Personal information leakage prevention and personalized response system and method based on equipment side

Abstract

According to the embodiment of the invention, the personal information leakage prevention and personalized response system based on the equipment terminal comprises a plurality of user terminals for detecting personal identity information PII in input user queries and sending user neutral queries of which the PII is converted into neutral information; and a management server that receives a user neutral query and learns a management language model that generates a common response pattern for each neutral query pattern.

Description

Technical Field

[0001] The present invention relates to a personal information anti-disclosure and personalized response system and method based on a device side. Background Art

[0002] Recently, although generative dialogue artificial intelligence such as ChatGPT has received wide attention, limitations such as the leakage of personal identity information (PII) of artificial intelligence and the lack of user-customizable functions have gradually emerged.

[0003] In the case of existing smartphones, little PII (such as fingerprint or facial recognition information) is processed within the user device without being transmitted to a management server. On the other hand, in the case of newly introduced generative dialogue AI, the amount of PII that needs to be stored is much larger, and models such as ChatGPT are emerging, which can not only answer questions through text questions, but also through images or videos with a large amount of data. The existing method of only managing a small part of PII on the device has limitations because the amount of information accumulated increases when the user uses the personal device for a long time.

[0004] In addition, the personalization of dialogue AI is crucial for providing responses tailored to user preferences, but the problem is that technologies for preventing PII leakage must be supplemented in this process.

[0005] Therefore, to solve the above problems, the present invention seeks to propose an artificial intelligence agent executed on a user terminal such as a smart phone without transmitting PII to a management server such as the cloud. Summary of the Invention

[0006]

Problems to be Solved

[0007] The present invention aims to solve the above problems, and the purpose is to provide a personal information anti-disclosure and personalized response system and method based on a device side, which is used to prevent PII from only existing on the user terminal without being leaked to the outside, and generate user-customizable responses through programs inside the user terminal.

[0008] However, the technical problems to be achieved in this embodiment are not limited to the above technical problems, and there may be other technical problems.

[0009]

Solutions to the Problems

[0010] As a technical solution to solve the above technical problems, the device - side personal information leakage prevention and personalized response system according to the first aspect of the present invention includes: a plurality of user terminals, which detect personal identifiable information (PII) in the input user query and send a user neutral query in which the PII is converted into neutral information; and a management server, which receives the user neutral query and learns a management language model, and the management language model generates a common response pattern for each neutral query pattern.

[0011] The device - side personal information leakage prevention and personalized response user terminal according to the second aspect of the present invention includes: a communication module; a memory for storing a personalized response program; and a processor for executing the personalized response program, and the personalized response program detects PII in the input user query, converts the PII into neutral information, and sends the user neutral query converted into neutral information to the management server.

[0012] The personalized response method executed in the device - side personal information leakage prevention and personalized response system according to the third aspect of the present invention includes the following steps: (a) The management server receives from a plurality of user terminals a user neutral query in which personal identifiable information PII included in the user query is converted into neutral information; and (b) The management server receives the user neutral query and learns a management language model, and the management language model generates a common response pattern for each neutral query pattern.

[0013]

Invention Effects

[0014] According to one of the above - mentioned solutions for solving the problems of the present invention, the PII only exists on the user terminal and will not be leaked to the outside, and a personalized response can be generated through a program inside the device. In addition, a user - customized AI Q&A service can also be provided. Brief Description of the Drawings

[0015] Figure 1 is a configuration diagram of a device - side personal information leakage prevention and personalized response system according to an embodiment of the present invention.

[0016] Figure 2 is a diagram showing a device - side personal information leakage prevention and personalized response system according to an embodiment of the present invention.

[0017] Figure 3 is a diagram showing an example of PII according to an embodiment of the present invention.

[0018] Figure 4 is a configuration diagram of a device - side personal information leakage prevention and personalized response user terminal according to another embodiment of the present invention.

[0019] Figure 5 is a flowchart showing a method for preventing personal information leakage and personalized response executed on a user terminal according to another embodiment of the present invention.

[0020] Figure 6 is a diagram showing a personalized response program executed on a user terminal according to another embodiment of the present invention.

[0021] Figure 7 is a diagram showing data interaction between a management server and a user terminal according to another embodiment of the present invention.

[0022] Figure 8 is a flowchart showing a personalized response method executed in a personal information leakage prevention and personalized response system based on the device side according to another embodiment of the present invention. Detailed implementation

[0023] Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. To clearly explain the present invention in the drawings, parts irrelevant to the description are omitted, and similar parts are given the same reference numerals throughout the specification.

[0024] Throughout the specification, when a component is referred to as being "connected" to another component, this includes not only the case where they are "directly connected", but also the case where they are "electrically connected" to another element in the middle. Additionally, when a component is referred to as "including" a certain component, this does not mean excluding other components, but may include other components unless explicitly stated to the contrary.

[0025] In this specification, a "unit" includes a unit implemented by hardware, a unit implemented by software, and a unit implemented using both. Additionally, one unit can be implemented using two or more hardware components, and two or more units can be implemented using one hardware component. Furthermore, a "~ unit" is not limited to software or hardware, and a "~ unit" can be configured to reside in an addressable storage medium or can be configured to be reproduced on one or more processors. Thus, by way of example, a "~ unit" includes: software components, object-oriented software components, class components, and task components, etc., processes, functions, attributes, procedures, subroutines, and program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within a component and a "~ unit" can be combined into a smaller number of components and "~ units", or can be further separated into additional components and "~ units". Additionally, a component and a "~ unit" can be implemented as one or more CPUs within a refreshing device.

[0026] A network refers to a connection structure that allows information exchange between nodes such as terminals and servers, including local area networks (LANs), wide area networks (WANs), the Internet (WWW: World Wide Web), wired and wireless data communication networks, telephone networks, wired and wireless television networks, etc. Examples of wireless data communication networks include 3G, 4G, 5G, 3GPP (Third Generation Partnership Project), LTE (Long Term Evolution), WiMAX (Worldwide Interoperability for Microwave Access), Wi-Fi, Bluetooth communication, infrared communication, ultrasonic communication, visible light communication (VLC), LiFi, etc., but are not limited thereto.

[0027] Figure 1 It is a configuration diagram of a device - side personal information leakage prevention and personalized response system according to an embodiment of the present invention. Figure 2 It is a diagram showing a device - side personal information leakage prevention and personalized response system according to an embodiment of the present invention.

[0028] Referring to Figure 1 , the personalized response system 1 may include a plurality of user terminals 10, a management server 20, and a database 30.

[0029] Referring to Figure 2 , the user terminal 10 can detect PII (Personally Identifiable Information) in the input user query and send a user - neutral query converted from the PII to the management server 20. Here, PII refers to a resident registration number, name, email address, or various other information that can be used to reveal the identity of a specific individual.

[0030] The management server 20 can receive neutral user queries from a plurality of user terminals 10 and learn a management language model 210, which generates a common response pattern for each neutral query pattern.

[0031] The database 30 can store or provide user - neutral query and response data generated based on the interaction between the management server 20 and the user terminal 10 as learning data. As an example, the learning data managed by the database 30 can be used for federated learning, in which data scattered in multiple locations does not need to be directly shared, but instead learns the model through mutual cooperation.

[0032] The management server 20 can be implemented as a computer or a portable terminal capable of accessing the network. Here, the computer includes a laptop, a desktop computer, a handheld computer, etc., and the portable terminal is, for example, a wireless communication device that ensures portability and mobility, and can include various types of handheld wireless communication devices, such as various smartphones, tablets, smartwatches, etc.

[0033] In addition, the management server 20 may provide a management language model based on user-neutral query learning to the user terminal 10. At this time, the management server 20 runs in a cloud computing service model such as SaaS (Software as a Service), PaaS (Platform as a Service), or IaaS (Infrastructure as a Service), or may also be deployed in the form of a private cloud, a public cloud, or a hybrid cloud.

[0034] Specifically, the management server 20 includes a communication module, a memory, and a processor. The communication module provides a communication interface required to combine a communication network to provide signals transmitted and received to the user terminal 10 in the form of packet data. Here, the communication module may be a device including hardware and software necessary to transmit and receive signals such as control signals or data signals through a wired or wireless connection with other network devices.

[0035] The memory stores a personalized response program. In addition, the memory performs the function of temporarily or permanently storing data processed by the processor. Here, the memory may include a volatile storage medium or a non-volatile storage medium, but the scope of the present invention is not limited thereto.

[0036] The memory may store a separate program such as an operating system for processing and controlling the processor, and may also perform the function of temporarily storing input or output data.

[0037] The memory includes at least one storage medium of a flash type, a hard disk type, a multimedia card micro type, a card type memory (such as an SD or XD memory, etc.), a RAM, and a ROM.

[0038] The processor executes an evacuation route analysis program and provides a function of controlling the hardware of the terminal according to the execution of the program. That is, the processor may perform hardware control functions such as a required file system, memory allocation, network, basic library, timer, device control (display, media, input device, 3D, etc.), and other utility programs when executing the program.

[0039] Here, the processor may include all types of devices capable of processing data such as a processor. Here, the "processor" may refer to, for example, a data processing device built into hardware, which has a circuit with a physical structure to execute a function expressed by code or instructions included in a program. Examples of data processing devices built into hardware may cover processing devices such as a microprocessor, a central processing unit (CPU), a processor core, a multi-processor, an application specific integrated circuit (ASIC), and a field programmable gate array (FPGA), but the scope of the present invention is not limited thereto.

[0040] The personalized response program according to an embodiment of the present invention may input a neutral user query received from a plurality of user terminals 10 into the management language model 210 to generate a common response pattern for each neutral query pattern. As an example, the management language model 210 is updated through joint continuous learning and may derive a common response pattern that is generally useful for multiple users. Thereafter, the update information of the management language model 210 is sent to each user terminal 10 and may be used to update the response generation model 132 of each user terminal 10.

[0041] As an example, the management language model 210 extracts elements of a sentence structure divided into words, phrases, and clauses from the text input as a user neutral query, classifies the elements of the sentence structure into preset neutral query patterns, and learns by matching the common response patterns corresponding to the neutral query patterns. As an example, the neutral query patterns may be classified in the same context based on a language distribution including the frequency and type of words in the user neutral query.

[0042] The management language model 210 may learn a PII pattern for each neutral query pattern by using the position of neutral information in the sentence structure of the user neutral query. At this time, neutral information means that the PII included in the user query has been converted into a representative neutral word or <mask>, which will be described in detail by the PII detection model later. Additionally, it can be determined by identifying neutral words or <mask>Determine the position of neutral information based on the position of the tag.

[0043] As an example, each user terminal 10 can generate various patterns including PII in the user query through conversation. As an example, the first user can generate a pattern (first PII pattern) that includes PII at position A in a neutral query pattern with the sentence structure "ABC". As another example, the second user can create a pattern (second PII pattern) that includes PII at position C in a neutral query pattern with the sentence structure "ABC". That is, depending on the user's situation, whether the same word is recognized as PII for each user may vary. For example, "Google" may be a simple word entity for the first user, but may be PII as workplace information for the second user.

[0044] Therefore, the management server 20 can determine for each neutral query pattern received from multiple user terminals 10 which position of PII is converted into neutral information in what context.

[0045] Figure 3 is a diagram showing an example of PII according to an embodiment of the present invention.

[0046] See Figure 3 , PII can be divided into direct identifiers and quasi-identifiers that can identify a specific individual. For example, a direct identifier refers to information that can directly identify a specific individual by itself in a PII dataset. For example, direct identifiers can include name, resident registration number, residential address, phone number, email address, etc. In addition, a quasi-identifier refers to information that can indirectly identify a specific individual through combination with other quasi-identifiers. For example, quasi-identifiers can include age, gender, political inclination, religion, habits, etc. At this time, a method for detecting PII will be described later with reference to Figures 5 to 7 Describe the method for detecting PII.

[0047] Figure 4 is a configuration diagram of a device-side personal information leakage prevention and personalized response user terminal according to another embodiment of the present invention, Figure 5 is a flowchart showing a personal information leakage prevention and personalized response method executed on a user terminal according to another embodiment of the present invention, Figure 6 is a diagram showing a personalized response program executed on a user terminal according to another embodiment of the present invention, Figure 7 is a diagram showing the data interaction between the management server and the user terminal according to another embodiment of the present invention.

[0048] Refer to Figure 4 , the user terminal 10 may include a communication module 110, a memory 120, a processor 130, and a database 140.

[0049] The communication module 110 can receive updated management language model information from the management server 20 and send it to the processor 130. Here, the communication module 110 can be a device including the hardware and software necessary for sending and receiving signals such as control signals or data signals through wired or wireless connections with other network devices.

[0050] The memory 120 can have a personalized response program recorded therein. The personalized response program detects PII in the input user query, converts the PII into neutral information, and sends the user neutral query converted into neutral information to the management server. Here, in addition to volatile storage devices that require power to maintain stored information, the memory 120 can also include magnetic storage media or flash storage media, but the scope of the present invention is not limited thereto.

[0051] The memory 120 can store separate programs such as an operating system for processing and controlling the processor 130, and can perform the function of temporarily storing input or output data.

[0052] The processor 130 executes the personalized response program (hereinafter referred to as the program) stored in the memory 120, and provides a function of controlling the hardware of the user terminal 10 according to the execution of the program. That is, when executing the program, the processor can execute hardware control functions such as a required file system, memory allocation, network, basic library, timer, device control (display, media, input device, 3D, etc.), and other utility programs.

[0053] Refer to Figure 5 , the processor 130 detects PII in the input user query (S110), converts the PII into neutral information (S120), and sends the user neutral query converted into neutral information to the management server 20 (S130).

[0054] The processor 130 can include all types of devices capable of processing data. For example, it can refer to, for example, a data processing device built into the hardware, which has a physical structure circuit to execute the functions expressed by the code or instructions included in the program. Examples of data processing devices built into such hardware can cover processing devices such as microprocessors, central processing units (CPUs), processor cores, multi-processors, and application-specific integrated circuits (ASICs) and field-programmable gate arrays (FPGAs), but the scope of the present invention is not limited thereto.

[0055] The database 140 stores or provides data required by the user terminal 10 under the control of the processor 130. For example, the database 140 can include a PII database 141. Refer to Figure 6 , the PII database 141 can store PII extracted from user queries. This database 140 can be included as a component independent of the memory 120, or can be built in a partial area of the memory 120.

[0056] Referring Figure 6 , the processor 130 can include detailed modules that perform various functions according to the execution of the personalized response program. At this time, the detailed modules can include a PII detection model 131 and a response generation model 132.

[0057] Referring Figure 7 , when the user terminal 10 receives update information of the management language model 210 from the management server 20, the response generation model 132 or the PII detection model 131 can be updated. As an example, the PII detection model 131 and the response generation model 132 can perform continuous learning so that the already learned models are updated for new data sets, environments, etc. while maintaining the existing performance.

[0058] As an example, the PII detection model 131 can use an existing language recognition model and can have an encoder structure. Similarly, the response generation model 132 can use an existing language generation model and can be composed of a decoder or an encoder-decoder structure. For example, each model is a basic component, and various elements such as a transformer model, a convolutional neural network (CNN), or a recurrent neural network (RNN) can be applied.

[0059] Referring Figure 7 , the PII detection model 131 can detect PII in the input user query, convert the PII into neutral information, and send the user neutral query converted into neutral information to the management server 20. At this time, the PII detection model 131 can mask the PII extracted from the user query or convert it into a predetermined word to generate a user neutral query with the PII filtered.

[0060] For example, when the PII detection model 131 converts PII into neutral information, the PII can be converted into the corresponding text by specifying a representative neutral word for each type of object classified as PII in the user query. For example, the query "My name is Yoo Sang-won. What's your name?" can be converted into a neutral query such as "My name is Hong Gil-dong. What's your name?" As another example, when the PII detection model 131 converts PII into neutral information, it can be simply by <mask>The token masking extracts PII from the user query to transform the query. For example, the query "My name is Yoo Sang-won. What's your name?" can be transformed into "My name is <mask>A neutral query such as "What is your name?"

[0061] In this way, the management server 20 uses the user-neutral query data with PII filtered to learn the management language model 210 and can provide a common response for each neutral query pattern.

[0062] As another example, the management server 20 can learn the PII pattern of each user while learning multiple user-neutral query patterns. At this time, as described above, neutral words in the sentence structure can be used or <mask>The positions of the tags are used to learn the PII patterns of each neutral query pattern. In one embodiment, the management server 20 may provide the PII pattern of the second user terminal to the first user terminal as update information. For example, in a first user query named "AB C", the first user usually includes PII at position A, while the second user includes PII at position C. Therefore, in a query with this structure, information indicating the possibility that PII is included in both position A and position C is provided to the first user terminal as update information. Subsequently, the first user terminal may input a first user query similar to the PII pattern of the second user terminal instead of its own way. At this time, the updated PII detection model 131 of the first user terminal can easily identify the PII in the first user query. Therefore, each user terminal 10 can improve the PII filtering function of the PII detection model 131. In addition, since what is updated is the pattern containing PII rather than the PII of other users themselves, personal information leakage can be prevented.

[0063] As an example, the PII detection model 131 determines personal information of an entity that is a character or word constituting a user query based on the initial PII pre-stored in the PII database 141, and can extract PII from the user query based on named entity recognition (NER) and store it in the PII database 141. For example, the process of constructing the PII database 141 must be performed first, and the PII database 141 can first extract some entities classified as public PII in the user query as the initial PII through a text recognition methodology such as named entity recognition (NER), or the initial PII can be constructed by the user directly registering. In addition, the PII detection model 131 can expand the PII database 141 by determining whether each object is PII based on the PII database 141 for the constituent entities of the newly input user query text, and can improve the accuracy of PII detection.

[0064] As an example, the PII detection model 131 can learn to identify PII patterns through a continuous learning method based on the PII database 141, which is updated according to user queries or usage patterns. Therefore, the PII detection model 131 can determine whether PII is included in the user query and send a user-neutral query with PII removed from the user query to the management server 20. That is, by using the PII database 141 as the learning data of the PII detection model 131, the text entities associated with the PII database 141 can also be detected by the PII detection model 131 to enhance the generalization ability of PII detection. The loss function applied at this time can be a loss function in a form commonly used for learning language encoder models. For example, in MLM (masked language modeling) learning, the NLL (negative log-likelihood) function can be applied as the loss function.

[0065] The response generation model 132 can identify the context of the user query through natural language processing analysis and generate a response. Additionally, the response generation model 132 can consist of a language model that learns the user query pattern using the user query. At this time, the user query pattern can be learned in the same context based on the language distribution including the frequency and type of words in the user query.

[0066] As an example, the response generation model 132 can use a set of user queries in the form of natural language text as learning data. At this time, as the user query data input by the user accumulates, the learning data can be continuously updated. Therefore, the response generation model 132 can learn the user query pattern based on the accumulated user queries, and each user terminal 10 can provide a personalized and natural response to the user.

[0067] As an example, the learning method of the response generation model 132 can be learned so that the distribution of the language generated in the initial state is close to the distribution of the query language input by the user. Additionally, a loss function that measures the distance between the distributions can be applied to the learning so that the text distribution of the data generated by the response generation model 132 is close to the text distribution composed of the user query data. And, in order to prevent catastrophic forgetting when continuously learning the continuously updated user query data, each time learning data is learned, a loss function that reduces the distance between the generation distribution of the model before learning and the generation distribution of the model after learning can be additionally applied to the learning. For example, crossentropy can be used as a loss function for measuring the difference between distributions.

[0068] In another embodiment, in the device-side personal information leakage prevention and personalized response system of the present invention, when the service providers of the management server 20 and the user terminal 10 are the same or cooperate, the management server (20) can provide joint learning of the management language model 210 and update the information of each model to the user terminal 10. That is, the management server 20 can exchange learning information with each user terminal 10 and interact.

[0069] In an additional embodiment, when there is no mutual access permission due to different service providers of the management server 20 and the user terminal 10, the user terminal 10 sends a neutral query to the management server 20 and can only receive a common response to the neutral query. That is, each user terminal 10 can use the response generation model 132 learned through the user query data to only provide a personalized response as a service. In this case, the management server 20, as a third party, can only play the role of additionally providing a common response when generating a response for the user terminal 10 through the management language model 210.

[0070] In the following, descriptions of the same configurations in the configurations shown above will be omitted. Figures 1 to 7

[0071] Figure 8 is a flowchart showing a personalized response method executed in a device - based personal information leakage prevention and personalized response system according to another embodiment of the present invention.

[0072] Referring to Figure 8 , the personalized response method executed in the device - based personal information leakage prevention and personalized response system 1 includes: a step (S210) in which the management server 20 receives a user - neutral query in which personal identity information PII included in a user query is converted into neutral information from a plurality of user terminals 10; and a step (S220) in which the management server 20 receives the user - neutral query and learns a management language model 210 that generates a common response pattern for each neutral query pattern.

[0073] Here, the management language model 210 can be configured to learn a PII pattern for each neutral query pattern by using the positions of neutral information in the sentence structure of the user - neutral query.

[0074] The personalized response method according to an embodiment of the present invention can also be implemented in the form of a recording medium containing computer - executable instructions (e.g., a program module executed by a computer). The computer - readable medium can be any available medium accessible by a computer and includes volatile and non - volatile media, removable and non - removable media. In addition, the computer - readable medium can include a computer - storage medium. The computer - storage medium includes volatile and non - volatile, removable and non - removable media implemented by any method or technology for storing information such as computer - readable instructions, data structures, program modules, or other data.

[0075] Although the devices and methods of the present invention have been described with respect to specific embodiments, some or all of their components or operations can be implemented using a computer system having a general - purpose hardware architecture.

[0076] The above description of the present invention is for illustrative purposes, and those skilled in the art will understand that the present invention can be easily modified into other specific forms without changing the technical idea or basic characteristics of the present invention. Therefore, the above embodiments should be understood as illustrative in all aspects rather than restrictive. For example, each component described as a single one can be implemented in a distributed manner, and similarly, components described as distributed can also be implemented in a combined form.

[0077] ​The scope of the present invention is indicated by the claims that follow, rather than the foregoing detailed description, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts shall be construed as being included within the scope of the present invention.< / mask> < / mask> < / mask> < / mask> < / mask>

Claims

1. A personalized response system, which is a personal information leakage prevention and personalized response system based on the device side. The personalized response system includes: Multiple user terminals, which detect personal identity information PII in the input user query and send a user neutral query in which the PII is converted into neutral information; And A management server, which receives the user neutral query and learns and manages a language model. The management language model generates a common response pattern for each neutral query pattern.

2. The personalized response system according to claim 1, wherein The user terminal includes: A response generation model, which identifies the context of the user query through natural language processing analysis and generates a response; and A PII detection model, which generates the user neutral query by masking or converting the PII extracted from the user query into a predetermined word.

3. The personalized response system according to claim 2, wherein The response generation model is a language model that learns the user query pattern using the user query, and The response generation model learns the user query pattern in the same context based on a language distribution including the frequency and type of words in the user query.

4. The personalized response system according to claim 2, wherein The PII detection model is used for: Judging whether an entity that is a character or word constituting the user query belongs to personal information according to the initial PII stored in the PII database, and Extracting the PII from the user query based on named entity recognition NER and storing it in the PII database.

5. The personalized response system according to claim 1, wherein The management language model is used for: Using the position of the neutral information in the sentence structure of the user neutral query to learn the PII pattern for each neutral query pattern.

6. The personalized response system according to claim 1, wherein The PII is divided into direct identifiers and quasi-identifiers that can identify specific individuals.

7. A user terminal, which is a personal information leakage prevention and personalized response user terminal based on the device side. The user terminal includes: A communication module; A memory for storing a personalized response program; And A processor for executing the personalized response program, The personalized response program detects PII in the input user query, converts the PII into neutral information, and sends the user neutral query converted into the neutral information to the management server.

8. The user terminal according to claim 7, wherein The personalized response program includes: A response generation model, which identifies the context of the user query through natural language processing analysis and generates a response; and A PII detection model, which generates the user neutral query by masking or converting the PII extracted from the user query into a predetermined word.

9. The user terminal according to claim 8, wherein The response generation model is a language model that learns user query patterns using the user query, and the response generation model learns the user query patterns for each identical context based on a language distribution including the frequency and type of words in the user query.

10. The user terminal according to claim 8, wherein the PII detection model is used for: judging whether an entity that is a character or word constituting the user query belongs to personal information according to initial PII stored in a PII database, and extracting the PII from the user query based on entity name recognition NER and storing the PII in the PII database.

11. A personalized response method executed in a device-side personal information leakage prevention and personalized response system, the personalized response method comprising the following steps: (a) A management server receives a user neutral query in which personal identity information PII included in a user query is converted into neutral information from a plurality of user terminals; and (b) The management server receives the user neutral query and learns a management language model, and the management language model generates a common response pattern for each neutral query pattern.

12. The personalized response method according to claim 11, wherein the management language model is used for: learning a PII pattern for each neutral query pattern by using the position of the neutral information in the sentence structure of the user neutral query.

Images