system

Abstract

We provide the system. [Solution] A means for training an artificial intelligence model using collected text data, A means for preprocessing user-input text using natural language processing technology, A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, A means for displaying the calculated probability to the user, A system that includes this.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, the method including: receiving a user utterance; adding the user utterance to a prompt including an instruction sentence related to an explanation of a character of the chatbot; 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

Summary of the Invention

Problems to be Solved by the Invention

[0004] The present invention relates to a technology for determining whether a text is written by a human or generated by a generative artificial intelligence. In the prior art, since text generation by generative artificial intelligence has advanced highly, it is difficult for a user to identify this by his / her own judgment. As a result, there is a possibility that a decision-making based on trust in incorrect information may occur. Regarding this point, it is an object of the present invention to provide a system that enables a user to easily determine the origin of a text.

Means for Solving the Problems

[0005] The present invention solves the above-mentioned problems by providing the following means: a system comprising means for training an artificial intelligence model using collected text data; means for preprocessing text input from a user using natural language processing technology; means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence; and means for displaying the calculated probability to the user. With this system, the user can easily determine the source of the text, thereby preventing decision-making based on the reliability of false information.

[0006] "Collected text data" refers to a wide variety of text data gathered from the internet and online sources.

[0007] An "artificial intelligence model" is a computer model that uses machine learning algorithms to learn from data and identify patterns.

[0008] "User-inputted text" refers to the text that a user provides to the system.

[0009] "Natural language processing technology" refers to the technology that enables computers to understand, interpret, and generate human language.

[0010] "Preprocessing" refers to the process of modifying text, such as tokenization, stop word removal, and word morphology, to make it easier to analyze.

[0011] "Generative artificial intelligence" refers to machine learning algorithms and models that have the ability to generate new text based on specific prompts.

[0012] "Means for calculating probability" refers to methods or devices that calculate the probability of an input text being created by a generative artificial intelligence.

[0013] "Means of display" refers to methods or devices for providing data and information to users visually. [Brief explanation of the drawing]

[0014] [Figure 1] This is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] This is a conceptual diagram showing an example of the essential functions of a data processing device and a smart device according to the first embodiment. [Figure 3] This is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4] This is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] This is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] This is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] This is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] This is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] This shows an emotion map where multiple emotions are mapped. [Figure 10] This shows an emotion map where multiple emotions are mapped. [Figure 11] This is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] This is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] This is a sequence diagram showing the processing flow of the data processing system in Example 2, when an emotion engine is combined. [Figure 14]It is a sequence diagram showing the processing flow of a data processing system in Application Example 2 when a sentiment engine is combined.

Embodiments for Carrying Out the Invention

[0015] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0016] First, the terms used in the following description will be explained.

[0017] In the following embodiments, a numbered processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Also, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), an APU (Accelerated Processing Unit), and the like.

[0018] In the following embodiments, a numbered RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a work memory by the processor.

[0019] In the following embodiments, a numbered storage is one or more non-volatile storage devices that store various programs and various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes, etc.

[0020] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0021] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0022] [First Embodiment]

[0023] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0024] As shown in Figure 1, the 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.

[0025] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0026] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0027] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and 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.

[0028] 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 perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0029] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0030] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0031] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0032] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0033] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0034] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0035] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0036] 1. Means for training an artificial intelligence model using collected text data:

[0037] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0038] For training, deep learning frameworks such as TENSORFLOW® or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or by a generative artificial intelligence.

[0039] 2. Means for preprocessing user-input text using natural language processing techniques:

[0040] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0041] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0042] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0043] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0044] 4. Means for displaying the calculated probability to the user:

[0045] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0046] To explain with a concrete example, it would be as follows:

[0047] Specific example 1:

[0048] User A enters the following sentence into the terminal: "This product is very easy to use."

[0049] The terminal sends the entered text to the server.

[0050] The server preprocesses the received text and extracts features.

[0051] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0052] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0053] Specific example 2:

[0054] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0055] The terminal sends the entered text to the server.

[0056] The server preprocesses the received text and extracts features.

[0057] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0058] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0059] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text and provide the results to the user.

[0060] The following describes the processing flow.

[0061] Step 1:

[0062] The server collects data. The server gathers diverse text data from the internet and other online sources. This includes news articles, blog entries, and social media posts.

[0063] Step 2:

[0064] The server constructs a dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and assigns labels to each.

[0065] Step 3:

[0066] The server performs preprocessing using natural language processing techniques. The server tokenizes the collected data and then processes it by removing stop words, stemming, rematizing, and other methods to prepare the data.

[0067] Step 4:

[0068] The server builds and trains a machine learning model. The server uses a deep learning framework to build an artificial intelligence model and trains it using a prepared dataset. Training involves inputting data features and performing supervised learning to enable the model to distinguish between human text and text generated by generative AI.

[0069] Step 5:

[0070] The user enters the text they want to evaluate. The user instructs their device to input the text and send it to the server.

[0071] Step 6:

[0072] The terminal sends the text to the server. The terminal collects the text entered by the user and sends it to the server.

[0073] Step 7:

[0074] The server performs preprocessing. The server preprocesses the received text again, performing tokenization, stemming, stop word removal, feature extraction, and other operations.

[0075] Step 8:

[0076] The server inputs text into a trained artificial intelligence model. The server inputs the pre-processed text into the AI ​​model and calculates the probability that it was generated by a generative AI.

[0077] Step 9:

[0078] The server sends the calculation result to the terminal. The server sends the calculated score to the terminal.

[0079] Step 10:

[0080] The device displays the results to the user. The device analyzes the received score and displays to the user, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0081] (Example 1)

[0082] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0083] Conventional methods make it difficult to determine whether text entered by a user was written by a human or generated by generative artificial intelligence. This leads to problems such as the inability to make accurate judgments, which can result in issues like the spread of misinformation and the inability to properly evaluate reliability. The present invention aims to solve these problems.

[0084] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0085] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, and means for displaying the calculated probability to the user. This makes it possible to determine with high accuracy whether the text input by the user was written by a human or generated by a generative artificial intelligence.

[0086] "Collected text data" refers to text written by humans and text generated by generative artificial intelligence, collected from various sources.

[0087] An "artificial intelligence model" refers to a model that is trained on collected data and has an algorithmic structure that can determine whether input text was written by a human or generated by a generative artificial intelligence.

[0088] "Natural language processing technology" refers to techniques that convert text into a machine-readable format using methods such as tokenization, stem extraction, stop word removal, and feature extraction.

[0089] "Preprocessing" refers to a series of processes performed to convert raw text data into a format that is easy to analyze, specifically steps such as tokenization, stemming, and stop word removal.

[0090] "Generative artificial intelligence" refers to artificial intelligence systems that perform natural language generation, specifically the technology that automatically generates text based on specific prompts.

[0091] "Means for calculating probability" refers to a method of inputting pre-processed text data into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence.

[0092] "Means of display" refers to a device or method that visually presents the calculated probability score to the user, specifically, means of displaying the results on the display of a terminal or the screen of a mobile device.

[0093] The present invention is a system that determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0094] 1. Means for training an artificial intelligence model using collected text data:

[0095] The server trains artificial intelligence models using text data collected from diverse sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0096] The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0097] For training, deep learning frameworks (such as TensorFlow or PyTorch) are used.

[0098] The server uses this data to perform supervised learning and build a model that can determine whether the text was written by a human or by a generative artificial intelligence.

[0099] 2. Means for preprocessing user-input text using natural language processing techniques:

[0100] The user enters a specific message into the device and sends it.

[0101] The device will send this message to the server.

[0102] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0103] 3. A means for inputting preprocessed text into an artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0104] The server inputs the pre-processed text into a trained artificial intelligence model.

[0105] The model analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative AI.

[0106] 4. Means for displaying the calculated probability to the user:

[0107] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0108] Specific example

[0109] Specific example 1:

[0110] User A enters the following sentence into the terminal: "This product is very easy to use."

[0111] The terminal sends the entered text to the server.

[0112] The server preprocesses the received text and extracts features.

[0113] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0114] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0115] Specific example 2:

[0116] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0117] The terminal sends the entered text to the server.

[0118] The server preprocesses the received text and extracts features.

[0119] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0120] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0121] The system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text through these processes and provide the results to the user.

[0122] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0123] Step 1:

[0124] Data collection

[0125] The server collects text data from various sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0126] Input: Text data from publicly available databases on the internet and independently collected data sources.

[0127] Output: A set of collected text data.

[0128] Step 2:

[0129] Data preprocessing

[0130] The server tokenizes the collected text data. This is the process of dividing the text into units of words or phrases.

[0131] The server then performs stemming and stop word removal. Stemming is the process of converting words into their base forms, while stop word removal is the process of removing common words that are not needed for analysis.

[0132] Input: Collected text data

[0133] Output: Preprocessed text data (data that has undergone tokenization, stemming, and stop word removal)

[0134] Step 3:

[0135] Training of artificial intelligence models

[0136] The server uses pre-processed data to create a training dataset, which includes labels for human-written text and text generated by generative artificial intelligence.

[0137] The server uses a deep learning framework (such as TensorFlow or PyTorch) to perform supervised learning and train an artificial intelligence model.

[0138] Input: Preprocessed text data and its corresponding labels

[0139] Output: Trained artificial intelligence model

[0140] Step 4:

[0141] Text input from the user

[0142] The user enters a specific sentence into the input field on the terminal and presses the send button.

[0143] Input: Text entered by the user

[0144] Output: Text data sent to the terminal

[0145] Step 5:

[0146] Sending and receiving input data

[0147] The terminal sends the text entered by the user to the server.

[0148] The server receives the text that was sent.

[0149] Input: Text data sent from the device

[0150] Output: Text data received by the server

[0151] Step 6:

[0152] Reprocessing of received data

[0153] The server reprocesses the received text. Specifically, it performs tokenization, stemming, stop word removal, and feature extraction.

[0154] Input: Text data received by the server

[0155] Output: Preprocessed text data

[0156] Step 7:

[0157] Score calculation

[0158] The server inputs the pre-processed text data into a trained artificial intelligence model.

[0159] The model analyzes the input text data and calculates a probability score generated by generative artificial intelligence. This score is expressed as a probability.

[0160] Input: Pre-processed text data

[0161] Output: Probability score generated by generative artificial intelligence

[0162] Step 8:

[0163] Submitting and displaying scores

[0164] The server sends the calculated score to the terminal.

[0165] The device displays a score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0166] Input: Calculated score

[0167] Output: Score displayed to the user

[0168] In this way, through each processing step, the system can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user.

[0169] (Application Example 1)

[0170] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0171] In today's internet environment, phishing attacks and fake news are rampant, threatening user security. Users need to be able to quickly detect these threats and receive appropriate warnings, but current technology lacks the means to do so effectively. In particular, with the widespread use of text created by generative artificial intelligence, there is a need for a reliable system that can distinguish between text written by humans and text generated by generative AI and warn users accordingly.

[0172] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0173] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, and means for detecting phishing attacks and fake news and issuing warnings. This enables users to quickly recognize the threat of phishing attacks and fake news and take appropriate measures.

[0174] "Collected text data" refers to text data collected from various sources and used to train artificial intelligence models.

[0175] An "artificial intelligence model" is trained on collected text data to determine whether a text was written by a human or generated by a generative artificial intelligence system.

[0176] "Preprocessing" refers to the process of preparing user-input text into a format that is easy to analyze, using techniques such as tokenization, stem extraction, and stop word removal.

[0177] "Generative artificial intelligence" refers to technology that automatically generates text using natural language generation models.

[0178] "Calculated probability" refers to the probability that the preprocessed text was generated by a generative artificial intelligence, and is calculated by the AI ​​model.

[0179] "Means of displaying to the user" refers to an interface for visually or audibly communicating the calculated probability to the user.

[0180] A "phishing attack" refers to a fraudulent message or email sent by a malicious third party to illegally obtain a user's personal information.

[0181] "Fake news" refers to news that is not based on actual facts and conveys false information, with the aim of misleading readers.

[0182] "Warning mechanisms" are part of a system designed to inform users of potential phishing attacks or fake news and to draw their attention to them.

[0183] Embodiments of this invention will now be described. First, the server trains an artificial intelligence model using text data collected from various sources. The collected text data includes both human-written text and text generated by generative artificial intelligence. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (e.g., TensorFlow or PyTorch) are used for preprocessing.

[0184] The user inputs a specific sentence into a device such as a smartphone or computer and sends it. The device then sends this sentence to a server. The server preprocesses the received sentence again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes it easier to analyze the sentence.

[0185] The server inputs pre-processed text into a trained artificial intelligence model, which analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[0186] The server then sends the calculated score to the terminal, which displays this score to the user. The display uses a format such as, "There is a XX% chance that this text was generated by a generative artificial intelligence." In addition, a warning is displayed if a phishing attack or fake news is detected.

[0187] As a concrete example, a user enters the following sentence into the terminal: "Click this link for a chance to win a free gift card." The terminal sends the entered sentence to the server, which preprocesses the received sentence and extracts features. The server then uses a trained model to analyze the sentence and calculate a score indicating the likelihood that it was generated by generative artificial intelligence. For example, if this score is 85%, the terminal displays to the user "There is an 85% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message is highly likely to have been generated by generative artificial intelligence."

[0188] As another example, a user enters the following sentence into the terminal: "Due to a new government decision, all bank accounts will be suspended." The terminal sends the entered sentence to a server, which preprocesses the received sentence and extracts features. The server analyzes the sentence using a trained model and calculates a score indicating the likelihood that it was generated by generative artificial intelligence. If this score is 90%, the terminal displays to the user "There is a 90% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message may be generated. Please verify whether it is genuine information."

[0189] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user. Furthermore, it enables rapid recognition and response to threats such as phishing attacks and fake news.

[0190] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0191] Step 1:

[0192] The server trains an artificial intelligence model using text data collected from diverse sources. This collected text data includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (TensorFlow and PyTorch) are used for preprocessing. The input is the collected text data, and the output is the preprocessed data.

[0193] Step 2:

[0194] The user inputs a specific sentence into a device such as a smartphone or computer and sends that sentence. The input is the sentence entered by the user, and the output is the raw data sent from the device to the server.

[0195] Step 3:

[0196] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier. The input is raw data sent from the terminal, and the output is preprocessed data.

[0197] Step 4:

[0198] The server inputs pre-processed text data into a trained artificial intelligence model, which analyzes the data to calculate a probability score of how likely it was generated by a generative AI. The input is pre-processed data, and the output is the probability score.

[0199] Step 5:

[0200] The server sends the calculated score to the terminal. The input is the probability score, and the output is the transmission of the score data to the terminal.

[0201] Step 6:

[0202] The device displays the received score to the user. The display format might be, for example, "There is an 85% chance that this text was generated by a generative artificial intelligence." In addition, if there is a high probability of a phishing attack or fake news, a warning will be displayed such as, "Warning: This message is highly likely to have been generated by a generative artificial intelligence." The input is the score data sent from the server, and the output is what is displayed to the user.

[0203] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0204] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of the following main components.

[0205] 1. Means for training an artificial intelligence model using collected text data:

[0206] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0207] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[0208] 2. Means for preprocessing user-input text using natural language processing techniques:

[0209] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0210] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0211] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0212] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0213] 4. Means for displaying the calculated probability to the user:

[0214] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0215] 5. Further includes an emotion engine to recognize user emotions:

[0216] The server incorporates an emotion engine to analyze the emotions in the text entered by the user. This emotion engine analyzes the emotional tone and nuances of the text and determines whether the emotion is positive, negative, neutral, etc.

[0217] 6. Means for displaying the results of user sentiment analysis:

[0218] The server sends the results analyzed by the emotion engine to the terminal, and the terminal displays the emotion analysis results to the user. The display takes the form of "The emotion contained in this text is positive."

[0219] To explain with a concrete example, it would be as follows:

[0220] Specific example 1:

[0221] User A enters the following sentence into the terminal: "This product is very easy to use."

[0222] The terminal sends the entered text to the server.

[0223] The server preprocesses the received text and extracts features.

[0224] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0225] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0226] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be positive, the terminal displays to the user, "The emotion contained in this text is positive."

[0227] Specific example 2:

[0228] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0229] The terminal sends the entered text to the server.

[0230] The server preprocesses the received text and extracts features.

[0231] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0232] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0233] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be neutral, the terminal displays to the user, "The emotion contained in this text is neutral."

[0234] As described above, the system of the present invention can effectively distinguish between text written by a generative artificial intelligence system and text written by a human, and further analyze the user's emotions and provide the results to the user.

[0235] The following describes the processing flow.

[0236] Step 1:

[0237] The server collects data. The server gathers diverse text data from the internet and other online sources. This data includes news articles, blog entries, and social media posts.

[0238] Step 2:

[0239] The server builds the dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and then creates a dataset by assigning labels to each.

[0240] Step 3:

[0241] The server performs data preprocessing using natural language processing techniques. The server tokenizes the data, removes stop words, extracts stems and rematization, and extracts features.

[0242] Step 4:

[0243] The server builds a machine learning model. The server trains the artificial intelligence model using a deep learning framework. This includes supervised learning using text data within the dataset.

[0244] Step 5:

[0245] The user enters the text they want to evaluate. The user enters a specific sentence into their device and clicks the send button.

[0246] Step 6:

[0247] The terminal sends the entered text to the server. The terminal converts the text entered by the user into a packet format and sends it to the server.

[0248] Step 7:

[0249] The server preprocesses the received text. The server tokenizes the received text, performs stemming, stop word removal, and feature extraction to make it easier to analyze.

[0250] Step 8:

[0251] The server inputs pre-processed text into an artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. The server then calculates this score and determines the probability that the text was generated by a generative AI.

[0252] Step 9:

[0253] The server uses an emotion engine to recognize the emotion of a text. The server analyzes the text, which has been preprocessed by the emotion engine, and identifies the emotional tone and nuances within the text. As a result, it determines the emotion, such as positive, negative, or neutral.

[0254] Step 10:

[0255] The server sends the calculation results and emotion analysis results to the terminal. The server stores the potential score of the generative artificial intelligence and the emotion analysis results in a packet and sends it to the terminal.

[0256] Step 11:

[0257] The device displays the results to the user. The device analyzes the received score and sentiment analysis results and displays to the user, "There is a XX% chance that this text was generated by generative artificial intelligence." It also displays, for example, "The sentiment contained in this text is positive."

[0258] The following are specific examples.

[0259] Specific example 1:

[0260] User A enters the following sentence into the terminal: "This product is very easy to use."

[0261] The terminal sends the entered text to the server.

[0262] The server preprocesses the received text and extracts features.

[0263] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0264] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0265] The server analyzes the text using an emotion engine and determines that the emotion is positive. The terminal then displays to the user, "The emotion contained in this text is positive."

[0266] Specific example 2:

[0267] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0268] The terminal sends the entered text to the server.

[0269] The server preprocesses the received text and extracts features.

[0270] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0271] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0272] The server analyzes the text using an emotion engine and determines that the emotion is neutral. The terminal then displays to the user, "The emotion contained in this text is neutral."

[0273] As a result, the system of the present invention can provide users with not only the ability to discriminate text using generative artificial intelligence, but also the results of sentiment analysis of the text's content.

[0274] (Example 2)

[0275] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0276] In modern text generation technology, it is difficult to distinguish between text written by humans and text generated by generative artificial intelligence. Furthermore, there is a need to analyze the sentiment of user-input text and provide feedback to more accurately understand the user's intent. However, existing technologies do not provide a single system that offers both of these functions. Therefore, the challenge is to provide a unified system that determines whether user-input text was generated by generative artificial intelligence or written by a human, and further analyzes and provides feedback on its sentiment.

[0277] The specific processing by the specific processing unit 290 of the data processing apparatus 12 in Example 2 is realized by the following means.

[0278] In this invention, the server includes means for training an artificial intelligence model using the collected text data, means for preprocessing the text input by the user by natural language processing technology, means for calculating the possibility of being generated by a generative artificial intelligence by inputting the preprocessed text into the artificial intelligence model, means for displaying the calculated possibility to the user, means for analyzing the sentiment of the text input by the user, and means for displaying the sentiment analysis result to the user. Thereby, it is possible to determine whether the text input by the user is by a generative artificial intelligence or by a human, perform sentiment analysis, and feedback the result to the user.

[0279] The "collected text data" is data of texts written by humans and texts generated by generative artificial intelligence collected from various sources.

[0280] The "artificial intelligence model" is a model constructed using deep learning technology and trained to distinguish between texts written by humans and texts generated by generative artificial intelligence.

[0281] The "natural language processing technology" is a technology for analyzing and preprocessing texts, such as tokenization, stemming, stop word removal, and feature extraction of texts.

[0282] The "generative artificial intelligence" is an artificial intelligence that has been pre-trained and has the ability to generate new texts based on the input text.

[0283] The "means for calculating the possibility" is means for inputting the preprocessed text into the artificial intelligence model and calculating the possibility that the text is generated by the generative artificial intelligence as a probability value.

[0284] The "means for displaying to the user" is means for displaying the calculated possibility and the sentiment analysis result to the user's terminal.

[0285] The "means for analyzing sentiment" is a means for analyzing the sentiment tone of the text input by the user and determining whether the text corresponds to positive, negative, or neutral.

[0286] The "means for displaying the sentiment analysis result" is a means for visually presenting the analyzed sentiment result to the user.

[0287] The system of the present invention combines a sentiment engine that determines whether the text input by the user is generated by a generative artificial intelligence or written by a human, and further recognizes the user's sentiment. This system is composed of a server, a terminal, and a user.

[0288] 1. Generation of the program

[0289] The server trains an artificial intelligence model using the collected text data. This data includes both texts written by humans and texts generated by generative artificial intelligence. The server preprocesses this data, performing tokenization (splitting of words), stemming (extraction of the root part of words), removal of stop words (frequently occurring words), etc. Based on the preprocessed data, supervised learning is performed using deep learning frameworks such as TensorFlow and PyTorch to construct an artificial intelligence model.

[0290] 2. Processing of the program

[0291] The server preprocesses the text input by the user using natural language processing technology. As a result, the newly input text is also converted into an analyzable format. The content of the preprocessing consists of tokenization, stemming, stop word removal, and feature extraction.

[0292] Next, the server inputs the preprocessed text into the trained artificial intelligence model and calculates the score of the possibility of being generated by a generative artificial intelligence. This score indicates the probability that the text is generated by a generative artificial intelligence.

[0293] 3. Hardware and Software

[0294] The servers utilize high-performance server clusters and cloud computing services. Furthermore, TensorFlow and PyTorch are used as software for natural language processing and deep learning. These software programs are specialized for preprocessing text data and training models.

[0295] 4. Data processing and calculations

[0296] The server first tokenizes the input text data, then performs stemming, stop word removal, and feature extraction. This allows for efficient text analysis. Subsequently, the pre-processed data is input into an artificial intelligence model to obtain analysis results. These results are expressed as a score and provided to the user.

[0297] 5. Specific Examples

[0298] For example, user A enters the following sentence into the terminal: "This product is very easy to use."

[0299] The terminal sends the entered text to the server.

[0300] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 20%, the terminal will display "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0301] The server analyzes the text using its sentiment engine, and if it determines that the sentiment is positive, the terminal displays the message "The sentiment contained in this text is positive." to the user.

[0302] As another example, user B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0303] The terminal sends the input text to the server.

[0304] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 75%, the terminal displays "There is a 75% chance that this text was generated by a generative AI."

[0305] When the server analyzes the text with an emotion engine and determines it to be neutral, the terminal displays to the user "The emotion contained in this text is neutral."

[0306] In this way, the present invention realizes a system that can effectively distinguish between texts generated by generative AI and those written by humans, and further analyze the user's emotion and provide the result to the user.

[0307] The flow of specific processing in Example 2 will be described using FIG. 13.

[0308] Flow of processing

[0309] Step 1: The user inputs and sends a text

[0310] Input: The user inputs a text into the input box.

[0311] Specific operation: The user inputs an arbitrary text into the input form of the terminal and clicks the send button. For example, input the sentence "This product is very easy to use."

[0312] Output: The input text is passed to the terminal.

[0313] Step 2: The terminal sends the text to the server

[0314] Input: The input text passed to the terminal.

[0315] Specific operation: The terminal sends the text entered by the user to the server as text data. The transmission is performed as an HTTP request.

[0316] Output: The input text is passed to the server.

[0317] Step 3: The server preprocesses the text.

[0318] Input: The text sent to the server.

[0319] Specific operation: The server tokenizes the received text and breaks it down into individual words. For example, the sentence "This product is very easy to use." is split into tokens such as "this," "product," "is," "very," "easy to use," and "is." Next, the server extracts the stems of the tokens and converts "easy to use" to "easy to use." Furthermore, unnecessary words (stop words) are removed, and words like "is" and "is" are deleted. Finally, features are extracted and formatted in a way that a machine learning model can understand.

[0320] Output: Preprocessed data.

[0321] Step 4: The server parses the pre-processed text.

[0322] Input: Pre-processed data.

[0323] Specific operation: The server inputs pre-processed data into a trained artificial intelligence model, which then analyzes the data.

[0324] Output: Analysis results (probability score of generation by generative AI).

[0325] Step 5: The server calculates the potential score of the generative AI.

[0326] Input: Analysis results.

[0327] Specific operation: The server obtains the output of the artificial intelligence model as a probability value. If the model determines "20%", it indicates that there is a 20% chance that the text was generated by a generative AI.

[0328] Output: Potential score for generative AI.

[0329] Step 6: The device displays the score to the user.

[0330] Input: Potential score for generative AI.

[0331] Specific operation: The server sends the calculated score to the terminal, and the terminal displays this score to the user. The display format will be "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0332] Output: The user checks the score.

[0333] Step 7: The server analyzes emotions using the emotion engine.

[0334] Input: Text entered by the user.

[0335] Specific operation: The server uses an emotion engine to analyze the emotional tone of a sentence. For example, the sentence "This product is very easy to use" is determined to have a positive emotion.

[0336] Output: Emotion analysis results.

[0337] Step 8: The device displays the sentiment analysis results to the user.

[0338] Input: Sentiment analysis results.

[0339] Specific operation: The server sends the analysis results to the terminal, and the terminal displays to the user in the form of "The emotions contained in this text are positive."

[0340] Output: The user reviews the sentiment analysis results.

[0341] Through these steps, the system of the present invention can determine whether the text entered by the user was generated by a generative artificial intelligence system or by a human, analyze the emotions contained in the text, and provide feedback to the user.

[0342] (Application Example 2)

[0343] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0344] Conventional systems have struggled to determine whether user-generated text is created by artificial intelligence or by a human, and to analyze the emotions contained within that text. In particular, the food delivery industry requires both authenticity verification and sentiment analysis of input text to efficiently manage customer support and reviews. This will enable more accurate and faster service delivery, leading to improved customer satisfaction.

[0345] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0346] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability of it being generated by generative artificial intelligence, means for displaying the calculated probability to the user, means for performing sentiment analysis on the text input from the user, and means for displaying the results of the sentiment analysis to the user. This enables more efficient customer support and review management in the food delivery industry, contributing to improved customer satisfaction.

[0347] "Collected text data" refers to text information obtained from various sources, including text written by humans and text generated by generative artificial intelligence.

[0348] An "artificial intelligence model" is an algorithm used to learn from collected data and perform a specific task.

[0349] "Training methods" refer to the process of inputting a wide variety of data into an artificial intelligence model to improve its accuracy and discriminatory capabilities.

[0350] "Natural language processing technology" refers to technologies that enable computers to understand, interpret, and generate human language, and includes techniques such as tokenization, stem extraction, and stop word removal.

[0351] "Preprocessing" refers to the process of converting raw text data into a parseable format, which includes tokenization and stop word removal.

[0352] "Generative artificial intelligence" refers to algorithms that use trained models to generate text input from users, and which have a high accuracy in text generation.

[0353] "Means for calculating the probability of generation" refers to the process of inputting a pre-processed text into an artificial intelligence model and calculating a probability score indicating whether or not the text was generated by a generative artificial intelligence.

[0354] "Means of displaying to the user" refers to an interface for providing the user with the results calculated by the artificial intelligence model, and includes screen display and audio output.

[0355] "Sentiment analysis" is the process of analyzing the emotional tone and nuances contained in text to determine whether the emotion is positive, negative, neutral, etc.

[0356] The system of the present invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further analyzes the emotions contained in that text. In order to specifically implement the system of the present invention, the following means and processes are necessary.

[0357] First, the server trains an artificial intelligence model using text data collected from diverse sources. During this process, the server uses deep learning frameworks such as TensorFlow and PyTorch to preprocess data containing both human-written text and text generated by generative AI. This preprocessing includes tokenization, stem extraction, and stop word removal to prepare the training data.

[0358] Next, the user uses a smartphone or other device to input and send a specific sentence. The device sends this sentence to the server, which then preprocesses the received sentence again. This re-preprocessing involves tokenization, stemming, stop word removal, and feature extraction.

[0359] The pre-processed text is input into a server-trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[0360] The calculated score is sent from the server to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0361] Furthermore, the server performs sentiment analysis on the text entered by the user. This process uses a sentiment engine that analyzes the emotional tone and nuances of the text and determines emotions such as positive, negative, or neutral. Natural language processing tools such as the nlptown / bert-base-multilingual-uncased-sentiment model are used for sentiment analysis.

[0362] The analysis results are sent from the server to the terminal, which then displays the sentiment analysis results to the user. The display takes the form of statements such as, "The sentiment contained in this text is positive," or "The sentiment contained in this text is neutral."

[0363] Specific example:

[0364] The user left the following review: "This delivery is fantastic! Fast and reliable."

[0365] In this case, the system will display the following result:

[0366] "There is a 15.50% probability that this text was generated by a generative artificial intelligence." (Example: Calculated using training data)

[0367] "The emotions conveyed in this text are positive."

[0368] Examples of prompts to input into a generative AI model:

[0369] User input: "This delivery is fantastic! Fast and reliable."

[0370] analysis:

[0371] 1. There is a 15.50% probability that this text was generated by a generative artificial intelligence.

[0372] 2. The emotions contained in this sentence are positive.

[0373] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0374] Step 1:

[0375] Users input and send specific text using their smartphones or other devices. The entered text is sent from the device to the server in its original form.

[0376] Input: Text entered by the user (e.g., "This delivery is fantastic! Fast and reliable.")

[0377] Output: Sent text data

[0378] Step 2:

[0379] The server preprocesses the received text using natural language processing techniques. This preprocessing step involves tokenization, stemming, and stop word removal. This converts the text into a format that is easy to analyze.

[0380] Input: Submitted text data

[0381] Output: Preprocessed text data

[0382] Specific operation: Use the BERT tokenizer to tokenize the text, remove unnecessary words, and format it for analysis.

[0383] Step 3:

[0384] The server inputs pre-processed text into a trained artificial intelligence model, which then calculates a probability score for the text as it was generated by the generative AI. The model analyzes the input text data and outputs the result as a probability score.

[0385] Input: Preprocessed text data

[0386] Output: Probability score generated by generative artificial intelligence (e.g., 15.50%)

[0387] Specific operation: Preprocessed text is input into a BERT-based model, and the classification result is output as a probability score.

[0388] Step 4:

[0389] The server sends the calculated score for the generative artificial intelligence to the terminal. The terminal displays this score to the user. The display format is "There is a XX% chance that this text was generated by generative artificial intelligence."

[0390] Input: Score of the Generative Artificial Intelligence

[0391] Output: Content displayed to the user

[0392] Specific action: The submitted score is displayed on the terminal's user interface.

[0393] Step 5:

[0394] The server performs sentiment analysis on text entered by the user. It uses a sentiment engine to analyze the emotional tone and nuances of the text. The analysis results are classified as positive, negative, neutral, etc.

[0395] Input: Preprocessed text data

[0396] Output: Sentiment analysis results (e.g., positive)

[0397] Specific operation: Classify the sentiment of a text using the nlptown / bert-base-multilingual-uncased-sentiment model and obtain the result.

[0398] Step 6:

[0399] The server sends the sentiment analysis results to the terminal, and the terminal displays the results to the user. The display format is something like, "The sentiment contained in this text is positive."

[0400] Input: Sentiment analysis results

[0401] Output: Content displayed to the user

[0402] Specific operation: The transmitted sentiment analysis results are displayed on the device's user interface.

[0403] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0404] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0405] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0406] [Second Embodiment]

[0407] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0408] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0409] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0410] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0411] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0412] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0413] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0414] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0415] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0416] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0417] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0418] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. 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".

[0419] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0420] 1. Means for training an artificial intelligence model using collected text data:

[0421] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0422] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[0423] 2. Means for preprocessing user-input text using natural language processing techniques:

[0424] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0425] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0426] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0427] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0428] 4. Means for displaying the calculated probability to the user:

[0429] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0430] To explain with a concrete example, it would be as follows:

[0431] Specific example 1:

[0432] User A enters the following sentence into the terminal: "This product is very easy to use."

[0433] The terminal sends the entered text to the server.

[0434] The server preprocesses the received text and extracts features.

[0435] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0436] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0437] Specific example 2:

[0438] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0439] The terminal sends the entered text to the server.

[0440] The server preprocesses the received text and extracts features.

[0441] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0442] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0443] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text and provide the results to the user.

[0444] The following describes the processing flow.

[0445] Step 1:

[0446] The server collects data. The server gathers diverse text data from the internet and other online sources. This includes news articles, blog entries, and social media posts.

[0447] Step 2:

[0448] The server constructs a dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and assigns labels to each.

[0449] Step 3:

[0450] The server performs preprocessing using natural language processing techniques. The server tokenizes the collected data and then processes it by removing stop words, stemming, rematizing, and other methods to prepare the data.

[0451] Step 4:

[0452] The server builds and trains a machine learning model. The server uses a deep learning framework to build an artificial intelligence model and trains it using a prepared dataset. Training involves inputting data features and performing supervised learning to enable the model to distinguish between human text and text generated by generative AI.

[0453] Step 5:

[0454] The user enters the text they want to evaluate. The user instructs their device to input the text and send it to the server.

[0455] Step 6:

[0456] The terminal sends the text to the server. The terminal collects the text entered by the user and sends it to the server.

[0457] Step 7:

[0458] The server performs preprocessing. The server preprocesses the received text again, performing tokenization, stemming, stop word removal, feature extraction, and other operations.

[0459] Step 8:

[0460] The server inputs text into a trained artificial intelligence model. The server inputs the pre-processed text into the AI ​​model and calculates the probability that it was generated by a generative AI.

[0461] Step 9:

[0462] The server sends the calculation result to the terminal. The server sends the calculated score to the terminal.

[0463] Step 10:

[0464] The device displays the results to the user. The device analyzes the received score and displays to the user, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0465] (Example 1)

[0466] Next, we will describe Example 1. 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."

[0467] Conventional methods make it difficult to determine whether text entered by a user was written by a human or generated by generative artificial intelligence. This leads to problems such as the inability to make accurate judgments, which can result in issues like the spread of misinformation and the inability to properly evaluate reliability. The present invention aims to solve these problems.

[0468] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0469] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, and means for displaying the calculated probability to the user. This makes it possible to determine with high accuracy whether the text input by the user was written by a human or generated by a generative artificial intelligence.

[0470] "Collected text data" refers to text written by humans and text generated by generative artificial intelligence, collected from various sources.

[0471] An "artificial intelligence model" refers to a model that is trained on collected data and has an algorithmic structure that can determine whether input text was written by a human or generated by a generative artificial intelligence.

[0472] "Natural language processing technology" refers to techniques that convert text into a machine-readable format using methods such as tokenization, stem extraction, stop word removal, and feature extraction.

[0473] "Preprocessing" refers to a series of processes performed to convert raw text data into a format that is easy to analyze, specifically steps such as tokenization, stemming, and stop word removal.

[0474] "Generative artificial intelligence" refers to artificial intelligence systems that perform natural language generation, specifically the technology that automatically generates text based on specific prompts.

[0475] "Means for calculating probability" refers to a method of inputting pre-processed text data into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence.

[0476] "Means of display" refers to a device or method that visually presents the calculated probability score to the user, specifically, means of displaying the results on the display of a terminal or the screen of a mobile device.

[0477] The present invention is a system that determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0478] 1. Means for training an artificial intelligence model using collected text data:

[0479] The server trains artificial intelligence models using text data collected from diverse sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0480] The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0481] For training, deep learning frameworks (such as TensorFlow or PyTorch) are used.

[0482] The server uses this data to perform supervised learning and build a model that can determine whether the text was written by a human or by a generative artificial intelligence.

[0483] 2. Means for preprocessing user-input text using natural language processing techniques:

[0484] The user enters a specific message into the device and sends it.

[0485] The device will send this message to the server.

[0486] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0487] 3. A means for inputting preprocessed text into an artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0488] The server inputs the pre-processed text into a trained artificial intelligence model.

[0489] The model analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative AI.

[0490] 4. Means for displaying the calculated probability to the user:

[0491] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0492] Specific example

[0493] Specific example 1:

[0494] User A enters the following sentence into the terminal: "This product is very easy to use."

[0495] The terminal sends the entered text to the server.

[0496] The server preprocesses the received text and extracts features.

[0497] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0498] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0499] Specific example 2:

[0500] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0501] The terminal sends the entered text to the server.

[0502] The server preprocesses the received text and extracts features.

[0503] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0504] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0505] The system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text through these processes and provide the results to the user.

[0506] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0507] Step 1:

[0508] Data collection

[0509] The server collects text data from various sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0510] Input: Text data from publicly available databases on the internet and independently collected data sources.

[0511] Output: A set of collected text data.

[0512] Step 2:

[0513] Data preprocessing

[0514] The server tokenizes the collected text data. This is the process of dividing the text into units of words or phrases.

[0515] The server then performs stemming and stop word removal. Stemming is the process of converting words into their base forms, while stop word removal is the process of removing common words that are not needed for analysis.

[0516] Input: Collected text data

[0517] Output: Preprocessed text data (data that has undergone tokenization, stemming, and stop word removal)

[0518] Step 3:

[0519] Training of artificial intelligence models

[0520] The server uses pre-processed data to create a training dataset, which includes labels for human-written text and text generated by generative artificial intelligence.

[0521] The server uses a deep learning framework (such as TensorFlow or PyTorch) to perform supervised learning and train an artificial intelligence model.

[0522] Input: Preprocessed text data and its corresponding labels

[0523] Output: Trained artificial intelligence model

[0524] Step 4:

[0525] Text input from the user

[0526] The user enters a specific sentence into the input field on the terminal and presses the send button.

[0527] Input: Text entered by the user

[0528] Output: Text data sent to the terminal

[0529] Step 5:

[0530] Sending and receiving input data

[0531] The terminal sends the text entered by the user to the server.

[0532] The server receives the text that was sent.

[0533] Input: Text data sent from the device

[0534] Output: Text data received by the server

[0535] Step 6:

[0536] Reprocessing of received data

[0537] The server reprocesses the received text. Specifically, it performs tokenization, stemming, stop word removal, and feature extraction.

[0538] Input: Text data received by the server

[0539] Output: Preprocessed text data

[0540] Step 7:

[0541] Score calculation

[0542] The server inputs the pre-processed text data into a trained artificial intelligence model.

[0543] The model analyzes the input text data and calculates a probability score generated by generative artificial intelligence. This score is expressed as a probability.

[0544] Input: Pre-processed text data

[0545] Output: Probability score generated by generative artificial intelligence

[0546] Step 8:

[0547] Submitting and displaying scores

[0548] The server sends the calculated score to the terminal.

[0549] The device displays a score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0550] Input: Calculated score

[0551] Output: Score displayed to the user

[0552] In this way, through each processing step, the system can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user.

[0553] (Application Example 1)

[0554] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0555] In today's internet environment, phishing attacks and fake news are rampant, threatening user security. Users need to be able to quickly detect these threats and receive appropriate warnings, but current technology lacks the means to do so effectively. In particular, with the widespread use of text created by generative artificial intelligence, there is a need for a reliable system that can distinguish between text written by humans and text generated by generative AI and warn users accordingly.

[0556] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0557] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, and means for detecting phishing attacks and fake news and issuing warnings. This enables users to quickly recognize the threat of phishing attacks and fake news and take appropriate measures.

[0558] "Collected text data" refers to text data collected from various sources and used to train artificial intelligence models.

[0559] An "artificial intelligence model" is trained on collected text data to determine whether a text was written by a human or generated by a generative artificial intelligence system.

[0560] "Preprocessing" refers to the process of preparing user-input text into a format that is easy to analyze, using techniques such as tokenization, stem extraction, and stop word removal.

[0561] "Generative artificial intelligence" refers to technology that automatically generates text using natural language generation models.

[0562] "Calculated probability" refers to the probability that the preprocessed text was generated by a generative artificial intelligence, and is calculated by the AI ​​model.

[0563] "Means of displaying to the user" refers to an interface for visually or audibly communicating the calculated probability to the user.

[0564] A "phishing attack" refers to a fraudulent message or email sent by a malicious third party to illegally obtain a user's personal information.

[0565] "Fake news" refers to news that is not based on actual facts and conveys false information, with the aim of misleading readers.

[0566] "Warning mechanisms" are part of a system designed to inform users of potential phishing attacks or fake news and to draw their attention to them.

[0567] Embodiments of this invention will now be described. First, the server trains an artificial intelligence model using text data collected from various sources. The collected text data includes both human-written text and text generated by generative artificial intelligence. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (e.g., TensorFlow or PyTorch) are used for preprocessing.

[0568] The user inputs a specific sentence into a device such as a smartphone or computer and sends it. The device then sends this sentence to a server. The server preprocesses the received sentence again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes it easier to analyze the sentence.

[0569] The server inputs pre-processed text into a trained artificial intelligence model, which analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[0570] The server then sends the calculated score to the terminal, which displays this score to the user. The display uses a format such as, "There is a XX% chance that this text was generated by a generative artificial intelligence." In addition, a warning is displayed if a phishing attack or fake news is detected.

[0571] As a concrete example, a user enters the following sentence into the terminal: "Click this link for a chance to win a free gift card." The terminal sends the entered sentence to the server, which preprocesses the received sentence and extracts features. The server then uses a trained model to analyze the sentence and calculate a score indicating the likelihood that it was generated by generative artificial intelligence. For example, if this score is 85%, the terminal displays to the user "There is an 85% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message is highly likely to have been generated by generative artificial intelligence."

[0572] As another example, a user enters the following sentence into the terminal: "Due to a new government decision, all bank accounts will be suspended." The terminal sends the entered sentence to a server, which preprocesses the received sentence and extracts features. The server analyzes the sentence using a trained model and calculates a score indicating the likelihood that it was generated by generative artificial intelligence. If this score is 90%, the terminal displays to the user "There is a 90% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message may be generated. Please verify whether it is genuine information."

[0573] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user. Furthermore, it enables rapid recognition and response to threats such as phishing attacks and fake news.

[0574] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0575] Step 1:

[0576] The server trains an artificial intelligence model using text data collected from diverse sources. This collected text data includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (TensorFlow and PyTorch) are used for preprocessing. The input is the collected text data, and the output is the preprocessed data.

[0577] Step 2:

[0578] The user inputs a specific sentence into a device such as a smartphone or computer and sends that sentence. The input is the sentence entered by the user, and the output is the raw data sent from the device to the server.

[0579] Step 3:

[0580] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier. The input is raw data sent from the terminal, and the output is preprocessed data.

[0581] Step 4:

[0582] The server inputs pre-processed text data into a trained artificial intelligence model, which analyzes the data to calculate a probability score of how likely it was generated by a generative AI. The input is pre-processed data, and the output is the probability score.

[0583] Step 5:

[0584] The server sends the calculated score to the terminal. The input is the probability score, and the output is the transmission of the score data to the terminal.

[0585] Step 6:

[0586] The device displays the received score to the user. The display format might be, for example, "There is an 85% chance that this text was generated by a generative artificial intelligence." In addition, if there is a high probability of a phishing attack or fake news, a warning will be displayed such as, "Warning: This message is highly likely to have been generated by a generative artificial intelligence." The input is the score data sent from the server, and the output is what is displayed to the user.

[0587] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0588] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of the following main components.

[0589] 1. Means for training an artificial intelligence model using collected text data:

[0590] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0591] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[0592] 2. Means for preprocessing user-input text using natural language processing techniques:

[0593] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0594] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0595] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0596] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0597] 4. Means for displaying the calculated probability to the user:

[0598] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0599] 5. Further includes an emotion engine to recognize user emotions:

[0600] The server incorporates an emotion engine to analyze the emotions in the text entered by the user. This emotion engine analyzes the emotional tone and nuances of the text and determines whether the emotion is positive, negative, neutral, etc.

[0601] 6. Means for displaying the results of user sentiment analysis:

[0602] The server sends the results analyzed by the emotion engine to the terminal, and the terminal displays the emotion analysis results to the user. The display takes the form of "The emotion contained in this text is positive."

[0603] To explain with a concrete example, it would be as follows:

[0604] Specific example 1:

[0605] User A enters the following sentence into the terminal: "This product is very easy to use."

[0606] The terminal sends the entered text to the server.

[0607] The server preprocesses the received text and extracts features.

[0608] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0609] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0610] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be positive, the terminal displays to the user, "The emotion contained in this text is positive."

[0611] Specific example 2:

[0612] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0613] The terminal sends the entered text to the server.

[0614] The server preprocesses the received text and extracts features.

[0615] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0616] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0617] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be neutral, the terminal displays to the user, "The emotion contained in this text is neutral."

[0618] As described above, the system of the present invention can effectively distinguish between text written by a generative artificial intelligence system and text written by a human, and further analyze the user's emotions and provide the results to the user.

[0619] The following describes the processing flow.

[0620] Step 1:

[0621] The server collects data. The server gathers diverse text data from the internet and other online sources. This data includes news articles, blog entries, and social media posts.

[0622] Step 2:

[0623] The server builds the dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and then creates a dataset by assigning labels to each.

[0624] Step 3:

[0625] The server performs data preprocessing using natural language processing techniques. The server tokenizes the data, removes stop words, extracts stems and rematization, and extracts features.

[0626] Step 4:

[0627] The server builds a machine learning model. The server trains the artificial intelligence model using a deep learning framework. This includes supervised learning using text data within the dataset.

[0628] Step 5:

[0629] The user enters the text they want to evaluate. The user enters a specific sentence into their device and clicks the send button.

[0630] Step 6:

[0631] The terminal sends the entered text to the server. The terminal converts the text entered by the user into a packet format and sends it to the server.

[0632] Step 7:

[0633] The server preprocesses the received text. The server tokenizes the received text, performs stemming, stop word removal, and feature extraction to make it easier to analyze.

[0634] Step 8:

[0635] The server inputs pre-processed text into an artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. The server then calculates this score and determines the probability that the text was generated by a generative AI.

[0636] Step 9:

[0637] The server uses an emotion engine to recognize the emotion of a text. The server analyzes the text, which has been preprocessed by the emotion engine, and identifies the emotional tone and nuances within the text. As a result, it determines the emotion, such as positive, negative, or neutral.

[0638] Step 10:

[0639] The server sends the calculation results and emotion analysis results to the terminal. The server stores the potential score of the generative artificial intelligence and the emotion analysis results in a packet and sends it to the terminal.

[0640] Step 11:

[0641] The device displays the results to the user. The device analyzes the received score and sentiment analysis results and displays to the user, "There is a XX% chance that this text was generated by generative artificial intelligence." It also displays, for example, "The sentiment contained in this text is positive."

[0642] The following are specific examples.

[0643] Specific example 1:

[0644] User A enters the following sentence into the terminal: "This product is very easy to use."

[0645] The terminal sends the entered text to the server.

[0646] The server preprocesses the received text and extracts features.

[0647] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0648] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0649] The server analyzes the text using an emotion engine and determines that the emotion is positive. The terminal then displays to the user, "The emotion contained in this text is positive."

[0650] Specific example 2:

[0651] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0652] The terminal sends the entered text to the server.

[0653] The server preprocesses the received text and extracts features.

[0654] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0655] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0656] The server analyzes the text using an emotion engine and determines that the emotion is neutral. The terminal then displays to the user, "The emotion contained in this text is neutral."

[0657] As a result, the system of the present invention can provide users with not only the ability to discriminate text using generative artificial intelligence, but also the results of sentiment analysis of the text's content.

[0658] (Example 2)

[0659] Next, we will describe Example 2. 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".

[0660] In modern text generation technology, it is difficult to distinguish between text written by humans and text generated by generative artificial intelligence. Furthermore, there is a need to analyze the sentiment of user-input text and provide feedback to more accurately understand the user's intent. However, existing technologies do not provide a single system that offers both of these functions. Therefore, the challenge is to provide a unified system that determines whether user-input text was generated by generative artificial intelligence or written by a human, and further analyzes and provides feedback on its sentiment.

[0661] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0662] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, means for analyzing the sentiment of the text input by the user, and means for displaying the sentiment analysis results to the user. This makes it possible to determine whether the text input by the user was generated by a generative artificial intelligence or by a human, perform sentiment analysis, and provide feedback to the user.

[0663] "Collected text data" refers to data consisting of texts written by humans and texts generated by generative artificial intelligence, collected from various sources.

[0664] An "artificial intelligence model" is a model built using deep learning technology and trained to distinguish between human-written text and text generated by a generative artificial intelligence system.

[0665] "Natural language processing technology" refers to techniques for analyzing and preprocessing text, such as tokenization, stem extraction, stop word removal, and feature extraction.

[0666] "Generative artificial intelligence" refers to artificial intelligence that has been pre-trained and has the ability to generate new text based on input text.

[0667] "Methods for calculating probability" refers to methods for inputting pre-processed text into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence as a probability value.

[0668] "Means of displaying to the user" refers to the means of displaying the calculated probability and sentiment analysis results on the user's device.

[0669] "Methods for analyzing emotions" refer to methods for analyzing the emotional tone of text entered by a user and determining whether the text is positive, negative, or neutral.

[0670] "Means for displaying emotion analysis results" refers to means of visually presenting the analyzed emotion results to the user.

[0671] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user.

[0672] 1. Program generation

[0673] The server trains an artificial intelligence model using collected text data. This data includes both human-written text and text generated by generative AI. The server preprocesses this data by performing tokenization (word splitting), stemming (extraction of the core part of words), and removal of stop words (frequently occurring words). Based on the preprocessed data, it performs supervised learning using deep learning frameworks such as TensorFlow and PyTorch to build an artificial intelligence model.

[0674] 2. Program Processing

[0675] The server preprocesses the text entered by the user using natural language processing techniques. This converts newly entered text into a format that can be parsed. The preprocessing consists of tokenization, stemming, stop word removal, and feature extraction.

[0676] Next, the server inputs the pre-processed text into a trained artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. This score represents the probability that the text was generated by a generative AI.

[0677] 3. Hardware and Software

[0678] The servers utilize high-performance server clusters and cloud computing services. Furthermore, TensorFlow and PyTorch are used as software for natural language processing and deep learning. These software programs are specialized for preprocessing text data and training models.

[0679] 4. Data processing and calculations

[0680] The server first tokenizes the input text data, then performs stemming, stop word removal, and feature extraction. This allows for efficient text analysis. Subsequently, the pre-processed data is input into an artificial intelligence model to obtain analysis results. These results are expressed as a score and provided to the user.

[0681] 5. Specific Examples

[0682] For example, user A enters the following sentence into the terminal: "This product is very easy to use."

[0683] The terminal sends the entered text to the server.

[0684] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 20%, the terminal will display "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0685] The server analyzes the text using its sentiment engine, and if it determines that the sentiment is positive, the terminal displays the message "The sentiment contained in this text is positive." to the user.

[0686] As another example, user B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0687] The terminal sends the entered text to the server.

[0688] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 75%, the terminal will display "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0689] If the server analyzes the text using its emotion engine and determines it to be neutral, the terminal will display the message "The emotion contained in this text is neutral." to the user.

[0690] In this way, the present invention realizes a system that can effectively distinguish between text written by a generative artificial intelligence system and a human, and further analyze the user's emotions and provide the results to the user.

[0691] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0692] Processing flow

[0693] Step 1: The user enters and submits the text.

[0694] Input: The user enters text into the input box.

[0695] Specific action: The user enters any text into the input form on the device and clicks the submit button. For example, the user enters the sentence, "This product is very easy to use."

[0696] Output: The input text is passed to the terminal.

[0697] Step 2: The device sends the text to the server.

[0698] Input: The text entered by the terminal.

[0699] Specific operation: The terminal sends the text entered by the user to the server as text data. The transmission is performed as an HTTP request.

[0700] Output: The input text is passed to the server.

[0701] Step 3: The server preprocesses the text.

[0702] Input: The text sent to the server.

[0703] Specific operation: The server tokenizes the received text and breaks it down into individual words. For example, the sentence "This product is very easy to use." is split into tokens such as "this," "product," "is," "very," "easy to use," and "is." Next, the server extracts the stems of the tokens and converts "easy to use" to "easy to use." Furthermore, unnecessary words (stop words) are removed, and words like "is" and "is" are deleted. Finally, features are extracted and formatted in a way that a machine learning model can understand.

[0704] Output: Preprocessed data.

[0705] Step 4: The server parses the pre-processed text.

[0706] Input: Pre-processed data.

[0707] Specific operation: The server inputs pre-processed data into a trained artificial intelligence model, which then analyzes the data.

[0708] Output: Analysis results (probability score of generation by generative AI).

[0709] Step 5: The server calculates the potential score of the generative AI.

[0710] Input: Analysis results.

[0711] Specific operation: The server obtains the output of the artificial intelligence model as a probability value. If the model determines "20%", it indicates that there is a 20% chance that the text was generated by a generative AI.

[0712] Output: Potential score for generative AI.

[0713] Step 6: The device displays the score to the user.

[0714] Input: Potential score for generative AI.

[0715] Specific operation: The server sends the calculated score to the terminal, and the terminal displays this score to the user. The display format will be "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0716] Output: The user checks the score.

[0717] Step 7: The server analyzes emotions using the emotion engine.

[0718] Input: Text entered by the user.

[0719] Specific operation: The server uses an emotion engine to analyze the emotional tone of a sentence. For example, the sentence "This product is very easy to use" is determined to have a positive emotion.

[0720] Output: Emotion analysis results.

[0721] Step 8: The device displays the sentiment analysis results to the user.

[0722] Input: Sentiment analysis results.

[0723] Specific operation: The server sends the analysis results to the terminal, and the terminal displays to the user in the form of "The emotions contained in this text are positive."

[0724] Output: The user reviews the sentiment analysis results.

[0725] Through these steps, the system of the present invention can determine whether the text entered by the user was generated by a generative artificial intelligence system or by a human, analyze the emotions contained in the text, and provide feedback to the user.

[0726] (Application Example 2)

[0727] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0728] Conventional systems have struggled to determine whether user-generated text is created by artificial intelligence or by a human, and to analyze the emotions contained within that text. In particular, the food delivery industry requires both authenticity verification and sentiment analysis of input text to efficiently manage customer support and reviews. This will enable more accurate and faster service delivery, leading to improved customer satisfaction.

[0729] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0730] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability of it being generated by generative artificial intelligence, means for displaying the calculated probability to the user, means for performing sentiment analysis on the text input from the user, and means for displaying the results of the sentiment analysis to the user. This enables more efficient customer support and review management in the food delivery industry, contributing to improved customer satisfaction.

[0731] "Collected text data" refers to text information obtained from various sources, including text written by humans and text generated by generative artificial intelligence.

[0732] An "artificial intelligence model" is an algorithm used to learn from collected data and perform a specific task.

[0733] "Training methods" refer to the process of inputting a wide variety of data into an artificial intelligence model to improve its accuracy and discriminatory capabilities.

[0734] "Natural language processing technology" refers to technologies that enable computers to understand, interpret, and generate human language, and includes techniques such as tokenization, stem extraction, and stop word removal.

[0735] "Preprocessing" refers to the process of converting raw text data into a parseable format, which includes tokenization and stop word removal.

[0736] "Generative artificial intelligence" refers to algorithms that use trained models to generate text input from users, and which have a high accuracy in text generation.

[0737] "Means for calculating the probability of generation" refers to the process of inputting a pre-processed text into an artificial intelligence model and calculating a probability score indicating whether or not the text was generated by a generative artificial intelligence.

[0738] "Means of displaying to the user" refers to an interface for providing the user with the results calculated by the artificial intelligence model, and includes screen display and audio output.

[0739] "Sentiment analysis" is the process of analyzing the emotional tone and nuances contained in text to determine whether the emotion is positive, negative, neutral, etc.

[0740] The system of the present invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further analyzes the emotions contained in that text. In order to specifically implement the system of the present invention, the following means and processes are necessary.

[0741] First, the server trains an artificial intelligence model using text data collected from diverse sources. During this process, the server uses deep learning frameworks such as TensorFlow and PyTorch to preprocess data containing both human-written text and text generated by generative AI. This preprocessing includes tokenization, stem extraction, and stop word removal to prepare the training data.

[0742] Next, the user uses a smartphone or other device to input and send a specific sentence. The device sends this sentence to the server, which then preprocesses the received sentence again. This re-preprocessing involves tokenization, stemming, stop word removal, and feature extraction.

[0743] The pre-processed text is input into a server-trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[0744] The calculated score is sent from the server to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0745] Furthermore, the server performs sentiment analysis on the text entered by the user. This process uses a sentiment engine that analyzes the emotional tone and nuances of the text and determines emotions such as positive, negative, or neutral. Natural language processing tools such as the nlptown / bert-base-multilingual-uncased-sentiment model are used for sentiment analysis.

[0746] The analysis results are sent from the server to the terminal, which then displays the sentiment analysis results to the user. The display takes the form of statements such as, "The sentiment contained in this text is positive," or "The sentiment contained in this text is neutral."

[0747] Specific example:

[0748] The user left the following review: "This delivery is fantastic! Fast and reliable."

[0749] In this case, the system will display the following result:

[0750] "There is a 15.50% probability that this text was generated by a generative artificial intelligence." (Example: Calculated using training data)

[0751] "The emotions conveyed in this text are positive."

[0752] Examples of prompts to input into a generative AI model:

[0753] User input: "This delivery is fantastic! Fast and reliable."

[0754] analysis:

[0755] 1. There is a 15.50% probability that this text was generated by a generative artificial intelligence.

[0756] 2. The emotions contained in this sentence are positive.

[0757] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0758] Step 1:

[0759] Users input and send specific text using their smartphones or other devices. The entered text is sent from the device to the server in its original form.

[0760] Input: Text entered by the user (e.g., "This delivery is fantastic! Fast and reliable.")

[0761] Output: Sent text data

[0762] Step 2:

[0763] The server preprocesses the received text using natural language processing techniques. This preprocessing step involves tokenization, stemming, and stop word removal. This converts the text into a format that is easy to analyze.

[0764] Input: Submitted text data

[0765] Output: Preprocessed text data

[0766] Specific operation: Use the BERT tokenizer to tokenize the text, remove unnecessary words, and format it for analysis.

[0767] Step 3:

[0768] The server inputs pre-processed text into a trained artificial intelligence model, which then calculates a probability score for the text as it was generated by the generative AI. The model analyzes the input text data and outputs the result as a probability score.

[0769] Input: Preprocessed text data

[0770] Output: Probability score generated by generative artificial intelligence (e.g., 15.50%)

[0771] Specific operation: Preprocessed text is input into a BERT-based model, and the classification result is output as a probability score.

[0772] Step 4:

[0773] The server sends the calculated score for the generative artificial intelligence to the terminal. The terminal displays this score to the user. The display format is "There is a XX% chance that this text was generated by generative artificial intelligence."

[0774] Input: Score of the Generative Artificial Intelligence

[0775] Output: Content displayed to the user

[0776] Specific action: The submitted score is displayed on the terminal's user interface.

[0777] Step 5:

[0778] The server performs sentiment analysis on text entered by the user. It uses a sentiment engine to analyze the emotional tone and nuances of the text. The analysis results are classified as positive, negative, neutral, etc.

[0779] Input: Preprocessed text data

[0780] Output: Sentiment analysis results (e.g., positive)

[0781] Specific operation: Classify the sentiment of a text using the nlptown / bert-base-multilingual-uncased-sentiment model and obtain the result.

[0782] Step 6:

[0783] The server sends the sentiment analysis results to the terminal, and the terminal displays the results to the user. The display format is something like, "The sentiment contained in this text is positive."

[0784] Input: Sentiment analysis results

[0785] Output: Content displayed to the user

[0786] Specific operation: The transmitted sentiment analysis results are displayed on the device's user interface.

[0787] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[0788] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0789] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[0790] [Third Embodiment]

[0791] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[0792] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[0793] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0794] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[0795] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0796] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0797] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0798] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0799] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0800] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0801] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0802] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[0803] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0804] 1. Means for training an artificial intelligence model using collected text data:

[0805] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0806] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[0807] 2. Means for preprocessing user-input text using natural language processing techniques:

[0808] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0809] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0810] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0811] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0812] 4. Means for displaying the calculated probability to the user:

[0813] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0814] To explain with a concrete example, it would be as follows:

[0815] Specific example 1:

[0816] User A enters the following sentence into the terminal: "This product is very easy to use."

[0817] The terminal sends the entered text to the server.

[0818] The server preprocesses the received text and extracts features.

[0819] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0820] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0821] Specific example 2:

[0822] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0823] The terminal sends the entered text to the server.

[0824] The server preprocesses the received text and extracts features.

[0825] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0826] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0827] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text and provide the results to the user.

[0828] The following describes the processing flow.

[0829] Step 1:

[0830] The server collects data. The server gathers diverse text data from the internet and other online sources. This includes news articles, blog entries, and social media posts.

[0831] Step 2:

[0832] The server constructs a dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and assigns labels to each.

[0833] Step 3:

[0834] The server performs preprocessing using natural language processing techniques. The server tokenizes the collected data and then processes it by removing stop words, stemming, rematizing, and other methods to prepare the data.

[0835] Step 4:

[0836] The server builds and trains a machine learning model. The server uses a deep learning framework to build an artificial intelligence model and trains it using a prepared dataset. Training involves inputting data features and performing supervised learning to enable the model to distinguish between human text and text generated by generative AI.

[0837] Step 5:

[0838] The user enters the text they want to evaluate. The user instructs their device to input the text and send it to the server.

[0839] Step 6:

[0840] The terminal sends the text to the server. The terminal collects the text entered by the user and sends it to the server.

[0841] Step 7:

[0842] The server performs preprocessing. The server preprocesses the received text again, performing tokenization, stemming, stop word removal, feature extraction, and other operations.

[0843] Step 8:

[0844] The server inputs text into a trained artificial intelligence model. The server inputs the pre-processed text into the AI ​​model and calculates the probability that it was generated by a generative AI.

[0845] Step 9:

[0846] The server sends the calculation result to the terminal. The server sends the calculated score to the terminal.

[0847] Step 10:

[0848] The device displays the results to the user. The device analyzes the received score and displays to the user, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0849] (Example 1)

[0850] Next, we will describe Example 1. 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."

[0851] Conventional methods make it difficult to determine whether text entered by a user was written by a human or generated by generative artificial intelligence. This leads to problems such as the inability to make accurate judgments, which can result in issues like the spread of misinformation and the inability to properly evaluate reliability. The present invention aims to solve these problems.

[0852] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0853] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, and means for displaying the calculated probability to the user. This makes it possible to determine with high accuracy whether the text input by the user was written by a human or generated by a generative artificial intelligence.

[0854] "Collected text data" refers to text written by humans and text generated by generative artificial intelligence, collected from various sources.

[0855] An "artificial intelligence model" refers to a model that is trained on collected data and has an algorithmic structure that can determine whether input text was written by a human or generated by a generative artificial intelligence.

[0856] "Natural language processing technology" refers to techniques that convert text into a machine-readable format using methods such as tokenization, stem extraction, stop word removal, and feature extraction.

[0857] "Preprocessing" refers to a series of processes performed to convert raw text data into a format that is easy to analyze, specifically steps such as tokenization, stemming, and stop word removal.

[0858] "Generative artificial intelligence" refers to artificial intelligence systems that perform natural language generation, specifically the technology that automatically generates text based on specific prompts.

[0859] "Means for calculating probability" refers to a method of inputting pre-processed text data into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence.

[0860] "Means of display" refers to a device or method that visually presents the calculated probability score to the user, specifically, means of displaying the results on the display of a terminal or the screen of a mobile device.

[0861] The present invention is a system that determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[0862] 1. Means for training an artificial intelligence model using collected text data:

[0863] The server trains artificial intelligence models using text data collected from diverse sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0864] The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0865] For training, deep learning frameworks (such as TensorFlow or PyTorch) are used.

[0866] The server uses this data to perform supervised learning and build a model that can determine whether the text was written by a human or by a generative artificial intelligence.

[0867] 2. Means for preprocessing user-input text using natural language processing techniques:

[0868] The user enters a specific message into the device and sends it.

[0869] The device will send this message to the server.

[0870] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0871] 3. A means for inputting preprocessed text into an artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0872] The server inputs the pre-processed text into a trained artificial intelligence model.

[0873] The model analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative AI.

[0874] 4. Means for displaying the calculated probability to the user:

[0875] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0876] Specific example

[0877] Specific example 1:

[0878] User A enters the following sentence into the terminal: "This product is very easy to use."

[0879] The terminal sends the entered text to the server.

[0880] The server preprocesses the received text and extracts features.

[0881] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0882] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0883] Specific example 2:

[0884] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0885] The terminal sends the entered text to the server.

[0886] The server preprocesses the received text and extracts features.

[0887] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0888] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[0889] The system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text through these processes and provide the results to the user.

[0890] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0891] Step 1:

[0892] Data collection

[0893] The server collects text data from various sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[0894] Input: Text data from publicly available databases on the internet and independently collected data sources.

[0895] Output: A set of collected text data.

[0896] Step 2:

[0897] Data preprocessing

[0898] The server tokenizes the collected text data. This is the process of dividing the text into units of words or phrases.

[0899] The server then performs stemming and stop word removal. Stemming is the process of converting words into their base forms, while stop word removal is the process of removing common words that are not needed for analysis.

[0900] Input: Collected text data

[0901] Output: Preprocessed text data (data that has undergone tokenization, stemming, and stop word removal)

[0902] Step 3:

[0903] Training of artificial intelligence models

[0904] The server uses pre-processed data to create a training dataset, which includes labels for human-written text and text generated by generative artificial intelligence.

[0905] The server uses a deep learning framework (such as TensorFlow or PyTorch) to perform supervised learning and train an artificial intelligence model.

[0906] Input: Preprocessed text data and its corresponding labels

[0907] Output: Trained artificial intelligence model

[0908] Step 4:

[0909] Text input from the user

[0910] The user enters a specific sentence into the input field on the terminal and presses the send button.

[0911] Input: Text entered by the user

[0912] Output: Text data sent to the terminal

[0913] Step 5:

[0914] Sending and receiving input data

[0915] The terminal sends the text entered by the user to the server.

[0916] The server receives the text that was sent.

[0917] Input: Text data sent from the device

[0918] Output: Text data received by the server

[0919] Step 6:

[0920] Reprocessing of received data

[0921] The server reprocesses the received text. Specifically, it performs tokenization, stemming, stop word removal, and feature extraction.

[0922] Input: Text data received by the server

[0923] Output: Preprocessed text data

[0924] Step 7:

[0925] Score calculation

[0926] The server inputs the pre-processed text data into a trained artificial intelligence model.

[0927] The model analyzes the input text data and calculates a probability score generated by generative artificial intelligence. This score is expressed as a probability.

[0928] Input: Pre-processed text data

[0929] Output: Probability score generated by generative artificial intelligence

[0930] Step 8:

[0931] Submitting and displaying scores

[0932] The server sends the calculated score to the terminal.

[0933] The device displays a score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0934] Input: Calculated score

[0935] Output: Score displayed to the user

[0936] In this way, through each processing step, the system can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user.

[0937] (Application Example 1)

[0938] Next, we will explain Application Example 1. In the following explanation, 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."

[0939] In today's internet environment, phishing attacks and fake news are rampant, threatening user security. Users need to be able to quickly detect these threats and receive appropriate warnings, but current technology lacks the means to do so effectively. In particular, with the widespread use of text created by generative artificial intelligence, there is a need for a reliable system that can distinguish between text written by humans and text generated by generative AI and warn users accordingly.

[0940] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0941] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, and means for detecting phishing attacks and fake news and issuing warnings. This enables users to quickly recognize the threat of phishing attacks and fake news and take appropriate measures.

[0942] "Collected text data" refers to text data collected from various sources and used to train artificial intelligence models.

[0943] An "artificial intelligence model" is trained on collected text data to determine whether a text was written by a human or generated by a generative artificial intelligence system.

[0944] "Preprocessing" refers to the process of preparing user-input text into a format that is easy to analyze, using techniques such as tokenization, stem extraction, and stop word removal.

[0945] "Generative artificial intelligence" refers to technology that automatically generates text using natural language generation models.

[0946] "Calculated probability" refers to the probability that the preprocessed text was generated by a generative artificial intelligence, and is calculated by the AI ​​model.

[0947] "Means of displaying to the user" refers to an interface for visually or audibly communicating the calculated probability to the user.

[0948] A "phishing attack" refers to a fraudulent message or email sent by a malicious third party to illegally obtain a user's personal information.

[0949] "Fake news" refers to news that is not based on actual facts and conveys false information, with the aim of misleading readers.

[0950] "Warning mechanisms" are part of a system designed to inform users of potential phishing attacks or fake news and to draw their attention to them.

[0951] Embodiments of this invention will now be described. First, the server trains an artificial intelligence model using text data collected from various sources. The collected text data includes both human-written text and text generated by generative artificial intelligence. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (e.g., TensorFlow or PyTorch) are used for preprocessing.

[0952] The user inputs a specific sentence into a device such as a smartphone or computer and sends it. The device then sends this sentence to a server. The server preprocesses the received sentence again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes it easier to analyze the sentence.

[0953] The server inputs pre-processed text into a trained artificial intelligence model, which analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[0954] The server then sends the calculated score to the terminal, which displays this score to the user. The display uses a format such as, "There is a XX% chance that this text was generated by a generative artificial intelligence." In addition, a warning is displayed if a phishing attack or fake news is detected.

[0955] As a concrete example, a user enters the following sentence into the terminal: "Click this link for a chance to win a free gift card." The terminal sends the entered sentence to the server, which preprocesses the received sentence and extracts features. The server then uses a trained model to analyze the sentence and calculate a score indicating the likelihood that it was generated by generative artificial intelligence. For example, if this score is 85%, the terminal displays to the user "There is an 85% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message is highly likely to have been generated by generative artificial intelligence."

[0956] As another example, a user enters the following sentence into the terminal: "Due to a new government decision, all bank accounts will be suspended." The terminal sends the entered sentence to a server, which preprocesses the received sentence and extracts features. The server analyzes the sentence using a trained model and calculates a score indicating the likelihood that it was generated by generative artificial intelligence. If this score is 90%, the terminal displays to the user "There is a 90% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message may be generated. Please verify whether it is genuine information."

[0957] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user. Furthermore, it enables rapid recognition and response to threats such as phishing attacks and fake news.

[0958] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0959] Step 1:

[0960] The server trains an artificial intelligence model using text data collected from diverse sources. This collected text data includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (TensorFlow and PyTorch) are used for preprocessing. The input is the collected text data, and the output is the preprocessed data.

[0961] Step 2:

[0962] The user inputs a specific sentence into a device such as a smartphone or computer and sends that sentence. The input is the sentence entered by the user, and the output is the raw data sent from the device to the server.

[0963] Step 3:

[0964] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier. The input is raw data sent from the terminal, and the output is preprocessed data.

[0965] Step 4:

[0966] The server inputs pre-processed text data into a trained artificial intelligence model, which analyzes the data to calculate a probability score of how likely it was generated by a generative AI. The input is pre-processed data, and the output is the probability score.

[0967] Step 5:

[0968] The server sends the calculated score to the terminal. The input is the probability score, and the output is the transmission of the score data to the terminal.

[0969] Step 6:

[0970] The device displays the received score to the user. The display format might be, for example, "There is an 85% chance that this text was generated by a generative artificial intelligence." In addition, if there is a high probability of a phishing attack or fake news, a warning will be displayed such as, "Warning: This message is highly likely to have been generated by a generative artificial intelligence." The input is the score data sent from the server, and the output is what is displayed to the user.

[0971] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0972] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of the following main components.

[0973] 1. Means for training an artificial intelligence model using collected text data:

[0974] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[0975] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[0976] 2. Means for preprocessing user-input text using natural language processing techniques:

[0977] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[0978] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[0979] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[0980] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[0981] 4. Means for displaying the calculated probability to the user:

[0982] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[0983] 5. Further includes an emotion engine to recognize user emotions:

[0984] The server incorporates an emotion engine to analyze the emotions in the text entered by the user. This emotion engine analyzes the emotional tone and nuances of the text and determines whether the emotion is positive, negative, neutral, etc.

[0985] 6. Means for displaying the results of user sentiment analysis:

[0986] The server sends the results analyzed by the emotion engine to the terminal, and the terminal displays the emotion analysis results to the user. The display takes the form of "The emotion contained in this text is positive."

[0987] To explain with a concrete example, it would be as follows:

[0988] Specific example 1:

[0989] User A enters the following sentence into the terminal: "This product is very easy to use."

[0990] The terminal sends the entered text to the server.

[0991] The server preprocesses the received text and extracts features.

[0992] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[0993] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[0994] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be positive, the terminal displays to the user, "The emotion contained in this text is positive."

[0995] Specific example 2:

[0996] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[0997] The terminal sends the entered text to the server.

[0998] The server preprocesses the received text and extracts features.

[0999] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1000] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1001] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be neutral, the terminal displays to the user, "The emotion contained in this text is neutral."

[1002] As described above, the system of the present invention can effectively distinguish between text written by a generative artificial intelligence system and text written by a human, and further analyze the user's emotions and provide the results to the user.

[1003] The following describes the processing flow.

[1004] Step 1:

[1005] The server collects data. The server gathers diverse text data from the internet and other online sources. This data includes news articles, blog entries, and social media posts.

[1006] Step 2:

[1007] The server builds the dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and then creates a dataset by assigning labels to each.

[1008] Step 3:

[1009] The server performs data preprocessing using natural language processing techniques. The server tokenizes the data, removes stop words, extracts stems and rematization, and extracts features.

[1010] Step 4:

[1011] The server builds a machine learning model. The server trains the artificial intelligence model using a deep learning framework. This includes supervised learning using text data within the dataset.

[1012] Step 5:

[1013] The user enters the text they want to evaluate. The user enters a specific sentence into their device and clicks the send button.

[1014] Step 6:

[1015] The terminal sends the entered text to the server. The terminal converts the text entered by the user into a packet format and sends it to the server.

[1016] Step 7:

[1017] The server preprocesses the received text. The server tokenizes the received text, performs stemming, stop word removal, and feature extraction to make it easier to analyze.

[1018] Step 8:

[1019] The server inputs pre-processed text into an artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. The server then calculates this score and determines the probability that the text was generated by a generative AI.

[1020] Step 9:

[1021] The server uses an emotion engine to recognize the emotion of a text. The server analyzes the text, which has been preprocessed by the emotion engine, and identifies the emotional tone and nuances within the text. As a result, it determines the emotion, such as positive, negative, or neutral.

[1022] Step 10:

[1023] The server sends the calculation results and emotion analysis results to the terminal. The server stores the potential score of the generative artificial intelligence and the emotion analysis results in a packet and sends it to the terminal.

[1024] Step 11:

[1025] The device displays the results to the user. The device analyzes the received score and sentiment analysis results and displays to the user, "There is a XX% chance that this text was generated by generative artificial intelligence." It also displays, for example, "The sentiment contained in this text is positive."

[1026] The following are specific examples.

[1027] Specific example 1:

[1028] User A enters the following sentence into the terminal: "This product is very easy to use."

[1029] The terminal sends the entered text to the server.

[1030] The server preprocesses the received text and extracts features.

[1031] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1032] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1033] The server analyzes the text using an emotion engine and determines that the emotion is positive. The terminal then displays to the user, "The emotion contained in this text is positive."

[1034] Specific example 2:

[1035] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1036] The terminal sends the entered text to the server.

[1037] The server preprocesses the received text and extracts features.

[1038] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1039] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1040] The server analyzes the text using an emotion engine and determines that the emotion is neutral. The terminal then displays to the user, "The emotion contained in this text is neutral."

[1041] As a result, the system of the present invention can provide users with not only the ability to discriminate text using generative artificial intelligence, but also the results of sentiment analysis of the text's content.

[1042] (Example 2)

[1043] Next, we will describe Example 2. 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."

[1044] In modern text generation technology, it is difficult to distinguish between text written by humans and text generated by generative artificial intelligence. Furthermore, there is a need to analyze the sentiment of user-input text and provide feedback to more accurately understand the user's intent. However, existing technologies do not provide a single system that offers both of these functions. Therefore, the challenge is to provide a unified system that determines whether user-input text was generated by generative artificial intelligence or written by a human, and further analyzes and provides feedback on its sentiment.

[1045] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1046] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, means for analyzing the sentiment of the text input by the user, and means for displaying the sentiment analysis results to the user. This makes it possible to determine whether the text input by the user was generated by a generative artificial intelligence or by a human, perform sentiment analysis, and provide feedback to the user.

[1047] "Collected text data" refers to data consisting of texts written by humans and texts generated by generative artificial intelligence, collected from various sources.

[1048] An "artificial intelligence model" is a model built using deep learning technology and trained to distinguish between human-written text and text generated by a generative artificial intelligence system.

[1049] "Natural language processing technology" refers to techniques for analyzing and preprocessing text, such as tokenization, stem extraction, stop word removal, and feature extraction.

[1050] "Generative artificial intelligence" refers to artificial intelligence that has been pre-trained and has the ability to generate new text based on input text.

[1051] "Methods for calculating probability" refers to methods for inputting pre-processed text into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence as a probability value.

[1052] "Means of displaying to the user" refers to the means of displaying the calculated probability and sentiment analysis results on the user's device.

[1053] "Methods for analyzing emotions" refer to methods for analyzing the emotional tone of text entered by a user and determining whether the text is positive, negative, or neutral.

[1054] "Means for displaying emotion analysis results" refers to means of visually presenting the analyzed emotion results to the user.

[1055] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user.

[1056] 1. Program generation

[1057] The server trains an artificial intelligence model using collected text data. This data includes both human-written text and text generated by generative AI. The server preprocesses this data by performing tokenization (word splitting), stemming (extraction of the core part of words), and removal of stop words (frequently occurring words). Based on the preprocessed data, it performs supervised learning using deep learning frameworks such as TensorFlow and PyTorch to build an artificial intelligence model.

[1058] 2. Program Processing

[1059] The server preprocesses the text entered by the user using natural language processing techniques. This converts newly entered text into a format that can be parsed. The preprocessing consists of tokenization, stemming, stop word removal, and feature extraction.

[1060] Next, the server inputs the pre-processed text into a trained artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. This score represents the probability that the text was generated by a generative AI.

[1061] 3. Hardware and Software

[1062] The servers utilize high-performance server clusters and cloud computing services. Furthermore, TensorFlow and PyTorch are used as software for natural language processing and deep learning. These software programs are specialized for preprocessing text data and training models.

[1063] 4. Data processing and calculations

[1064] The server first tokenizes the input text data, then performs stemming, stop word removal, and feature extraction. This allows for efficient text analysis. Subsequently, the pre-processed data is input into an artificial intelligence model to obtain analysis results. These results are expressed as a score and provided to the user.

[1065] 5. Specific Examples

[1066] For example, user A enters the following sentence into the terminal: "This product is very easy to use."

[1067] The terminal sends the entered text to the server.

[1068] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 20%, the terminal will display "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1069] The server analyzes the text using its sentiment engine, and if it determines that the sentiment is positive, the terminal displays the message "The sentiment contained in this text is positive." to the user.

[1070] As another example, user B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1071] The terminal sends the entered text to the server.

[1072] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 75%, the terminal will display "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1073] If the server analyzes the text using its emotion engine and determines it to be neutral, the terminal will display the message "The emotion contained in this text is neutral." to the user.

[1074] In this way, the present invention realizes a system that can effectively distinguish between text written by a generative artificial intelligence system and a human, and further analyze the user's emotions and provide the results to the user.

[1075] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1076] Processing flow

[1077] Step 1: The user enters and submits the text.

[1078] Input: The user enters text into the input box.

[1079] Specific action: The user enters any text into the input form on the device and clicks the submit button. For example, the user enters the sentence, "This product is very easy to use."

[1080] Output: The input text is passed to the terminal.

[1081] Step 2: The device sends the text to the server.

[1082] Input: The text entered by the terminal.

[1083] Specific operation: The terminal sends the text entered by the user to the server as text data. The transmission is performed as an HTTP request.

[1084] Output: The input text is passed to the server.

[1085] Step 3: The server preprocesses the text.

[1086] Input: The text sent to the server.

[1087] Specific operation: The server tokenizes the received text and breaks it down into individual words. For example, the sentence "This product is very easy to use." is split into tokens such as "this," "product," "is," "very," "easy to use," and "is." Next, the server extracts the stems of the tokens and converts "easy to use" to "easy to use." Furthermore, unnecessary words (stop words) are removed, and words like "is" and "is" are deleted. Finally, features are extracted and formatted in a way that a machine learning model can understand.

[1088] Output: Preprocessed data.

[1089] Step 4: The server parses the pre-processed text.

[1090] Input: Pre-processed data.

[1091] Specific operation: The server inputs pre-processed data into a trained artificial intelligence model, which then analyzes the data.

[1092] Output: Analysis results (probability score of generation by generative AI).

[1093] Step 5: The server calculates the potential score of the generative AI.

[1094] Input: Analysis results.

[1095] Specific operation: The server obtains the output of the artificial intelligence model as a probability value. If the model determines "20%", it indicates that there is a 20% chance that the text was generated by a generative AI.

[1096] Output: Potential score for generative AI.

[1097] Step 6: The device displays the score to the user.

[1098] Input: Potential score for generative AI.

[1099] Specific operation: The server sends the calculated score to the terminal, and the terminal displays this score to the user. The display format will be "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1100] Output: The user checks the score.

[1101] Step 7: The server analyzes emotions using the emotion engine.

[1102] Input: Text entered by the user.

[1103] Specific operation: The server uses an emotion engine to analyze the emotional tone of a sentence. For example, the sentence "This product is very easy to use" is determined to have a positive emotion.

[1104] Output: Emotion analysis results.

[1105] Step 8: The device displays the sentiment analysis results to the user.

[1106] Input: Sentiment analysis results.

[1107] Specific operation: The server sends the analysis results to the terminal, and the terminal displays to the user in the form of "The emotions contained in this text are positive."

[1108] Output: The user reviews the sentiment analysis results.

[1109] Through these steps, the system of the present invention can determine whether the text entered by the user was generated by a generative artificial intelligence system or by a human, analyze the emotions contained in the text, and provide feedback to the user.

[1110] (Application Example 2)

[1111] Next, we will explain application example 2. In the following explanation, 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."

[1112] Conventional systems have struggled to determine whether user-generated text is created by artificial intelligence or by a human, and to analyze the emotions contained within that text. In particular, the food delivery industry requires both authenticity verification and sentiment analysis of input text to efficiently manage customer support and reviews. This will enable more accurate and faster service delivery, leading to improved customer satisfaction.

[1113] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1114] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability of it being generated by generative artificial intelligence, means for displaying the calculated probability to the user, means for performing sentiment analysis on the text input from the user, and means for displaying the results of the sentiment analysis to the user. This enables more efficient customer support and review management in the food delivery industry, contributing to improved customer satisfaction.

[1115] "Collected text data" refers to text information obtained from various sources, including text written by humans and text generated by generative artificial intelligence.

[1116] An "artificial intelligence model" is an algorithm used to learn from collected data and perform a specific task.

[1117] "Training methods" refer to the process of inputting a wide variety of data into an artificial intelligence model to improve its accuracy and discriminatory capabilities.

[1118] "Natural language processing technology" refers to technologies that enable computers to understand, interpret, and generate human language, and includes techniques such as tokenization, stem extraction, and stop word removal.

[1119] "Preprocessing" refers to the process of converting raw text data into a parseable format, which includes tokenization and stop word removal.

[1120] "Generative artificial intelligence" refers to algorithms that use trained models to generate text input from users, and which have a high accuracy in text generation.

[1121] "Means for calculating the probability of generation" refers to the process of inputting a pre-processed text into an artificial intelligence model and calculating a probability score indicating whether or not the text was generated by a generative artificial intelligence.

[1122] "Means of displaying to the user" refers to an interface for providing the user with the results calculated by the artificial intelligence model, and includes screen display and audio output.

[1123] "Sentiment analysis" is the process of analyzing the emotional tone and nuances contained in text to determine whether the emotion is positive, negative, neutral, etc.

[1124] The system of the present invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further analyzes the emotions contained in that text. In order to specifically implement the system of the present invention, the following means and processes are necessary.

[1125] First, the server trains an artificial intelligence model using text data collected from diverse sources. During this process, the server uses deep learning frameworks such as TensorFlow and PyTorch to preprocess data containing both human-written text and text generated by generative AI. This preprocessing includes tokenization, stem extraction, and stop word removal to prepare the training data.

[1126] Next, the user uses a smartphone or other device to input and send a specific sentence. The device sends this sentence to the server, which then preprocesses the received sentence again. This re-preprocessing involves tokenization, stemming, stop word removal, and feature extraction.

[1127] The pre-processed text is input into a server-trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[1128] The calculated score is sent from the server to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1129] Furthermore, the server performs sentiment analysis on the text entered by the user. This process uses a sentiment engine that analyzes the emotional tone and nuances of the text and determines emotions such as positive, negative, or neutral. Natural language processing tools such as the nlptown / bert-base-multilingual-uncased-sentiment model are used for sentiment analysis.

[1130] The analysis results are sent from the server to the terminal, which then displays the sentiment analysis results to the user. The display takes the form of statements such as, "The sentiment contained in this text is positive," or "The sentiment contained in this text is neutral."

[1131] Specific example:

[1132] The user left the following review: "This delivery is fantastic! Fast and reliable."

[1133] In this case, the system will display the following result:

[1134] "There is a 15.50% probability that this text was generated by a generative artificial intelligence." (Example: Calculated using training data)

[1135] "The emotions conveyed in this text are positive."

[1136] Examples of prompts to input into a generative AI model:

[1137] User input: "This delivery is fantastic! Fast and reliable."

[1138] analysis:

[1139] 1. There is a 15.50% probability that this text was generated by a generative artificial intelligence.

[1140] 2. The emotions contained in this sentence are positive.

[1141] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1142] Step 1:

[1143] Users input and send specific text using their smartphones or other devices. The entered text is sent from the device to the server in its original form.

[1144] Input: Text entered by the user (e.g., "This delivery is fantastic! Fast and reliable.")

[1145] Output: Sent text data

[1146] Step 2:

[1147] The server preprocesses the received text using natural language processing techniques. This preprocessing step involves tokenization, stemming, and stop word removal. This converts the text into a format that is easy to analyze.

[1148] Input: Submitted text data

[1149] Output: Preprocessed text data

[1150] Specific operation: Use the BERT tokenizer to tokenize the text, remove unnecessary words, and format it for analysis.

[1151] Step 3:

[1152] The server inputs pre-processed text into a trained artificial intelligence model, which then calculates a probability score for the text as it was generated by the generative AI. The model analyzes the input text data and outputs the result as a probability score.

[1153] Input: Preprocessed text data

[1154] Output: Probability score generated by generative artificial intelligence (e.g., 15.50%)

[1155] Specific operation: Preprocessed text is input into a BERT-based model, and the classification result is output as a probability score.

[1156] Step 4:

[1157] The server sends the calculated score for the generative artificial intelligence to the terminal. The terminal displays this score to the user. The display format is "There is a XX% chance that this text was generated by generative artificial intelligence."

[1158] Input: Score of the Generative Artificial Intelligence

[1159] Output: Content displayed to the user

[1160] Specific action: The submitted score is displayed on the terminal's user interface.

[1161] Step 5:

[1162] The server performs sentiment analysis on text entered by the user. It uses a sentiment engine to analyze the emotional tone and nuances of the text. The analysis results are classified as positive, negative, neutral, etc.

[1163] Input: Preprocessed text data

[1164] Output: Sentiment analysis results (e.g., positive)

[1165] Specific operation: Classify the sentiment of a text using the nlptown / bert-base-multilingual-uncased-sentiment model and obtain the result.

[1166] Step 6:

[1167] The server sends the sentiment analysis results to the terminal, and the terminal displays the results to the user. The display format is something like, "The sentiment contained in this text is positive."

[1168] Input: Sentiment analysis results

[1169] Output: Content displayed to the user

[1170] Specific operation: The transmitted sentiment analysis results are displayed on the device's user interface.

[1171] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1172] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1173] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1174] [Fourth Embodiment]

[1175] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1176] As shown in Figure 7, the 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.

[1177] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1178] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1179] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1180] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1181] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1182] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive 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 robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1183] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1184] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1185] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1186] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1187] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1188] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[1189] 1. Means for training an artificial intelligence model using collected text data:

[1190] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[1191] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[1192] 2. Means for preprocessing user-input text using natural language processing techniques:

[1193] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[1194] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[1195] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[1196] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[1197] 4. Means for displaying the calculated probability to the user:

[1198] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1199] To explain with a concrete example, it would be as follows:

[1200] Specific example 1:

[1201] User A enters the following sentence into the terminal: "This product is very easy to use."

[1202] The terminal sends the entered text to the server.

[1203] The server preprocesses the received text and extracts features.

[1204] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1205] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1206] Specific example 2:

[1207] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1208] The terminal sends the entered text to the server.

[1209] The server preprocesses the received text and extracts features.

[1210] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1211] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1212] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text and provide the results to the user.

[1213] The following describes the processing flow.

[1214] Step 1:

[1215] The server collects data. The server gathers diverse text data from the internet and other online sources. This includes news articles, blog entries, and social media posts.

[1216] Step 2:

[1217] The server constructs a dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and assigns labels to each.

[1218] Step 3:

[1219] The server performs preprocessing using natural language processing techniques. The server tokenizes the collected data and then processes it by removing stop words, stemming, rematizing, and other methods to prepare the data.

[1220] Step 4:

[1221] The server builds and trains a machine learning model. The server uses a deep learning framework to build an artificial intelligence model and trains it using a prepared dataset. Training involves inputting data features and performing supervised learning to enable the model to distinguish between human text and text generated by generative AI.

[1222] Step 5:

[1223] The user enters the text they want to evaluate. The user instructs their device to input the text and send it to the server.

[1224] Step 6:

[1225] The terminal sends the text to the server. The terminal collects the text entered by the user and sends it to the server.

[1226] Step 7:

[1227] The server performs preprocessing. The server preprocesses the received text again, performing tokenization, stemming, stop word removal, feature extraction, and other operations.

[1228] Step 8:

[1229] The server inputs text into a trained artificial intelligence model. The server inputs the pre-processed text into the AI ​​model and calculates the probability that it was generated by a generative AI.

[1230] Step 9:

[1231] The server sends the calculation result to the terminal. The server sends the calculated score to the terminal.

[1232] Step 10:

[1233] The device displays the results to the user. The device analyzes the received score and displays to the user, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1234] (Example 1)

[1235] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1236] Conventional methods make it difficult to determine whether text entered by a user was written by a human or generated by generative artificial intelligence. This leads to problems such as the inability to make accurate judgments, which can result in issues like the spread of misinformation and the inability to properly evaluate reliability. The present invention aims to solve these problems.

[1237] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1238] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, and means for displaying the calculated probability to the user. This makes it possible to determine with high accuracy whether the text input by the user was written by a human or generated by a generative artificial intelligence.

[1239] "Collected text data" refers to text written by humans and text generated by generative artificial intelligence, collected from various sources.

[1240] An "artificial intelligence model" refers to a model that is trained on collected data and has an algorithmic structure that can determine whether input text was written by a human or generated by a generative artificial intelligence.

[1241] "Natural language processing technology" refers to techniques that convert text into a machine-readable format using methods such as tokenization, stem extraction, stop word removal, and feature extraction.

[1242] "Preprocessing" refers to a series of processes performed to convert raw text data into a format that is easy to analyze, specifically steps such as tokenization, stemming, and stop word removal.

[1243] "Generative artificial intelligence" refers to artificial intelligence systems that perform natural language generation, specifically the technology that automatically generates text based on specific prompts.

[1244] "Means for calculating probability" refers to a method of inputting pre-processed text data into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence.

[1245] "Means of display" refers to a device or method that visually presents the calculated probability score to the user, specifically, means of displaying the results on the display of a terminal or the screen of a mobile device.

[1246] The present invention is a system that determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human. This system consists of the following main components.

[1247] 1. Means for training an artificial intelligence model using collected text data:

[1248] The server trains artificial intelligence models using text data collected from diverse sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[1249] The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[1250] For training, deep learning frameworks (such as TensorFlow or PyTorch) are used.

[1251] The server uses this data to perform supervised learning and build a model that can determine whether the text was written by a human or by a generative artificial intelligence.

[1252] 2. Means for preprocessing user-input text using natural language processing techniques:

[1253] The user enters a specific message into the device and sends it.

[1254] The device will send this message to the server.

[1255] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[1256] 3. A means for inputting preprocessed text into an artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[1257] The server inputs the pre-processed text into a trained artificial intelligence model.

[1258] The model analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative AI.

[1259] 4. Means for displaying the calculated probability to the user:

[1260] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1261] Specific example

[1262] Specific example 1:

[1263] User A enters the following sentence into the terminal: "This product is very easy to use."

[1264] The terminal sends the entered text to the server.

[1265] The server preprocesses the received text and extracts features.

[1266] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1267] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1268] Specific example 2:

[1269] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1270] The terminal sends the entered text to the server.

[1271] The server preprocesses the received text and extracts features.

[1272] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1273] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1274] The system of the present invention can effectively distinguish between generative artificial intelligence and human-generated text through these processes and provide the results to the user.

[1275] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1276] Step 1:

[1277] Data collection

[1278] The server collects text data from various sources. This data includes both text written by humans and text generated by generative artificial intelligence.

[1279] Input: Text data from publicly available databases on the internet and independently collected data sources.

[1280] Output: A set of collected text data.

[1281] Step 2:

[1282] Data preprocessing

[1283] The server tokenizes the collected text data. This is the process of dividing the text into units of words or phrases.

[1284] The server then performs stemming and stop word removal. Stemming is the process of converting words into their base forms, while stop word removal is the process of removing common words that are not needed for analysis.

[1285] Input: Collected text data

[1286] Output: Preprocessed text data (data that has undergone tokenization, stemming, and stop word removal)

[1287] Step 3:

[1288] Training of artificial intelligence models

[1289] The server uses pre-processed data to create a training dataset, which includes labels for human-written text and text generated by generative artificial intelligence.

[1290] The server uses a deep learning framework (such as TensorFlow or PyTorch) to perform supervised learning and train an artificial intelligence model.

[1291] Input: Preprocessed text data and its corresponding labels

[1292] Output: Trained artificial intelligence model

[1293] Step 4:

[1294] Text input from the user

[1295] The user enters a specific sentence into the input field on the terminal and presses the send button.

[1296] Input: Text entered by the user

[1297] Output: Text data sent to the terminal

[1298] Step 5:

[1299] Sending and receiving input data

[1300] The terminal sends the text entered by the user to the server.

[1301] The server receives the text that was sent.

[1302] Input: Text data sent from the device

[1303] Output: Text data received by the server

[1304] Step 6:

[1305] Reprocessing of received data

[1306] The server reprocesses the received text. Specifically, it performs tokenization, stemming, stop word removal, and feature extraction.

[1307] Input: Text data received by the server

[1308] Output: Preprocessed text data

[1309] Step 7:

[1310] Score calculation

[1311] The server inputs the pre-processed text data into a trained artificial intelligence model.

[1312] The model analyzes the input text data and calculates a probability score generated by generative artificial intelligence. This score is expressed as a probability.

[1313] Input: Pre-processed text data

[1314] Output: Probability score generated by generative artificial intelligence

[1315] Step 8:

[1316] Submitting and displaying scores

[1317] The server sends the calculated score to the terminal.

[1318] The device displays a score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1319] Input: Calculated score

[1320] Output: Score displayed to the user

[1321] In this way, through each processing step, the system can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user.

[1322] (Application Example 1)

[1323] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1324] In today's internet environment, phishing attacks and fake news are rampant, threatening user security. Users need to be able to quickly detect these threats and receive appropriate warnings, but current technology lacks the means to do so effectively. In particular, with the widespread use of text created by generative artificial intelligence, there is a need for a reliable system that can distinguish between text written by humans and text generated by generative AI and warn users accordingly.

[1325] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1326] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, and means for detecting phishing attacks and fake news and issuing warnings. This enables users to quickly recognize the threat of phishing attacks and fake news and take appropriate measures.

[1327] "Collected text data" refers to text data collected from various sources and used to train artificial intelligence models.

[1328] An "artificial intelligence model" is trained on collected text data to determine whether a text was written by a human or generated by a generative artificial intelligence system.

[1329] "Preprocessing" refers to the process of preparing user-input text into a format that is easy to analyze, using techniques such as tokenization, stem extraction, and stop word removal.

[1330] "Generative artificial intelligence" refers to technology that automatically generates text using natural language generation models.

[1331] "Calculated probability" refers to the probability that the preprocessed text was generated by a generative artificial intelligence, and is calculated by the AI ​​model.

[1332] "Means of displaying to the user" refers to an interface for visually or audibly communicating the calculated probability to the user.

[1333] A "phishing attack" refers to a fraudulent message or email sent by a malicious third party to illegally obtain a user's personal information.

[1334] "Fake news" refers to news that is not based on actual facts and conveys false information, with the aim of misleading readers.

[1335] "Warning mechanisms" are part of a system designed to inform users of potential phishing attacks or fake news and to draw their attention to them.

[1336] Embodiments of this invention will now be described. First, the server trains an artificial intelligence model using text data collected from various sources. The collected text data includes both human-written text and text generated by generative artificial intelligence. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (e.g., TensorFlow or PyTorch) are used for preprocessing.

[1337] The user inputs a specific sentence into a device such as a smartphone or computer and sends it. The device then sends this sentence to a server. The server preprocesses the received sentence again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes it easier to analyze the sentence.

[1338] The server inputs pre-processed text into a trained artificial intelligence model, which analyzes the input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[1339] The server then sends the calculated score to the terminal, which displays this score to the user. The display uses a format such as, "There is a XX% chance that this text was generated by a generative artificial intelligence." In addition, a warning is displayed if a phishing attack or fake news is detected.

[1340] As a concrete example, a user enters the following sentence into the terminal: "Click this link for a chance to win a free gift card." The terminal sends the entered sentence to the server, which preprocesses the received sentence and extracts features. The server then uses a trained model to analyze the sentence and calculate a score indicating the likelihood that it was generated by generative artificial intelligence. For example, if this score is 85%, the terminal displays to the user "There is an 85% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message is highly likely to have been generated by generative artificial intelligence."

[1341] As another example, a user enters the following sentence into the terminal: "Due to a new government decision, all bank accounts will be suspended." The terminal sends the entered sentence to a server, which preprocesses the received sentence and extracts features. The server analyzes the sentence using a trained model and calculates a score indicating the likelihood that it was generated by generative artificial intelligence. If this score is 90%, the terminal displays to the user "There is a 90% chance that this sentence was generated by generative artificial intelligence," and issues a warning: "Warning: This message may be generated. Please verify whether it is genuine information."

[1342] As described above, the system of the present invention can effectively distinguish between generative artificial intelligence and human-written text and provide the results to the user. Furthermore, it enables rapid recognition and response to threats such as phishing attacks and fake news.

[1343] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1344] Step 1:

[1345] The server trains an artificial intelligence model using text data collected from diverse sources. This collected text data includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stem extraction, and stop word removal. Deep learning frameworks (TensorFlow and PyTorch) are used for preprocessing. The input is the collected text data, and the output is the preprocessed data.

[1346] Step 2:

[1347] The user inputs a specific sentence into a device such as a smartphone or computer and sends that sentence. The input is the sentence entered by the user, and the output is the raw data sent from the device to the server.

[1348] Step 3:

[1349] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier. The input is raw data sent from the terminal, and the output is preprocessed data.

[1350] Step 4:

[1351] The server inputs pre-processed text data into a trained artificial intelligence model, which analyzes the data to calculate a probability score of how likely it was generated by a generative AI. The input is pre-processed data, and the output is the probability score.

[1352] Step 5:

[1353] The server sends the calculated score to the terminal. The input is the probability score, and the output is the transmission of the score data to the terminal.

[1354] Step 6:

[1355] The device displays the received score to the user. The display format might be, for example, "There is an 85% chance that this text was generated by a generative artificial intelligence." In addition, if there is a high probability of a phishing attack or fake news, a warning will be displayed such as, "Warning: This message is highly likely to have been generated by a generative artificial intelligence." The input is the score data sent from the server, and the output is what is displayed to the user.

[1356] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1357] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of the following main components.

[1358] 1. Means for training an artificial intelligence model using collected text data:

[1359] The server uses text data collected from diverse sources to train artificial intelligence models. This includes both human-written text and text generated by generative AI. The server preprocesses this data, preparing it using techniques such as tokenization, stemming, and stop word removal.

[1360] For training, deep learning frameworks such as TensorFlow or PyTorch are used. The server uses this data to perform supervised learning and build a model that can determine whether a text was written by a human or generated by a generative artificial intelligence.

[1361] 2. Means for preprocessing user-input text using natural language processing techniques:

[1362] The user enters a specific message into the terminal and sends it. The terminal then sends this message to the server.

[1363] The server preprocesses the received text again, performing tokenization, stemming, stop word removal, and feature extraction. This preprocessing makes text analysis easier.

[1364] 3. Means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence:

[1365] The server inputs pre-processed text into a trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score, expressed as a probability, indicates whether or not the text was generated by a generative artificial intelligence.

[1366] 4. Means for displaying the calculated probability to the user:

[1367] The server sends the calculated score to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1368] 5. Further includes an emotion engine to recognize user emotions:

[1369] The server incorporates an emotion engine to analyze the emotions in the text entered by the user. This emotion engine analyzes the emotional tone and nuances of the text and determines whether the emotion is positive, negative, neutral, etc.

[1370] 6. Means for displaying the results of user sentiment analysis:

[1371] The server sends the results analyzed by the emotion engine to the terminal, and the terminal displays the emotion analysis results to the user. The display takes the form of "The emotion contained in this text is positive."

[1372] To explain with a concrete example, it would be as follows:

[1373] Specific example 1:

[1374] User A enters the following sentence into the terminal: "This product is very easy to use."

[1375] The terminal sends the entered text to the server.

[1376] The server preprocesses the received text and extracts features.

[1377] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1378] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1379] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be positive, the terminal displays to the user, "The emotion contained in this text is positive."

[1380] Specific example 2:

[1381] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1382] The terminal sends the entered text to the server.

[1383] The server preprocesses the received text and extracts features.

[1384] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1385] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1386] The server analyzes the text using an emotion engine and determines the emotion. If the emotion is determined to be neutral, the terminal displays to the user, "The emotion contained in this text is neutral."

[1387] As described above, the system of the present invention can effectively distinguish between text written by a generative artificial intelligence system and text written by a human, and further analyze the user's emotions and provide the results to the user.

[1388] The following describes the processing flow.

[1389] Step 1:

[1390] The server collects data. The server gathers diverse text data from the internet and other online sources. This data includes news articles, blog entries, and social media posts.

[1391] Step 2:

[1392] The server builds the dataset. The server classifies the collected text data into those written by humans and those generated by generative artificial intelligence, and then creates a dataset by assigning labels to each.

[1393] Step 3:

[1394] The server performs data preprocessing using natural language processing techniques. The server tokenizes the data, removes stop words, extracts stems and rematization, and extracts features.

[1395] Step 4:

[1396] The server builds a machine learning model. The server trains the artificial intelligence model using a deep learning framework. This includes supervised learning using text data within the dataset.

[1397] Step 5:

[1398] The user enters the text they want to evaluate. The user enters a specific sentence into their device and clicks the send button.

[1399] Step 6:

[1400] The terminal sends the entered text to the server. The terminal converts the text entered by the user into a packet format and sends it to the server.

[1401] Step 7:

[1402] The server preprocesses the received text. The server tokenizes the received text, performs stemming, stop word removal, and feature extraction to make it easier to analyze.

[1403] Step 8:

[1404] The server inputs pre-processed text into an artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. The server then calculates this score and determines the probability that the text was generated by a generative AI.

[1405] Step 9:

[1406] The server uses an emotion engine to recognize the emotion of a text. The server analyzes the text, which has been preprocessed by the emotion engine, and identifies the emotional tone and nuances within the text. As a result, it determines the emotion, such as positive, negative, or neutral.

[1407] Step 10:

[1408] The server sends the calculation results and emotion analysis results to the terminal. The server stores the potential score of the generative artificial intelligence and the emotion analysis results in a packet and sends it to the terminal.

[1409] Step 11:

[1410] The device displays the results to the user. The device analyzes the received score and sentiment analysis results and displays to the user, "There is a XX% chance that this text was generated by generative artificial intelligence." It also displays, for example, "The sentiment contained in this text is positive."

[1411] The following are specific examples.

[1412] Specific example 1:

[1413] User A enters the following sentence into the terminal: "This product is very easy to use."

[1414] The terminal sends the entered text to the server.

[1415] The server preprocesses the received text and extracts features.

[1416] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1417] If the score is 20%, the device will display to the user, "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1418] The server analyzes the text using an emotion engine and determines that the emotion is positive. The terminal then displays to the user, "The emotion contained in this text is positive."

[1419] Specific example 2:

[1420] User B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1421] The terminal sends the entered text to the server.

[1422] The server preprocesses the received text and extracts features.

[1423] The server uses a trained model to perform the analysis and calculates a probability score generated by generative artificial intelligence.

[1424] If the score is 75%, the device will display to the user, "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1425] The server analyzes the text using an emotion engine and determines that the emotion is neutral. The terminal then displays to the user, "The emotion contained in this text is neutral."

[1426] As a result, the system of the present invention can provide users with not only the ability to discriminate text using generative artificial intelligence, but also the results of sentiment analysis of the text's content.

[1427] (Example 2)

[1428] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1429] In modern text generation technology, it is difficult to distinguish between text written by humans and text generated by generative artificial intelligence. Furthermore, there is a need to analyze the sentiment of user-input text and provide feedback to more accurately understand the user's intent. However, existing technologies do not provide a single system that offers both of these functions. Therefore, the challenge is to provide a unified system that determines whether user-input text was generated by generative artificial intelligence or written by a human, and further analyzes and provides feedback on its sentiment.

[1430] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1431] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input by a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, means for displaying the calculated probability to the user, means for analyzing the sentiment of the text input by the user, and means for displaying the sentiment analysis results to the user. This makes it possible to determine whether the text input by the user was generated by a generative artificial intelligence or by a human, perform sentiment analysis, and provide feedback to the user.

[1432] "Collected text data" refers to data consisting of texts written by humans and texts generated by generative artificial intelligence, collected from various sources.

[1433] An "artificial intelligence model" is a model built using deep learning technology and trained to distinguish between human-written text and text generated by a generative artificial intelligence system.

[1434] "Natural language processing technology" refers to techniques for analyzing and preprocessing text, such as tokenization, stem extraction, stop word removal, and feature extraction.

[1435] "Generative artificial intelligence" refers to artificial intelligence that has been pre-trained and has the ability to generate new text based on input text.

[1436] "Methods for calculating probability" refers to methods for inputting pre-processed text into an artificial intelligence model and calculating the probability that the text was generated by a generative artificial intelligence as a probability value.

[1437] "Means of displaying to the user" refers to the means of displaying the calculated probability and sentiment analysis results on the user's device.

[1438] "Methods for analyzing emotions" refer to methods for analyzing the emotional tone of text entered by a user and determining whether the text is positive, negative, or neutral.

[1439] "Means for displaying emotion analysis results" refers to means of visually presenting the analyzed emotion results to the user.

[1440] The system of this invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further combines this with an emotion engine that recognizes the user's emotions. This system consists of a server, a terminal, and a user.

[1441] 1. Program generation

[1442] The server trains an artificial intelligence model using collected text data. This data includes both human-written text and text generated by generative AI. The server preprocesses this data by performing tokenization (word splitting), stemming (extraction of the core part of words), and removal of stop words (frequently occurring words). Based on the preprocessed data, it performs supervised learning using deep learning frameworks such as TensorFlow and PyTorch to build an artificial intelligence model.

[1443] 2. Program Processing

[1444] The server preprocesses the text entered by the user using natural language processing techniques. This converts newly entered text into a format that can be parsed. The preprocessing consists of tokenization, stemming, stop word removal, and feature extraction.

[1445] Next, the server inputs the pre-processed text into a trained artificial intelligence model and calculates a score indicating the likelihood that the text was generated by a generative AI. This score represents the probability that the text was generated by a generative AI.

[1446] 3. Hardware and Software

[1447] The servers utilize high-performance server clusters and cloud computing services. Furthermore, TensorFlow and PyTorch are used as software for natural language processing and deep learning. These software programs are specialized for preprocessing text data and training models.

[1448] 4. Data processing and calculations

[1449] The server first tokenizes the input text data, then performs stemming, stop word removal, and feature extraction. This allows for efficient text analysis. Subsequently, the pre-processed data is input into an artificial intelligence model to obtain analysis results. These results are expressed as a score and provided to the user.

[1450] 5. Specific Examples

[1451] For example, user A enters the following sentence into the terminal: "This product is very easy to use."

[1452] The terminal sends the entered text to the server.

[1453] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 20%, the terminal will display "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1454] The server analyzes the text using its sentiment engine, and if it determines that the sentiment is positive, the terminal displays the message "The sentiment contained in this text is positive." to the user.

[1455] As another example, user B enters the following sentence into the terminal: "Performance has been improved thanks to the introduction of a new algorithm."

[1456] The terminal sends the entered text to the server.

[1457] The server preprocesses the text and analyzes it using a trained model. For example, if the score is 75%, the terminal will display "There is a 75% chance that this text was generated by a generative artificial intelligence."

[1458] If the server analyzes the text using its emotion engine and determines it to be neutral, the terminal will display the message "The emotion contained in this text is neutral." to the user.

[1459] In this way, the present invention realizes a system that can effectively distinguish between text written by a generative artificial intelligence system and a human, and further analyze the user's emotions and provide the results to the user.

[1460] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1461] Processing flow

[1462] Step 1: The user enters and submits the text.

[1463] Input: The user enters text into the input box.

[1464] Specific action: The user enters any text into the input form on the device and clicks the submit button. For example, the user enters the sentence, "This product is very easy to use."

[1465] Output: The input text is passed to the terminal.

[1466] Step 2: The device sends the text to the server.

[1467] Input: The text entered by the terminal.

[1468] Specific operation: The terminal sends the text entered by the user to the server as text data. The transmission is performed as an HTTP request.

[1469] Output: The input text is passed to the server.

[1470] Step 3: The server preprocesses the text.

[1471] Input: The text sent to the server.

[1472] Specific operation: The server tokenizes the received text and breaks it down into individual words. For example, the sentence "This product is very easy to use." is split into tokens such as "this," "product," "is," "very," "easy to use," and "is." Next, the server extracts the stems of the tokens and converts "easy to use" to "easy to use." Furthermore, unnecessary words (stop words) are removed, and words like "is" and "is" are deleted. Finally, features are extracted and formatted in a way that a machine learning model can understand.

[1473] Output: Preprocessed data.

[1474] Step 4: The server parses the pre-processed text.

[1475] Input: Pre-processed data.

[1476] Specific operation: The server inputs pre-processed data into a trained artificial intelligence model, which then analyzes the data.

[1477] Output: Analysis results (probability score of generation by generative AI).

[1478] Step 5: The server calculates the potential score of the generative AI.

[1479] Input: Analysis results.

[1480] Specific operation: The server obtains the output of the artificial intelligence model as a probability value. If the model determines "20%", it indicates that there is a 20% chance that the text was generated by a generative AI.

[1481] Output: Potential score for generative AI.

[1482] Step 6: The device displays the score to the user.

[1483] Input: Potential score for generative AI.

[1484] Specific operation: The server sends the calculated score to the terminal, and the terminal displays this score to the user. The display format will be "There is a 20% chance that this text was generated by a generative artificial intelligence."

[1485] Output: The user checks the score.

[1486] Step 7: The server analyzes emotions using the emotion engine.

[1487] Input: Text entered by the user.

[1488] Specific operation: The server uses an emotion engine to analyze the emotional tone of a sentence. For example, the sentence "This product is very easy to use" is determined to have a positive emotion.

[1489] Output: Emotion analysis results.

[1490] Step 8: The device displays the sentiment analysis results to the user.

[1491] Input: Sentiment analysis results.

[1492] Specific operation: The server sends the analysis results to the terminal, and the terminal displays to the user in the form of "The emotions contained in this text are positive."

[1493] Output: The user reviews the sentiment analysis results.

[1494] Through these steps, the system of the present invention can determine whether the text entered by the user was generated by a generative artificial intelligence system or by a human, analyze the emotions contained in the text, and provide feedback to the user.

[1495] (Application Example 2)

[1496] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1497] Conventional systems have struggled to determine whether user-generated text is created by artificial intelligence or by a human, and to analyze the emotions contained within that text. In particular, the food delivery industry requires both authenticity verification and sentiment analysis of input text to efficiently manage customer support and reviews. This will enable more accurate and faster service delivery, leading to improved customer satisfaction.

[1498] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1499] In this invention, the server includes means for training an artificial intelligence model using collected text data, means for preprocessing text input from a user using natural language processing technology, means for inputting the preprocessed text into the artificial intelligence model and calculating the probability of it being generated by generative artificial intelligence, means for displaying the calculated probability to the user, means for performing sentiment analysis on the text input from the user, and means for displaying the results of the sentiment analysis to the user. This enables more efficient customer support and review management in the food delivery industry, contributing to improved customer satisfaction.

[1500] "Collected text data" refers to text information obtained from various sources, including text written by humans and text generated by generative artificial intelligence.

[1501] An "artificial intelligence model" is an algorithm used to learn from collected data and perform a specific task.

[1502] "Training methods" refer to the process of inputting a wide variety of data into an artificial intelligence model to improve its accuracy and discriminatory capabilities.

[1503] "Natural language processing technology" refers to technologies that enable computers to understand, interpret, and generate human language, and includes techniques such as tokenization, stem extraction, and stop word removal.

[1504] "Preprocessing" refers to the process of converting raw text data into a parseable format, which includes tokenization and stop word removal.

[1505] "Generative artificial intelligence" refers to algorithms that use trained models to generate text input from users, and which have a high accuracy in text generation.

[1506] "Means for calculating the probability of generation" refers to the process of inputting a pre-processed text into an artificial intelligence model and calculating a probability score indicating whether or not the text was generated by a generative artificial intelligence.

[1507] "Means of displaying to the user" refers to an interface for providing the user with the results calculated by the artificial intelligence model, and includes screen display and audio output.

[1508] "Sentiment analysis" is the process of analyzing the emotional tone and nuances contained in text to determine whether the emotion is positive, negative, neutral, etc.

[1509] The system of the present invention determines whether text entered by a user was generated by a generative artificial intelligence system or written by a human, and further analyzes the emotions contained in that text. In order to specifically implement the system of the present invention, the following means and processes are necessary.

[1510] First, the server trains an artificial intelligence model using text data collected from diverse sources. During this process, the server uses deep learning frameworks such as TensorFlow and PyTorch to preprocess data containing both human-written text and text generated by generative AI. This preprocessing includes tokenization, stem extraction, and stop word removal to prepare the training data.

[1511] Next, the user uses a smartphone or other device to input and send a specific sentence. The device sends this sentence to the server, which then preprocesses the received sentence again. This re-preprocessing involves tokenization, stemming, stop word removal, and feature extraction.

[1512] The pre-processed text is input into a server-trained artificial intelligence model. The model analyzes this input text and calculates a score indicating the likelihood that it was generated by a generative artificial intelligence. This score is expressed as a probability and indicates whether or not the text was generated by a generative artificial intelligence.

[1513] The calculated score is sent from the server to the terminal, which then displays this score to the user. The display takes the form of, "There is a XX% chance that this text was generated by a generative artificial intelligence."

[1514] Furthermore, the server performs sentiment analysis on the text entered by the user. This process uses a sentiment engine that analyzes the emotional tone and nuances of the text and determines emotions such as positive, negative, or neutral. Natural language processing tools such as the nlptown / bert-base-multilingual-uncased-sentiment model are used for sentiment analysis.

[1515] The analysis results are sent from the server to the terminal, which then displays the sentiment analysis results to the user. The display takes the form of statements such as, "The sentiment contained in this text is positive," or "The sentiment contained in this text is neutral."

[1516] Specific example:

[1517] The user left the following review: "This delivery is fantastic! Fast and reliable."

[1518] In this case, the system will display the following result:

[1519] "There is a 15.50% probability that this text was generated by a generative artificial intelligence." (Example: Calculated using training data)

[1520] "The emotions conveyed in this text are positive."

[1521] Examples of prompts to input into a generative AI model:

[1522] User input: "This delivery is fantastic! Fast and reliable."

[1523] analysis:

[1524] 1. There is a 15.50% probability that this text was generated by a generative artificial intelligence.

[1525] 2. The emotions contained in this sentence are positive.

[1526] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1527] Step 1:

[1528] Users input and send specific text using their smartphones or other devices. The entered text is sent from the device to the server in its original form.

[1529] Input: Text entered by the user (e.g., "This delivery is fantastic! Fast and reliable.")

[1530] Output: Sent text data

[1531] Step 2:

[1532] The server preprocesses the received text using natural language processing techniques. This preprocessing step involves tokenization, stemming, and stop word removal. This converts the text into a format that is easy to analyze.

[1533] Input: Submitted text data

[1534] Output: Preprocessed text data

[1535] Specific operation: Use the BERT tokenizer to tokenize the text, remove unnecessary words, and format it for analysis.

[1536] Step 3:

[1537] The server inputs pre-processed text into a trained artificial intelligence model, which then calculates a probability score for the text as it was generated by the generative AI. The model analyzes the input text data and outputs the result as a probability score.

[1538] Input: Preprocessed text data

[1539] Output: Probability score generated by generative artificial intelligence (e.g., 15.50%)

[1540] Specific operation: Preprocessed text is input into a BERT-based model, and the classification result is output as a probability score.

[1541] Step 4:

[1542] The server sends the calculated score for the generative artificial intelligence to the terminal. The terminal displays this score to the user. The display format is "There is a XX% chance that this text was generated by generative artificial intelligence."

[1543] Input: Score of the Generative Artificial Intelligence

[1544] Output: Content displayed to the user

[1545] Specific action: The submitted score is displayed on the terminal's user interface.

[1546] Step 5:

[1547] The server performs sentiment analysis on text entered by the user. It uses a sentiment engine to analyze the emotional tone and nuances of the text. The analysis results are classified as positive, negative, neutral, etc.

[1548] Input: Preprocessed text data

[1549] Output: Sentiment analysis results (e.g., positive)

[1550] Specific operation: Classify the sentiment of a text using the nlptown / bert-base-multilingual-uncased-sentiment model and obtain the result.

[1551] Step 6:

[1552] The server sends the sentiment analysis results to the terminal, and the terminal displays the results to the user. The display format is something like, "The sentiment contained in this text is positive."

[1553] Input: Sentiment analysis results

[1554] Output: Content displayed to the user

[1555] Specific operation: The transmitted sentiment analysis results are displayed on the device's user interface.

[1556] 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 controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1557] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1558] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[1559] Furthermore, the emotion identification model 59, acting 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 a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[1560] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[1561] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[1562] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[1563] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[1564] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is 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 the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[1565] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[1566] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[1567] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[1568] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[1569] 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.

[1570] Furthermore, it is not necessary to store the entirety 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 the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[1571] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[1572] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of 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). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[1573] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[1574] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[1575] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[1576] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted as being incorporated by reference.

[1577] The following is further disclosed regarding the embodiments described above.

[1578] (Claim 1)

[1579] A method for training an artificial intelligence model using collected text data,

[1580] A means for preprocessing user-input text using natural language processing technology,

[1581] A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence,

[1582] A means for displaying the calculated probability to the user,

[1583] A system that includes this.

[1584] (Claim 2)

[1585] The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning technology.

[1586] (Claim 3)

[1587] The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence.

[1588] "Example 1"

[1589] (Claim 1)

[1590] A method for training an artificial intelligence model using collected text data,

[1591] A means for preprocessing text input by a user using natural language processing technology,

[1592] A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence,

[1593] A means for displaying the calculated probability to the user,

[1594] A system that includes this.

[1595] (Claim 2)

[1596] The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning techniques.

[1597] (Claim 3)

[1598] The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence.

[1599] "Application Example 1"

[1600] (Claim 1)

[1601] A method for training an artificial intelligence model using collected text data,

[1602] A means for preprocessing user-input text using natural language processing technology,

[1603] A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence,

[1604] A means for displaying the calculated probability to the user,

[1605] A means to detect and warn about phishing attacks and fake news,

[1606] A system that includes this.

[1607] (Claim 2)

[1608] The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning technology.

[1609] (Claim 3)

[1610] The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence.

[1611] "Example 2 of combining an emotion engine"

[1612] (Claim 1)

[1613] A method for training an artificial intelligence model using collected text data,

[1614] A means for preprocessing user-input text using natural language processing technology,

[1615] A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence,

[1616] A means for displaying the calculated probability to the user,

[1617] A means of analyzing the sentiment of text entered by the user,

[1618] A means for displaying the emotion analysis results to the user,

[1619] A system that includes this.

[1620] (Claim 2)

[1621] The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning technology.

[1622] (Claim 3)

[1623] The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence.

[1624] "Application example 2 when combining with an emotional engine"

[1625] (Claim 1)

[1626] A method for training an artificial intelligence model using collected text data,

[1627] A means for preprocessing user-input text using natural language processing technology,

[1628] A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence,

[1629] A means for displaying the calculated probability to the user,

[1630] A means of performing sentiment analysis on text input by a user,

[1631] A means for displaying the results of the aforementioned emotion analysis to the user,

[1632] A system that includes this.

[1633] (Claim 2)

[1634] The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning technology.

[1635] (Claim 3)

[1636] The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence. [Explanation of Symbols]

[1637] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Devices 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robots< / url:> < / url:> < / url:> < / url:>

Claims

1. A method for training an artificial intelligence model using collected text data, A means for preprocessing user-input text using natural language processing technology, A means for inputting preprocessed text into the artificial intelligence model and calculating the probability that it was generated by a generative artificial intelligence, A means for displaying the calculated probability to the user, A system that includes this.

2. The system according to claim 1, wherein the artificial intelligence model is constructed using deep learning technology.

3. The system according to claim 1, wherein the collected text data includes both text written by humans and text generated by a generative artificial intelligence.

Images