Method and device for estimating personality of text using artificial intelligence

Abstract

A method according to an embodiment of the present invention may include: a step of generating a first agent induced to a first personality trait based on a first prompt; a step of generating a second agent induced to a second personality trait opposite to the first personality trait based on a second prompt; a step of generating a psycholinguistic description of a target text through each of the first agent and the second agent; and a step of estimating a personality trait of the author of the target text through a third agent based on the psycholinguistic descriptions generated through each of the first agent and the second agent.

Description

Technology Field

[0001] The present disclosure describes a method and apparatus for estimating text characteristics using artificial intelligence. Background Technology

[0003] Since individual personality manifests in various aspects, outcomes in daily life and work can vary significantly depending on one's personality. Consequently, research is continuously being conducted to leverage these personalities to personalize various services tailored to individual characteristics.

[0004] In particular, estimating an individual's personality based on user-generated text, rather than identifying personality through direct personality tests, requires much more complex and difficult calculations because it does not represent direct personality.

[0005] Therefore, much research is being conducted to perform complex and difficult calculations using artificial intelligence models to estimate individual personalities based on text.

[0006] However, while a large dataset of labels is important for estimating personality based on text using an artificial intelligence model, there is a problem that it is costly because a significant amount of labels are required.

[0007] Under these circumstances, the development of large language models has improved the performance of natural language processing tasks, and the emergence of agent-based approaches has further enhanced computational processing capabilities more effectively. An agent refers to a system or interface designed around a language model to perform specific tasks or goals. Agents can play the role of executing actions or generating responses based on user prompt input.

[0008] However, language models have the problem of potentially exhibiting cognitive biases, which limits the ability to acquire diverse perspectives by relying solely on agent-based approaches.

[0009] Therefore, this paper proposes a framework for reliably detecting individual personalities by utilizing various agents. The problem to be solved

[0011] The present disclosure aims to provide a method and apparatus for estimating personality based on text using an agent.

[0012] The present disclosure aims to provide a method and apparatus for improving text-based personality estimation performance using artificial intelligence. means of solving the problem

[0014] A method according to various embodiments of the present disclosure may include: generating a first agent induced by a first personality trait; generating a second agent induced by a second personality trait opposite to the first personality trait; generating a psycholinguistic description of a target text through each of the first agent and the second agent; and estimating a personality trait of the author of the target text through a third agent based on the psycholinguistic descriptions generated through each of the first agent and the second agent.

[0015] In one embodiment, the first personality trait and the second personality trait are determined from the same personality type, and the personality type may include openness, conscientiousness, extraversion, agreeableness, and neuroticism.

[0016] In one embodiment, the first agent, the second agent, and the third agent may be generated in a large language model.

[0017] In one embodiment, the step of generating a psycholinguistic description of a target text through each of the first agent and the second agent may include: generating a first description of the target text in emotional, cognitive, and social aspects based on the first personality trait in the first agent; and generating a second description of the target text in emotional, cognitive, and social aspects based on the second personality trait in the second agent.

[0018] In one embodiment, the step of estimating personality traits of the author of the target text may include: determining similarity by comparing points of agreement and points of disagreement based on the target text, the first description, and the second description in the third agent; and estimating personality traits of the author of the target text based on the similarity.

[0019] A personality estimation device according to various embodiments of the present disclosure may be configured to include: a memory; a modem; and a processor connected to the modem and the memory, wherein the processor generates a first agent induced to a first personality trait based on a first prompt, generates a second agent induced to a second personality trait opposite to the first personality trait based on a second prompt, generates a psycholinguistic description of a target text through each of the first agent and the second agent, and estimates a personality trait of the author of the target text through a third agent based on the psycholinguistic descriptions generated through each of the first agent and the second agent. Effects of the invention

[0021] According to one embodiment of the present disclosure, the performance of a model for estimating an individual's personality based on text can be improved.

[0022] According to one embodiment of the present disclosure, efficient performance can be achieved in training a personality estimation model while consuming little cost. Brief explanation of the drawing

[0024] A brief description of each drawing is provided to help to better understand the drawings cited in the detailed description of the present disclosure. FIG. 1 is a conceptual diagram illustrating the basic principles of artificial intelligence technology according to one embodiment of the present disclosure. FIG. 2 is a diagram illustrating a process for a personality induction step performed in a language model according to one embodiment of the present disclosure. FIG. 3 is a diagram illustrating a process for a psycholinguistic explanation step performed in a language model according to one embodiment of the present disclosure. FIG. 4 is a diagram illustrating a process for a comparison evaluation step performed in a language model according to one embodiment of the present disclosure. FIG. 5 is an example showing a pseudocode for a personality estimation method according to one embodiment of the present disclosure. Figure 6 shows the result of performing personality estimation in a language model according to the present disclosure. FIG. 7 is a block diagram of an electronic device to which an artificial intelligence algorithm model according to one embodiment of the present disclosure is applied. FIG. 8 is a flowchart illustrating a method for performing character estimation according to one embodiment of the present disclosure. Specific details for implementing the invention

[0025] The technical concept of the present disclosure is subject to various modifications and may have various embodiments. Specific embodiments are illustrated in the drawings and described in detail through the detailed description. However, this is not intended to limit the technical concept of the present invention to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the scope of the technical concept of the present invention.

[0026] In describing the technical concept of the present disclosure, detailed descriptions of related prior art are omitted if it is determined that such descriptions may unnecessarily obscure the essence of the present invention. Furthermore, numbers used in the description of this specification (e.g., first, second, etc.) are merely identification symbols to distinguish one component from another.

[0027] In addition, when a component is described in this specification as being "connected" or "connected" to another component, it should be understood that the component may be directly connected to or directly connected to the other component, but unless otherwise specifically stated, it may also be connected or connected through another component in between.

[0028] In addition, terms such as “~part,” “~device,” “~device,” and “~module” described in this specification refer to a unit that processes at least one function or operation, and may be implemented as hardware or software or a combination of hardware and software such as a processor, microprocessor, microcontroller, CPU (Central Processing Unit), GPU (Graphics Processing Unit), APU (Accelerate Processor Unit), DSP (Drive Signal Processor), ASIC (Application Specific Integrated Circuit), and FPGA (Field Programmable Gate Array), and may also be implemented in a form combined with memory that stores data necessary for processing at least one function or operation.

[0029] Furthermore, it is intended to clarify that the classification of components in this specification is merely based on the primary function each component is responsible for. That is, two or more components described below may be combined into a single component, or a single component may be divided into two or more components based on more subdivided functions. Additionally, each component described below may additionally perform some or all of the functions of other components in addition to the primary function it is responsible for, and it is obvious that some of the primary functions of each component may be exclusively performed by other components.

[0030] In describing the embodiments of the present disclosure, specific descriptions of related functions or configurations are omitted if it is determined that such detailed descriptions would unnecessarily obscure the essence of the present disclosure. Furthermore, terms used below are defined in consideration of their functions within the present disclosure, and these definitions may vary depending on the intent or practices of the user or operator. Therefore, such definitions should be based on the content throughout this specification.

[0031] For the same reason, some components in the attached drawings may be exaggerated, omitted, or schematically depicted. Additionally, the size of each component does not entirely reflect its actual size. Identical or corresponding components in each drawing have been assigned the same reference number.

[0032] The advantages and features of the present disclosure 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 disclosure is not limited to the embodiments disclosed below but may be implemented in various different forms. The embodiments are provided merely to make the description of the present disclosure complete and to fully inform those skilled in the art of the scope of the invention to which the embodiments of the present disclosure belong, and the scope of the claims of the present disclosure is defined only by the scope of the claims.

[0033] At this point, it will be understood that each block of the drawings showing the process flow diagram and the combinations of the process flow diagrams can be executed by computer program instructions. Since these computer program instructions can be loaded into the processor of a general-purpose computer, a specialized computer, or other programmable data processing equipment, the instructions executed through the processor of the computer or other programmable data processing equipment create means to perform the functions described in the flow diagram block(s). Since these computer program instructions can also be stored in computer-available or computer-readable memory that can be directed toward the computer or other programmable data processing equipment to implement the function in a specific way, the instructions stored in computer-available or computer-readable memory can also produce a manufactured item containing the means of instruction to perform the function described in the flow diagram block(s). Since computer program instructions can be loaded onto a computer or other programmable data processing equipment, instructions that perform a series of operation steps on the computer or other programmable data processing equipment to create a process executed by the computer can also provide steps for executing the functions described in the flowchart block(s).

[0034] Additionally, each block may represent a module, segment, or part of code containing one or more executable instructions for executing a specified logical function(s). It should also be noted that in some alternative execution examples, the functions mentioned in the blocks may occur out of order. For instance, two blocks described in succession may actually be executed substantially simultaneously, or the blocks may be executed in reverse order according to their corresponding functions.

[0035] As used in this disclosure, the term “unit or part” refers to a software or hardware component, such as a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), and the “part” may be configured to perform specific roles. However, the “part” is not limited to software or hardware. The “part” may be configured to reside in an addressable storage medium or to execute one or more processors. Thus, by example, the “part” 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” may be combined into a smaller number of components and “parts” or further separated into additional components and “parts.” In addition, the components and 'parts' may be implemented to utilize one or more CPUs within the device or secure multimedia card. Also, in the embodiments, 'parts' may include one or more processors and / or devices.

[0037] Hereinafter, embodiments according to the technical concept of the present invention will be described in detail in turn.

[0039] FIG. 1 is a conceptual diagram illustrating the basic principles of artificial intelligence technology according to one embodiment of the present disclosure.

[0040] Referring to Figure 1, the basic principle of how learning is performed in an artificial intelligence structure is illustrated.

[0041] Artificial intelligence (AI) technology refers to techniques designed to solve cognitive problems primarily associated with human intelligence, such as learning, problem-solving, and perception. AI can be trained through machine learning (ML) and deep learning (DL). Machine learning is primarily used in techniques for pattern recognition and learning, representing algorithms that learn from recorded data to predict future data. It represents a technology that learns autonomously from data rather than relying on predefined rules or patterns. On the other hand, deep learning is a subfield of machine learning that differs in that it processes data based on Artificial Neural Networks (ANN). Because deep learning utilizes artificial neural networks, it can handle more complex and sophisticated computations than machine learning. Types of algorithms for deep learning include Convolutional Neural Networks (CNN), Artificial Neural Networks (ANN), and Recurrent Neural Networks (RNN).

[0042] Referring to FIG. 1, the artificial intelligence structure can be represented as an artificial intelligence module (110). The artificial intelligence module (110) receives predetermined input data (105), performs learning through a predetermined method determined in the module, and outputs output data (115) for the learning result. According to one embodiment, the input data (105) may include predetermined data, prompt data, etc. The output data (115) may include personality characteristic information, information about attributes and objects, output sequence, etc.

[0044] In this disclosure, a method for estimating personality based on text through multiple agents in three stages in a large language model (LLM) will be described.

[0045] The three stages of personality estimation can be divided into the personality induction stage, the psycholinguistic explanation stage, and the comparative evaluation stage.

[0046] FIG. 2 is a diagram illustrating a process for a personality induction step performed in a language model according to one embodiment of the present disclosure.

[0047] The language model of Fig. 2 may be one of the types of artificial intelligence modules (110) of Fig. 1.

[0048] An individual's personality for estimation can be classified into five categories using the Big 5 theory, a highly reliable personality theory that has been used for a long time in personality psychology.

[0049] The five factors of personality can be classified into openness (205a), conscientiousness (205b), extraversion (205c), agrrableness (205d), and neuroticism (205e). These personalities can be classified by numerical values, and people with relatively high scores and people with low scores can be interpreted as having contrasting personalities. Since each personality includes widely known information, a detailed explanation will be omitted. In the following disclosure, personalities may be described using expressions such as "high" or "low," which do not represent specific numerical values ​​but rather indicate relative high or low values, or represent extreme cases where the numerical value is 0 or 100.

[0050] In the personality induction stage (200), all stages can be performed separately for each of the five personality factors. That is, to determine the openness personality, the personality induction stage, the psycholinguistic explanation stage, and the comparative evaluation stage are performed, and to estimate sincerity (205b), the personality induction stage, the psycholinguistic explanation stage, and the comparative evaluation stage can be performed.

[0051] FIG. 2 describes a method for estimating openness (205a) in the personality induction stage (200) as one example, and the remaining personality can be estimated in the same or similar way.

[0052] First, as the first step of personality estimation, a personality induction step (200) may be performed for personality estimation. The personality induction step (200) may include analyzing personality from various perspectives by creating an agent with distinct personality traits. To induce personality traits in the language model, prompts may be constructed using sentences and keywords related to personality traits so that the agent internalizes specific personality traits.

[0053] Referring to FIG. 2, a high openness agent (240) and a low openness agent (250) can be generated by inputting a prompt (hereinafter, high openness prompt) (210) for inducing a high level of openness (205a) and a prompt (hereinafter, low openness prompt) (220) for inducing a low level of openness (230a, 230b), respectively, into language models (230a, 230b).

[0054] For example, a high openness agent (240) can be generated by inputting a high openness prompt (210) such as “You are an agent for analyzing human personality, and you are an open person” into a language model (230a).

[0055] Conversely, a low openness agent (250) can be generated by inputting a low openness prompt (220) into a language model (230b) that says, "You are an agent for analyzing human personality, and you are a cautious and realistic person."

[0056] In this way, a total of 10 personality agents can be generated for the five factors of personality.

[0058] FIG. 3 is a diagram illustrating a process for a psycholinguistic explanation step performed in a language model according to one embodiment of the present disclosure.

[0059] The psycholinguistic explanation step of Fig. 3 may represent a step performed after generating a personality agent through the personality induction step of Fig. 2.

[0060] As a second step of personality estimation, a psycholinguistic explanation step (300) can be performed. In the explanation step (300), two personality agents may be asked to explain one personality factor regarding the target text (305) written by a person.

[0061] Personality agents can enable the analysis of the relationship between text and personality traits using three major psycholinguistic elements—emotional, cognitive, and social aspects—that have been previously studied, through prompt design in advance.

[0062] Each personality agent can be provided with definitions for three psycholinguistic factors and, based on these, be asked to generate explanations for criteria for predicting personality for each factor.

[0063] Each personality agent can interpret the text from a different perspective regarding the target text (305) and analyze the personality of the author of the target text (305) in emotional, cognitive, and social aspects.

[0064] For example, regarding the same personality factor, an analysis and explanation of the target text (305) can be requested from each of the high personality agent (310) and the low personality agent (320).

[0065] The high personality agent (310) and the low personality agent (320) can analyze the target text (305) in terms of emotional aspects (335), cognitive aspects (340), and social aspects (345), and output psycholinguistic descriptions (330, 350) according to each aspect.

[0067] FIG. 4 is a diagram illustrating a process for a comparison evaluation step performed in a language model according to one embodiment of the present disclosure.

[0068] The comparative evaluation step of Fig. 4 may represent a step performed after generating an explanation through the psycholinguistic explanation step of Fig. 3.

[0069] As a third step of personality estimation, a comparative evaluation step (400) can be performed. In the comparative evaluation step (400), the descriptions of the high personality agent and the low personality agent are analyzed from a neutral perspective through a judge agent (405) to determine which description is more accurate and to determine the personality of the user who wrote the target text.

[0070] The judge agent (405) can make a decision through three stages of the decision-making process: a comparative analysis stage (430), an overall evaluation stage (450), and a final judgment stage (460).

[0071] The judge agent (405) can identify points of agreement and disagreement regarding the analysis of each key element—emotional, cognitive, and social—written by the two personality agents in the comparative analysis step (430), determine how well each analysis matches a specific example of user text, and evaluate the depth and evidence supporting the final conclusion.

[0072] The judge agent (405) can synthesize the results of the judge agent (405) in the overall evaluation stage (450) and determine which personality agent's description better reflects the personality traits of the text writer.

[0073] The judge agent (405) can determine whether the personality trait is high or low based on the accumulated evidence and analysis at the final judgment stage (460).

[0074] For example, in the comparison evaluation stage (400), a judgment may be requested from the judge agent (405) regarding the description (410) generated by the high-personality agent and the description (420) generated by the low-personality agent.

[0075] The judge agent (405) can determine similarity by comparing the high personality description (410) and the low personality description (420) in the comparative analysis step (430) (435), analyzing the psycholinguistic factor descriptions (440a, 440b) of each description, and comparing the points of agreement and disagreement.

[0076] The judge agent (405) can synthesize the analysis results based on the similarity of the descriptions explained by each personality agent in the overall evaluation stage (450).

[0077] Finally, the judge agent (405) can determine the final personality (465) by determining the numerical value of the text writer's personality at the final judgment stage (460).

[0079] FIG. 5 is an example showing a pseudocode for a personality estimation method according to one embodiment of the present disclosure.

[0080] FIG. 5 shows pseudo code for utilizing the personality estimation method according to the present disclosure in a language model.

[0081] Referring to FIG. 5, user text (T) (505), a prompt (H) (510) inducing a high personality, a prompt (L) (520) inducing a low personality, a large language model (M) (525), a linguistic psycho-explanatory prompt (E) (530), and a judge prompt (535) can be input. A final decision (D) (540) can be derived as output.

[0082] First, in the first step, an explanation generation step (550) can be performed. A first prompt (552a) can be generated by combining a high personality-inducing prompt (510), a linguistic psychological explanation prompt (530), and user text (505). The first prompt (552a) can be input into a large language model (525) to output a first explanation (555a). In the same way, a second prompt (552b) can be generated using a low personality-inducing prompt (520), and a second explanation (555b) can be output based on this.

[0083] In the second stage, an arbitrary explanation order step (560) may be performed. The arbitrary explanation order step (560) is intended to perform a more unbiased and accurate personality analysis by ensuring that the judge agent does not know the order of the generated explanations. The first explanation (555a) and the second explanation (555b) may be arbitrarily assigned to the A explanation (565a) and the B explanation (565b).

[0084] In the third stage, the judge evaluation generation step (570) can be performed.

[0085] To generate an evaluation of the judge agent, the judge prompt (535), user text (505), explanation A (565a), and explanation B (565b) can be combined and input into a large language model (525). The large language model (525) can output an evaluation based on the input.

[0086] In the fourth step, the final decision extraction step (580) can be performed.

[0087] A final decision (540) can be output based on the evaluation output from the large language model (525).

[0089] Figure 6 shows the result of performing personality estimation in a language model according to the present disclosure.

[0090] Figure 6 shows the results obtained when the personality estimation method described in Figures 2 to 5 is actually applied.

[0091] The input text (605) is a text written by the author who is the subject of the personality analysis and may contain various contents.

[0092] CoT (Chain of thought) (610) is a commonly used technique as one of the prompt learning methods. As a result of performing a personality analysis on the input text (605) using the CoT (610) technique, the writer is interpreted to have a low level of sincerity.

[0093] The first reasoner (615) performed a psycholinguistic analysis on the input text (605) as an agent induced to a high level of sincerity.

[0094] The second inferor (620) performed a psycholinguistic analysis on the input text (605) as an agent induced to a low level of sincerity.

[0095] The judge (625) makes a judgment on the input text (605) based on the explanations of the first inferor (615) and the second inferor (620), and finally determines that the author has high sincerity.

[0096] In this way, it can be seen that the final result differs when using CoT (610) and when using the nature estimation method of the present disclosure.

[0097] When considering the actual input text (605), since it indicates that the willpower for food is weak but efforts are being made to overcome it, it is desirable to interpret it as having high sincerity, so it can be confirmed that the result estimated by the personality estimation method of the present disclosure is more accurate.

[0099] FIG. 7 is a block diagram of an electronic device to which an artificial intelligence algorithm model according to one embodiment of the present disclosure is applied.

[0100] Referring to FIG. 7, the electronic device (710) may include a modem (MODEM, 720), a memory (MEMORY, 740), and a processor (PROCESSOR, 730).

[0101] The modem (720) may be a communication modem that is electrically connected to other electronic devices to enable mutual communication. In particular, the modem (720) may receive data input and transmit it to the processor (730), and the processor (730) may store the input data in memory (740). Additionally, information output by an artificial intelligence algorithm learned in the system may be transmitted to other electronic devices.

[0102] The memory (740) is a configuration in which various information and program instructions for the operation of the electronic device (710) are stored, and may be a storage device such as a hard disk or a solid state drive (SSD). In particular, the memory (740) may store one or more data inputs from the modem (720) under the control of the processor (730). Additionally, the memory (740) may store program instructions, such as an artificial intelligence algorithm for personality estimation that can be executed by the processor (730).

[0103] The processor (730) is composed of at least one processor and can calculate data by utilizing a personality estimation artificial intelligence algorithm and a large language model to which the personality estimation algorithm is applied, using data and program instructions stored in memory (740). The processor (730) can control and calculate all artificial intelligence algorithm models (e.g., personality estimation artificial intelligence algorithm models) described in FIGS. 1 to 6.

[0105] FIG. 8 is a flowchart illustrating a method for performing character estimation according to one embodiment of the present disclosure.

[0106] With reference to FIG. 8, the learning operation and method of the artificial intelligence algorithm of the large language model and the personality estimation artificial intelligence algorithm described with reference to FIG. 1 to 7 will be summarized and explained below. Each operation is not an operation that must be necessarily included in the series of processes, and only some of them may be configured and operated depending on the situation.

[0107] In step S810, the personality estimation device (e.g., the electronic device (710) of FIG. 7) can generate a first agent induced by a first personality trait (e.g., the high openness agent (240) of FIG. 2).

[0108] In step S820, the personality estimation device may generate a second agent (e.g., the low openness agent (250) of FIG. 2) induced by a second personality trait opposite to the first personality trait.

[0109] In one embodiment, the first personality trait and the second personality trait are determined from the same personality type, and the personality type may include openness (e.g., openness (205a) of FIG. 2), conscientiousness (e.g., conscientiousness (205b) of FIG. 2), extraversion (e.g., extraversion (205c) of FIG. 2), agreeableness (e.g., agreeableness (205d) of FIG. 2), and neuroticism (e.g., neuroticism (205e) of FIG. 2).

[0110] In step S830, a psycholinguistic description of the target text (e.g., the psycholinguistic description of FIG. 3 (330, 350)) can be generated through each of the first agent and the second agent.

[0111] In one embodiment, when generating a psycholinguistic description of a target text through each of the first agent and the second agent, the first agent may generate a first description of the target text based on the first personality trait in terms of emotional (e.g., emotional aspect (335) of FIG. 3), cognitive (e.g., cognitive aspect (340) of FIG. 3), and social (e.g., social aspect (345) of FIG. 3), and the second agent may generate a second description of the target text in terms of emotional, cognitive, and social aspects based on the second personality trait.

[0112] In step S840, personality traits of the author of the target text (e.g., final personality (465) in FIG. 4) can be estimated through a third agent (e.g., judge agent (405) in FIG. 4) based on psycholinguistic descriptions generated through each of the first agent and the second agent (e.g., psycholinguistic factor descriptions (440a, 440b) in FIG. 4).

[0113] In one embodiment, the first agent, the second agent, and the third agent may be generated in a large language model (e.g., the language model of FIG. 2 (230a, 230b)).

[0114] In one embodiment, when estimating personality traits of the author of the target text, the third agent determines similarity by comparing points of agreement and points of disagreement based on the target text, the first description, and the second description, and can estimate personality traits of the author of the target text based on the similarity.

[0116] Although the technical concept of the present invention has been described in detail with reference to various embodiments, the technical concept of the present invention is not limited to the above embodiments, and various modifications and changes are possible by those skilled in the art within the scope of the technical concept of the present invention.

Claims

Claim 1 A method comprising: generating a first agent induced by a first personality trait; generating a second agent induced by a second personality trait opposite to the first personality trait; generating a psycholinguistic description of a target text through each of the first agent and the second agent; and estimating a personality trait of the author of the target text through a third agent based on the psycholinguistic descriptions generated through each of the first agent and the second agent. Claim 2 A method according to claim 1, wherein the first personality trait and the second personality trait are determined from the same personality type, and the personality type includes openness, conscientiousness, extraversion, agreeableness, and neuroticism. Claim 3 In claim 1, the first agent, the second agent, and the third agent are generated in a large language model. Claim 4 A method according to claim 1, wherein the step of generating a psycholinguistic description of a target text through each of the first agent and the second agent comprises: a step of generating a first description of the target text in emotional, cognitive, and social aspects based on the first personality trait in the first agent; and a step of generating a second description of the target text in emotional, cognitive, and social aspects based on the second personality trait in the second agent. Claim 5 In claim 4, the step of estimating personality traits of the author of the target text comprises: a step of determining similarity by comparing points of agreement and points of disagreement based on the target text, the first description, and the second description in the third agent; and a step of estimating personality traits of the author of the target text based on the similarity. Claim 6 A personality estimation device comprising: a memory; a modem; and a processor connected to the modem and the memory, wherein the processor is configured to: generate a first agent induced to a first personality trait based on a first prompt, generate a second agent induced to a second personality trait opposite to the first personality trait based on a second prompt, generate a psycholinguistic description of a target text through each of the first agent and the second agent, and estimate a personality trait of the author of the target text through a third agent based on the psycholinguistic descriptions generated through each of the first agent and the second agent. Claim 7 In paragraph 6, the first personality trait and the second personality trait are determined from the same personality type, and the personality type includes openness, conscientiousness, extraversion, agreeableness, and neuroticism. Claim 8 In paragraph 6, the first agent, the second agent, and the third agent are devices generated in a large language model. Claim 9 In claim 6, the processor is configured to: generate a first description of the target text in emotional, cognitive, and social aspects based on the first personality trait in the first agent, and generate a second description of the target text in emotional, cognitive, and social aspects based on the second personality trait in the second agent. Claim 10 In claim 9, the processor is configured to determine similarity by comparing points of agreement and points of disagreement based on the target text, the first description, and the second description in the third agent, and to estimate personality traits of the author of the target text based on the similarity.

Images