Method, program and apparatus for conducting surveys based on generative artificial intelligence
The generative artificial intelligence-based survey method addresses inefficiencies in conventional survey methods by generating human-like response data, ensuring reliability and accuracy through a pre-trained model, thereby reducing costs and time while providing consistent and reliable survey insights.
Patent Information
- Application Number
- KR1020230144076
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2023-09-12
- Filing Date
- 2023-10-25
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2043-10-25
AI Technical Summary
Conventional survey methods, both traditional and online, face challenges in efficiently recruiting a diverse group of respondents, ensuring response reliability, and analyzing responses considering diverse personas, leading to distorted results and high costs.
A survey method utilizing generative artificial intelligence generates human-like response data based on survey and persona information through a pre-trained language model, incorporating a hierarchical classification system and dynamic retraining, enabling accurate and efficient analysis of response patterns, reliability, and relevance.
This approach reduces time and costs, provides consistent and reliable survey results, and allows for rapid analysis and response to survey trends, enhancing the effectiveness and reliability of survey data.
Smart Images

Figure 112023117525987-PAT00002_ABST
Abstract
Description
Technology Field
[0001] The present disclosure relates to a survey method based on generative artificial intelligence, and more specifically to a method and apparatus capable of providing response data based on survey information and persona information regarding survey respondents by utilizing a pre-trained artificial intelligence-based model. Background Technology
[0003] Generally, survey methods include traditional methods in which an investigator selects random subjects, calls them for interviews, or meets them in person to ask questions and compiles the results based on their responses, as well as online survey methods that utilize survey systems connected via networks such as the Internet and multiple user computers.
[0004] In the online survey method, the survey system generates a questionnaire containing survey items in the form of an email or similar, based on a request from the survey client, and transmits it to panelists or respondents. The survey responses provided by the respondents are received by the survey system, which analyzes the results to generate a final survey report and provides it to the survey client.
[0005] Conventional survey methods require significant time and cost to recruit a large number of respondents, and the process of distributing questionnaires and collecting responses is not only time-consuming but also cumbersome.
[0006] Traditional survey methods that obtain responses through telephone or in-person interviews require significantly more effort and cost compared to online survey methods. Consequently, survey clients or researchers with limited budgets or time constraints face many difficulties in obtaining the desired survey results.
[0007] Online surveys have the advantage of being able to be conducted quickly and inexpensively compared to traditional offline or telephone surveys, and are widely used to identify consumer preferences for various products and services or for other marketing purposes.
[0008] However, online surveys currently lack clear measures to ensure response reliability other than securing personal information and continuously updating it, and there is a problem in that it is difficult to generate or analyze responses while considering the diverse personas of survey respondents.
[0009] Most online surveys fail to adequately consider the diversity of respondents, such as gender, age, occupation, and cultural background, and if the number of respondents is insufficient, the results may be distorted by the subjectivity of individual respondents. Consequently, there is a problem in that it is impossible to secure high-quality data for reliable and meaningful analysis of the survey results. Prior art literature
[0011] Republic of Korea Registered Patent No. 10-2288995 (2021.08.05) The problem to be solved
[0012] The present disclosure is devised in response to the aforementioned background technology and aims to provide a method and apparatus based on generative artificial intelligence that can generate human-like response data based on survey information and persona information regarding survey respondents in an agent based on an artificial intelligence-based model, and derive meaningful information by analyzing the patterns, reliability, and relevance of the response data through various analysis techniques.
[0013] However, the problems to be solved in this disclosure are not limited to those mentioned above, and other unmentioned problems may be clearly understood based on the description below. means of solving the problem
[0015] A survey method based on generative artificial intelligence is disclosed, which is performed by a computing device comprising at least one processor according to an embodiment of the present disclosure for realizing the aforementioned tasks. The method comprises: a step of obtaining survey information and persona information regarding a survey respondent; and a step of generating response data based on the survey information and persona information regarding the survey respondent using a pre-trained first language model.
[0016] Alternatively, the step of acquiring the survey information and persona information regarding the survey respondents comprises: a step of extracting detailed characteristic information regarding the survey respondents based on basic information regarding the survey respondents and question information related to the basic information, using a pre-trained second language model; a step of selecting key characteristic information to be used for acquiring the persona information from among the detailed characteristic information generated by the second language model based on user input; and a step of generating the persona information based on the key characteristic information and additional question information related to the key characteristic information, using the second language model.
[0017] Alternatively, if the above-mentioned key characteristic information is selected in multiple quantities, the above-mentioned persona information is generated by combining the multiple key characteristic information.
[0018] Alternatively, the second language model is dynamically retrained based on user feedback regarding the generated persona information or performance indicators of the second language model.
[0019] Alternatively, the above persona information is based on a hierarchical classification system according to the characteristics of the survey respondents.
[0020] Alternatively, the above classification system comprises: a first classification system regarding basic characteristics including demographic information, occupation and professional information, and education level information; a second classification system regarding lifestyle including hobbies and interests, consumption habits, and health and physical information; a third classification system regarding values and psychological characteristics including values, personality, decision-making style, and communication style; a fourth classification system regarding technology and media usage habits including technology consumption preferences and media consumption patterns; and a fifth classification system regarding social networks and relationships including social relationships and social networking habits.
[0021] Alternatively, the response data generated through the first language model is used to adjust the language distribution of the first language model or to retrain the first language model based on user feedback.
[0022] Alternatively, if there are multiple persona informations, the response data generated through the first language model includes response information and statistical information generated for each of the multiple persona informations.
[0023] Alternatively, the method further comprises the step of analyzing at least one of the pattern, reliability, or relevance of the response data using a decision-making technique.
[0024] Meanwhile, a computing device for implementing a survey method based on generative artificial intelligence according to one embodiment of the present disclosure comprises: a processor including at least one core; and a memory including program codes executable on the processor; wherein the processor acquires survey information and persona information about a survey respondent according to the execution of the program codes, and generates response data based on the survey information and persona information about the survey respondent using a pre-trained first language model. Effects of the invention
[0026] The present disclosure aims to provide a method and apparatus that can generate human-like response data based on survey information and respondent persona information using a pre-trained artificial intelligence-based model, and derive meaningful information by analyzing the patterns, reliability, and relevance of the response data through various analysis techniques, thereby enabling faster and more consistent survey results, significantly reducing time and costs compared to existing survey methods, quickly identifying respondents' opinions or trend changes, enabling rapid analysis and response to survey results, and allowing conversational interaction methods utilizing agents to become commonplace in various fields. Brief explanation of the drawing
[0028] FIG. 1 is a block diagram of a computing device according to one embodiment of the present disclosure. FIG. 2 is a flowchart for explaining a survey method based on generative artificial intelligence according to one embodiment of the present disclosure. FIG. 3 is a flowchart showing in detail the steps for obtaining persona information according to one embodiment of the present invention. Specific details for implementing the invention
[0029] Embodiments of the present disclosure are described below with reference to the attached drawings so that those skilled in the art (hereinafter, those skilled in the art) can easily implement them. The embodiments presented in the present disclosure are provided to enable those skilled in the art to use or implement the contents of the present disclosure. Accordingly, various modifications to the embodiments of the present disclosure will be apparent to those skilled in the art. That is, the present disclosure may be embodied in various different forms and is not limited to the embodiments below.
[0030] Throughout the specification of the present disclosure, identical or similar reference numerals refer to identical or similar components. Additionally, to clearly explain the present disclosure, reference numerals in the drawings that are unrelated to the description of the present disclosure may be omitted.
[0031] The term “or” as used in this disclosure is intended to mean an implicit “or” rather than an exclusive “or.” That is, unless otherwise specified in this disclosure or its meaning is not clear from the context, “X uses A or B” should be understood to mean one of the natural implicit substitutions. For example, unless otherwise specified in this disclosure or its meaning is not clear from the context, “X uses A or B” may be interpreted as X using A, X using B, or X using both A and B.
[0032] The term “and / or” as used in this disclosure should be understood to refer to and include all possible combinations of one or more of the enumerated related concepts.
[0033] The terms “comprising” and / or “comprising” as used in this disclosure should be understood to mean the presence of certain features and / or components. However, the terms “comprising” and / or “comprising” should be understood not to exclude the presence or addition of one or more other features, other components and / or combinations thereof.
[0034] Where not otherwise specified in the present disclosure or where it is not clear from the context that the singular form indicates, the singular should generally be interpreted as including “one or more.”
[0035] The term “the N (N is a natural number)” used in this disclosure may be understood as an expression used to distinguish the components of this disclosure from one another according to certain criteria, such as functional perspectives, structural perspectives, or convenience of explanation. For example, components performing different functional roles in this disclosure may be distinguished as a first component or a second component. However, components that are substantially identical within the technical scope of this disclosure but must be distinguished for the convenience of explanation may also be distinguished as a first component or a second component.
[0036] The term “connection” as used in the present disclosure should be interpreted to include not only cases where the components are “directly connected,” but also cases where other components are “present” in between, and cases where they are “electrically connected” with other components in between.
[0037] The term “acquisition” as used in this disclosure may be understood to refer not only to receiving data through a wireless communication network with an external device or system, but also to generating or receiving data in an on-device form.
[0038] Meanwhile, the terms "module" or "unit" used in this disclosure may be understood as referring to an independent functional unit that processes computing resources, such as a computer-related entity, firmware, software or a part thereof, hardware or a part thereof, or a combination of software and hardware. In this case, "module" or "unit" may be a unit composed of a single element, or a unit expressed as a combination or set of multiple elements. For example, in a narrow sense, "module" or "unit" may refer to a hardware element of a computing device or a set thereof, an application program that performs a specific function of software, a procedure implemented through software execution, or a set of instructions for program execution. Furthermore, in a broad sense, "module" or "unit" may refer to the computing device itself that constitutes the system, or an application executed on the computing device. However, since the above-described concept is merely an example, the concept of "module" or "part" may be defined in various ways within the scope understandable to those skilled in the art based on the contents of this disclosure.
[0039] As used in this disclosure, the term "model" may be understood as a system implemented using mathematical concepts and language to solve a specific problem, a set of software units to solve a specific problem, or an abstract model regarding a processing process to solve a specific problem. For example, a deep learning "model" may refer to an overall system implemented as a neural network that possesses problem-solving capabilities through learning. In this case, the neural network may possess problem-solving capabilities by optimizing parameters connecting nodes or neurons through learning. A deep learning "model" may include a single neural network or a set of neural networks composed of multiple neural networks.
[0040] The explanation of the foregoing terms is intended to aid in understanding the present disclosure. Accordingly, it should be noted that unless a foregoing term is explicitly stated as a matter limiting the content of the present disclosure, it is not to be used in the sense of limiting the technical concept of the content of the present disclosure.
[0041] FIG. 1 is a block diagram of a computing device according to one embodiment of the present disclosure.
[0042] A computing device (100) according to one embodiment of the present disclosure may be a hardware device or a part of a hardware device that performs comprehensive processing and computation of data, or it may be a software-based computing environment connected to a communication network. For example, the computing device (100) may be a server that performs intensive data processing functions and shares resources through communication with clients. Additionally, the computing device (100) may be a cloud system that enables multiple servers and clients to process data comprehensively. Since the above description is merely one example regarding the type of computing device (100), the type of computing device (100) may be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.
[0043] Referring to FIG. 1, a computing device (100) according to one embodiment of the present disclosure may include a processor (110), a memory (120), and a network unit (130). However, since FIG. 1 is merely an example, the computing device (100) may include other configurations for implementing a computing environment. Additionally, only some of the disclosed configurations may be included in the computing device (100).
[0044] A processor (110) according to one embodiment of the present disclosure may be understood as a constituent unit comprising hardware and / or software for performing computing operations. For example, the processor (110) may process instructions generated as a result of user interaction through a user interface. Additionally, the processor (110) may read a computer program and perform data processing for machine learning. The processor (110) may process computational processes such as processing input data for machine learning, feature extraction for machine learning, and error calculation based on backpropagation. A processor (110) for performing such data processing and operations may include a central processing unit (CPU), a general purpose graphics processing unit (GPGPU), a tensor processing unit (TPU), an application specific integrated circuit (ASIC), or a field programmable gate array (FPGA). Since the above-described type of processor (110) is merely an example, the type of processor (110) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.
[0045] The processor (110) can generate an AI-based agent that responds to a survey using a pre-trained first language model, and can generate response data based on survey information and persona information about the survey respondent using the agent.
[0046] In this case, the AI-based agent is a dedicated program for providing AI-based services (e.g., voice recognition service, assistant service, translation service, search service, survey service, etc.) and can be executed by an existing general-purpose processor or a separate AI-dedicated processor (e.g., GPU, etc.). In particular, the agent can control various modules described below.
[0047] The processor (110) can generate response data based on survey information and persona information using a pre-trained first language model. Additionally, the processor (110) can extract features necessary for generating persona information based on a training dataset and train a second language model that generates persona information based on the extracted features. Then, the processor (110) can generate persona information using the trained second language model.
[0048] For example, the first language model and the second language model may be Generative AI models, specifically Large Language Models (LLM). Accordingly, the processor (110) can generate an agent that responds to a survey based on persona information by using a Generative AI model based on an LLM.
[0049] The processor (110) can extract characteristics to be used for acquiring persona information using a second language model generated through the aforementioned learning process, and can estimate persona information by dynamically performing persona profiling based on the extracted characteristics. The processor (110) can input survey information and persona information into a first language model learned through the aforementioned process, and generate response data representing the estimated result by reflecting the persona information for the survey questions and the knowledge information at the time of the survey.
[0050] In addition to the examples described above, the types of training datasets according to language distribution and the outputs of the first language model and the second language model may be configured in various ways within a range understandable to those skilled in the art based on the contents of this disclosure.
[0051] A memory (120) according to one embodiment of the present disclosure may be understood as a configuration unit comprising hardware and / or software for storing and managing data processed by a computing device (100). That is, the memory (120) may store data of any form generated or determined by a processor (110) and data of any form received by a network unit (130). For example, the memory (120) may include at least one type of storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory, RAM (random access memory), SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read-only memory), PROM (programmable read-only memory), magnetic memory, a magnetic disk, or an optical disk. Additionally, the memory (120) may include a database system that controls and manages data in a predetermined system. Since the above-described type of memory (120) is merely an example, the type of memory (120) can be configured in various ways within a range understandable to those skilled in the art based on the contents of the present disclosure.
[0052] Memory (120) can manage data, combinations of data, and program code executable by the processor (110) by structuring and organizing them for the processor (110) to perform operations. For example, memory (120) can store program code that enables the processor (110) to process data, program code that enables the processor (110) to process commands based on user input through a user interface, and various data generated as the program code is executed.
[0053] A network unit (130) according to one embodiment of the present disclosure may be understood as a configuration unit that transmits and receives data through any known form of wired or wireless communication system. For example, the network unit (130) may perform data transmission and reception using wired or wireless communication systems such as a local area network (LAN), wideband code division multiple access (WCDMA), long term evolution (LTE), wireless broadband internet (WiBro), 5th generation mobile communication (5G), ultra-wide-band wireless communication, ZigBee, radio frequency (RF) communication, wireless LAN, wireless fidelity (Wi-Fi), near field communication (NFC), or Bluetooth. Since the communication systems described above are merely examples, wired or wireless communication systems for data transmission and reception of the network unit (130) may be applied in various ways other than those described above.
[0054] The network unit (130) can receive data necessary for the processor (110) to perform calculations through wired or wireless communication with any system or any client. Additionally, the network unit (130) can transmit data generated through the processor (110)'s calculations through wired or wireless communication with any system or any client. The network unit (130) can transmit and receive data generated through interaction with any entity through communication with the entity. For example, when user input regarding survey information, etc. is received by a user terminal through a user interface, the network unit (130) can obtain the user input through communication with the user terminal. Then, the network unit (130) can transmit data processed by the processor (110) to the user terminal.
[0055] FIG. 2 is a flowchart for explaining a survey method based on generative artificial intelligence according to one embodiment of the present disclosure.
[0056] Referring to FIG. 2, the computing device (100) obtains survey information and persona information about the survey respondents (S100).
[0057] A computing device (100) generates response data based on survey information and persona information about survey respondents using a pre-trained first language model (S200). The response data generated through the first language model can be used to adjust the language distribution of the first language model or to retrain the first language model based on user feedback.
[0058] The agent can collect survey information, such as the survey topic, purpose, and target audience, based on the surveyor's input, and utilize this information to understand the context when generating response data. Since this agent is generated based on a pre-trained first language model, it can use persona information as a prompt or initial value for the first language model. Additionally, when multiple persona information is reflected in a single survey, the agent can generate different response data for each persona. In this case, the response data generated by the agent includes response and statistical information generated for each persona, allowing for different questions and responses based on each persona to respond to various scenarios or situations set by the surveyor for each type.
[0059] At this time, since the agent has been pre-learned with vast knowledge at the time of the survey, it can generate response data for the survey information by reflecting the knowledge at the time of the survey in real time. To this end, the computing device (100) can perform the process of learning a new learning dataset for the first language model by utilizing a pre-learned second language model. Therefore, since the first language model is learned using a learning dataset consisting of a vast amount of text data, it can possess information on various contexts and topics. Through this learning process, the agent based on the first language model can generate appropriate response data for various questions. In addition, the first language model has the ability to analyze and understand the question context of the pre-set survey information, so the agent can generate response data that matches the intent or purpose of the question.
[0060] In this way, an agent can be created based on a first language model linked to a second language model that dynamically generates personas. The agent created through this pre-trained first language model acquires language understanding and generation capabilities, enabling it to generate response data.
[0061] Meanwhile, persona information regarding the survey respondents targeted by the agent can be set by the survey terminal conducting the survey. This set persona information can be utilized by the agent to generate response data by considering various characteristics such as gender, age, occupation, residential area, nationality, hobbies, and family.
[0062] For example, if the configured persona information includes a specific overseas country or place name, age, or occupation, knowledge information at the time of the survey is collected based on that country, place name, age, or occupation. By retraining the first language model based on this collected knowledge information, the agent can develop a response style tailored to the configured persona information (e.g., gender, age, occupation, nationality, etc.). Through this process, the agent can generate response data that better reflects the reactions or tendencies of the specific target group. Furthermore, because the agent generates response data based on given conditions and information—namely, knowledge information at the time of the survey, survey data, and configured persona information—it can provide consistent response data without human subjectivity or bias. Additionally, as the first and second language models underlying the agent are continuously retrained using new training datasets, response data reflecting the latest knowledge information and trends can be generated.
[0063] The computing device (100) can apply feedback technology to the generated response data to adjust or improve the agent's response data in real time according to the response of the user or surveyor.
[0064] The computing device (100) can analyze at least one of the pattern, reliability, or relevance of the response data using a decision-making technique (S300). At this time, any one of KANO, TOPSIS, AHP, Conjoint, QFD (Quality Function Deployment), and AB testing may be used as the decision-making technique.
[0065] The computing device (100) can provide reliable information, such as response characteristics and distribution by persona information, as a result of analyzing response data. Accordingly, the computing device (100) can store response data and feedback information generated by the agent in a database and utilize them for future analysis and verification. The computing device (100) derives meaningful insights by applying various analysis techniques based on the response data stored in the database. Through this analysis process, the value of the response data generated by the agent can be maximized, and information that can be utilized for decision-making can be derived.
[0066] FIG. 3 is a flowchart showing in detail the steps for obtaining persona information according to one embodiment of the present invention.
[0067] A step of obtaining persona information (S100) according to one embodiment of the present invention may specifically include steps such as extracting characteristics necessary for generating persona information and dynamic persona profiling.
[0068] The computing device (100) constructs a training dataset through a data collection and preprocessing process for text data, such as knowledge information, questions, and response data at the time of the survey, based on user basic information and a second language model (S110). At this time, the data collection and preprocessing process performs the processes of text data cleaning, tokenization, and vectorization.
[0069] A computing device (100) uses a pre-trained second language model to extract detailed characteristic information about a survey respondent based on basic information about the survey respondent and question information related to the basic information (S120).
[0070] For example, when a computing device (100) inputs question information related to basic user information, such as "What hobbies is the user likely to have?", into a second language model, which is a large language model (LLM), it can receive response data in text form for the question from the second language model, and this response data can be used as lifestyle information among the detailed characteristic information of the persona information.
[0071] The computing device (100) selects key characteristic information to be used for obtaining persona information from among the detailed characteristic information generated by the second language model based on user input (S130). For example, the key characteristic information may include information such as hobbies, values, and communication style.
[0072] The computing device (100) can generate persona information through dynamic persona profiling based on main characteristic information and additional question information related to the main characteristic information using a second language model (S140). For example, a question such as "What characteristics will the communication style of the user have?" is input into the second language model as additional question information related to the main characteristic information, and the persona information can be configured in more detail based on the prediction result of the second language model, i.e., the response data.
[0073] Additionally, the computing device (100) can generate persona information by combining multiple key characteristic information when multiple key characteristic information is selected. For example, two key characteristic information, such as ‘an active college student’ and ‘an introvert who likes books,’ can be combined to generate new persona information such as ‘a college student who likes books and is active.’
[0074] Meanwhile, the computing device (100) can dynamically retrain the second language model based on user feedback on the generated persona information or performance indicators of the second language model, thereby increasing the performance of the second language model, i.e., the accuracy of persona generation.
[0075] Additionally, the computing device (100) can increase the performance of the second language model, i.e., the accuracy of persona generation, by generating automatic feedback information based on the performance of the agent (e.g., response accuracy) and providing the feedback information from the agent to the training dataset, thereby retraining the second language model using the newly constructed training dataset.
[0076] In this way, the computing device (100) can continuously monitor the performance of the second language model required for persona generation by utilizing the performance evaluation indicator of the agent, and improve the accuracy of persona generation through the retraining process of the second language model when necessary. And, as a result, the performance of the first language model can also be improved based on the second language model.
[0077] Meanwhile, persona information may include various types of information, ranging from basic characteristic information such as age, gender, occupation, hobbies, personality, and consumption patterns to deep-level information such as the user's technology preferences, decision-making processes, and values. Accordingly, the computing device (100) can generate persona information based on a hierarchical classification system according to the characteristics of the survey respondents. This classification system may include first to fifth classification systems.
[0078] The first classification system relates to basic characteristics including demographic information, occupational and professional information, and education level information, and may include demographic information such as age, gender, nationality, and place of residence; occupational and professional information such as occupational group, professional field, length of career, and work environment; and education level information such as highest level of education, major, additional education, and certifications.
[0079] The second classification system relates to a lifestyle that includes hobbies and interests, consumption habits, and health and physical information. It may include information on hobbies and interests such as art, sports, travel, music, and books; information on consumption habits such as consumption preferences, major purchased items, purchase frequency, and choice of purchasing channels (online vs. offline); and information on health and physical information such as physical characteristics, health status, dietary habits, and exercise habits. In particular, information on hobbies and interests may be subdivided by category.
[0080] The third classification system relates to values and psychological characteristics, including values, personality, decision-making styles, and communication styles, and may include value information regarding the perception of values concerning the environment, society, politics, economy, ethics, etc.; personality information regarding personality types such as MBTI or the Big Five; decision-making style information such as intuitive vs. analytical or open vs. conservative; and communication style information such as passive, active, aggressive, and friendly.
[0081] The fourth classification system relates to technology and media usage habits, including technology consumption preferences and media consumption patterns, and may include information on technology consumption preferences such as smartphone usage frequency, degree of computer and internet usage, and preference for SNS platforms, as well as information on media consumption patterns such as major media channels watched, preferred content genres, and media consumption time.
[0082] The fifth classification system relates to social networks and relationships, including social relationships and social networking habits, and may include information on social relationships such as the depth and frequency of relationships with family, friends, and colleagues, and information on social networking habits such as the frequency of online and offline network activities and major network platforms.
[0083] Meanwhile, the computing device (100) can collect user feedback information regarding the generated persona information, and by providing the persona information and user feedback information to the training dataset, the performance of the second language model, i.e., the accuracy of persona generation, can be improved by retraining the second language model using the newly constructed training dataset.
[0084] Additionally, the computing device (100) can increase the performance of the second language model, i.e., the accuracy of persona generation, by generating automatic feedback information based on the performance of the agent (e.g., response accuracy) and providing the feedback information from the agent to the training dataset, thereby retraining the second language model using the newly constructed training dataset.
[0085] In this way, the computing device (100) can continuously monitor the performance of the second language model required for persona generation by utilizing the performance evaluation indicator of the agent, and improve the accuracy of persona generation through the retraining process of the second language model when necessary, and as a result, the performance of the first language model can also be improved based on the second language model.
[0087] As such, the present disclosure allows for the generation of personalized response data by reflecting the persona information of survey respondents in an agent using an artificial intelligence-based model, thereby providing response data that is more aligned with the interests and characteristics of survey respondents, and thus enhancing the effectiveness and reliability of the response data.
[0088] In addition, the present disclosure collects the latest knowledge information at the time of the survey in the form of text data through crawling, preprocesses the collected text data to construct a training dataset, and then performs a retraining process for a second language model using the constructed training dataset. Since an agent created using a first language model linked with the second language model can generate response data by reflecting the knowledge information at the time of the survey in real time, the freshness and accuracy of the response data can be ensured.
[0089] In the present disclosure, response data generated by an agent is stored in a database, and the response data stored in the database can be efficiently analyzed using various analysis techniques. Accordingly, the present disclosure can derive insights related to persona information regarding survey respondents more precisely and provide information useful for decision-making.
[0090] Above all, compared to survey operations that involve asking respondents questions via telephone or online and waiting for responses to those questions, the present disclosure enables an agent to automatically generate response data by combining survey information and persona information, thereby enabling faster and more consistent survey results, saving time and costs, and allowing for the rapid identification of respondents' opinions or changes in trends, as well as rapid analysis and response to survey results.
[0091] Accordingly, a survey method based on generative artificial intelligence according to one embodiment of the present disclosure can be applied to various fields such as market research and marketing, medical fields, education fields, entertainment fields, and policy decisions of companies or information.
[0093] The various embodiments of the present disclosure described above may be combined with additional embodiments and modified to the extent understandable to those skilled in the art in light of the detailed description above. The embodiments of the present disclosure are illustrative in all respects and should be understood as not restrictive. For example, each component described as a single unit may be implemented in a distributed manner, and components described as distributed may likewise be implemented in a combined form. Accordingly, all modifications or variations derived from the meaning, scope, and equivalents of the claims of the present disclosure should be interpreted as being included within the scope of the present disclosure.
Claims
Claim 1 A survey method based on generative artificial intelligence, performed by a computing device comprising at least one processor, comprising: a step of acquiring survey information and persona information about a survey respondent; and a step of generating response data based on the survey information and persona information about the survey respondent using a pre-trained first language model, wherein the persona information is based on a classification system hierarchically classified by characteristics of the survey respondent, and the classification system comprises: a first classification system regarding basic characteristics including demographic information, occupation and expertise information, and education level information; a second classification system regarding lifestyle including hobbies and interests, consumption habits, health and physical information; a third classification system regarding values and psychological characteristics including values, personality, decision-making style, and communication style; a fourth classification system regarding technology and media usage habits including technology consumption preferences and media consumption patterns; and a fifth classification system regarding social networks and relationships including social relationships and social networking habits. Claim 2 A method according to claim 1, wherein the step of acquiring the survey information and persona information for the survey respondent comprises: a step of extracting detailed characteristic information for the survey respondent based on basic information for the survey respondent and question information related to the basic information using a pre-trained second language model; a step of selecting key characteristic information to be used for acquiring the persona information from among the detailed characteristic information generated by the second language model based on user input; and a step of generating the persona information based on the key characteristic information and additional question information related to the key characteristic information using the second language model. Claim 3 In paragraph 2, when the above-mentioned major characteristic information is selected in multiple quantities, the persona information is generated by combining the multiple major characteristic information. Claim 4 In paragraph 2, the method wherein the second language model dynamically performs retraining based on user feedback regarding the generated persona information or performance indicators of the second language model. Claim 5 delete Claim 6 delete Claim 7 A method according to claim 1, wherein the response data generated through the first language model is used to adjust the language distribution of the first language model or to retrain the first language model based on user feedback. Claim 8 A method according to claim 1, wherein, when the persona information is multiple, the response data generated through the first language model includes response information and statistical information generated for each of the multiple persona information. Claim 9 A method according to claim 1, further comprising the step of analyzing at least one of the pattern, reliability, or relevance of the response data using a decision-making technique. Claim 10 A computing device for implementing a survey method based on generative artificial intelligence comprises: a processor including at least one core; and a memory including program codes executable on the processor; wherein the processor acquires survey information and persona information regarding a survey respondent according to the execution of the program code, and generates response data based on the survey information and persona information regarding the survey respondent using a pre-trained first language model, wherein the persona information is based on a hierarchical classification system according to the characteristics of the survey respondent, and wherein the classification system comprises: a first classification system regarding basic characteristics including demographic information, occupation and expertise information, and education level information; a second classification system regarding lifestyle including hobbies and interests, consumption habits, and health and physical information; a third classification system regarding values and psychological characteristics including values, personality, decision-making style, and communication style; and a fourth classification system regarding technology and media usage habits including technology consumption preferences and media consumption patterns. A device comprising a fifth classification system regarding social networks and relationships, including social relationships and social networking habits.
Citation Information
Patent Citations
Automatic disposition / preprocessing system of disposition data for persoanl disposition analysis, artificial intelligence disposition analysis system, and method thereof
KR1020230091570A
Personal disposition big data collection / analysis service and artificial intelligence-based disposition data processing system
KR1020230091568A
