Method for generating ai hyper persona

Abstract

The AI ​​hyper persona generation method of the present invention comprises the steps of: extracting personal information related to a virtual natural person from previously stored personal data; generating an AI person representing the natural person by learning the biological characteristics, social characteristics, psychological characteristics, and cultural characteristics of the natural person; and generating an AI hyper persona by applying business characteristic information to the AI ​​person.

Description

Technology Field

[0001] The present invention relates to a method for generating AI hyper personas. Background Technology

[0002] A persona is a profile of an individual with specific characteristics; in the field of marketing, it refers to a fictional customer profile that depicts the ideal customer group for a particular product or service. Personas are created based on factors such as customer gender, age, occupation, income level, interests, purchasing motivations, and consumption patterns, and play a crucial role in marketing strategies, content creation, advertising campaigns, and product development.

[0003] Existing traditional persona generation methods have limitations, such as being time-consuming due to manual generation, potentially low representativeness due to limited sample sizes, the possibility of reflecting researcher biases, and difficulty in responding immediately to market changes.

[0004] Meanwhile, AI Human is a virtual human that embodies human-like form and behavior based on artificial intelligence technology, enabling interaction with humans primarily through the use of natural language processing, computer vision, speech recognition, sentiment analysis, and 3D animation technologies. Through this, it can be utilized in various fields such as virtual assistants, customer service, education, healthcare, entertainment, and marketing.

[0005] These AI humans operate based on Natural Language Processing (NLP), speech recognition, computer vision, and 3D animation technologies, but they have difficulty perfectly understanding and responding to human linguistic nuances. For example, AI humans struggle to grasp context or naturally handle humor, satire, and emotional expressions, and their ability to respond to new situations or unstructured data that deep learning models have not learned is limited.

[0006] Recently, with the advancement of artificial intelligence technology, AI-based chatbots are being developed that utilize Natural Language Processing (NLP) and Machine Learning (ML) technologies to understand questions, recognize the intent and key entities of user utterances, and flexibly generate appropriate responses. While tools such as OpenAI’s ChatGPT and Google Gemini replicate conversations close to those of real people based on Large Language Models (LLM), existing AI-based chatbots are designed to handle questions from a large number of users and therefore cannot provide answers based on specific conditions or the thoughts of specific individuals. In other words, existing AI-based chatbots face the problem of being unable to obtain the opinions of a minority of individuals and struggling to deeply incorporate their views. The problem to be solved

[0007] The technical problem that the present invention aims to solve is to provide a method for generating an AI hyper-persona that utilizes artificial intelligence technology to implement an AI-based virtual consumer model possessing the identity of a specific individual, supporting data-driven decision-making, and overcoming temporal and spatial constraints. means of solving the problem

[0008] A method for generating an AI hyper persona according to an embodiment of the present invention comprises the steps of: extracting personal information related to a virtual natural person from previously stored personal data; generating an AI person representing the natural person by learning the biological characteristics, social characteristics, psychological characteristics, and cultural characteristics of the natural person; and generating an AI hyper persona by applying business characteristic information to the AI ​​person.

[0009] According to an embodiment, the step of generating the AI ​​person comprises: generating a first answer based on the personal information; providing a first question to a raw AI person regarding a first profile item related to biological and social characteristics and a second profile item corresponding to psychological characteristics; training the raw AI person with the first answer in response to the first question to generate a first preliminary AI person; providing a second question regarding a third profile item to the first preliminary AI person; training the first preliminary AI person with a second answer that maintains consistency with the first answer in response to the second question to generate a second preliminary AI person; providing a third question regarding a fourth profile item to the second preliminary AI person; the second preliminary AI person self-generating a third answer that maintains consistency with the first and second answers through a pre-trained Large Language Model (LM) in response to the third question; training the second preliminary AI person with the self-generated third answer to generate the AI ​​person; and the first to It may include a step of storing the third answers as characteristic information of the AI ​​person.

[0010] According to an embodiment, the first profile item may include information related to age, gender, occupation, education level, residential area, family composition, and income level, and the second profile item may include information related to openness, conscientiousness, extraversion, agreeableness, and neuroticism based on the Five Factor Theory (FFT) of personality.

[0011] According to an embodiment, the third profile item is generated by expanding the frame defining human characteristics for each of the first profile item and the second profile item, and includes biological characteristics, social characteristics, psychological characteristics, and cultural characteristics, and the fourth profile item is generated by expanding the frame defining human characteristics for the third profile item, and may include biological characteristics, social characteristics, psychological characteristics, and cultural characteristics.

[0012] According to an embodiment, the step of generating the first preliminary AI person may include providing the first answer to the raw AI person in response to the first question, selecting the raw AI person that selected the first answer as the first preliminary AI person, and storing the first answer as the first characteristic information of the first preliminary AI person.

[0013] According to an embodiment, the step of generating the second preliminary AI person may include providing selectable answers corresponding to the second question to the second preliminary AI person, selecting the first preliminary AI person who selected the second answer among the answers as the second preliminary AI person, and storing the first answer and the second answer as second characteristic information of the second preliminary AI person.

[0014] According to an embodiment, the step of selecting the second preliminary AI person may include a step in which one of the answers is randomly selected by the first preliminary AI person, a step in which a validation checker determines consistency between the selected answer and the first answer, and a step of repeating the answer selection of the first preliminary AI person if the first preliminary AI person selects an inconsistent answer.

[0015] According to an embodiment, the step of self-generating the third answer may include the step of the second preliminary AI person generating an answer corresponding to the third question based on the first and second answers through the LLM, the step of a validation checker determining consistency between the generated answer and the first and second answers, and the step of repeating the generation of an answer through the LLM if the second preliminary AI person generates an inconsistent answer.

[0016] According to an embodiment, the step of generating the AI ​​hyper persona may include: generating a business question using the business characteristic information; the AI ​​person generating a fourth answer corresponding to the business question through LLM; determining the consistency between the generated fourth answer and the characteristic information of the AI ​​person by a validation checker; and selecting the AI ​​person exhibiting a consistent answer pattern as the AI ​​hyper persona.

[0017] According to an embodiment, the step of selecting the AI ​​hyper persona may remove an AI person that fails to generate a consistent fourth answer within a preset number of times. Effects of the invention

[0018] According to the AI ​​hyper-persona generation method of an embodiment of the present invention, a virtual consumer model possessing the identity of a specific individual can be provided by utilizing artificial intelligence technology, thereby supporting data-based decision-making and enabling interviews and market research without spatial and temporal constraints.

[0019] In addition, according to the AI ​​hyper persona generation method of the embodiment of the present invention, rather than a simple static persona, it is possible to provide an AI hyper persona that evolves according to market changes by continuously performing data learning, and thereby enable real-time data updates by reflecting the latest user trends.

[0020] In addition, according to the AI ​​hyper-persona generation method of the embodiment of the present invention, multiple AI persons with different identities can be generated, and an AI hyper-persona with applied business characteristics can be generated by selecting an AI person that meets desired conditions.

[0021] In addition, according to the AI ​​hyper-persona generation method of the embodiment of the present invention, by automatically generating an AI hyper-persona based on large-scale data analysis and AI learning, a more objective and reliable consumer profile can be provided.

[0022] In addition, according to the AI ​​hyper persona generation method of the embodiment of the present invention, interview subjects desired by the user can be generated as virtual AI hyper personas, thereby enabling interviews to be conducted beyond time and spatial constraints.

[0023] In addition, according to the AI ​​hyper-persona generation method of the embodiment of the present invention, the user research process is automated to reduce time and costs, and data-driven decision-making is supported. Users can intervene in real time and modify data through the system, thereby experiencing a more flexible and efficient research process.

[0024] Furthermore, according to the AI ​​hyper-persona generation method of the embodiment of the present invention, more sophisticated interviews are possible by reflecting the psychological characteristics and purchasing behavior of a specific market group through the AI ​​hyper-persona, and it is possible to derive deeper insights beyond simple responses. Brief explanation of the drawing

[0025] FIG. 1 is a schematic block diagram of a user research automation system utilizing artificial intelligence technology according to an embodiment of the present invention. FIG. 2 is a schematic block diagram of an AI hyper-persona generation device according to an embodiment of the present invention. FIG. 3 is a block diagram of a processor according to an embodiment of the present invention. FIG. 4 is a diagram illustrating a method for generating an AI person according to an embodiment of the present invention. FIG. 5 is a diagram illustrating an AI hyper-persona generation method according to an embodiment of the present invention. FIGS. 6 to 9 are drawings illustrating a method for automating user research using artificial intelligence technology according to an embodiment of the present invention. Specific details for implementing the invention

[0026] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the present invention, and the present invention is defined only by the scope of the claims.

[0027] The terms used in this specification are for describing embodiments and are not intended to limit the invention. In this specification, the singular form includes the plural form unless specifically stated otherwise in the text. The terms "comprises" and / or "comprising" as used in this specification do not exclude the presence or addition of one or more other components in addition to the components mentioned. Throughout the specification, the same reference numerals refer to the same components, and "and / or" includes each of the mentioned components and all combinations of one or more. Although terms such as "first," "second," etc., are used to describe various components, these components are not limited by these terms. These terms are used merely to distinguish one component from another. Accordingly, the first component mentioned below may be the second component within the technical scope of the invention.

[0028] Unless otherwise defined, all terms used herein (including technical and scientific terms) may be used in a meaning commonly understood by those skilled in the art to which the present invention pertains. Additionally, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise.

[0029] The terms “part” or “module” as used in the specification refer to software or hardware components, such as FPGAs or ASICs, and “part” or “module” perform certain roles. However, “part” or “module” is not limited to software or hardware. “Part” or “module” may be configured to reside in an addressable storage medium or configured to run on one or more processors. Thus, by example, “part” or “module” includes components such as software components, object-oriented software components, class components, and task components, as well as processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and “parts” or “modules” may be combined into a smaller number of components and “parts” or “modules,” or further separated into additional components and “parts” or “modules.”

[0030] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0031] FIG. 1 is a schematic block diagram of a user research automation system utilizing artificial intelligence technology according to an embodiment of the present invention.

[0032] Referring to FIG. 1, an AI hyper-persona generation system (10) according to an embodiment of the present invention includes an AI hyper-persona generation device (100) and a user terminal (200).

[0033] The AI ​​hyper persona generation device (100) is a network-addressable device capable of hosting an online network and can communicate with the user terminal (200) to provide a business interview to the user terminal (200).

[0034] Here, the network communication refers to a connection structure capable of exchanging information between the AI ​​hyper-persona generation device (100) and the user terminal (200), and may include a local area network (LAN), a wide area network (WAN), the internet (WWW), a wired and wireless data communication network, a telephone network, a wired and wireless television communication network, etc.

[0035] According to an embodiment, the AI ​​hyper persona generation device (100) may be a cloud computing model that operates a service platform for providing business interviews, builds the service platform on-premise, and provides a Platform as a Service (PaaS) capable of running and managing related applications.

[0036] The AI ​​hyper persona generation device (100) can generate an AI person using a pre-trained artificial intelligence model and generate an AI hyper persona by applying business characteristic information to the AI ​​person. Here, the artificial intelligence model may be a deep learning model. Deep learning models (e.g., deep neural networks (DNN)) may refer to multilayer perceptrons that include multiple hidden layers in addition to input and output layers. Deep neural networks can be used to identify the latent structures of data. Deep neural networks may include, but are not limited to, convolutional neural networks (CNN), recurrent neural networks (RNN), autoencoders, Generative Adversarial Networks (GAN), restricted Boltzmann machines (RBM), deep belief networks (DBN), Q networks, U networks, and Siamese networks.

[0037] The AI ​​hyper persona generation device (100) can generate an AI Person representing a virtual natural person. The AI ​​Person is a digital being that thinks and acts like a human based on artificial intelligence, and unlike a simple chatbot or virtual assistant, it has an individual personality and identity, and can acquire the individual personality and identity of the natural person by learning the biological characteristics, social characteristics, psychological characteristics, and cultural characteristics of the natural person.

[0038] To this end, the AI ​​hyper persona generation device (100) can generate an AI person by extracting personal information related to a virtual natural person from previously stored personal data and assigning an identity related to the personal information to the raw AI person through an interview with the raw AI person. Here, the raw AI person is a virtual natural person capable of conducting an interview, and may be in a state where an identity has not been formed as it has not learned biological characteristics, social characteristics, psychological characteristics, and cultural characteristics.

[0039] For example, the AI ​​hyper persona generation device (100) can extract personal information related to a natural person by using data from the National Statistical Office census, social security system statistics, job information from the Ministry of Employment and Labor, labor market survey data, personal data from the Korea Educational Development Institute, family statistics data from the National Statistical Office, National Health and Nutrition Survey data, data from the National Health Insurance Service, data from the Korea Disease Control and Prevention Agency, etc.

[0040] Personal information extracted from personal data can be used as personal information of a virtual natural person to create an AI person, and may include information regarding age, gender, occupation, education level, residential area, family composition, and income level.

[0041] For example, the AI ​​hyper persona generation device (100) can extract personal information of a natural person from personal data, and the personal information may include 63 years old, female, unemployed, high school graduate, resident of Seoul, family of four, and no income. The AI ​​hyper persona generation device (100) can perform an interview with a raw AI person based on the personal information and can generate an AI person by controlling the raw AI person to learn its own answers.

[0042] The AI ​​hyper-persona generation device (100) can manage an AI person data pool. Multiple AI persons with different identities can be stored in the AI ​​person data pool, and the AI ​​hyper-persona generation device (100) can add and store the AI ​​person in the AI ​​person data pool whenever an AI person is created.

[0043] The AI ​​Person data pool serves as a virtual space where AI Persons exist, and the AI ​​Hyper Persona generation device (100) extracts an AI Person from the AI ​​Person data pool that matches the user's needs or business characteristics, and can utilize the extracted AI Person as an AI Hyper Persona.

[0044] The AI ​​hyper persona generation device (100) can generate an AI hyper persona by applying business characteristic information to an AI person. An AI hyper persona is a type of AI human implemented based on artificial intelligence, and refers to a state in which individual characteristics such as personality, speech, values, behavioral patterns, consumption habits, age, place of residence, and occupation are implemented in relation to a specific business.

[0045] AI Hyper Personas are a type of AI-based virtual consumer model that reflects actual consumer data and evolves through continuous learning in response to market changes. Unlike traditional static personas, AI Hyper Personas are generated through large-scale data analysis and AI learning; they can incorporate real-time feedback and the latest trends, and by learning from user data and online behavioral data in real time, they can rapidly respond to consumer inquiries related to the business.

[0046] In particular, the AI ​​hyper persona according to an embodiment of the present invention refers to a virtual persona designed to simulate the mindset of a specific individual or group by reflecting more in-depth demographic and psychological characteristics for a specific purpose and situation than conventional AI humans.

[0047] According to an embodiment, the AI ​​hyper persona generation device (100) may receive business characteristic information related to a business field from a user terminal (200) and may generate an AI hyper persona based on the business characteristic information. The AI ​​hyper persona generation device (100) may receive business characteristic information directly from the user terminal (200), but may also generate business characteristic information by analyzing the input user data. Here, the business characteristic information may be provided in various formats, such as text files, image files, PDF files, URL links, and shared documents.

[0048] According to an embodiment, the AI ​​hyper persona generation device (100) may provide a virtual interview room for conducting an interview between a user and an AI hyper persona.

[0049] The AI ​​hyper persona can provide answers that correspond to information learned based on personal information. The AI ​​hyper persona generation device (100) performs a process of specifying what tone of voice and attitude the AI ​​hyper persona will respond with in what situation, and through this, the AI ​​hyper persona can be designed to perform consistent answers.

[0050] According to another embodiment, the AI ​​hyper persona generation device (100) may provide additional questions to assist in an interview between the user and the AI ​​hyper persona, and may analyze the interview and provide the results of the analysis to the user.

[0051] The user terminal (200) is connected to the service platform and can input business characteristic information related to the business field to perform an interview with an AI hyper persona.

[0052] According to an embodiment, the user terminal (200) may be any one of a personal computer, a smartphone, a tablet PC, a mobile internet device (MID), an internet tablet, an IoT (internet of things) device, an IoE (internet of everything) device, a desktop computer, a laptop computer, a workstation computer, a Wibro (Wireless Broadband Internet) terminal, and a PDA (Personal Digital Assistant), but is not limited thereto and may include all types of communication devices.

[0053] FIG. 2 is a schematic block diagram of an AI hyper-persona generation device according to an embodiment of the present invention.

[0054] Referring to FIG. 2, the AI ​​hyper persona generation device (100) operates a service platform and can conduct interviews by generating an AI hyper persona related to the user's business field through the service platform.

[0055] To this end, an AI hyper persona generation device (100) according to an embodiment of the present invention includes a processor (110), memory (120), a communication interface (130), and storage (140).

[0056] The processor (110) can control the overall operation of each component of the AI ​​hyper-persona generation device (100). The processor (110) may be configured to include a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphic Processing Unit), or any form of processor (110) well known in the art of the present invention.

[0057] Additionally, the processor (110) can perform operations for at least one application or program for executing the method according to embodiments of the present invention, and the server (100) may have one or more processors (110).

[0058] According to an embodiment, the processor (110) may further include RAM (Random Access Memory, not shown) and ROM (Read-Only Memory, not shown) for temporarily and / or permanently storing signals (or data) processed within the processor (110). Additionally, the processor (110) may be implemented in the form of a system-on-chip (SoC) comprising at least one of a graphics processing unit, RAM, and ROM.

[0059] Memory (120) stores various data, commands, or information. Memory (120) may load a program for operating a service platform from storage (140) to execute a method according to various embodiments of the present invention. When a computer program is loaded into memory (120), the processor (110) may perform the method by executing one or more instructions constituting the computer program. Memory (120) may be implemented as volatile memory such as RAM, but the technical scope of the present disclosure is not limited thereto.

[0060] The communication interface (130) supports wired or wireless communication of the AI ​​hyper persona generation device (100). Additionally, the communication interface (130) may support various communication methods other than internet communication. To this end, the communication interface (130) may be configured to include a communication module well known in the art of the present invention. In some embodiments, the communication interface (130) may be omitted.

[0061] Storage (140) can store computer programs non-temporarily. When operating a service platform through an AI hyper-persona generation device (100), storage (140) can store various information required for generation and processing during the execution of a process.

[0062] The storage (140) may be configured to include non-volatile memory such as ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), flash memory, a hard disk, a removable disk, or any form of computer-readable recording medium well known in the art to which the present invention belongs.

[0063] The bus (150) provides communication functions between components of the AI ​​hyper-persona generation device (100). The bus (150) can be implemented as various types of buses, such as an address bus, a data bus, and a control bus.

[0064] FIG. 3 is a block diagram of a processor according to an embodiment of the present invention.

[0065] Referring to FIG. 3, the processor (110) includes a data processing unit (111), an AI person generation unit (112), an AI hyper persona generation unit (113), an interview execution unit (114), and an analysis unit (115), each component being implemented by a software module, a hardware module, or a combination thereof, and may mean a unit that processes at least one function or operation connected through at least one communication path.

[0066] The data processing unit (111) can preprocess business characteristic information received from the user terminal (200), and the business characteristic information can be provided in various formats such as short text, long text, PDF file, URL link, shared document link, etc.

[0067] The data processing unit (111) can determine important words in the text by analyzing business characteristic information.

[0068] For example, the data processing unit (111) can extract keywords based on the word frequency included in business characteristic information and determine the frequency of occurrence of a specific word to determine frequently occurring words as important words.

[0069] For example, the data processing unit (111) can learn vectors that reflect semantic relationships between words using natural language processing models such as Word2Vec, GloVe, and FastText, and can determine important words within business characteristic information using the learned natural language processing model.

[0070] And, the data processing unit (111) can derive major topics by considering the association between important words.

[0071] For example, the data processing unit (111) can use Latent Dirichlet Allocation (LDA) to assume that business characteristic information consists of multiple topics and probabilistically model which topic each word included in the business characteristic information belongs to. The data processing unit (111) can randomly assign the words included in the business characteristic information to topics and extract the distribution for the topics to which the words are assigned. The data processing unit (111) can use the distribution for the topics to determine which topics are emphasized in the business characteristic information.

[0072] And, the data processing unit (111) can determine the business field by analyzing the context based on important words and topics, and can determine the main consumer base of the service product by generating an overview of the service product of the determined business field.

[0073] For example, the data processing unit (111) can receive a business plan PDF file as business characteristic information from the user terminal (200), analyze the contents of the PDF file to determine the business field as the men's cosmetics field, and generate an overview of men's skin lotion as a service product to determine men in their 20s to 40s as the main consumer group.

[0074] For example, the data processing unit (111) can receive a URL link of a shopping mall purchase site as business characteristic information from the user terminal (200), analyze the webpage connected to the URL link to determine the business field as the men's cosmetics field, and generate an overview of men's skin lotion as a service product to determine men in their 20s to 40s as the main consumer group.

[0075] The AI ​​person generation unit (112) can generate AI persons and manage an AI person data pool containing multiple generated AI persons.

[0076] First, the AI ​​person generation unit (112) can generate a first answer for a raw AI person by extracting personal information related to a virtual natural person from previously stored personal data to generate an AI person, and based on the extracted personal information.

[0077] For example, the AI ​​person generation unit (112) can extract personal information related to a natural person by using data from the National Statistical Office's population survey, social security system statistics, employment information from the Ministry of Employment and Labor, labor market survey data, personal data from the Korea Educational Development Institute, family statistics data from the National Statistical Office, National Health and Nutrition Survey data, National Health Insurance Service data, Korea Disease Control and Prevention Agency data, etc.

[0078] The first answer may include information related to age, gender, occupation, education level, residential area, family composition, and income level, generated based on personal information. For example, the first answer may include information such as 'I am 63 years old', 'I am a woman', 'I currently have no job', 'I only graduated from high school', 'I live in Seoul', 'My family has 4 members', and 'I currently have no income'.

[0079] And, the AI ​​person generation unit (112) can provide the raw AI person with a first profile item related to biological characteristics and social characteristics, and a first question regarding a second profile item corresponding to psychological characteristics.

[0080] According to an embodiment, the first profile item may include information related to age, gender, occupation, education level, residential area, family composition, and income level, and the second profile item may include information related to openness, conscientiousness, extraversion, agreeableness, and neuroticism based on the Five Factor Theory (FFT) of personality.

[0081] For example, the AI ​​person generation unit (112) can provide the raw AI person with first questions in relation to the first profile item, such as 'What is your current age?', 'What is your gender?', 'What is your current job?', 'What is your highest level of education?', 'What are your family members?', 'What is your annual income range?'

[0082] For example, the AI ​​person generation unit (112) can provide the raw AI person with first questions in relation to the second profile item, such as 'Do you think of yourself as an open person?', 'Do you think of yourself as a sincere person?', 'Do you think of yourself as an extroverted person?', 'Do you think of yourself as a friendly person?', 'How do you handle stress or anxiety?'.

[0083] And, the AI ​​person generation unit (112) can generate a first preliminary AI person by training a first answer in response to a first question on a raw AI person.

[0084] The AI ​​person generation unit (112) can control the raw AI person to select a first answer in response to a first profile item. For example, the AI ​​person generation unit (112) can control the raw AI person to select a first answer, "I am 63 years old," in response to a first question, "What is your current age?"

[0085] According to an embodiment, the AI ​​person generation unit (112) can control the raw AI person to randomly select one of the previously generated answers in response to a second profile item. For example, the AI ​​person generation unit (112) can control the raw AI person to randomly select one of the following answers in response to a first question, "Do you consider yourself an open person?": "I think I am open-minded," "I am somewhat conservative," and "It depends on the situation."

[0086] In this way, the AI ​​person generation unit (112) can select the raw AI person that selected the first answer in response to the first question as the first preliminary AI person. And, the first answer selected by the raw AI person can be stored as the first characteristic information of the first preliminary AI person.

[0087] The AI ​​person generation unit (112) can provide a second question regarding a third profile item to the first preliminary AI person, and can generate a second preliminary AI person by training the first preliminary AI person with a second answer that maintains consistency with the first answer in response to the second question. Here, the third profile item is generated by expanding the frame defining human characteristics for each of the first profile item and the second profile item, and may include biological characteristics, social characteristics, psychological characteristics, and cultural characteristics.

[0088] For example, the AI ​​person generation unit (112) can expand the frame regarding biological characteristics to provide a second question regarding health status to the first preliminary AI person, such as ‘Do you know your current body mass index (BMI)? What is the index?’, ‘Have you been diagnosed with a chronic disease in a recent health checkup?’

[0089] At this time, the AI ​​person generation unit (112) can provide selectable answers to the second preliminary AI person in response to the second question. The selectable answers are answers generated in advance by the manager and may include answer information that is not awkward as a response to the second question.

[0090] For example, if the second question is 'Have you been diagnosed with a chronic disease in a recent health checkup?', the possible answers may include 'Yes, I was diagnosed with hyperlipidemia in a recent health checkup', 'No, I haven't had a health checkup recently', 'I was diagnosed with early-stage diabetes, but I don't know if it is a chronic disease', etc.

[0091] The AI ​​person generation unit (112) can control one of the selectable answers in response to the second question to be selected by the first preliminary AI person. At this time, the answer selected by the first preliminary AI person may conflict with the first characteristic information.

[0092] The Validation Checker can determine the consistency between the answer selected by the first preliminary AI person and the first characteristic information, i.e., the first answer. If there is a conflict and inconsistency, the AI ​​person generation unit (112) can induce the second preliminary AI person to re-select an answer.

[0093] The AI ​​person generation unit (112) can select a first preliminary AI person that selects a second answer that maintains consistency with the first answer in response to the second question as a second preliminary AI person, and can store the first answer and the second answer as second characteristic information of the second preliminary AI person.

[0094] The AI ​​person generation unit (112) can provide a third question regarding a fourth profile item to a second preliminary AI person, and can generate an AI person by training the second preliminary AI person with a third answer that maintains consistency with the first and second answers in response to the third question. Here, the fourth profile item is generated by expanding the frame defining human characteristics for the third profile item, and may include biological characteristics, social characteristics, psychological characteristics, and cultural characteristics.

[0095] The second preliminary AI person can generate a third answer in response to the third question through a pre-trained Large Language Model (LM). The second preliminary AI person can generate a third answer corresponding to the third question based on the first and second answers through the LLM. In other words, before generating an answer corresponding to the third question, the second preliminary AI person can generate a third answer that does not conflict with the first and second answers by referring to the first and second answers it selected itself.

[0096] The Validation Checker can determine the consistency between the answer generated by the LLM and the first and second answers. If the answer generated through the LLM conflicts with the first and second answers and is inconsistent, the AI ​​Person Generation Unit (112) can induce the regeneration of the answer of the third preliminary AI Person.

[0097] The AI ​​person generation unit (112) can select a second preliminary AI person that has generated a third answer that maintains consistency with the first and second answers in response to the third question as an AI person, and can store the first to third answers as characteristic information of the AI ​​person.

[0098] The characteristic information of an AI person, including the first to third answers, is a set of answers that the AI ​​person has selected or generated itself since its initial creation, and can represent the identity of the AI ​​person.

[0099] The AI ​​person generation unit (112) can set and store tags in the characteristic information of the AI ​​person. For example, the first AI person may include characteristic information such as 'female', 'resident of Seoul', '20s', 'open personality', 'prefers outdoor exercise', and for each profile, tags such as 'female', 'Seoul', '20s', 'open', and 'exercise' can be set.

[0100] The AI ​​Person Generation Unit (112) can continuously generate different AI Persons to generate indicators for people residing in specific countries, regions, etc., and the generated AI Persons can be stored for the generation of AI Hyper Personas. For example, the AI ​​Person Generation Unit (112) can generate and store 1 million AI Persons with different profiles applied, and these AI Persons can represent 1 million virtual people.

[0101] The AI ​​hyper persona generation unit (113) can generate an AI hyper persona by selecting an AI person that is suitable for the main consumer group among the AI ​​persons that have already been generated, and by synthesizing business characteristic information related to the business field to the selected AI person.

[0102] The AI ​​hyper persona generation unit (113) can determine a first profile corresponding to the main consumer group derived from the user data processing unit (111), and can select an AI person that corresponds to the first profile among the AI ​​persons.

[0103] At this time, the AI ​​hyper persona generation unit (113) can search for tags corresponding to the first profile in order to select an AI person that matches the first profile.

[0104] For example, if the main consumer base is 'men in their 20s or older who have been discharged from the military,' the AI ​​hyper persona generation unit (113) can select AI persons that include the corresponding tags using keywords such as 'military,' '20s,' and 'male.' The AI ​​hyper persona generation unit (113) can determine an AI person by considering past interview answers among the AI ​​persons selected through the tags. That is, the AI ​​hyper persona generation unit (113) can finally select an AI person that has performed answers consistent with 'men in their 20s or older who have been discharged from the military' by considering the answers given by the selected AI persons during the generation process.

[0105] And, the AI ​​hyper persona generation unit (113) can generate an AI hyper persona by conducting a preliminary interview with a selected AI person regarding a business field and having the selected AI person give thoughts on the business field.

[0106] Specifically, the AI ​​hyper-persona generation unit (113) can generate business questions using business characteristic information, and the AI ​​person can generate a fourth answer corresponding to the business question through LLM. That is, the selected AI person provides a self-generated answer to the business question of the AI ​​hyper-persona generation unit (113), and the fourth answer must maintain consistency with the answer from the past AI person generation process.

[0107] The validation checker can determine the consistency between the fourth answer and the characteristic information of the AI ​​person, and can select the AI ​​person exhibiting a consistent answer pattern as the AI ​​hyper persona. The AI ​​hyper persona generation unit (113) can provide guidance on past answers when the AI ​​person gives an answer that conflicts with past answers.

[0108] For example, the AI ​​hyper-persona generator (113) may perform a preliminary interview question, "Have you ever used men's skin lotion?" on an AI person selected as having a lot of interest in skin care, and the AI ​​person may provide a conflicting answer. In this case, the AI ​​hyper-persona generator (113) may point out that the AI ​​person previously answered that it had a lot of interest in skin care, and may subsequently provide a preset number of preliminary interview questions to monitor whether a consistent answer pattern is provided.

[0109] According to an embodiment, the AI ​​hyper persona generation unit (113) can remove an AI person that fails to generate a consistent fourth answer within a preset number of times from the AI ​​person data pool.

[0110] Through this process, an AI person who has successfully completed a preliminary interview can function as an AI hyper-persona.

[0111] The interview execution unit (114) operates a virtual interview room that displays questions entered by the user and answers provided by the AI ​​hyper persona in response to them, conducts an interview between the AI ​​hyper persona and the user in the interview room, and can save the conversation.

[0112] It may occur when the user has difficulty deciding what questions to ask the subject. To address this, the interview execution unit (114) can automatically generate various questions related to business fields and service products for the user and recommend them to the user.

[0113] At this time, the interview execution unit (114) can support various interview methods. For example, the interview execution unit (114) can perform NPS (Net Promoter Score) analysis, concept testing, problem identification, purchasing method investigation, etc., and can automatically generate appropriate questions for each methodology.

[0114] The interview execution unit (114) may receive a request for additional questions from the user during the interview. In this case, the interview execution unit (114) may analyze the interview context to generate questions that are natural to continue the conversation and may verify the consistency of the AI ​​hyper persona's answers in real time to maintain interview quality.

[0115] According to an embodiment, the interview execution unit (114) may involve the user as an observer and provide questions related to the business field directly to the AI ​​hyper persona on behalf of the user. The AI ​​hyper persona may provide answers in response to the questions from the interview execution unit (114), and the interview between the interview execution unit (114) and the AI ​​hyper persona may be displayed to the user through an interview room.

[0116] The analysis unit (115) can perform an in-depth analysis of the interview to derive final insights. In this process, the analysis unit (115) can further verify the consistency of the interview responses and evaluate the degree of alignment between the AI ​​hyper-persona characteristics and the business field.

[0117] The analysis unit (115) can provide feedback to the interview execution unit (114) to automatically generate re-questions for responses with low reliability and supplement them. The analysis unit (115) can correct the interview results by accepting feedback from the user and can generate a research report reflecting the user's feedback and provide it to the user terminal (200). The research report may include analysis results regarding market requirements and trends.

[0118] Additionally, the analysis unit (115) can automatically identify unmet needs regarding the user's business field based on the interview results and recommend a new AI hyper persona to resolve them. For example, if the business field can be broadly divided into Part A and Part B, and the first AI hyper persona mainly provides answers regarding Part A, the analysis unit (115) can recommend a second AI hyper persona that can provide answers regarding Part B. The user can continue the interview with the new AI hyper persona recommended by the analysis unit (115).

[0119] FIG. 4 is a diagram illustrating a method for generating an AI person according to an embodiment of the present invention.

[0120] Referring to FIG. 4, the AI ​​hyper persona generation device (100) can extract personal information related to a natural person from stored personal data to generate an AI person (S100). The personal information may include information related to the age, gender, occupation, education level, residential area, family composition, and income level related to the natural person.

[0121] The AI ​​hyper persona generation device (100) can generate a first answer for a raw AI person based on extracted person information (S110). The first answer may include information related to age, gender, occupation, education level, residential area, family composition, and income level generated based on the person information.

[0122] For example, the first answer may include information such as 'I am 63 years old', 'I am a woman', 'I currently have no job', 'I only graduated from high school', 'I live in Seoul', 'My family has 4 members', and 'I currently have no income'.

[0123] The AI ​​hyper persona generation device (100) can provide a first question to the raw AI person regarding a first profile item related to biological characteristics and social characteristics, and a second profile item corresponding to psychological characteristics (S120).

[0124] The AI ​​hyper persona generation device (100) can generate a first preliminary AI person by training a raw AI person with a first answer in response to a first question (S130). At this time, the AI ​​hyper persona generation device (100) can control the raw AI person to select a first answer in response to a first profile item, and the first answer selected by the raw AI person can be stored as the first characteristic information of the first preliminary AI person.

[0125] The AI ​​hyper persona generation device (100) may provide a second question regarding a third profile item to the first preliminary AI person (S140). The third profile item is generated by expanding a frame defining human characteristics for each of the first profile item and the second profile item, and the second question may include content regarding the third profile item.

[0126] The AI ​​hyper persona generation device (100) can generate a second preliminary AI person by training a second answer selected in response to a second question on a first preliminary AI person (S150). At this time, the second answer is an answer that maintains consistency with the first answer, and the first answer and the second answer can be stored as second characteristic information of the second preliminary AI person.

[0127] The AI ​​hyper persona generation device (100) can provide a third question regarding a fourth profile item to a second preliminary AI person (S160). The fourth profile item is generated by expanding a frame that defines human characteristics for the third profile item, and the third question may include content regarding the fourth profile item.

[0128] The AI ​​hyper persona generation device (100) can generate an AI person that self-generates a third answer that maintains consistency with the first and second answers in response to a third question (S170). At this time, the first to third answers can be stored as characteristic information of the AI ​​person.

[0129] FIG. 5 is a diagram illustrating an AI hyper-persona generation method according to an embodiment of the present invention.

[0130] Referring to FIG. 5, an AI hyper persona generation device (100) according to an embodiment of the present invention can receive business characteristic information from a user terminal (200) through a service platform (S100). The business characteristic information includes information related to a business field, and since there are no restrictions on the data format, it can be provided as a text file, image file, PDF file, URL link, shared document, etc.

[0131] And, the AI ​​hyper persona generation device (100) can determine the user's business field by analyzing business characteristic information and determine the main consumer base of the service product by generating an overview of the service product of the determined business field (S110).

[0132] And, the AI ​​hyper persona generation device (100) can select an AI person that corresponds to the determined main consumer group (S120). The AI ​​person is a virtual person to whom a pre-set profile is applied and individual characteristics corresponding to each profile are assigned. The profile applies social indicators such as gender, place of residence, education level, occupation, marital status, presence of children, and income level, as well as psychological indicators such as openness, conscientiousness, extraversion, agreeableness, and neuroticism. A

[0133] The AI ​​person is generated in advance by the AI ​​hyper persona generation device (100), and the AI ​​hyper persona generation device (100) can select an AI person that includes a profile suitable for the main consumer group among the generated AI persons.

[0134] And, the AI ​​hyper persona generation device (100) can generate an AI hyper persona by conducting a preliminary interview with a selected AI person (S130). Here, the preliminary interview is a process of generating an AI hyper persona by applying business characteristic information to the AI ​​person, and can be conducted by asking questions related to the business field and having the AI ​​person answer.

[0135] And, the AI ​​hyper persona generation device (100) can conduct an interview between the AI ​​hyper persona and the user in a virtual interview room (S140). The AI ​​hyper persona generation device (100) can operate various interview methods, such as automatic question generation, question recommendation, and observer mode, in response to the user's request.

[0136] Additionally, the AI ​​hyper persona generation device (100) can analyze an interview and provide the analysis results to a user terminal (200). The analysis results may be provided in the form of a research report containing analysis results on market requirements and trends, but are not limited thereto.

[0137] FIGS. 6 to 9 are drawings illustrating a method for automating user research using artificial intelligence technology according to an embodiment of the present invention.

[0138] Referring to FIGS. 6 through 9, a user can access a service platform operated by an AI hyper-persona generation device (100) through a user terminal (200). The user can input business characteristic information related to a business field into the service platform.

[0139] For example, a user can enter a short text, "I want to sell shoes," into the input field of a service platform.

[0140] A service platform can determine the business field by analyzing business characteristic information and identify the key consumer base for the service products by generating an overview of the service products within that field. For example, the service platform can determine the business field as a 'shoe sales platform' by analyzing user data and identify the key consumer base by generating an overview of the service products for the 'shoe sales platform'.

[0141] The service platform can generate AI hyper-personas that align with the main consumer base of the 'shoe sales platform'. For example, the service platform can generate AI hyper-personas for the main consumer base of the 'shoe sales platform', such as 'overseas direct purchase users', 'professionals who prefer limited editions', 'consumers who thoroughly compare prices', and 'hobby tool collectors'.

[0142] The service platform can conduct interviews with an AI hyper-persona selected by the user in a virtual interview room. Depending on the interview method selected by the user, it can receive questions directly from the user or automatically generate questions and answers in observer mode.

[0143] Although embodiments of the present invention have been described above with reference to the attached drawings, those skilled in the art will understand that the present invention may be implemented in other specific forms without altering its technical concept or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. Explanation of the symbols

[0144] 10: AI Hyper Persona Generation System 100: AI Hyper Persona Generator 110: Processor 111: Data processing unit 112: AI Person Generation Unit 113: AI Hyper Persona Generation Unit 114: Interview Execution Department 115: Analysis Department 120: Memory 130: Communication interface 140: Storage 150: Bus 200: User terminal

Claims

Claim 1 A method for generating an AI hyper persona, comprising: a step of extracting personal information related to a virtual natural person from previously stored personal data; a step of generating an AI person representing the natural person by learning the biological characteristics, social characteristics, psychological characteristics, and cultural characteristics of the natural person; and a step of generating an AI hyper persona by applying business characteristic information to the AI ​​person. Claim 2 In claim 1, the step of generating the AI ​​person comprises: generating a first answer based on the personal information; providing a first question to a raw AI person regarding a first profile item related to biological and social characteristics and a second profile item corresponding to psychological characteristics; training the raw AI person with the first answer in response to the first question to generate a first preliminary AI person; providing a second question regarding a third profile item to the first preliminary AI person; training the first preliminary AI person with a second answer that maintains consistency with the first answer in response to the second question to generate a second preliminary AI person; providing a third question regarding a fourth profile item to the second preliminary AI person; the second preliminary AI person self-generating a third answer that maintains consistency with the first and second answers through a pre-trained Large Language Model (LM) in response to the third question; and training the second preliminary AI person with the self-generated third answer to generate the AI ​​person. A method for generating an AI hyper persona, comprising the step of storing the first to third answers as characteristic information of the AI ​​person. Claim 3 A method for generating an AI hyper-persona according to paragraph 2, wherein the first profile item includes information related to age, gender, occupation, education level, residential area, family composition, and income level, and the second profile item includes information related to openness, conscientiousness, extraversion, agreeableness, and neuroticism based on the Five Factor Theory (FFT) of personality. Claim 4 A method for generating an AI hyper persona according to paragraph 2, wherein the third profile item is generated by expanding a frame defining human characteristics for each of the first profile item and the second profile item, and includes biological characteristics, social characteristics, psychological characteristics, and cultural characteristics, and the fourth profile item is generated by expanding a frame defining human characteristics for the third profile item, and includes biological characteristics, social characteristics, psychological characteristics, and cultural characteristics. Claim 5 In paragraph 2, the step of generating the first preliminary AI person comprises: providing the first answer to the raw AI person in response to the first question; selecting the raw AI person that selected the first answer as the first preliminary AI person; and storing the first answer as the first characteristic information of the first preliminary AI person. Claim 6 In claim 2, the step of generating the second preliminary AI person comprises: providing selectable answers corresponding to the second question to the second preliminary AI person; selecting the first preliminary AI person who selected the second answer among the answers as the second preliminary AI person; and storing the first answer and the second answer as second characteristic information of the second preliminary AI person. Claim 7 In claim 6, the step of selecting the second preliminary AI person comprises: a step in which one of the answers is randomly selected by the first preliminary AI person; a step in which a Validation Checker determines consistency between the selected answer and the first answer; and a step of repeating the answer selection of the first preliminary AI person if the first preliminary AI person selects an inconsistent answer. Claim 8 In paragraph 2, the step of self-generating the third answer comprises: the step of the second preliminary AI person generating an answer corresponding to the third question based on the first and second answers through the LLM; the step of a validation checker determining consistency between the generated answer and the first and second answers; and the step of repeating the generation of an answer through the LLM if the second preliminary AI person generates an inconsistent answer. Claim 9 A method for generating an AI hyperpersona according to claim 1, wherein the step of generating the AI ​​hyperpersona comprises: a step of generating a business question using the business characteristic information; a step of the AI ​​person generating a fourth answer corresponding to the business question through LLM; a step of determining consistency between the generated fourth answer and the characteristic information of the AI ​​person by a validation checker; and a step of selecting the AI ​​person exhibiting a consistent answer pattern as the AI ​​hyperpersona. Claim 10 In claim 9, the step of selecting the above AI hyper persona is an AI hyper persona generation method that removes an AI person that fails to generate a consistent fourth answer within a preset number of times.

Images