User profile generation method utilizing artificial intelligence, and device therefor

Abstract

A method, according to one embodiment of the present invention, may comprise the steps of: receiving target text; generating a preference profile on the basis of the target text through a first agent; generating a personality profile on the basis of the target text through a second agent; generating a social characteristic profile on the basis of the target text through a third agent; and generating a user profile for the target text on the basis of the preference profile, the personality profile, and the social characteristic profile.

Description

Method and apparatus for generating a user profile using artificial intelligence

[0001] The present disclosure describes a method and apparatus for estimating a user profile using artificial intelligence.

[0002] Recently, there has been an increasing number of cases where large language models (LLMs) are used to make recommendations regarding user preferences via natural language and to provide explanations for the reasons behind the recommendations.

[0003] However, current models focus on explaining the characteristics of recommended items, and in such cases, there is an aspect where the preferences of specific users are not reflected.

[0004] Therefore, in order to expand the scope of user-centered recommendations, research is continuing to enhance the reliability of recommendation system explanations by constructing user profiles based on user-generated data.

[0005] However, because user-generated data includes latent characteristics that are not clearly apparent, methods that simply analyze combinations of specific words in the text cannot capture the characteristics that users actually prefer.

[0006] Therefore, we propose a method to generate user profiles by analyzing not only clearly revealed preferences but also latent preferences through analysis from various aspects based on data written by users.

[0007] The present disclosure aims to provide a method and apparatus for generating a profile by analyzing a user based on text using a plurality of agents.

[0008] The present disclosure aims to provide a method and apparatus for constructing a recommendation system utilizing user profiles and performing an evaluation thereof.

[0009] A method according to various embodiments of the present disclosure may include: receiving a target text; generating a preference profile based on the target text through a first agent; generating a personality profile based on the target text through a second agent; generating a social characteristic profile based on the target text through a third agent; and generating a user profile for the target text based on the preference profile, the personality profile, and the social characteristic profile.

[0010] In one embodiment, the method may further include the steps of: identifying a recommended item; evaluating the recommended item based on the user profile; and estimating a rating for the recommended item based on a fourth agent.

[0011] In one embodiment, the first agent may be guided to determine explicit and implicit preferences by analyzing the repetitive use of a specific word or phrase.

[0012] In one embodiment, the second agent is induced through a prompt generated based on words representing personality factors, and the personality factors may include openness, conscientiousness, extraversion, agreeableness, and neuroticism.

[0013] In one embodiment, the third agent may be induced through a prompt generated based on a parameter determined based on the number of interactions with the recommended item.

[0014] In one embodiment, the fourth agent may be configured to determine the user's preferences, personality, and social characteristics based on the user profile to assign a persona, and to analyze the characteristics of the recommendation items based on the persona.

[0015] In one embodiment, the fourth agent may be configured to estimate the rating by comparing it with the persona based on the text associated with the recommendation item.

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

[0017] An electronic device according to various embodiments of the present disclosure comprises: a memory; a modem; and a processor connected to the modem and the memory, wherein the processor may be configured to generate a preference profile based on the target text through a first agent, generate a personality profile based on the target text through a second agent, generate a social characteristic profile based on the target text through a third agent, and generate a user profile for the target text based on the preference profile, the personality profile, and the social characteristic profile.

[0018] According to one embodiment of the present disclosure, the performance of a model that generates a user profile based on text can be improved.

[0019] According to one embodiment of the present disclosure, in training a recommendation model, in-depth analysis and recommendation may be possible through a multidimensional approach.

[0020] A brief description of each drawing is provided to help to better understand the drawings cited in the detailed description of the present disclosure.

[0021] FIG. 1 is a conceptual diagram illustrating the basic principles of artificial intelligence technology according to one embodiment of the present disclosure.

[0022] FIG. 2 is a diagram illustrating a profile generation process performed in a large language model according to one embodiment of the present disclosure.

[0023] FIG. 3 is a diagram illustrating the process of a rating prediction model performed in a language model according to one embodiment of the present disclosure.

[0024] FIG. 4 is an example of pseudocode for user profile generation and rating prediction performed in a large language model according to one embodiment of the present disclosure.

[0025] FIG. 5 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.

[0026] FIG. 6 is a flowchart illustrating a method for generating a user profile and performing a rating prediction according to one embodiment of the present disclosure.

[0027] The technical concept of the present invention 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.

[0028] In describing the technical concept of the present invention, detailed descriptions of related prior art are omitted if it is determined that such descriptions may unnecessarily obscure the essence of the 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] 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.

[0033] 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.

[0034] 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, and the scope of the claims of the present disclosure is defined only by the scope of the claims.

[0035] At this point, it will be understood that each block of the drawings showing the process flow diagram and 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).

[0036] 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.

[0037] 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.

[0038]

[0039] Hereinafter, embodiments according to the technical concept of the present disclosure will be described in detail in turn.

[0040]

[0041] FIG. 1 is a conceptual diagram illustrating the basic principles of artificial intelligence technology according to one embodiment of the present disclosure.

[0042] Referring to Figure 1, the basic principle of how learning is performed in an artificial intelligence structure is illustrated.

[0043] 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).

[0044] 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, reviews, text, etc. The output data (115) may include user profiles, ratings, recommended items, output sequences, etc.

[0045]

[0046] The present disclosure describes a method for estimating a user profile based on user-generated data in a large language model (LLM), and a method for recommending desired items and evaluating them by utilizing said profile.

[0047]

[0048] FIG. 2 is a diagram illustrating a profile generation process performed in a large language model according to one embodiment of the present disclosure.

[0049] The large language model of Fig. 2 may be one of the types of artificial intelligence modules (110) of Fig. 1.

[0050] Large language models can create a framework for generating user profiles using multiple agents. This can be simply referred to as the MAPLE (Multi-Agent ProfiLEr) model (hereinafter MAPLE) (200).

[0051] MAPLE (200) may include three LLM-based agents. The three agents of MAPLE (200) may include a preference agent (210), a personality agent (220), and a social trait agent (230).

[0052] MAPLE (200) can generate a profile corresponding to the agent by using three agents separately for the target text (205).

[0053] First, the preference agent (210) generates a profile for the user's personal preferences and explicit and implicit evaluations of specific items.

[0054] The preference agent (210) can analyze the repeated use of specific words or phrases to infer the value of the words or phrases and identify preferred topics based on where the author of the target text (205) expresses positive, neutral, or negative emotions. Additionally, the preference agent (210) can analyze the length of the target text (205) to infer topics important to the author.

[0055] The preference agent (210) can be configured to determine the author's preference based on text by configuring prompts in a large language model.

[0056] The preference agent (210) can generate a preference profile (250) of the author based on an analysis of the target text (205). The preference profile (250) may include a summary of the analysis of the target text (205) in terms of preference.

[0057] Next, various personality theories can be applied to the personality agent (220), and in this disclosure, the Big 5 theory, which is a highly reliable personality theory that has been used for a long time in personality psychology for personality estimation, was used.

[0058] In the Big 5 theory, the five personality factors can be classified as openness, conscientiousness, extraversion, agreeableness, and neuroticism. Since these personality traits can be distinguished by numerical values, individuals with relatively high scores and those with low scores can be interpreted as having contrasting personalities.

[0059] A personality agent (220) can be configured to analyze text based on Big 5 theory by constructing prompts based on words representing personality in a large language model.

[0060] The personality agent (220) can generate a personality profile (260) of the author of the target text (205) by analyzing the target text (205) and interpreting the linguistic expressions of the words. The personality profile (260) can represent the author's personality within the Big 5 theory as shown in the target text (205).

[0061] Finally, the social characteristic agent (230) may be an agent that analyzes social characteristics representing behaviors or tendencies that an individual exhibits in a social context. In particular, activity may be considered as a key element of social characteristics. Activity may be a parameter indicating how frequently a user interacts with recommended items.

[0062] A social characteristic agent (230) can be configured by describing the definition of an activity in a recommendation context to a large language model.

[0063] A social characteristic agent (230) can generate a social characteristic profile (270) by analyzing the target text (205) and analyzing how often the author interacts with various items.

[0064] MAPLE (200) can create a user profile (240) based on profiles created using three agents.

[0065] MAPLE (200) can determine and notify the user of recommended items based on the user profile (240).

[0066]

[0067] FIG. 3 is a diagram illustrating the process of a rating prediction model performed in a language model according to one embodiment of the present disclosure.

[0068] The rating prediction model of Fig. 3 can predict the rating of the content recommended to the user based on the user profile (240) generated by MAPLE in Fig. 2.

[0069] As described in Fig. 2, MAPLE can generate a user profile based on the author's text using three agents. Based on the generated user profile, various recommendation systems including MAPLE can perform recommendations for items desired by the author.

[0070] The rating prediction agent (300) of Fig. 3 can perform an evaluation of whether the recommendation for the item desired by the author was properly recommended based on the user profile generated in MAPLE and predict the rating.

[0071] A rating prediction agent (300) receives a user profile and can assign a persona in terms of preference (305), personality (310), and social characteristics (315).

[0072] The rating prediction agent (300) can analyze the features of the recommended items (320) based on the received user profile.

[0073] The rating prediction agent (300) can perform a comparison with the user profile based on the review (325) of the recommended item (320) and determine the rating (330) corresponding to the user.

[0074]

[0075] FIG. 4 is an example of pseudocode for user profile generation and rating prediction performed in a large language model according to one embodiment of the present disclosure.

[0076] FIG. 4 shows pseudo code for utilizing the user profile generation method and rating prediction method according to the present disclosure in a language model.

[0077] Referring to FIG. 4, user reviews (R) (405), a large language model (M) (410), a preference agent prompt (Preference) (415), a personality agent prompt (Personality) (420), a social trait agent prompt (Social trait) (425), and a rating prediction agent prompt (Rating) (430) can be input. A rating prediction result (P) (435) can be produced as output.

[0078] First, a user profile creation step (440) can be performed. A preference profile (445) can be created by inputting a user review (405) into a preference agent created by applying a preference agent prompt (415) to a large language model (410). A personality profile (450) can be created by inputting a user review (405) into a personality agent created by applying a personality agent prompt (420) to a large language model (410). A social trait profile (455) can be created by inputting a user review (405) into a personality agent created by applying a social trait agent prompt (425) to a large language model (410).

[0079] Next, a rating prediction step can be performed using the combined user profile. A user profile (465) can be created by combining the generated preference profile (445), personality profile (450), and social characteristic profile (455).

[0080] A rating prediction agent prompt (430) can be applied to a large language model (410) to input a user profile (465) into the generated rating prediction agent to output a rating prediction result (435).

[0081]

[0082] FIG. 5 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.

[0083] Referring to FIG. 5, the electronic device (510) may include a modem (MODEM, 520), a memory (MEMORY, 540), and a processor (PROCESSOR, 530).

[0084] The modem (520) may be a communication modem that is electrically connected to other electronic devices to enable mutual communication. In particular, the modem (520) may receive data input and transmit it to the processor (530), and the processor (530) may store the input data in memory (540). Additionally, information output by an artificial intelligence algorithm learned in the system may be transmitted to other electronic devices.

[0085] The memory (540) is configured to store various information and program instructions for the operation of the electronic device (510), and may be a storage device such as a hard disk or a solid state drive (SSD). In particular, the memory (540) may store one or more data inputs from the modem (520) under the control of the processor (530). Additionally, the memory (540) may store program instructions, such as an artificial intelligence algorithm for personality estimation that can be executed by the processor (530).

[0086] The processor (530) is composed of at least one processor and can compute and process data by utilizing data and program instructions stored in memory (540) and large language models to which MAPLE, a profile generation AI algorithm, and a rating prediction AI algorithm are applied. The processor (730) can control and compute all AI algorithm models described in FIGS. 1 to 4 (e.g., MAPLE, a rating prediction AI algorithm model).

[0087]

[0088] FIG. 6 is a flowchart illustrating a method for generating a user profile and performing a rating prediction according to one embodiment of the present disclosure.

[0089] Hereinafter, the electronic device, the learning operation and method of the artificial intelligence algorithm of a large language model, the MAPLE model, and the rating prediction artificial intelligence algorithm described with reference to FIG. 1 to FIG. 5 are summarized and explained in FIG. 6. Each operation is not an operation that must be necessarily included in a series of processes, and only some of them may be configured and operated depending on the situation.

[0090] In step S610, the electronic device can receive the target text (e.g., the target text (205) of FIG. 2).

[0091] In step S620, an electronic device (e.g., MAPLE (200) of FIG. 2, electronic device (510) of FIG. 5) can generate a preference profile (e.g., preference profile (250) of FIG. 2) based on the target text through a first agent (e.g., preference agent (210) of FIG. 2).

[0092] In one embodiment, the first agent may be guided to determine explicit and implicit preferences by analyzing the repetitive use of a specific word or phrase.

[0093] In step S630, the electronic device can generate a personality profile (e.g., personality profile (260) of FIG. 2) based on the target text through a second agent (e.g., personality agent (220) of FIG. 2).

[0094] In one embodiment, the second agent is induced through a prompt generated based on words representing personality factors, and the personality factors may include openness, conscientiousness, extraversion, agreeableness, and neuroticism.

[0095] In step S640, the electronic device can generate a social characteristic profile (e.g., the social characteristic profile (270) of FIG. 2) based on the target text through a third agent (e.g., the social characteristic agent (230) of FIG. 2).

[0096] In one embodiment, the third agent may be induced through a prompt generated based on a parameter determined based on the number of interactions with the recommended item.

[0097] In step S650, the electronic device can generate a user profile for the target text (e.g., the user profile (240) of FIG. 2) based on the preference profile, the personality profile, and the social characteristic profile.

[0098] In one embodiment, the electronic device may identify a recommended item (e.g., a recommended item (320) in FIG. 3), evaluate the recommended item based on the user profile, and estimate a rating for the recommended item (e.g., a rating (330) in FIG. 3) based on a fourth agent (e.g., a rating prediction agent (300) in FIG. 3).

[0099] In one embodiment, the fourth agent may be configured to determine the user's preference (e.g., preference aspect (305) of FIG. 3), personality (e.g., personality aspect (310) of FIG. 3), and social characteristics (e.g., social characteristic aspect (315) of FIG. 3) based on the user profile to assign a persona, and to analyze the characteristics of the recommendation item based on the persona.

[0100] In one embodiment, the fourth agent may be configured to estimate the rating by comparing it with the persona based on text related to the recommendation item (e.g., review (325) of FIG. 3).

[0101] In one embodiment, the first agent, the second agent, the third agent, and the fourth agent may be generated based on prompts in a large language model.

[0102]

[0103] Although the technical concept of the present disclosure has been described in detail with reference to various embodiments, the technical concept of the present disclosure is not limited to the above embodiments, and various modifications and changes can be made by those skilled in the art within the scope of the technical concept of the present disclosure.

Claims

1. Step of receiving target text; A step of generating a preference profile based on the target text through a first agent; A step of generating a personality profile based on the target text through a second agent; A step of generating a social characteristic profile based on the target text through a third agent; and A method comprising the step of generating a user profile for the target text based on the preference profile, the personality profile, and the social characteristic profile.

2. In Paragraph 1, Step to identify recommended items; A step of evaluating the recommendation items based on the above user profile; and A method further comprising the step of estimating a rating for the above-mentioned recommendation item based on a fourth agent.

3. In Paragraph 1, A method in which the first agent above is guided to determine explicit and implicit preferences by analyzing the repetitive use of specific words or phrases.

4. In Paragraph 1, The above-mentioned second agent is induced through a prompt generated based on words representing personality factors, and The above personality factors include openness, conscientiousness, extraversion, agreeableness, and neuroticism.

5. In Paragraph 1, The above third agent is a method induced through a prompt generated based on a parameter determined based on the number of interactions with the recommended item.

6. In Paragraph 2, A method configured such that the fourth agent determines the user's preferences, personality, and social characteristics based on the user profile to assign a persona, and analyzes the characteristics of the recommendation items based on the persona.

7. In Paragraph 6, A method configured such that the above-mentioned fourth agent estimates the above-mentioned rating by comparing it with the above-mentioned persona based on text related to the above-mentioned recommendation item.

8. In Paragraph 6, The first agent, the second agent, the third agent, and the fourth agent are generated based on prompts in a large language model.

9. In electronic devices, Memory; Modem; and It includes the above modem and a processor connected to the above memory, The above processor is: A preference profile is generated based on the above target text through the first agent, and A personality profile is generated based on the above target text through a second agent, and A social characteristic profile is generated based on the above target text through a third agent, and An electronic device configured to generate a user profile for the target text based on the preference profile, the personality profile, and the social characteristic profile.

10. In Paragraph 9, The above processor is: Identify recommended items, Evaluate the above recommendation items based on the above user profile, and An electronic device further configured to estimate a rating for the above-mentioned recommendation item based on a fourth agent.

11. In Paragraph 9, The above-mentioned first agent is an electronic device that is induced to determine explicit and implicit preferences by analyzing the repetitive use of specific words or phrases.

12. In Paragraph 9, The above-mentioned second agent is induced through a prompt generated based on words representing personality factors, and The above personality factors are electronic devices including openness, conscientiousness, extraversion, agreeableness, and neuroticism.

13. In Paragraph 9, The above third agent is an electronic device induced through a prompt generated based on a parameter determined based on the number of interactions with a recommended item.

14. In Paragraph 10, The above-mentioned fourth agent is an electronic device configured to determine the user's preferences, personality, and social characteristics based on the user profile to assign a persona, and to analyze the characteristics of the recommendation items based on the persona.

15. In Paragraph 10, The above-mentioned fourth agent is an electronic device configured to estimate the rating by comparing it with the persona based on the text related to the above-mentioned recommendation item.

Images