promotional content that are inserted into the fictional news as a means of earning advertising revenue. The server obtains advertising data from an advertising network and randomly combines advertisements for each generated fictional news article to create the final content.

[0275] Finally, the server publishes the fictional news with the inserted advertisements on the website. Publishing is the act of distributing the generated fictional news and the inserted advertisements on the Internet or other media in a form that is accessible to users. This allows the advertisements to be displayed along with the fictional news when users visit the website, making it possible to earn advertising revenue.

[0276] Specific examples

[0277] The server loads the generative AI model "some-ai-model." For example, a generative language model is loaded from a hard disk or cloud storage and deployed in memory.

[0278] The server retrieves news data from "http: / / example.com / news-data." The retrieved data includes categories such as politics, economics, and sports.

[0279] Based on the acquired news data, the server generates fictional news stories such as "A mysterious event happened in New York City last night" or "New technology may change the future."

[0280] The server retrieves advertisements such as "Buy this product now!" from the advertising network and randomly inserts them into the generated fictional news.

[0281] The server publishes the fictional news containing the inserted advertisements on a website, so that the fictional news and advertisements can be viewed when a user accesses the website.

[0282] This invention makes it possible to effectively utilize the "hallucination" function of generative language models, thereby enabling the creation of fictional news as a new form of entertainment and information provision while simultaneously generating advertising revenue.

[0283] The processing flow will be explained below.

[0284] Step 1:

[0285] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name. This operation deploys the model in memory and prepares it for generating fictional news.

[0286] Step 2:

[0287] The server obtains training data from the information provider. Specifically, it sends an HTTP request to obtain data from the specified URL. For example, it obtains news data from "http: / / example.com / news-data." This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0288] Step 3:

[0289] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the fictional news output by the model is obtained. This fictional news has a sense of verisimilitude, but is fictional news that does not exist in reality.

[0290] Step 4:

[0291] The server inserts advertisements into the generated fictional news. Specifically, it obtains advertisement data from an advertising network and randomly inserts advertisements into each fictional news article. For example, it obtains an advertisement "Buy this product now!" from the advertising network and inserts it into the generated news article.

[0292] Step 5:

[0293] The server publishes the fictional news articles, interspersed with advertisements, to a website by sending a POST request to a publishing API endpoint and uploading each article to the website, where it can be accessed by users over the internet.

[0294] Step 6:

[0295] Users access the website and view the published fiction news and advertisements, thereby generating advertising revenue along with the viewing of the fiction news.

[0296] Example 1

[0297] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0298] Providing information quickly and generating revenue are important challenges in today's world. However, while there are existing systems that combine real news with advertising, there are no systems that generate fictional articles, insert advertisements into them, and publish them, creating new forms of entertainment and revenue opportunities. Therefore, there is a need for a system that uses generative artificial intelligence models to automatically generate fictional articles and effectively insert advertisements into them for publication.

[0299] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0300] In this invention, the server includes means for loading a generative artificial intelligence model, means for acquiring training data from an information provider, means for generating fictional articles using the generative artificial intelligence model based on the acquired training data, means for inserting advertisements into the generated fictional articles, and means for publishing the fictional articles including the inserted advertisements, thereby enabling the automatic generation of fictional articles and the effective insertion and publication of advertisements.

[0301] A "generative artificial intelligence model" is an artificial intelligence model that learns from large datasets and enables language generation.

[0302] An "information provider" refers to a data provider that has training data and provides the data necessary to train and generate a generative model.

[0303] "Training data" refers to large sets of text and other data used to train generative models.

[0304] "Fictional articles" are fictional news articles or article-style content generated by a generative model based on training data.

[0305] "Advertising" means marketing messages or promotional content that are inserted into and displayed alongside fictional articles.

[0306] "Publication" refers to the act of distributing the generated fictional article and inserted advertisements over the Internet or other media in a manner that makes it accessible to users.

[0307] To implement the present invention, the following system configuration and processes are involved: The components of the server, generative AI model, information provider, advertising network, and website work together.

[0308] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model that enables language generation based on a large dataset and is specialized for generating news articles. The server downloads this model from cloud storage, expands it into memory, and makes it executable.

[0309] Next, the server obtains training data from information providers. Information providers are data providers that hold news articles and other text data, and the server obtains the data by sending an HTTP request to them. The obtained data is provided in JSON format, so the server parses it and converts it into a usable format.

[0310] The server then uses the acquired training data to generate fictional articles using a generative AI model. For example, it provides prompt sentences such as "A strange event happened in New York City last night" and "New technology may change the future" as input, and generates fictional news articles based on these prompts.

[0311] The server then inserts advertisements into the generated fictional stories. Advertisements are pre-prepared marketing messages or promotional content obtained from an advertising network. For example, an advertising message such as "Buy this product now!" is inserted randomly into the fictional stories by the server.

[0312] Finally, the server publishes the fictional story with the inserted advertisements on a website. Publishing is the act of distributing the generated fictional story and the inserted advertisements in a user-accessible form over the Internet or other media, so that users can view the fictional story and advertisements when they visit the website, and advertising revenue can be generated.

[0313] In this way, generative AI models can be used to generate fictional articles and effectively insert advertisements, providing new forms of entertainment and revenue.

[0314] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0315] Step 1:

[0316] The server loads the generative AI model from cloud storage. First, the server accesses cloud storage and downloads a model file called "some-ai-model". Then, it expands the downloaded model file into memory and makes it executable. The input is the cloud storage URL, and the output is the generative AI model expanded in memory.

[0317] Step 2:

[0318] The server obtains training data from information providers. This means accessing an API endpoint using an HTTP request to obtain the training data. Specifically, the server sends a GET request to "http: / / example.com / news-data" and receives news data in JSON format as a response. The input is the URL of the API endpoint, and the output is the parsed news data.

[0319] Step 3:

[0320] The server generates a fictional article using a generative AI model based on the acquired training data. First, the server creates a prompt sentence from the news data and inputs it into the generative AI model. For example, the server inputs the prompt sentence "A strange event happened in New York City last night" and obtains the article output by the model. The input is the prompt sentence, and the output is the generated fictional article.

[0321] Step 4:

[0322] The server inserts advertisements into the generated fictional article. The server obtains advertising data from an advertising network and randomly inserts it into the fictional article. For example, it inserts an advertising message such as "Buy this product now!" The input is the generated fictional article and advertising data, and the output is the fictional article with the advertisement inserted.

[0323] Step 5:

[0324] The server publishes the fictional article with the advertisements inserted on a website. Specifically, the server uploads the generated content, which is a combination of the fictional article and the advertisements, to a web server and distributes it in a form that can be accessed by users. The input is the fictional article with the advertisements inserted, and the output is the content published on the website.

[0325] (Application example 1)

[0326] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0327] Conventional advertising display systems were unable to blend in with the real world, limiting the entertainment value and advertising effectiveness for users. In particular, in situations where augmented reality technology such as smart glasses and head-mounted displays is utilized, it is necessary to display instantly generated content in a natural way, but the technological means to achieve this have not yet been established.

[0328] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0329] In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for publishing the fictional news containing the inserted advertisements, and means for displaying the generated fictional news and advertisements using augmented reality technology. This makes it possible to display the fictional news and advertisements in a manner that harmonizes with real space, thereby improving the entertainment value and advertising effectiveness for users.

[0330] A "generative language model" is an artificial intelligence model trained to generate natural-sounding sentences in the same way as humans do, based on input text data.

[0331] An "information provider" is a data holder that holds news articles and various data and from which the server obtains learning data.

[0332] "Training data" is a collection of news articles and other information used to train a generative language model.

[0333] "Fictional news" refers to fictional news that is automatically generated by a generative language model and has a sense of authenticity but does not exist in reality.

[0334] "Advertising" means marketing messages or promotional content used to promote a particular product or service.

[0335] "Publishing" refers to the act of distributing the generated fictional news and advertisements over the Internet or other media in a form accessible to users.

[0336] "Augmented reality technology" is a technology that overlays computer-generated information onto real space, and is often used with smart glasses or head-mounted displays.

[0337] A "server" is a computer system that has the functionality to load generative language models, obtain training data, generate fictional news, insert advertisements, and publish them.

[0338] Program Generation and Processing Description

[0339] 1. Load the generative language model:

[0340] The server loads a generative language model, which has been trained to generate natural-sounding sentences, using a deep learning library such as TensorFlow or PyTorch. The generative language model is loaded from cloud storage (e.g., AWS S3 or Google Cloud Storage) and deployed in memory.

[0341] 2. Obtaining training data:

[0342] The server obtains training data from the information provider (data holder). This is done using a communication library (e.g., requests) by sending an HTTP request from a specific API endpoint. The obtained data is then parsed from JSON format to extract the necessary information.

[0343] 3. Fictional News Generation:

[0344] The server provides input to a generative language model based on the training data to generate fictional news. The generated fictional news is fictional news that has a sense of authenticity but does not exist in reality. For example, a prompt sentence could be, "A mysterious event occurred in New York City last night that shocked many people. Here are the details about this event."

[0345] 4. Ad Insertion:

[0346] The server retrieves advertising data from the advertising network's API (e.g., Google Ads API) and inserts advertisements into the generated fictional news. The advertisements are inserted randomly, incorporating marketing messages and promotional content into the generated fictional news.

[0347] 5. Publication of generated fictional news and advertisements:

[0348] The server publishes fictional news stories with inserted advertisements using augmented reality technology. AR development kits such as Unity and Vuforia are used to display the stories superimposed on the real world, and the stories are visually presented to users through smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest).

[0349] Adding specific examples

[0350] The hardware used is smart glasses or a head-mounted display, which allows users to experience fictional news and advertisements generated in real space in real time.

[0351] Using the example of a specific prompt sentence, "Last night, a mysterious event occurred in New York City that shocked many people. Here are the details about the event," the generative language model generates natural-sounding fictional news.

[0352] In this way, through a series of processes, the server can provide fictional news and advertisements in a manner that is in harmony with the real world, thereby improving the entertainment value for users and the effectiveness of advertisements.

[0353] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0354] Step 1:

[0355] The server loads the generative AI model. It retrieves the generative AI model as input from cloud storage (e.g., AWS S3) and deploys it in memory using a deep learning library (e.g., TensorFlow or PyTorch). This process makes the generative AI model ready for use.

[0356] Step 2:

[0357] The server obtains training data from information providers. As input, it sends an HTTP request to a specific API endpoint to obtain news data in JSON format. The obtained data is processed using a data parsing library (e.g., json). The output of this step is structured training data.

[0358] Step 3:

[0359] The server inputs a prompt sentence into a generative language model based on the acquired training data to generate fictional news. Specifically, it inputs a prompt sentence (e.g., "A strange event happened in New York City last night") into a generative language model (e.g., GPT-3) and generates fictional news text. The output is the generated fictional news.

[0360] Step 4:

[0361] The server retrieves ad data from the ad network. It sends a request to the ad network's API (e.g., Google Ads API) as input to retrieve ad message data. The retrieved ad data is structured and converted into text format. The output of this step is the ad data.

[0362] Step 5:

[0363] The server inserts advertisements into the generated fictional news. It randomly combines the fictional news with advertisement data. Specifically, it inserts advertisement messages into appropriate positions in the fictional news. The output of this step is fictional news with advertisements inserted.

[0364] Step 6:

[0365] The server publishes the fictional news with the inserted advertisements. Specifically, the server distributes the generated content in a user-accessible form on a website or application. The output of this step is the published fictional news in a user-accessible form.

[0366] Step 7:

[0367] The server displays the generated fictional news and advertisements using augmented reality technology. Specifically, it uses an AR development kit such as Unity or Vuforia to overlay the content on smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest). The output of this step is the fictional news and advertisements visually presented to the user on the AR device.

[0368] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0369] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0370] First, the server loads a generative language model, which uses pre-trained models and is deployed in memory to enable the generation of fictional news.

[0371] Next, the server retrieves training data from information providers by sending an HTTP request to a specific API endpoint to retrieve news data from major data holders. This data is provided in JSON format, which the server parses and converts into a usable format.

[0372] The acquired training data is fed into a generative language model, and the server generates fictional news stories that appear realistic but are based on information that does not actually exist.

[0373] This is where the emotion engine comes in. Based on input data (e.g., text, voice, and images) provided by the device or user, the emotion engine recognizes the user's emotions. Based on this recognition result, the server adjusts the content of the generated fictional news. This adjustment includes the news title, body text, and even the overall tone and emotional nuances.

[0374] The server then inserts advertisements into the generated fictional news. The display of advertisements is optimized based on the recognition results of the emotion engine. For example, if the user has positive emotions, advertisements that stimulate purchasing desire are displayed, and if the user has negative emotions, advertisements for relaxation items are displayed.

[0375] Finally, the server publishes the fiction news with the advertisements inserted on the website, and users can access the website and view the tailored fiction news and the optimized advertisements.

[0376] Specific examples

[0377] The server loads the generative AI model "some-ai-model". The model is read from storage and loaded into memory.

[0378] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0379] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0380] The device analyzes the user's emotions through an emotion engine based on emotional input from the user (e.g., text message or voice). For example, if the user feels "fun," the device adjusts the news article accordingly.

[0381] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0382] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0383] The present invention enables the provision of fictional news and advertisements based on user emotions, improving the user experience and optimizing advertising revenue.

[0384] The processing flow will be explained below.

[0385] Step 1:

[0386] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name, "some-ai-model." This operation deploys the model in memory and prepares it to generate fictional news.

[0387] Step 2:

[0388] The server obtains training data from the information provider by sending an HTTP request to retrieve news data from the specified URL (e.g., "http: / / example.com / news-data"). This data is provided in JSON format, which the server parses and converts into a usable format.

[0389] Step 3:

[0390] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the model outputs fictional news, including news articles that appear believable but do not actually exist.

[0391] Step 4:

[0392] The device uses an emotion engine to recognize the user's emotions. Specifically, it analyzes emotions from text, voice, images, etc. input by the user. The analysis results in the user's emotional state (e.g., joy, sadness, surprise, etc.).

[0393] Step 5:

[0394] The server adjusts the content of the generated fictional news based on the user's emotions recognized by the emotion engine. For example, if the user is feeling "happy," the news will be tailored to a positive story that matches that emotion. On the other hand, if the user is feeling "sad," the news will be tailored to include comforting and encouraging content.

[0395] Step 6:

[0396] The server inserts advertisements into the generated fictional news. Appropriate advertisements are selected based on the recognition results of the emotion engine. For example, advertisements that stimulate purchasing desire are inserted for positive emotions, and advertisements for relaxation items are inserted for negative emotions. Advertisements are obtained from an advertising network and inserted randomly into articles.

[0397] Step 7:

[0398] The server publishes the fictional news articles, interspersed with advertisements, to the website by sending a POST request to the publishing API endpoint and uploading each article to the website, where it can then be accessed by users.

[0399] Step 8:

[0400] Users access the website and view published fiction news and advertisements. This allows the website to display fiction news and advertisements based on the user's emotions, providing a more personalized experience. Furthermore, advertising revenue is generated by maximizing the effectiveness of advertisements.

[0401] Example 2

[0402] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0403] In today's information-saturated society, the quantity and quality of content viewed by users are important. However, conventional systems have difficulty optimizing content and advertisements based on user emotions, making it difficult to improve user experience and maximize advertising revenue. Furthermore, randomly inserted advertisements may not attract user attention and may reduce advertising effectiveness. Therefore, there is a need for a system that can adjust content and advertisements based on user emotional data.

[0404] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for acquiring emotion data from a user, means for adjusting the content of the fictional news based on the emotion data, means for inserting advertisements into the generated fictional news, means for optimizing the advertisements based on the emotion data, and means for publishing the fictional news including the inserted advertisements. This makes it possible to provide fictional news and advertisements according to the user's emotions, thereby improving the user experience and optimizing advertising revenue.

[0405] A "generative language model" is a machine learning model that has been trained on a large amount of text data in advance and is used to generate new text data and complete text.

[0406] An "information provider" is an individual or organization that serves to provide training data to the server.

[0407] "Training data" refers to text data and related information used to train a generative language model.

[0408] "Fictional news" refers to fictional news articles that are based on events or phenomena that do not actually exist.

[0409] "Emotion data" is information that represents the user's emotional state and is provided in the form of text, audio, images, or the like.

[0410] An "emotion engine" is a system that has the functionality to analyze emotion data provided by a user and identify the user's emotional state.

[0411] "Advertising" means promotional content, such as messages, images, banners, etc., created to advertise a product or service.

[0412] "Advertising optimization" is the process of selecting and displaying the most effective advertisements based on a user's emotional state and preferences.

[0413] "Publishing" means making the generated fictional news and inserted advertisements available to users on a website or other platform.

[0414] A "server" is a computer system and associated software that processes, stores, and provides various types of data.

[0415] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0416] First, the server loads the generative AI model. The generative AI model to be used is "some-ai-model", which is read from the server's storage and deployed in memory to become operational. This allows new text data to be generated immediately.

[0417] Next, the server obtains training data from the information provider. In this case, the server sends an HTTP request to the API endpoint "http: / / example.com / news-data" to obtain news data from major data holders. The obtained data is provided in JSON format, which the server parses and converts into a usable format. The obtained news data includes a variety of categories, such as politics, economics, and sports.

[0418] The server inputs the acquired news data into a generative language model to generate fictional news stories, such as "A mysterious event occurred in New York City last night." Although the news stories generated are not real, they are detailed and realistically described.

[0419] Next, the emotion engine is introduced. The device or user provides emotion data such as text, voice, and images. For example, if the user feels "happy," the device sends that emotion to the emotion engine and receives the analysis results. Based on this, the emotion engine recognizes the user's emotional state.

[0420] The server receives the analysis results of the emotion engine and adjusts the content of the generated fictional news accordingly: for example, if positive emotion is detected, the news title, text, overall tone, and emotional nuances will be changed to be more cheerful.

[0421] The server then inserts advertisements into the generated fictional news. Based on the analysis results of the emotion engine, the content and display method of the advertisements are optimized. For example, if positive emotions are detected, a high-energy advertisement such as "Buy this product now!" is selected.

[0422] Finally, the server publishes the fiction news with advertisements inserted on the website, and users access the website through their browsers to view the tailored fiction news and optimized advertisements, thereby improving the user experience and increasing the effectiveness of the advertisements.

[0423] Specific examples

[0424] The server loads the generative AI model "some-ai-model", which is read from the server's storage and loaded into memory.

[0425] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0426] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0427] The device analyzes emotional input from the user (e.g., text message or voice) using an emotion engine. For example, if the user feels "fun," the device adjusts the news article based on that emotion.

[0428] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0429] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0430] Prompt Sentence Examples

[0431] "Generate a detailed fictional news story about an event. For example, about a mysterious event that happened in New York City last night."

[0432] "Tailor news articles based on users' positive emotions (e.g., 'I feel happy')."

[0433] "Generate ad copy to display when positive sentiment is detected."

[0434] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0435] Step 1:

[0436] The server reads the generative AI model "some-ai-model" from storage and expands it into memory. The input is the path to the model file in storage, and the output is the generative AI model loaded into memory. The server first retrieves the model file from a specific path in storage and then loads it into memory.

[0437] Step 2:

[0438] The server retrieves news data by sending an HTTP request to an API endpoint, such as "http: / / example.com / news-data." The input is the API endpoint URL, and the output is the news data in JSON format. The server generates an HTTP request, sends it to the API endpoint, and then parses the JSON data received as a response and converts it into a usable format.

[0439] Step 3:

[0440] The server inputs the acquired news data into a generative AI model to generate fictional news. The news data and prompt sentences are used as input, and fictional news is generated as output. Specifically, the server sends the news data and prompt sentences (e.g., "A mysterious event occurred in New York City last night") to the generative AI model and receives the fictional news generated by the model.

[0441] Step 4:

[0442] The terminal collects emotion data from the user. As input, it receives emotion data such as text messages and voice from the user, and prepares to send the emotion data to the emotion engine as output. The terminal provides an interface for, for example, speech recognition or text input, and acquires emotion data from the user.

[0443] Step 5:

[0444] The device sends the collected emotion data to the emotion engine and receives the analysis results. The device sends emotion data to the emotion engine as input and receives the emotion analysis results (positive, negative, neutral, etc.) as output. The device sends an HTTP request to the emotion engine and receives the emotion analysis results.

[0445] Step 6:

[0446] The server adjusts the content of the generated fictional news based on the results of the emotion engine. It uses the fictional news and the results of the emotion analysis as input, and obtains adjusted fictional news as output. For example, the server modifies the title, body, and tone of the news according to the emotion.

[0447] Step 7:

[0448] The server selects advertisements based on the analysis results of the emotion engine and inserts them into the fictional news. The inputs are the fictional news, the emotion analysis results, and the advertisement data obtained from the advertising network, and the output is the fictional news with the advertisement inserted. The server communicates with the advertising network to obtain the appropriate advertisements and insert them into the news article.

[0449] Step 8:

[0450] The server publishes the fictional news with advertisements inserted on a website. The server uses the fictional news with advertisements inserted as input and obtains a web page ready for user viewing as output. The server generates a web page, embeds the fictional news and advertisements, and publishes it.

[0451] Step 9:

[0452] A user accesses a website in a browser and views tailored fiction news and advertisements. The website URL is used as input, and tailored fiction news and optimized advertisements are displayed as output. The user opens the web page using a web browser and views its content.

[0453] (Application example 2)

[0454] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0455] Existing fiction news generation systems have a problem in that news content and advertisement display do not correspond to user emotions, limiting the improvement of user experience. In addition, since advertisement display is not optimized based on user emotions, it is difficult to maximize advertising revenue.

[0456] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for analyzing emotions from user input data, means for adjusting the content of the generated fictional news based on the analyzed emotions, means for inserting advertisements selected based on the analyzed emotions, and means for publishing the adjusted fictional news and the selected advertisements. This makes it possible to display news content that corresponds to the user's emotions and optimized advertisements.

[0457] A "generative language model" is an artificial intelligence model that has the ability to generate, translate, and summarize text based on a pre-trained dataset.

[0458] "Training data" is a collection of data, such as text, images, and audio, that is used to train and apply a generative language model.

[0459] "Fictional news" is fictional, artificially generated news stories that are not based on actual events.

[0460] "Advertising" means information or a message created to promote a particular product or service.

[0461] "User input data" refers to data such as text messages, voice, images, etc. that a user provides to the system.

[0462] "Analyzing emotions" refers to identifying the user's current emotional state based on input data.

[0463] "Adjusting the news content" means changing the title, text, and overall tone and emotional nuances of the generated fictional news based on the user's emotions.

[0464] "Inserting advertisements" means incorporating advertisements into the generated fictional news.

[0465] "Publishing" means making the generated fictional news and inserted advertisements accessible to users.

[0466] The present invention is a system that generates fictional news using a generative language model, analyzes user emotions using an emotion engine, and adjusts the news content and advertisement display based on the results. This system is composed of a server and a terminal.

[0467] First, the server loads the generative language model. This uses a high-performance natural language processing model such as Hugging Face's T5 model. The generative language model is loaded from storage and deployed in memory. Next, training data is obtained from information providers. This training data is often obtained using web APIs and includes a wide range of datasets from news providers, such as politics, economics, and sports. The obtained training data is provided in a format such as JSON, and the server parses it appropriately and converts it into a usable format.

[0468] The server then uses a generative language model to generate fictional news stories based on the acquired training data. The generated news stories appear realistic at first glance, but are based on events that do not actually exist. For example, if you provide a prompt such as "A mysterious event occurred in New York City last night," a detailed fictional news story will be generated based on that content.

[0469] After the fictional news is generated, a user sentiment analysis is performed. The sentiment engine analyzes sentiment based on input data (text, audio, images, etc.) provided by the user from their device (smartphone, PC, etc.). TextBlob and other sentiment analysis libraries are used for sentiment analysis. The analyzed sentiment information is classified as positive or negative.

[0470] The server adjusts the content of the generated fictional news based on the analyzed emotion information, for example, adjusting the tone of the news article to be lighthearted and fun if positive emotion is detected, or to be calmer and more soothing if negative emotion is detected.

[0471] Furthermore, based on the results of the sentiment analysis, appropriate advertisements are selected and inserted from the advertising network. For positive emotions, advertisements that stimulate purchasing desire are selected, and for negative emotions, advertisements for relaxation items are selected.

[0472] Finally, the server publishes the tailored fictional news and inserted advertisements on websites and applications, allowing users to view the news and advertisements optimized based on their emotions.

[0473] For example, if a user types the text message "I'm feeling happy!" into their device, the emotion engine will interpret this as a positive emotion. This will allow the server to adjust the tone of the news article generated based on the prompt "A strange event happened in New York City last night" to be lighthearted, and insert a high-energy advertisement saying "Buy this product now!" These will then be displayed when the user visits the site.

[0474] In this way, the present invention achieves improved user experience and optimized advertising revenue by individually optimizing news content and advertisements based on user sentiment.

[0475] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0476] Step 1:

[0477] The server loads the generative language model. This is done by extracting a high-performance natural language processing model, such as Hugging Face's T5 model, from storage and storing it in memory. The file path of the model is specified as input, and the generative language model extracted in memory is obtained as output.

[0478] Step 2:

[0479] The server obtains training data from information providers. In this step, the server sends an HTTP request to the specified API endpoint to obtain news data from the data holder. The URL of the API endpoint is given as input, and the obtained news data in JSON format is obtained as output.

[0480] Step 3:

[0481] The server generates fictional news using a generative language model based on the acquired training data. It receives a prompt sentence provided by the user (e.g., "A strange event happened in New York City last night") as input and generates fictional news using the generative language model. The prompt sentence and model input are given as input, and the generated fictional news is obtained as output.

[0482] Step 4:

[0483] The device receives user input data (e.g., text, voice, image) and analyzes the emotion using an emotion engine. The input is the user input, and the output is the analyzed emotion information (positive, negative, etc.). In this step, an emotion analysis library such as TextBlob is used.

[0484] Step 5:

[0485] The server adjusts the content of the generated fictional news based on the analyzed emotional information. For example, if positive emotions are detected, the tone of the news article is adjusted to be lighthearted and fun. The input is given as emotional information and the generated news, and the output is the adjusted news article.

[0486] Step 6:

[0487] The server selects advertisements based on the analyzed emotional information and inserts them into the generated fictional news. Advertisements that stimulate purchasing motivation are selected for positive emotions, and advertisements for relaxation items are selected for negative emotions. The input is given as emotional information and a list of available advertisements, and the output is the selected advertisements and the inserted news article.

[0488] Step 7:

[0489] The server publishes the tailored fictional news and selected advertisements on a website or application, which is then displayed when a user accesses the website or application. The input is the tailored news article and advertisements, and the output is the published news article and advertisements.

[0490] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0491] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0492] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0493] [Third embodiment]

[0494] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0495] 5, the data processing system 310 includes the data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0496] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0497] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0498] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0499] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0500] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0501] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0502] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0503] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0504] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0505] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0506] To implement the present invention, the following system configuration and processing must be included.

[0507] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model trained to generate text in a specific format based on a large dataset. By loading a pre-trained generative language model onto the server, it becomes possible to generate fictional news.

[0508] Next, the server obtains training data from information providers. Information providers are data holders that hold news articles and various data, and this data will be used to train the generative AI model. The server sends an HTTP request from a specific API endpoint to obtain the news data. This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0509] The server then generates fictional news using a generative language model based on the acquired training data. Fictional news is automatically generated by the generative language model and is fictional news that has a sense of authenticity but does not exist in reality. The server provides the training data as input to the generative language model to generate the fictional news.

[0510] Next, the server inserts advertisements into the generated fictional news. Advertisements are prepared marketing messages or promotional content that are inserted into the fictional news as a means of earning advertising revenue. The server obtains advertising data from an advertising network and randomly combines advertisements for each generated fictional news article to create the final content.

[0511] Finally, the server publishes the fictional news with the inserted advertisements on the website. Publishing is the act of distributing the generated fictional news and the inserted advertisements on the Internet or other media in a form that is accessible to users. This allows the advertisements to be displayed along with the fictional news when users visit the website, making it possible to earn advertising revenue.

[0512] Specific examples

[0513] The server loads the generative AI model "some-ai-model." For example, a generative language model is loaded from a hard disk or cloud storage and deployed in memory.

[0514] The server retrieves news data from "http: / / example.com / news-data." The retrieved data includes categories such as politics, economics, and sports.

[0515] Based on the acquired news data, the server generates fictional news stories such as "A mysterious event happened in New York City last night" or "New technology may change the future."

[0516] The server retrieves advertisements such as "Buy this product now!" from the advertising network and randomly inserts them into the generated fictional news.

[0517] The server publishes the fictional news containing the inserted advertisements on a website, so that the fictional news and advertisements can be viewed when a user accesses the website.

[0518] This invention makes it possible to effectively utilize the "hallucination" function of generative language models, thereby enabling the creation of fictional news as a new form of entertainment and information provision while simultaneously generating advertising revenue.

[0519] The processing flow will be explained below.

[0520] Step 1:

[0521] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name. This operation deploys the model in memory and prepares it for generating fictional news.

[0522] Step 2:

[0523] The server obtains training data from the information provider. Specifically, it sends an HTTP request to obtain data from the specified URL. For example, it obtains news data from "http: / / example.com / news-data." This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0524] Step 3:

[0525] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the fictional news output by the model is obtained. This fictional news has a sense of verisimilitude, but is fictional news that does not exist in reality.

[0526] Step 4:

[0527] The server inserts advertisements into the generated fictional news. Specifically, it obtains advertisement data from an advertising network and randomly inserts advertisements into each fictional news article. For example, it obtains an advertisement "Buy this product now!" from the advertising network and inserts it into the generated news article.

[0528] Step 5:

[0529] The server publishes the fictional news articles, interspersed with advertisements, to a website by sending a POST request to a publishing API endpoint and uploading each article to the website, where it can be accessed by users over the internet.

[0530] Step 6:

[0531] Users access the website and view the published fiction news and advertisements, thereby generating advertising revenue along with the viewing of the fiction news.

[0532] Example 1

[0533] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0534] Providing information quickly and generating revenue are important challenges in today's world. However, while there are existing systems that combine real news with advertising, there are no systems that generate fictional articles, insert advertisements into them, and publish them, creating new forms of entertainment and revenue opportunities. Therefore, there is a need for a system that uses generative artificial intelligence models to automatically generate fictional articles and effectively insert advertisements into them for publication.

[0535] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0536] In this invention, the server includes means for loading a generative artificial intelligence model, means for acquiring training data from an information provider, means for generating fictional articles using the generative artificial intelligence model based on the acquired training data, means for inserting advertisements into the generated fictional articles, and means for publishing the fictional articles including the inserted advertisements, thereby enabling the automatic generation of fictional articles and the effective insertion and publication of advertisements.

[0537] A "generative artificial intelligence model" is an artificial intelligence model that learns from large datasets and enables language generation.

[0538] An "information provider" refers to a data provider that has training data and provides the data necessary to train and generate a generative model.

[0539] "Training data" refers to large sets of text and other data used to train generative models.

[0540] "Fictional articles" are fictional news articles or article-style content generated by a generative model based on training data.

[0541] "Advertising" means marketing messages or promotional content that are inserted into and displayed alongside fictional articles.

[0542] "Publication" refers to the act of distributing the generated fictional article and inserted advertisements over the Internet or other media in a manner that makes it accessible to users.

[0543] To implement the present invention, the following system configuration and processes are involved: The components of the server, generative AI model, information provider, advertising network, and website work together.

[0544] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model that enables language generation based on a large dataset and is specialized for generating news articles. The server downloads this model from cloud storage, expands it into memory, and makes it executable.

[0545] Next, the server obtains training data from information providers. Information providers are data providers that hold news articles and other text data, and the server obtains the data by sending an HTTP request to them. The obtained data is provided in JSON format, so the server parses it and converts it into a usable format.

[0546] The server then uses the acquired training data to generate fictional articles using a generative AI model. For example, it provides prompt sentences such as "A strange event happened in New York City last night" and "New technology may change the future" as input, and generates fictional news articles based on these prompts.

[0547] The server then inserts advertisements into the generated fictional stories. Advertisements are pre-prepared marketing messages or promotional content obtained from an advertising network. For example, an advertising message such as "Buy this product now!" is inserted randomly into the fictional stories by the server.

[0548] Finally, the server publishes the fictional story with the inserted advertisements on a website. Publishing is the act of distributing the generated fictional story and the inserted advertisements in a user-accessible form over the Internet or other media, so that users can view the fictional story and advertisements when they visit the website, and advertising revenue can be generated.

[0549] In this way, generative AI models can be used to generate fictional articles and effectively insert advertisements, providing new forms of entertainment and revenue.

[0550] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0551] Step 1:

[0552] The server loads the generative AI model from cloud storage. First, the server accesses cloud storage and downloads a model file called "some-ai-model". Then, it expands the downloaded model file into memory and makes it executable. The input is the cloud storage URL, and the output is the generative AI model expanded in memory.

[0553] Step 2:

[0554] The server obtains training data from information providers. This means accessing an API endpoint using an HTTP request to obtain the training data. Specifically, the server sends a GET request to "http: / / example.com / news-data" and receives news data in JSON format as a response. The input is the URL of the API endpoint, and the output is the parsed news data.

[0555] Step 3:

[0556] The server generates a fictional article using a generative AI model based on the acquired training data. First, the server creates a prompt sentence from the news data and inputs it into the generative AI model. For example, the server inputs the prompt sentence "A strange event happened in New York City last night" and obtains the article output by the model. The input is the prompt sentence, and the output is the generated fictional article.

[0557] Step 4:

[0558] The server inserts advertisements into the generated fictional article. The server obtains advertising data from an advertising network and randomly inserts it into the fictional article. For example, it inserts an advertising message such as "Buy this product now!" The input is the generated fictional article and advertising data, and the output is the fictional article with the advertisement inserted.

[0559] Step 5:

[0560] The server publishes the fictional article with the advertisements inserted on a website. Specifically, the server uploads the generated content, which is a combination of the fictional article and the advertisements, to a web server and distributes it in a form that can be accessed by users. The input is the fictional article with the advertisements inserted, and the output is the content published on the website.

[0561] (Application example 1)

[0562] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0563] Conventional advertising display systems were unable to blend in with the real world, limiting the entertainment value and advertising effectiveness for users. In particular, in situations where augmented reality technology such as smart glasses and head-mounted displays is utilized, it is necessary to display instantly generated content in a natural way, but the technological means to achieve this have not yet been established.

[0564] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0565] In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for publishing the fictional news containing the inserted advertisements, and means for displaying the generated fictional news and advertisements using augmented reality technology. This makes it possible to display the fictional news and advertisements in a manner that harmonizes with real space, thereby improving the entertainment value and advertising effectiveness for users.

[0566] A "generative language model" is an artificial intelligence model trained to generate natural-sounding sentences in the same way as humans do, based on input text data.

[0567] An "information provider" is a data holder that holds news articles and various data and from which the server obtains learning data.

[0568] "Training data" is a collection of news articles and other information used to train a generative language model.

[0569] "Fictional news" refers to fictional news that is automatically generated by a generative language model and has a sense of authenticity but does not exist in reality.

[0570] "Advertising" means marketing messages or promotional content used to promote a particular product or service.

[0571] "Publishing" refers to the act of distributing the generated fictional news and advertisements over the Internet or other media in a form accessible to users.

[0572] "Augmented reality technology" is a technology that overlays computer-generated information onto real space, and is often used with smart glasses or head-mounted displays.

[0573] A "server" is a computer system that has the functionality to load generative language models, obtain training data, generate fictional news, insert advertisements, and publish them.

[0574] Program Generation and Processing Description

[0575] 1. Load the generative language model:

[0576] The server loads a generative language model, which has been trained to generate natural-sounding sentences, using a deep learning library such as TensorFlow or PyTorch. The generative language model is loaded from cloud storage (e.g., AWS S3 or Google Cloud Storage) and deployed in memory.

[0577] 2. Obtaining training data:

[0578] The server obtains training data from the information provider (data holder). This is done using a communication library (e.g., requests) by sending an HTTP request from a specific API endpoint. The obtained data is then parsed from JSON format to extract the necessary information.

[0579] 3. Fictional News Generation:

[0580] The server provides input to a generative language model based on the training data to generate fictional news. The generated fictional news is fictional news that has a sense of authenticity but does not exist in reality. For example, a prompt sentence could be, "A mysterious event occurred in New York City last night that shocked many people. Here are the details about this event."

[0581] 4. Ad Insertion:

[0582] The server retrieves advertising data from the advertising network's API (e.g., Google Ads API) and inserts advertisements into the generated fictional news. The advertisements are inserted randomly, incorporating marketing messages and promotional content into the generated fictional news.

[0583] 5. Publication of generated fictional news and advertisements:

[0584] The server publishes fictional news stories with inserted advertisements using augmented reality technology. AR development kits such as Unity and Vuforia are used to display the stories superimposed on the real world, and the stories are visually presented to users through smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest).

[0585] Adding specific examples

[0586] The hardware used is smart glasses or a head-mounted display, which allows users to experience fictional news and advertisements generated in real space in real time.

[0587] Using the example of a specific prompt sentence, "Last night, a mysterious event occurred in New York City that shocked many people. Here are the details about the event," the generative language model generates natural-sounding fictional news.

[0588] In this way, through a series of processes, the server can provide fictional news and advertisements in a manner that is in harmony with the real world, thereby improving the entertainment value for users and the effectiveness of advertisements.

[0589] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0590] Step 1:

[0591] The server loads the generative AI model. It retrieves the generative AI model as input from cloud storage (e.g., AWS S3) and deploys it in memory using a deep learning library (e.g., TensorFlow or PyTorch). This process makes the generative AI model ready for use.

[0592] Step 2:

[0593] The server obtains training data from information providers. As input, it sends an HTTP request to a specific API endpoint to obtain news data in JSON format. The obtained data is processed using a data parsing library (e.g., json). The output of this step is structured training data.

[0594] Step 3:

[0595] The server inputs a prompt sentence into a generative language model based on the acquired training data to generate fictional news. Specifically, it inputs a prompt sentence (e.g., "A strange event happened in New York City last night") into a generative language model (e.g., GPT-3) and generates fictional news text. The output is the generated fictional news.

[0596] Step 4:

[0597] The server retrieves ad data from the ad network. It sends a request to the ad network's API (e.g., Google Ads API) as input to retrieve ad message data. The retrieved ad data is structured and converted into text format. The output of this step is the ad data.

[0598] Step 5:

[0599] The server inserts advertisements into the generated fictional news. It randomly combines the fictional news with advertisement data. Specifically, it inserts advertisement messages into appropriate positions in the fictional news. The output of this step is fictional news with advertisements inserted.

[0600] Step 6:

[0601] The server publishes the fictional news with the inserted advertisements. Specifically, the server distributes the generated content in a user-accessible form on a website or application. The output of this step is the published fictional news in a user-accessible form.

[0602] Step 7:

[0603] The server displays the generated fictional news and advertisements using augmented reality technology. Specifically, it uses an AR development kit such as Unity or Vuforia to overlay the content on smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest). The output of this step is the fictional news and advertisements visually presented to the user on the AR device.

[0604] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0605] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0606] First, the server loads a generative language model, which uses pre-trained models and is deployed in memory to enable the generation of fictional news.

[0607] Next, the server retrieves training data from information providers by sending an HTTP request to a specific API endpoint to retrieve news data from major data holders. This data is provided in JSON format, which the server parses and converts into a usable format.

[0608] The acquired training data is fed into a generative language model, and the server generates fictional news stories that appear realistic but are based on information that does not actually exist.

[0609] This is where the emotion engine comes in. Based on input data (e.g., text, voice, and images) provided by the device or user, the emotion engine recognizes the user's emotions. Based on this recognition result, the server adjusts the content of the generated fictional news. This adjustment includes the news title, body text, and even the overall tone and emotional nuances.

[0610] The server then inserts advertisements into the generated fictional news. The display of advertisements is optimized based on the recognition results of the emotion engine. For example, if the user has positive emotions, advertisements that stimulate purchasing desire are displayed, and if the user has negative emotions, advertisements for relaxation items are displayed.

[0611] Finally, the server publishes the fiction news with the advertisements inserted on the website, and users can access the website and view the tailored fiction news and the optimized advertisements.

[0612] Specific examples

[0613] The server loads the generative AI model "some-ai-model". The model is read from storage and loaded into memory.

[0614] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0615] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0616] The device analyzes the user's emotions through an emotion engine based on emotional input from the user (e.g., text message or voice). For example, if the user feels "fun," the device adjusts the news article accordingly.

[0617] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0618] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0619] The present invention enables the provision of fictional news and advertisements based on user emotions, improving the user experience and optimizing advertising revenue.

[0620] The processing flow will be explained below.

[0621] Step 1:

[0622] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name, "some-ai-model." This operation deploys the model in memory and prepares it to generate fictional news.

[0623] Step 2:

[0624] The server obtains training data from the information provider by sending an HTTP request to retrieve news data from the specified URL (e.g., "http: / / example.com / news-data"). This data is provided in JSON format, which the server parses and converts into a usable format.

[0625] Step 3:

[0626] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the model outputs fictional news, including news articles that appear believable but do not actually exist.

[0627] Step 4:

[0628] The device uses an emotion engine to recognize the user's emotions. Specifically, it analyzes emotions from text, voice, images, etc. input by the user. The analysis results in the user's emotional state (e.g., joy, sadness, surprise, etc.).

[0629] Step 5:

[0630] The server adjusts the content of the generated fictional news based on the user's emotions recognized by the emotion engine. For example, if the user is feeling "happy," the news will be tailored to a positive story that matches that emotion. On the other hand, if the user is feeling "sad," the news will be tailored to include comforting and encouraging content.

[0631] Step 6:

[0632] The server inserts advertisements into the generated fictional news. Appropriate advertisements are selected based on the recognition results of the emotion engine. For example, advertisements that stimulate purchasing desire are inserted for positive emotions, and advertisements for relaxation items are inserted for negative emotions. Advertisements are obtained from an advertising network and inserted randomly into articles.

[0633] Step 7:

[0634] The server publishes the fictional news articles, interspersed with advertisements, to the website by sending a POST request to the publishing API endpoint and uploading each article to the website, where it can then be accessed by users.

[0635] Step 8:

[0636] Users access the website and view published fiction news and advertisements. This allows the website to display fiction news and advertisements based on the user's emotions, providing a more personalized experience. Furthermore, advertising revenue is generated by maximizing the effectiveness of advertisements.

[0637] Example 2

[0638] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0639] In today's information-saturated society, the quantity and quality of content viewed by users are important. However, conventional systems have difficulty optimizing content and advertisements based on user emotions, making it difficult to improve user experience and maximize advertising revenue. Furthermore, randomly inserted advertisements may not attract user attention and may reduce advertising effectiveness. Therefore, there is a need for a system that can adjust content and advertisements based on user emotional data.

[0640] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for acquiring emotion data from a user, means for adjusting the content of the fictional news based on the emotion data, means for inserting advertisements into the generated fictional news, means for optimizing the advertisements based on the emotion data, and means for publishing the fictional news including the inserted advertisements. This makes it possible to provide fictional news and advertisements according to the user's emotions, thereby improving the user experience and optimizing advertising revenue.

[0641] A "generative language model" is a machine learning model that has been trained on a large amount of text data in advance and is used to generate new text data and complete text.

[0642] An "information provider" is an individual or organization that serves to provide training data to the server.

[0643] "Training data" refers to text data and related information used to train a generative language model.

[0644] "Fictional news" refers to fictional news articles that are based on events or phenomena that do not actually exist.

[0645] "Emotion data" is information that represents the user's emotional state and is provided in the form of text, audio, images, or the like.

[0646] An "emotion engine" is a system that has the functionality to analyze emotion data provided by a user and identify the user's emotional state.

[0647] "Advertising" means promotional content, such as messages, images, banners, etc., created to advertise a product or service.

[0648] "Advertising optimization" is the process of selecting and displaying the most effective advertisements based on a user's emotional state and preferences.

[0649] "Publishing" means making the generated fictional news and inserted advertisements available to users on a website or other platform.

[0650] A "server" is a computer system and associated software that processes, stores, and provides various types of data.

[0651] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0652] First, the server loads the generative AI model. The generative AI model to be used is "some-ai-model", which is read from the server's storage and deployed in memory to become operational. This allows new text data to be generated immediately.

[0653] Next, the server obtains training data from the information provider. In this case, the server sends an HTTP request to the API endpoint "http: / / example.com / news-data" to obtain news data from major data holders. The obtained data is provided in JSON format, which the server parses and converts into a usable format. The obtained news data includes a variety of categories, such as politics, economics, and sports.

[0654] The server inputs the acquired news data into a generative language model to generate fictional news stories, such as "A mysterious event occurred in New York City last night." Although the news stories generated are not real, they are detailed and realistically described.

[0655] Next, the emotion engine is introduced. The device or user provides emotion data such as text, voice, and images. For example, if the user feels "happy," the device sends that emotion to the emotion engine and receives the analysis results. Based on this, the emotion engine recognizes the user's emotional state.

[0656] The server receives the analysis results of the emotion engine and adjusts the content of the generated fictional news accordingly: for example, if positive emotion is detected, the news title, text, overall tone, and emotional nuances will be changed to be more cheerful.

[0657] The server then inserts advertisements into the generated fictional news. Based on the analysis results of the emotion engine, the content and display method of the advertisements are optimized. For example, if positive emotions are detected, a high-energy advertisement such as "Buy this product now!" is selected.

[0658] Finally, the server publishes the fiction news with advertisements inserted on the website, and users access the website through their browsers to view the tailored fiction news and optimized advertisements, thereby improving the user experience and increasing the effectiveness of the advertisements.

[0659] Specific examples

[0660] The server loads the generative AI model "some-ai-model", which is read from the server's storage and loaded into memory.

[0661] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0662] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0663] The device analyzes emotional input from the user (e.g., text message or voice) using an emotion engine. For example, if the user feels "fun," the device adjusts the news article based on that emotion.

[0664] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0665] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0666] Prompt Sentence Examples

[0667] "Generate a detailed fictional news story about an event. For example, about a mysterious event that happened in New York City last night."

[0668] "Tailor news articles based on users' positive emotions (e.g., 'I feel happy')."

[0669] "Generate ad copy to display when positive sentiment is detected."

[0670] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0671] Step 1:

[0672] The server reads the generative AI model "some-ai-model" from storage and expands it into memory. The input is the path to the model file in storage, and the output is the generative AI model loaded into memory. The server first retrieves the model file from a specific path in storage and then loads it into memory.

[0673] Step 2:

[0674] The server retrieves news data by sending an HTTP request to an API endpoint, such as "http: / / example.com / news-data." The input is the API endpoint URL, and the output is the news data in JSON format. The server generates an HTTP request, sends it to the API endpoint, and then parses the JSON data received as a response and converts it into a usable format.

[0675] Step 3:

[0676] The server inputs the acquired news data into a generative AI model to generate fictional news. The news data and prompt sentences are used as input, and fictional news is generated as output. Specifically, the server sends the news data and prompt sentences (e.g., "A mysterious event occurred in New York City last night") to the generative AI model and receives the fictional news generated by the model.

[0677] Step 4:

[0678] The terminal collects emotion data from the user. As input, it receives emotion data such as text messages and voice from the user, and prepares to send the emotion data to the emotion engine as output. The terminal provides an interface for, for example, speech recognition or text input, and acquires emotion data from the user.

[0679] Step 5:

[0680] The device sends the collected emotion data to the emotion engine and receives the analysis results. The device sends emotion data to the emotion engine as input and receives the emotion analysis results (positive, negative, neutral, etc.) as output. The device sends an HTTP request to the emotion engine and receives the emotion analysis results.

[0681] Step 6:

[0682] The server adjusts the content of the generated fictional news based on the results of the emotion engine. It uses the fictional news and the results of the emotion analysis as input, and obtains adjusted fictional news as output. For example, the server modifies the title, body, and tone of the news according to the emotion.

[0683] Step 7:

[0684] The server selects advertisements based on the analysis results of the emotion engine and inserts them into the fictional news. The inputs are the fictional news, the emotion analysis results, and the advertisement data obtained from the advertising network, and the output is the fictional news with the advertisement inserted. The server communicates with the advertising network to obtain the appropriate advertisements and insert them into the news article.

[0685] Step 8:

[0686] The server publishes the fictional news with advertisements inserted on a website. The server uses the fictional news with advertisements inserted as input and obtains a web page ready for user viewing as output. The server generates a web page, embeds the fictional news and advertisements, and publishes it.

[0687] Step 9:

[0688] A user accesses a website in a browser and views tailored fiction news and advertisements. The website URL is used as input, and tailored fiction news and optimized advertisements are displayed as output. The user opens the web page using a web browser and views its content.

[0689] (Application example 2)

[0690] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0691] Existing fiction news generation systems have a problem in that news content and advertisement display do not correspond to user emotions, limiting the improvement of user experience. In addition, since advertisement display is not optimized based on user emotions, it is difficult to maximize advertising revenue.

[0692] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for analyzing emotions from user input data, means for adjusting the content of the generated fictional news based on the analyzed emotions, means for inserting advertisements selected based on the analyzed emotions, and means for publishing the adjusted fictional news and the selected advertisements. This makes it possible to display news content that corresponds to the user's emotions and optimized advertisements.

[0693] A "generative language model" is an artificial intelligence model that has the ability to generate, translate, and summarize text based on a pre-trained dataset.

[0694] "Training data" is a collection of data, such as text, images, and audio, that is used to train and apply a generative language model.

[0695] "Fictional news" is fictional, artificially generated news stories that are not based on actual events.

[0696] "Advertising" means information or a message created to promote a particular product or service.

[0697] "User input data" refers to data such as text messages, voice, images, etc. that a user provides to the system.

[0698] "Analyzing emotions" refers to identifying the user's current emotional state based on input data.

[0699] "Adjusting the news content" means changing the title, text, and overall tone and emotional nuances of the generated fictional news based on the user's emotions.

[0700] "Inserting advertisements" means incorporating advertisements into the generated fictional news.

[0701] "Publishing" means making the generated fictional news and inserted advertisements accessible to users.

[0702] The present invention is a system that generates fictional news using a generative language model, analyzes user emotions using an emotion engine, and adjusts the news content and advertisement display based on the results. This system is composed of a server and a terminal.

[0703] First, the server loads the generative language model. This uses a high-performance natural language processing model such as Hugging Face's T5 model. The generative language model is loaded from storage and deployed in memory. Next, training data is obtained from information providers. This training data is often obtained using web APIs and includes a wide range of datasets from news providers, such as politics, economics, and sports. The obtained training data is provided in a format such as JSON, and the server parses it appropriately and converts it into a usable format.

[0704] The server then uses a generative language model to generate fictional news stories based on the acquired training data. The generated news stories appear realistic at first glance, but are based on events that do not actually exist. For example, if you provide a prompt such as "A mysterious event occurred in New York City last night," a detailed fictional news story will be generated based on that content.

[0705] After the fictional news is generated, a user sentiment analysis is performed. The sentiment engine analyzes sentiment based on input data (text, audio, images, etc.) provided by the user from their device (smartphone, PC, etc.). TextBlob and other sentiment analysis libraries are used for sentiment analysis. The analyzed sentiment information is classified as positive or negative.

[0706] The server adjusts the content of the generated fictional news based on the analyzed emotion information, for example, adjusting the tone of the news article to be lighthearted and fun if positive emotion is detected, or to be calmer and more soothing if negative emotion is detected.

[0707] Furthermore, based on the results of the sentiment analysis, appropriate advertisements are selected and inserted from the advertising network. For positive emotions, advertisements that stimulate purchasing desire are selected, and for negative emotions, advertisements for relaxation items are selected.

[0708] Finally, the server publishes the tailored fictional news and inserted advertisements on websites and applications, allowing users to view the news and advertisements optimized based on their emotions.

[0709] For example, if a user types the text message "I'm feeling happy!" into their device, the emotion engine will interpret this as a positive emotion. This will allow the server to adjust the tone of the news article generated based on the prompt "A strange event happened in New York City last night" to be lighthearted, and insert a high-energy advertisement saying "Buy this product now!" These will then be displayed when the user visits the site.

[0710] In this way, the present invention achieves improved user experience and optimized advertising revenue by individually optimizing news content and advertisements based on user sentiment.

[0711] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0712] Step 1:

[0713] The server loads the generative language model. This is done by extracting a high-performance natural language processing model, such as Hugging Face's T5 model, from storage and storing it in memory. The file path of the model is specified as input, and the generative language model extracted in memory is obtained as output.

[0714] Step 2:

[0715] The server obtains training data from information providers. In this step, the server sends an HTTP request to the specified API endpoint to obtain news data from the data holder. The URL of the API endpoint is given as input, and the obtained news data in JSON format is obtained as output.

[0716] Step 3:

[0717] The server generates fictional news using a generative language model based on the acquired training data. It receives a prompt sentence provided by the user (e.g., "A strange event happened in New York City last night") as input and generates fictional news using the generative language model. The prompt sentence and model input are given as input, and the generated fictional news is obtained as output.

[0718] Step 4:

[0719] The device receives user input data (e.g., text, voice, image) and analyzes the emotion using an emotion engine. The input is the user input, and the output is the analyzed emotion information (positive, negative, etc.). In this step, an emotion analysis library such as TextBlob is used.

[0720] Step 5:

[0721] The server adjusts the content of the generated fictional news based on the analyzed emotional information. For example, if positive emotions are detected, the tone of the news article is adjusted to be lighthearted and fun. The input is given as emotional information and the generated news, and the output is the adjusted news article.

[0722] Step 6:

[0723] The server selects advertisements based on the analyzed emotional information and inserts them into the generated fictional news. Advertisements that stimulate purchasing motivation are selected for positive emotions, and advertisements for relaxation items are selected for negative emotions. The input is given as emotional information and a list of available advertisements, and the output is the selected advertisements and the inserted news article.

[0724] Step 7:

[0725] The server publishes the tailored fictional news and selected advertisements on a website or application, which is then displayed when a user accesses the website or application. The input is the tailored news article and advertisements, and the output is the published news article and advertisements.

[0726] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0727] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0728] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[0729] [Fourth embodiment]

[0730] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[0731] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[0732] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0733] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[0734] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0735] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0736] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0737] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[0738] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0739] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0740] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0741] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0742] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0743] To implement the present invention, the following system configuration and processing must be included.

[0744] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model trained to generate text in a specific format based on a large dataset. By loading a pre-trained generative language model onto the server, it becomes possible to generate fictional news.

[0745] Next, the server obtains training data from information providers. Information providers are data holders that hold news articles and various data, and this data will be used to train the generative AI model. The server sends an HTTP request from a specific API endpoint to obtain the news data. This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0746] The server then generates fictional news using a generative language model based on the acquired training data. Fictional news is automatically generated by the generative language model and is fictional news that has a sense of authenticity but does not exist in reality. The server provides the training data as input to the generative language model to generate the fictional news.

[0747] Next, the server inserts advertisements into the generated fictional news. Advertisements are prepared marketing messages or promotional content that are inserted into the fictional news as a means of earning advertising revenue. The server obtains advertising data from an advertising network and randomly combines advertisements for each generated fictional news article to create the final content.

[0748] Finally, the server publishes the fictional news with the inserted advertisements on the website. Publishing is the act of distributing the generated fictional news and the inserted advertisements on the Internet or other media in a form that is accessible to users. This allows the advertisements to be displayed along with the fictional news when users visit the website, making it possible to earn advertising revenue.

[0749] Specific examples

[0750] The server loads the generative AI model "some-ai-model." For example, a generative language model is loaded from a hard disk or cloud storage and deployed in memory.

[0751] The server retrieves news data from "http: / / example.com / news-data." The retrieved data includes categories such as politics, economics, and sports.

[0752] Based on the acquired news data, the server generates fictional news stories such as "A mysterious event happened in New York City last night" or "New technology may change the future."

[0753] The server retrieves advertisements such as "Buy this product now!" from the advertising network and randomly inserts them into the generated fictional news.

[0754] The server publishes the fictional news containing the inserted advertisements on a website, so that the fictional news and advertisements can be viewed when a user accesses the website.

[0755] This invention makes it possible to effectively utilize the "hallucination" function of generative language models, thereby enabling the creation of fictional news as a new form of entertainment and information provision while simultaneously generating advertising revenue.

[0756] The processing flow will be explained below.

[0757] Step 1:

[0758] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name. This operation deploys the model in memory and prepares it for generating fictional news.

[0759] Step 2:

[0760] The server obtains training data from the information provider. Specifically, it sends an HTTP request to obtain data from the specified URL. For example, it obtains news data from "http: / / example.com / news-data." This data is provided in JSON format, so the server parses the obtained data and converts it into a usable format.

[0761] Step 3:

[0762] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the fictional news output by the model is obtained. This fictional news has a sense of verisimilitude, but is fictional news that does not exist in reality.

[0763] Step 4:

[0764] The server inserts advertisements into the generated fictional news. Specifically, it obtains advertisement data from an advertising network and randomly inserts advertisements into each fictional news article. For example, it obtains an advertisement "Buy this product now!" from the advertising network and inserts it into the generated news article.

[0765] Step 5:

[0766] The server publishes the fictional news articles, interspersed with advertisements, to a website by sending a POST request to a publishing API endpoint and uploading each article to the website, where it can be accessed by users over the internet.

[0767] Step 6:

[0768] Users access the website and view the published fiction news and advertisements, thereby generating advertising revenue along with the viewing of the fiction news.

[0769] Example 1

[0770] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0771] Providing information quickly and generating revenue are important challenges in today's world. However, while there are existing systems that combine real news with advertising, there are no systems that generate fictional articles, insert advertisements into them, and publish them, creating new forms of entertainment and revenue opportunities. Therefore, there is a need for a system that uses generative artificial intelligence models to automatically generate fictional articles and effectively insert advertisements into them for publication.

[0772] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0773] In this invention, the server includes means for loading a generative artificial intelligence model, means for acquiring training data from an information provider, means for generating fictional articles using the generative artificial intelligence model based on the acquired training data, means for inserting advertisements into the generated fictional articles, and means for publishing the fictional articles including the inserted advertisements, thereby enabling the automatic generation of fictional articles and the effective insertion and publication of advertisements.

[0774] A "generative artificial intelligence model" is an artificial intelligence model that learns from large datasets and enables language generation.

[0775] An "information provider" refers to a data provider that has training data and provides the data necessary to train and generate a generative model.

[0776] "Training data" refers to large sets of text and other data used to train generative models.

[0777] "Fictional articles" are fictional news articles or article-style content generated by a generative model based on training data.

[0778] "Advertising" means marketing messages or promotional content that are inserted into and displayed alongside fictional articles.

[0779] "Publication" refers to the act of distributing the generated fictional article and inserted advertisements over the Internet or other media in a manner that makes it accessible to users.

[0780] To implement the present invention, the following system configuration and processes are involved: The components of the server, generative AI model, information provider, advertising network, and website work together.

[0781] First, the server loads a generative AI model. A generative AI model is an artificial intelligence model that enables language generation based on a large dataset and is specialized for generating news articles. The server downloads this model from cloud storage, expands it into memory, and makes it executable.

[0782] Next, the server obtains training data from information providers. Information providers are data providers that hold news articles and other text data, and the server obtains the data by sending an HTTP request to them. The obtained data is provided in JSON format, so the server parses it and converts it into a usable format.

[0783] The server then uses the acquired training data to generate fictional articles using a generative AI model. For example, it provides prompt sentences such as "A strange event happened in New York City last night" and "New technology may change the future" as input, and generates fictional news articles based on these prompts.

[0784] The server then inserts advertisements into the generated fictional stories. Advertisements are pre-prepared marketing messages or promotional content obtained from an advertising network. For example, an advertising message such as "Buy this product now!" is inserted randomly into the fictional stories by the server.

[0785] Finally, the server publishes the fictional story with the inserted advertisements on a website. Publishing is the act of distributing the generated fictional story and the inserted advertisements in a user-accessible form over the Internet or other media, so that users can view the fictional story and advertisements when they visit the website, and advertising revenue can be generated.

[0786] In this way, generative AI models can be used to generate fictional articles and effectively insert advertisements, providing new forms of entertainment and revenue.

[0787] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0788] Step 1:

[0789] The server loads the generative AI model from cloud storage. First, the server accesses cloud storage and downloads a model file called "some-ai-model". Then, it expands the downloaded model file into memory and makes it executable. The input is the cloud storage URL, and the output is the generative AI model expanded in memory.

[0790] Step 2:

[0791] The server obtains training data from information providers. This means accessing an API endpoint using an HTTP request to obtain the training data. Specifically, the server sends a GET request to "http: / / example.com / news-data" and receives news data in JSON format as a response. The input is the URL of the API endpoint, and the output is the parsed news data.

[0792] Step 3:

[0793] The server generates a fictional article using a generative AI model based on the acquired training data. First, the server creates a prompt sentence from the news data and inputs it into the generative AI model. For example, the server inputs the prompt sentence "A strange event happened in New York City last night" and obtains the article output by the model. The input is the prompt sentence, and the output is the generated fictional article.

[0794] Step 4:

[0795] The server inserts advertisements into the generated fictional article. The server obtains advertising data from an advertising network and randomly inserts it into the fictional article. For example, it inserts an advertising message such as "Buy this product now!" The input is the generated fictional article and advertising data, and the output is the fictional article with the advertisement inserted.

[0796] Step 5:

[0797] The server publishes the fictional article with the advertisements inserted on a website. Specifically, the server uploads the generated content, which is a combination of the fictional article and the advertisements, to a web server and distributes it in a form that can be accessed by users. The input is the fictional article with the advertisements inserted, and the output is the content published on the website.

[0798] (Application example 1)

[0799] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0800] Conventional advertising display systems were unable to blend in with the real world, limiting the entertainment value and advertising effectiveness for users. In particular, in situations where augmented reality technology such as smart glasses and head-mounted displays is utilized, it is necessary to display instantly generated content in a natural way, but the technological means to achieve this have not yet been established.

[0801] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0802] In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for publishing the fictional news containing the inserted advertisements, and means for displaying the generated fictional news and advertisements using augmented reality technology. This makes it possible to display the fictional news and advertisements in a manner that harmonizes with real space, thereby improving the entertainment value and advertising effectiveness for users.

[0803] A "generative language model" is an artificial intelligence model trained to generate natural-sounding sentences in the same way as humans do, based on input text data.

[0804] An "information provider" is a data holder that holds news articles and various data and from which the server obtains learning data.

[0805] "Training data" is a collection of news articles and other information used to train a generative language model.

[0806] "Fictional news" refers to fictional news that is automatically generated by a generative language model and has a sense of authenticity but does not exist in reality.

[0807] "Advertising" means marketing messages or promotional content used to promote a particular product or service.

[0808] "Publishing" refers to the act of distributing the generated fictional news and advertisements over the Internet or other media in a form accessible to users.

[0809] "Augmented reality technology" is a technology that overlays computer-generated information onto real space, and is often used with smart glasses or head-mounted displays.

[0810] A "server" is a computer system that has the functionality to load generative language models, obtain training data, generate fictional news, insert advertisements, and publish them.

[0811] Program Generation and Processing Description

[0812] 1. Load the generative language model:

[0813] The server loads a generative language model, which has been trained to generate natural-sounding sentences, using a deep learning library such as TensorFlow or PyTorch. The generative language model is loaded from cloud storage (e.g., AWS S3 or Google Cloud Storage) and deployed in memory.

[0814] 2. Obtaining training data:

[0815] The server obtains training data from the information provider (data holder). This is done using a communication library (e.g., requests) by sending an HTTP request from a specific API endpoint. The obtained data is then parsed from JSON format to extract the necessary information.

[0816] 3. Fictional News Generation:

[0817] The server provides input to a generative language model based on the training data to generate fictional news. The generated fictional news is fictional news that has a sense of authenticity but does not exist in reality. For example, a prompt sentence could be, "A mysterious event occurred in New York City last night that shocked many people. Here are the details about this event."

[0818] 4. Ad Insertion:

[0819] The server retrieves advertising data from the advertising network's API (e.g., Google Ads API) and inserts advertisements into the generated fictional news. The advertisements are inserted randomly, incorporating marketing messages and promotional content into the generated fictional news.

[0820] 5. Publication of generated fictional news and advertisements:

[0821] The server publishes fictional news stories with inserted advertisements using augmented reality technology. AR development kits such as Unity and Vuforia are used to display the stories superimposed on the real world, and the stories are visually presented to users through smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest).

[0822] Adding specific examples

[0823] The hardware used is smart glasses or a head-mounted display, which allows users to experience fictional news and advertisements generated in real space in real time.

[0824] Using the example of a specific prompt sentence, "Last night, a mysterious event occurred in New York City that shocked many people. Here are the details about the event," the generative language model generates natural-sounding fictional news.

[0825] In this way, through a series of processes, the server can provide fictional news and advertisements in a manner that is in harmony with the real world, thereby improving the entertainment value for users and the effectiveness of advertisements.

[0826] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0827] Step 1:

[0828] The server loads the generative AI model. It retrieves the generative AI model as input from cloud storage (e.g., AWS S3) and deploys it in memory using a deep learning library (e.g., TensorFlow or PyTorch). This process makes the generative AI model ready for use.

[0829] Step 2:

[0830] The server obtains training data from information providers. As input, it sends an HTTP request to a specific API endpoint to obtain news data in JSON format. The obtained data is processed using a data parsing library (e.g., json). The output of this step is structured training data.

[0831] Step 3:

[0832] The server inputs a prompt sentence into a generative language model based on the acquired training data to generate fictional news. Specifically, it inputs a prompt sentence (e.g., "A strange event happened in New York City last night") into a generative language model (e.g., GPT-3) and generates fictional news text. The output is the generated fictional news.

[0833] Step 4:

[0834] The server retrieves ad data from the ad network. It sends a request to the ad network's API (e.g., Google Ads API) as input to retrieve ad message data. The retrieved ad data is structured and converted into text format. The output of this step is the ad data.

[0835] Step 5:

[0836] The server inserts advertisements into the generated fictional news. It randomly combines the fictional news with advertisement data. Specifically, it inserts advertisement messages into appropriate positions in the fictional news. The output of this step is fictional news with advertisements inserted.

[0837] Step 6:

[0838] The server publishes the fictional news with the inserted advertisements. Specifically, the server distributes the generated content in a user-accessible form on a website or application. The output of this step is the published fictional news in a user-accessible form.

[0839] Step 7:

[0840] The server displays the generated fictional news and advertisements using augmented reality technology. Specifically, it uses an AR development kit such as Unity or Vuforia to overlay the content on smart glasses or head-mounted displays (e.g., Microsoft HoloLens or Oculus Quest). The output of this step is the fictional news and advertisements visually presented to the user on the AR device.

[0841] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0842] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0843] First, the server loads a generative language model, which uses pre-trained models and is deployed in memory to enable the generation of fictional news.

[0844] Next, the server retrieves training data from information providers by sending an HTTP request to a specific API endpoint to retrieve news data from major data holders. This data is provided in JSON format, which the server parses and converts into a usable format.

[0845] The acquired training data is fed into a generative language model, and the server generates fictional news stories that appear realistic but are based on information that does not actually exist.

[0846] This is where the emotion engine comes in. Based on input data (e.g., text, voice, and images) provided by the device or user, the emotion engine recognizes the user's emotions. Based on this recognition result, the server adjusts the content of the generated fictional news. This adjustment includes the news title, body text, and even the overall tone and emotional nuances.

[0847] The server then inserts advertisements into the generated fictional news. The display of advertisements is optimized based on the recognition results of the emotion engine. For example, if the user has positive emotions, advertisements that stimulate purchasing desire are displayed, and if the user has negative emotions, advertisements for relaxation items are displayed.

[0848] Finally, the server publishes the fiction news with the advertisements inserted on the website, and users can access the website and view the tailored fiction news and the optimized advertisements.

[0849] Specific examples

[0850] The server loads the generative AI model "some-ai-model". The model is read from storage and loaded into memory.

[0851] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0852] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0853] The device analyzes the user's emotions through an emotion engine based on emotional input from the user (e.g., text message or voice). For example, if the user feels "fun," the device adjusts the news article accordingly.

[0854] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0855] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0856] The present invention enables the provision of fictional news and advertisements based on user emotions, improving the user experience and optimizing advertising revenue.

[0857] The processing flow will be explained below.

[0858] Step 1:

[0859] The server loads the generative language model. Specifically, it imports the library that manages generative AI models and loads the specified model name, "some-ai-model." This operation deploys the model in memory and prepares it to generate fictional news.

[0860] Step 2:

[0861] The server obtains training data from the information provider by sending an HTTP request to retrieve news data from the specified URL (e.g., "http: / / example.com / news-data"). This data is provided in JSON format, which the server parses and converts into a usable format.

[0862] Step 3:

[0863] The server generates fictional news using a generative language model based on the acquired training data. Specifically, the parsed news data is provided as input to the generative language model, and the model outputs fictional news, including news articles that appear believable but do not actually exist.

[0864] Step 4:

[0865] The device uses an emotion engine to recognize the user's emotions. Specifically, it analyzes emotions from text, voice, images, etc. input by the user. The analysis results in the user's emotional state (e.g., joy, sadness, surprise, etc.).

[0866] Step 5:

[0867] The server adjusts the content of the generated fictional news based on the user's emotions recognized by the emotion engine. For example, if the user is feeling "happy," the news will be tailored to a positive story that matches that emotion. On the other hand, if the user is feeling "sad," the news will be tailored to include comforting and encouraging content.

[0868] Step 6:

[0869] The server inserts advertisements into the generated fictional news. Appropriate advertisements are selected based on the recognition results of the emotion engine. For example, advertisements that stimulate purchasing desire are inserted for positive emotions, and advertisements for relaxation items are inserted for negative emotions. Advertisements are obtained from an advertising network and inserted randomly into articles.

[0870] Step 7:

[0871] The server publishes the fictional news articles, interspersed with advertisements, to the website by sending a POST request to the publishing API endpoint and uploading each article to the website, where it can then be accessed by users.

[0872] Step 8:

[0873] Users access the website and view published fiction news and advertisements. This allows the website to display fiction news and advertisements based on the user's emotions, providing a more personalized experience. Furthermore, advertising revenue is generated by maximizing the effectiveness of advertisements.

[0874] Example 2

[0875] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0876] In today's information-saturated society, the quantity and quality of content viewed by users are important. However, conventional systems have difficulty optimizing content and advertisements based on user emotions, making it difficult to improve user experience and maximize advertising revenue. Furthermore, randomly inserted advertisements may not attract user attention and may reduce advertising effectiveness. Therefore, there is a need for a system that can adjust content and advertisements based on user emotional data.

[0877] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for acquiring emotion data from a user, means for adjusting the content of the fictional news based on the emotion data, means for inserting advertisements into the generated fictional news, means for optimizing the advertisements based on the emotion data, and means for publishing the fictional news including the inserted advertisements. This makes it possible to provide fictional news and advertisements according to the user's emotions, thereby improving the user experience and optimizing advertising revenue.

[0878] A "generative language model" is a machine learning model that has been trained on a large amount of text data in advance and is used to generate new text data and complete text.

[0879] An "information provider" is an individual or organization that serves to provide training data to the server.

[0880] "Training data" refers to text data and related information used to train a generative language model.

[0881] "Fictional news" refers to fictional news articles that are based on events or phenomena that do not actually exist.

[0882] "Emotion data" is information that represents the user's emotional state and is provided in the form of text, audio, images, or the like.

[0883] An "emotion engine" is a system that has the functionality to analyze emotion data provided by a user and identify the user's emotional state.

[0884] "Advertising" means promotional content, such as messages, images, banners, etc., created to advertise a product or service.

[0885] "Advertising optimization" is the process of selecting and displaying the most effective advertisements based on a user's emotional state and preferences.

[0886] "Publishing" means making the generated fictional news and inserted advertisements available to users on a website or other platform.

[0887] A "server" is a computer system and associated software that processes, stores, and provides various types of data.

[0888] The present invention combines a fictional news generation system using a generative language model with an emotion engine, optimizing user experience and advertising revenue by adjusting news content and advertising display based on user emotions.

[0889] First, the server loads the generative AI model. The generative AI model to be used is "some-ai-model", which is read from the server's storage and deployed in memory to become operational. This allows new text data to be generated immediately.

[0890] Next, the server obtains training data from the information provider. In this case, the server sends an HTTP request to the API endpoint "http: / / example.com / news-data" to obtain news data from major data holders. The obtained data is provided in JSON format, which the server parses and converts into a usable format. The obtained news data includes a variety of categories, such as politics, economics, and sports.

[0891] The server inputs the acquired news data into a generative language model to generate fictional news stories, such as "A mysterious event occurred in New York City last night." Although the news stories generated are not real, they are detailed and realistically described.

[0892] Next, the emotion engine is introduced. The device or user provides emotion data such as text, voice, and images. For example, if the user feels "happy," the device sends that emotion to the emotion engine and receives the analysis results. Based on this, the emotion engine recognizes the user's emotional state.

[0893] The server receives the analysis results of the emotion engine and adjusts the content of the generated fictional news accordingly: for example, if positive emotion is detected, the news title, text, overall tone, and emotional nuances will be changed to be more cheerful.

[0894] The server then inserts advertisements into the generated fictional news. Based on the analysis results of the emotion engine, the content and display method of the advertisements are optimized. For example, if positive emotions are detected, a high-energy advertisement such as "Buy this product now!" is selected.

[0895] Finally, the server publishes the fiction news with advertisements inserted on the website, and users access the website through their browsers to view the tailored fiction news and optimized advertisements, thereby improving the user experience and increasing the effectiveness of the advertisements.

[0896] Specific examples

[0897] The server loads the generative AI model "some-ai-model", which is read from the server's storage and loaded into memory.

[0898] The server retrieves news data from "http: / / example.com / news-data", which includes politics, economics, sports, etc.

[0899] Based on the acquired news data, fictional news such as "A mysterious event occurred in New York City last night" is generated.

[0900] The device analyzes emotional input from the user (e.g., text message or voice) using an emotion engine. For example, if the user feels "fun," the device adjusts the news article based on that emotion.

[0901] The server retrieves "Buy this product now!" ads from the ad network and inserts them into the generated fictional news. The emotion engine detects positive emotions, so the high-energy ads are selected.

[0902] The server publishes the fictional news with the advertisements inserted on a website, and when a user accesses the website, the adjusted fictional news and the advertisements selected based on the user's emotions are displayed.

[0903] Prompt Sentence Examples

[0904] "Generate a detailed fictional news story about an event. For example, about a mysterious event that happened in New York City last night."

[0905] "Tailor news articles based on users' positive emotions (e.g., 'I feel happy')."

[0906] "Generate ad copy to display when positive sentiment is detected."

[0907] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0908] Step 1:

[0909] The server reads the generative AI model "some-ai-model" from storage and expands it into memory. The input is the path to the model file in storage, and the output is the generative AI model loaded into memory. The server first retrieves the model file from a specific path in storage and then loads it into memory.

[0910] Step 2:

[0911] The server retrieves news data by sending an HTTP request to an API endpoint, such as "http: / / example.com / news-data." The input is the API endpoint URL, and the output is the news data in JSON format. The server generates an HTTP request, sends it to the API endpoint, and then parses the JSON data received as a response and converts it into a usable format.

[0912] Step 3:

[0913] The server inputs the acquired news data into a generative AI model to generate fictional news. The news data and prompt sentences are used as input, and fictional news is generated as output. Specifically, the server sends the news data and prompt sentences (e.g., "A mysterious event occurred in New York City last night") to the generative AI model and receives the fictional news generated by the model.

[0914] Step 4:

[0915] The terminal collects emotion data from the user. As input, it receives emotion data such as text messages and voice from the user, and prepares to send the emotion data to the emotion engine as output. The terminal provides an interface for, for example, speech recognition or text input, and acquires emotion data from the user.

[0916] Step 5:

[0917] The device sends the collected emotion data to the emotion engine and receives the analysis results. The device sends emotion data to the emotion engine as input and receives the emotion analysis results (positive, negative, neutral, etc.) as output. The device sends an HTTP request to the emotion engine and receives the emotion analysis results.

[0918] Step 6:

[0919] The server adjusts the content of the generated fictional news based on the results of the emotion engine. It uses the fictional news and the results of the emotion analysis as input, and obtains adjusted fictional news as output. For example, the server modifies the title, body, and tone of the news according to the emotion.

[0920] Step 7:

[0921] The server selects advertisements based on the analysis results of the emotion engine and inserts them into the fictional news. The inputs are the fictional news, the emotion analysis results, and the advertisement data obtained from the advertising network, and the output is the fictional news with the advertisement inserted. The server communicates with the advertising network to obtain the appropriate advertisements and insert them into the news article.

[0922] Step 8:

[0923] The server publishes the fictional news with advertisements inserted on a website. The server uses the fictional news with advertisements inserted as input and obtains a web page ready for user viewing as output. The server generates a web page, embeds the fictional news and advertisements, and publishes it.

[0924] Step 9:

[0925] A user accesses a website in a browser and views tailored fiction news and advertisements. The website URL is used as input, and tailored fiction news and optimized advertisements are displayed as output. The user opens the web page using a web browser and views its content.

[0926] (Application example 2)

[0927] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[0928] Existing fiction news generation systems have a problem in that news content and advertisement display do not correspond to user emotions, limiting the improvement of user experience. In addition, since advertisement display is not optimized based on user emotions, it is difficult to maximize advertising revenue.

[0929] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes means for loading a generative language model, means for acquiring training data from an information provider, means for generating fictional news using the generative language model based on the acquired training data, means for inserting advertisements into the generated fictional news, means for analyzing emotions from user input data, means for adjusting the content of the generated fictional news based on the analyzed emotions, means for inserting advertisements selected based on the analyzed emotions, and means for publishing the adjusted fictional news and the selected advertisements. This makes it possible to display news content that corresponds to the user's emotions and optimized advertisements.

[0930] A "generative language model" is an artificial intelligence model that has the ability to generate, translate, and summarize text based on a pre-trained dataset.

[0931] "Training data" is a collection of data, such as text, images, and audio, that is used to train and apply a generative language model.

[0932] "Fictional news" is fictional, artificially generated news stories that are not based on actual events.

[0933] "Advertising" means information or a message created to promote a particular product or service.

[0934] "User input data" refers to data such as text messages, voice, images, etc. that a user provides to the system.

[0935] "Analyzing emotions" refers to identifying the user's current emotional state based on input data.

[0936] "Adjusting the news content" means changing the title, text, and overall tone and emotional nuances of the generated fictional news based on the user's emotions.

[0937] "Inserting advertisements" means incorporating advertisements into the generated fictional news.

[0938] "Publishing" means making the generated fictional news and inserted advertisements accessible to users.

[0939] The present invention is a system that generates fictional news using a generative language model, analyzes user emotions using an emotion engine, and adjusts the news content and advertisement display based on the results. This system is composed of a server and a terminal.

[0940] First, the server loads the generative language model. This uses a high-performance natural language processing model such as Hugging Face's T5 model. The generative language model is loaded from storage and deployed in memory. Next, training data is obtained from information providers. This training data is often obtained using web APIs and includes a wide range of datasets from news providers, such as politics, economics, and sports. The obtained training data is provided in a format such as JSON, and the server parses it appropriately and converts it into a usable format.

[0941] The server then uses a generative language model to generate fictional news stories based on the acquired training data. The generated news stories appear realistic at first glance, but are based on events that do not actually exist. For example, if you provide a prompt such as "A mysterious event occurred in New York City last night," a detailed fictional news story will be generated based on that content.

[0942] After the fictional news is generated, a user sentiment analysis is performed. The sentiment engine analyzes sentiment based on input data (text, audio, images, etc.) provided by the user from their device (smartphone, PC, etc.). TextBlob and other sentiment analysis libraries are used for sentiment analysis. The analyzed sentiment information is classified as positive or negative.

[0943] The server adjusts the content of the generated fictional news based on the analyzed emotion information, for example, adjusting the tone of the news article to be lighthearted and fun if positive emotion is detected, or to be calmer and more soothing if negative emotion is detected.

[0944] Furthermore, based on the results of the sentiment analysis, appropriate advertisements are selected and inserted from the advertising network. For positive emotions, advertisements that stimulate purchasing desire are selected, and for negative emotions, advertisements for relaxation items are selected.

[0945] Finally, the server publishes the tailored fictional news and inserted advertisements on websites and applications, allowing users to view the news and advertisements optimized based on their emotions.

[0946] For example, if a user types the text message "I'm feeling happy!" into their device, the emotion engine will interpret this as a positive emotion. This will allow the server to adjust the tone of the news article generated based on the prompt "A strange event happened in New York City last night" to be lighthearted, and insert a high-energy advertisement saying "Buy this product now!" These will then be displayed when the user visits the site.

[0947] In this way, the present invention achieves improved user experience and optimized advertising revenue by individually optimizing news content and advertisements based on user sentiment.

[0948] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0949] Step 1:

[0950] The server loads the generative language model. This is done by extracting a high-performance natural language processing model, such as Hugging Face's T5 model, from storage and storing it in memory. The file path of the model is specified as input, and the generative language model extracted in memory is obtained as output.

[0951] Step 2:

[0952] The server obtains training data from information providers. In this step, the server sends an HTTP request to the specified API endpoint to obtain news data from the data holder. The URL of the API endpoint is given as input, and the obtained news data in JSON format is obtained as output.

[0953] Step 3:

[0954] The server generates fictional news using a generative language model based on the acquired training data. It receives a prompt sentence provided by the user (e.g., "A strange event happened in New York City last night") as input and generates fictional news using the generative language model. The prompt sentence and model input are given as input, and the generated fictional news is obtained as output.

[0955] Step 4:

[0956] The device receives user input data (e.g., text, voice, image) and analyzes the emotion using an emotion engine. The input is the user input, and the output is the analyzed emotion information (positive, negative, etc.). In this step, an emotion analysis library such as TextBlob is used.

[0957] Step 5:

[0958] The server adjusts the content of the generated fictional news based on the analyzed emotional information. For example, if positive emotions are detected, the tone of the news article is adjusted to be lighthearted and fun. The input is given as emotional information and the generated news, and the output is the adjusted news article.

[0959] Step 6:

[0960] The server selects advertisements based on the analyzed emotional information and inserts them into the generated fictional news. Advertisements that stimulate purchasing motivation are selected for positive emotions, and advertisements for relaxation items are selected for negative emotions. The input is given as emotional information and a list of available advertisements, and the output is the selected advertisements and the inserted news article.

[0961] Step 7:

[0962] The server publishes the tailored fictional news and selected advertisements on a website or application, which is then displayed when a user accesses the website or application. The input is the tailored news article and advertisements, and the output is the published news article and advertisements.

[0963] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[0964] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0965] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[0966] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[0967] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[0968] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[0969] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[0970] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[0971] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[0972] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[0973] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[0974] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[0975] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[0976] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[0977] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[0978] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[0979] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[0980] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[0981] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[0982] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[0983] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[0984] The following is further disclosed regarding the above embodiment.

[0985] (Claim 1)

[0986] means for loading a generative language model;

[0987] A means for acquiring training data from an information provider;

[0988] A means for generating fictional news using a generative language model based on the acquired training data;

[0989] means for inserting advertisements into the generated fictional news;

[0990] a means for publishing fictional news with inserted advertisements;

[0991] A system including:

[0992] (Claim 2)

[0993] The system according to claim 1, wherein the information provider according to claim 1 is a means for acquiring learning data from a data holder.

[0994] (Claim 3)

[0995] 10. The system of claim 1, further comprising means for randomly inserting advertisements into the generated fictional news of claim 1.

[0996] "Example 1"

[0997] (Claim 1)

[0998] means for loading a generative artificial intelligence model;

[0999] A means for acquiring training data from an information provider;

[1000] A means for generating fictional articles using a generative artificial intelligence model based on the acquired learning data;

[1001] means for inserting advertisements into the generated fictional articles;

[1002] a means for publishing fictional articles containing inserted advertisements;

[1003] A system including:

[1004] (Claim 2)

[1005] 2. The system according to claim 1, wherein the information provider is a means for acquiring training data from a data supplier.

[1006] (Claim 3)

[1007] 10. The system of claim 1, further comprising means for randomly inserting advertisements into the generated fictional articles.

[1008] "Application Example 1"

[1009] (Claim 1)

[1010] means for loading a generative language model;

[1011] A means for acquiring training data from an information provider;

[1012] A means for generating fictional news using a generative language model based on the acquired training data;

[1013] means for inserting advertisements into the generated fictional news;

[1014] a means for publishing fictional news with inserted advertisements;

[1015] a means for displaying the generated fictional news and advertisements using augmented reality technology;

[1016] A system including:

[1017] (Claim 2)

[1018] 2. The system according to claim 1, wherein the information provider is a means for acquiring training data from a data holder.

[1019] (Claim 3)

[1020] 10. The system of claim 1, further comprising means for randomly inserting advertisements into the generated fictional news.

[1021] "Example 2: Combining Emotion Engines"

[1022] (Claim 1)

[1023] means for loading a generative language model;

[1024] A means for acquiring training data from an information provider;

[1025] A means for generating fictional news using a generative language model based on the acquired training data;

[1026] means for acquiring emotion data from a user;

[1027] a means for adjusting the content of fictional news based on the sentiment data;

[1028] means for inserting advertisements into the generated fictional news;

[1029] a means for optimizing advertising based on sentiment data;

[1030] a means for publishing fictional news with inserted advertisements;

[1031] A system including:

[1032] (Claim 2)

[1033] The system according to claim 1, wherein the information provider is a means for acquiring news data from major data holders.

[1034] (Claim 3)

[1035] 10. The system of claim 1, further comprising means for adjusting the tone and nuance of the generated fictional news based on the emotion data.

[1036] "Application example 2 when combining emotion engines"

[1037] (Claim 1)

[1038] means for loading a generative language model;

[1039] A means for acquiring training data from an information provider;

[1040] A means for generating fictional news using a generative language model based on the acquired training data;

[1041] means for inserting advertisements into the generated fictional news;

[1042] means for analyzing emotions from user input data;

[1043] A means for adjusting the content of the generated fictional news based on the analyzed sentiment;

[1044] means for inserting selected advertisements based on the analyzed sentiment;

[1045] means for publishing tailored fiction news and selected advertisements;

[1046] A system including:

[1047] (Claim 2)

[1048] The system of claim 1, wherein the training data is obtained from a data holder.

[1049] (Claim 3)

[1050] 10. The system of claim 1, wherein advertisements are inserted into the generated fictional news based on user sentiment. [Explanation of symbols]

[1051] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for loading a generative language model; A means for acquiring training data from an information provider; A means for generating fictional news using a generative language model based on the acquired training data; means for inserting advertisements into the generated fictional news; a means for publishing fictional news with inserted advertisements; A system including:

2. 2. The system according to claim 1, wherein the information provider according to claim 1 is a means for acquiring learning data from a data holder.

3. 2. The system of claim 1, further comprising means for randomly inserting advertisements into the generated fictional news of claim 1.

Images