WO2019103199A1 - Personalized intelligent system and method for operating same - Google Patents

Personalized intelligent system and method for operating same Download PDF

Info

Publication number
WO2019103199A1
WO2019103199A1 PCT/KR2017/013509 KR2017013509W WO2019103199A1 WO 2019103199 A1 WO2019103199 A1 WO 2019103199A1 KR 2017013509 W KR2017013509 W KR 2017013509W WO 2019103199 A1 WO2019103199 A1 WO 2019103199A1
Authority
WO
WIPO (PCT)
Prior art keywords
intelligent
data
user
big data
model
Prior art date
Application number
PCT/KR2017/013509
Other languages
French (fr)
Korean (ko)
Inventor
정종일
김용진
Original Assignee
주식회사 모다
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 주식회사 모다 filed Critical 주식회사 모다
Publication of WO2019103199A1 publication Critical patent/WO2019103199A1/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/22—Indexing; Data structures therefor; Storage structures
    • G06F16/2219—Large Object storage; Management thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N99/00—Subject matter not provided for in other groups of this subclass
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4662—Learning process for intelligent management, e.g. learning user preferences for recommending movies characterized by learning algorithms
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/466—Learning process for intelligent management, e.g. learning user preferences for recommending movies
    • H04N21/4668—Learning process for intelligent management, e.g. learning user preferences for recommending movies for recommending content, e.g. movies

Definitions

  • the present invention relates to a customized intelligent system, and more particularly, to a customized intelligent system capable of providing an immediate and active intelligent service to a user by using a customized intelligent model for each user apparatus and an operation method thereof.
  • AI artificial intelligence
  • Machine learning is a technique in which a computer learns data on its own and performs predictions based on the learning results to derive useful information.
  • Machine learning generally increases the accuracy of prediction as the quantity and quality of learning data is richer. Therefore, machine learning has a close relationship with Big Data, which is a technology for processing large data, and a typical example is machine learning based personalized service using Big Data.
  • An intelligent service is a service that provides customized content according to the user's personal information. For example, a service that constructs a customized initial screen for each user by analyzing usage patterns of portal site users, or analyzes a search pattern and suggests a recommendation word to other users. Such an intelligent service is provided to a user by an intelligent system. Referring to FIG. 1, a general intelligent system according to the prior art will be described.
  • FIG. 1 is a schematic view of a general intelligent system according to the prior art.
  • the general purpose intelligent system 100 may include a user device 110 and an intelligent server 130.
  • the user device 110 may include a smart phone or a tablet PC as an agent for providing an intelligent service to a user.
  • the intelligence server 130 collects a large amount of user data stored in at least one user equipment 110, for example, service usage histories, and performs machine learning based on the big data to generate a general intelligence model.
  • an intelligent model refers to a software platform that includes a prediction function or pattern information generated through learning about big data.
  • the intelligent server 130 distributes the predicted data, for example, information predictive of a potential user's intention, to the user device 110 using the general intelligence model.
  • the distributed prediction data is provided to the user through the user device 110.
  • generation of predicted data is performed by intelligent server 130, so that real-time service provision may be delayed due to increase in user connection traffic or network failure, and service may be disconnected.
  • the general-purpose intelligent system 100 since the general-purpose intelligent system 100 generates and provides the predictive data only when there is a request from the user (i.e., user data transmission), there is a problem that it is difficult to provide the instantaneous service.
  • the embodiments of the present invention have been made to solve the above-described problems, and in particular, the present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a user data management system capable of providing an immediate and active intelligent service based on small data using a customized intelligent model created and distributed by an intelligent server Intelligent systems that allow them to do so.
  • an intelligent system for providing an intelligent service based on Small Data comprising: an intelligent server for generating and distributing intelligent models from Big Data; And at least one user device for generating predictive data from the user data using the distributed intelligence model and providing intelligent services to the user through a decision based on the predictive data,
  • a big data collection unit for collecting big data from a big data source including a device;
  • a general intelligence model generation unit for generating a predictive function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data and analyzing the predictive function or the pattern information to generate a general intelligence model;
  • a customized intelligent model generation unit for extracting a customized intelligent model from the general intelligent model using profile data of each of the user devices.
  • an operation method of an intelligent system including an intelligent server for generating and distributing an intelligent model and at least one user device for providing an intelligent service based on Small Data, Collecting big data from a Big Data source including the user device; Generating a general function intelligence model by analyzing the prediction function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data to generate a prediction function or pattern information; Generating a customized intelligence model from the general intelligence model using profile data of each of the user devices; Distributing the customized intelligence model to each of the user devices; And each of the user devices generates predictive data from the user data using the distributed customized intelligent model and provides intelligent services to the user through a decision based on the predictive data. .
  • the intelligent system according to the present embodiment is effective in allowing the user device to provide an instant and active intelligent service based on small data using a customized intelligent model generated and distributed by the intelligent server.
  • FIG. 1 is a schematic view of a general intelligent system according to the prior art.
  • FIG. 2 schematically shows a customized intelligent system according to the present embodiment.
  • FIG. 3 is a diagram showing a configuration of a user apparatus according to the present embodiment.
  • FIG. 4 is a diagram showing a configuration of an intelligent server according to the present embodiment.
  • FIG. 5 is a diagram illustrating a process of generating and distributing a customized intelligent model by the intelligent server shown in FIG.
  • FIG. 6 is a diagram illustrating a process in which the user equipment shown in FIG. 3 provides an intelligent service to a user.
  • module refers to a unit that processes at least one function or operation, which may be implemented in hardware, software, or a combination of hardware and software.
  • FIG. 2 schematically shows a customized intelligent system according to the present embodiment.
  • the customized intelligent system 200 may include a user device 210 and an intelligent server 210.
  • the user device 210 provides the user with his / her own intelligence service using the customized intelligence model generated and distributed by the intelligence server 210.
  • the user device 210 may be an electronic device having various kinds of sensors for measuring the biometric information (e.g., heart rate, blood pressure) of the user IOT device, the user 110, a smart phone, A mobile digital device such as a smart watch, a tablet PC and the like, a fixed digital device such as a smart TV, a smart car, a drone, and the like.
  • biometric information e.g., heart rate, blood pressure
  • User device 210 is a big data source and data stored in user device 210 is collected by intelligence server 230 and used for intelligent model generation.
  • the data stored in the user device 210 may include user data and profile data of the user device.
  • the user profile data includes the service purpose and timing of the user device 210, the S / W performance analysis result (for example, the type and version of the operating system framework or library) and the H / W performance analysis result RAM, capacity of the storage device), and the like.
  • the intelligent model is generated by the intelligence server 230 and distributed to the user device 210 and may be updated by the user device 210 or the intelligence server 230 as the user data is added.
  • the user data provided to the intelligent server 230 and the profile data of the user device are encrypted and decrypted using various encryption algorithms (e.g., Advanced Encryption Standard (AES), Cipher-based Message Authentication Code (CMAC) Sex can be strengthened.
  • AES Advanced Encryption Standard
  • CMAC Cipher-based Message Authentication Code
  • the user device 210 may also send a learning algorithm stored in a memory (not shown) to the intelligence server 230.
  • the memory may include non-volatile memory (e.g., NAND flash) and / or volatile memory (e.g., DRAM, etc.).
  • the transmitted learning algorithm can be used by the intelligence server 230 to learn the big data and generate the intelligence model.
  • the intelligence server 130 collects a large amount of user data stored in at least one user equipment 110, for example, service usage histories, and performs machine learning based on the big data to generate a general intelligence model.
  • the intelligence server 230 can generate a customized intelligence model for each user device 210 by preprocessing, learning, and analyzing the user profile data received from the user device 210 using the general purpose intelligence model.
  • performance degradation such as slowing down the predicted speed, may occur when the big data-based general purpose intelligent model is directly applied to the user device 210 degradation may occur.
  • the intelligent service provided by the user device 210 is based on Small Data, when the general-purpose intelligent model is directly applied, resources (e.g., CPU, memory capacity, etc.) It can be wasted more than necessary.
  • the intelligent server 230 acquires, from the general intelligence model, personalized intelligent models based on the device characteristics, performance, and service information of each of the user devices 210 Create and distribute intelligence models.
  • the user device 210 generates prediction data for the intelligent service, that is, the small data, using the distributed customized intelligent model, and provides it to the user.
  • the customized intelligent system 200 can provide an active active intelligent service to the user by using the customized intelligent model itself, unlike the general intelligent system 100 according to the related art .
  • FIG. 3 is a diagram showing a configuration of a user apparatus according to the present embodiment.
  • the user device 210 includes an intelligent service unit 310, a communication unit 330, and a memory 350 as components necessary for providing an intelligent service to a user using a customized intelligent model .
  • the intelligent service unit 310 The intelligent service unit 310,
  • the intelligent service unit 310 can generate the predictive data from the user data using the customized intelligent model distributed from the intelligent server 230 and provide the intelligent service to the user. Specifically, the intelligent service unit 310 receives the customized intelligence model distributed from the intelligence server 230 through the communication unit 330.
  • the intelligent service unit 310 may update the customized intelligent model distributed from the intelligent server 230 by preprocessing, learning and analyzing the user data received from the user.
  • preprocessing learning and analysis operations
  • the intelligent service unit 310 performs prediction using the customized intelligence model for new user data received from the user, and generates predetermined prediction data. Then, the intelligent service unit 310 provides the intelligent service to the user through the decision process based on the generated predictive data.
  • the intelligent service unit 310 may be implemented by at least one processor.
  • the intelligent service unit 310 may include a plurality of GPUs (Graphics Processing Units) that perform parallel processing to improve the real- . ≪ / RTI >
  • GPUs Graphics Processing Units
  • the communication unit 330 The communication unit 330,
  • the communication unit 330 includes a wireless communication module (LTE, Wi-Fi, etc.), a short-range communication module (such as Bluetooth, ZigBee, etc.) as components necessary for exchanging data with the intelligent server 230 via a wired / (ZigBee), Near Field Communication (NFC), etc.).
  • LTE Long Term Evolution
  • Wi-Fi Wireless Fidelity
  • NFC Near Field Communication
  • the communication unit 330 functions to support communication connection with the intelligent server 230 so as to provide the intelligent server 230 with the big data.
  • the communication unit 330 performs a function of receiving the customized intelligence model distributed from the intelligence server 230.
  • the memory 350 may include non-volatile memory (e.g., NAND flash) and / or volatile memory (e.g., DRAM, etc.) as components necessary to store learning algorithms required for intelligent model generation.
  • the learning algorithm stored in the memory is transmitted to the intelligence server 230 through the communication unit 330, and can be used for big data learning for intelligent model generation.
  • FIG. 4 is a diagram showing a configuration of an intelligent server according to the present embodiment.
  • the intelligent server 230 includes components such as a big data collection unit 410, a general intelligent model generation unit 430, a customized intelligent model A generating unit 450, a communication unit 470, and a storage unit 490.
  • the big data collection unit 410 The big data collection unit 410,
  • the big data collecting unit 410 collects big data actively or passively from a big data source connected through the communication unit 470.
  • the big data includes structured data such as a database, a spreadsheet, semi-structured data such as XML and HTML, and unstructured data such as text documents, (unstructured data).
  • the big data source may be an internal source (e.g., an internal database management system (DBMS) or the like) existing in the intelligence server 230 and an external source (e.g., a user device, the Internet, etc.) . ≪ / RTI >
  • DBMS database management system
  • the big data collection unit 410 may store the collected big data in the storage unit 490 or a separate external storage (not shown).
  • the general intelligence model generation unit 430 The general intelligence model generation unit 430
  • the general intelligence model generation unit 430 performs a preprocessing operation for cleaning data in order to easily learn and analyze the collected big data in real time or at the request of the user device 210.
  • the big data refinement means an operation of detecting and correcting data where a missing value or an error exists in order to correct the inconsistency of the collected data.
  • the big data preprocessing operation may include a filtering operation, a data transformation operation, and a data integration operation.
  • the filtering operation means an operation of removing data that does not fit the purpose of collecting big data in order to shorten learning time and efficiently utilize the storage space.
  • the filtering operation means an operation (i.e., data mining) for eliminating errors or duplication of irregular data.
  • the data conversion operation refers to operations such as switching the unstructured data into a structured form or normalizing the data as a premise of the big data learning.
  • the data integration operation refers to the operation of merging data from multiple sources.
  • the big data collecting unit 410 may store the preprocessed big data in the storage unit 490 or a separate external storage (not shown).
  • the general intelligence model generation unit 430 performs a big data processing operation for learning and analyzing the big data in order to generate an intelligent model from the big data.
  • the general intelligence model generation unit 430 learns big data using various learning algorithms to generate a prediction function from big data or infer pattern information.
  • the learning algorithm may be an algorithm stored in a memory (not shown) of intelligence server 230 or an algorithm received from user device 210.
  • the learning algorithms can be classified into supervised learning algorithms (eg, multilayer neural networks (Perceptron), support vector machines (SVM), kernel machines, decision trees, etc.), unsupervised learning algorithms : K-Means, hierarchical clustering, self-organizing maps (SOM), etc.) and reinforcement learning algorithms.
  • supervised learning algorithms eg, multilayer neural networks (Perceptron), support vector machines (SVM), kernel machines, decision trees, etc.
  • unsupervised learning algorithms K-Means, hierarchical clustering, self-organizing maps (SOM), etc.
  • reinforcement learning algorithms eg., this is merely an example, and the present embodiment is not limited thereto.
  • the general intelligent model generation unit 430 may learn (i.e., perform ensemble learning) of the big data using two or more learning algorithms in order to improve the prediction performance of the intelligent model.
  • the general intelligent model generation unit 430 analyzes the error of the prediction function or the pattern information generated as a result of the big data learning. For example, the general intelligence model generation unit 430 may perform cross-validation to solve an overfitting problem of learning results.
  • the general intelligence model generation unit 430 generates a general intelligence model based on the error function or the pattern information.
  • the general purpose intelligent model may be distributed to the user device 210 as a prediction model for generating predictive data based on new user data.
  • the customized intelligent model generation unit 450 generates,
  • the customized intelligence model generation unit 450 prepares, learns, and analyzes the user profile data received from the user device 210 using the general-purpose intelligence model to generate a customized intelligence model for each user device.
  • the user profile data includes the service purpose and timing of the user device 210, the S / W performance analysis result (for example, the type and version of the operating system framework or library) and the H / W performance analysis result RAM, capacity of the storage device), and the like.
  • the user device 210 Since the user device 210 has a relatively low computing performance compared to the intelligent server 230, when the big data-based general purpose intelligent model is directly applied to the user device 210, degradation ). Since the intelligent service provided by the user device 210 is based on Small Data, when the general-purpose intelligent model is directly applied, resources (e.g., CPU, memory capacity, etc.) Waste more than necessary.
  • resources e.g., CPU, memory capacity, etc.
  • the customized intelligent model generation unit 450 generates the customized intelligent model from the general intelligence model based on the device characteristics, performance, and service information (i.e., user profile data) of each of the user devices 210 ) Create customized intelligence models. Then, the user device 210 generates prediction data for the intelligent service, that is, the small data, using the customized intelligence model distributed from the intelligence server 230, and provides it to the user.
  • the communication unit 470 The communication unit 470
  • the communication unit 470 includes a wireless communication module (LTE, Wi-Fi, etc.), a short-range communication module (such as Bluetooth, ZigBee, etc.) as components necessary for exchanging data with the user device 210 via a wired / (ZigBee), Near Field Communication (NFC), etc.).
  • LTE Long Term Evolution
  • Wi-Fi Wireless Fidelity
  • NFC Near Field Communication
  • the communication unit 470 functions to support a communication connection with a big data source so that the big data collection unit 410 can collect big data.
  • the communication unit 470 then distributes the generated customized intelligence model to the user device 210.
  • the storage unit 490 stores the collected big data or the preprocessed big data.
  • the storage unit 490 may be a relational database (e.g., MS-SQL), a non-relational database (e.g., NoSQL, etc.), or a distributed file system For example, Hadoop DFS, etc.).
  • FIG. 5 is a diagram illustrating a process of generating and distributing a customized intelligent model by the intelligent server shown in FIG.
  • step S510 the big data collection unit 410 collects big data actively or passively from a big data source connected through the communication unit 470.
  • the big data collection unit 410 may store the collected big data in the storage unit 490 or a separate external storage (not shown).
  • step S520 the general intelligent model generation unit 430 performs a pre-processing operation for cleaning data in order to easily learn and analyze the collected big data in real time.
  • the big data preprocessing operation may include a filtering operation, a data transformation operation, and a data integration operation.
  • step S530 the general intelligence model generation unit 430 learns big data in real time using various learning algorithms to generate a prediction function from big data or infer pattern information.
  • the learning algorithm may be an algorithm stored in the storage unit 490 of the intelligence server 230 or an algorithm received from the user device 210.
  • step S540 the general-purpose intelligent model generator 430 analyzes errors of the predictive function or pattern information generated as a result of the big data learning in real time in order to improve the prediction performance of the intelligent model.
  • step S550 the general-purpose intelligent model generation unit 430 generates the general intelligent model on the basis of the error function analyzed or the pattern information in real time.
  • step S560 the customized intelligent model generation unit 450 generates a customized intelligent model based on the device characteristics, performance, and service information (i.e., user profile data) of each user device 210, Generate intelligent models in real time.
  • step S570 the communication unit 470 distributes the customized intelligent model to the user device 210 through the wired / wireless communication connection in real time. Then, the user device 210 generates prediction data for the intelligent service, that is, the small data, using the distributed customized intelligent model, and provides it to the user.
  • FIG. 6 is a diagram illustrating a process in which the user equipment shown in FIG. 3 provides an intelligent service to a user.
  • step S610 the intelligent service unit 310 can receive the customized intelligent model distributed from the intelligent server 230 through the communication unit 350.
  • the intelligent service unit 310 may generate predetermined prediction data by performing prediction on the user data newly received from the user based on the customized intelligence model.
  • the intelligent service unit 310 may provide an intelligent service to the user through a decision process based on the generated predictive data.
  • FIGS. 5 and 6 illustrate that a plurality of processes are performed sequentially, this is merely illustrative of the technical idea of this embodiment. In other words, those skilled in the art will understand that the present invention can be implemented by changing the order described in FIGS. 5 and 6 or by changing some of the plurality of processes in parallel It should be understood that the present invention is not limited to the above-described embodiments.
  • a computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. That is, a computer-readable recording medium includes a magnetic storage medium (e.g., ROM, floppy disk, hard disk, etc.), an optical reading medium (e.g., CD ROM, And the like).
  • the computer-readable recording medium may also be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner.

Abstract

An embodiment provides an intelligent system, for providing a small data-based intelligent service, comprising: an intelligent server for generating an intelligent model from big data and distributing same; and at least one user equipment for generating prediction data from user data, by means of the distributed intelligent model, and providing an intelligent service to a user by means of decision-making on the basis of the prediction data. The intelligent server comprises: a big data collection unit which is for collecting big data from a big data source comprising the user equipment; a general intelligent model generation unit which is for generating a prediction function or pattern information by performing learning by means of a learning algorithm and preprocessing with respect to the collected big data, and is for generating a general intelligent model by analyzing the prediction function or the pattern information; and a personalized intelligent model generation unit which is for extracting a personalized intelligent model from the general intelligent model by means of the respective profile data of each user equipment.

Description

맞춤형 지능시스템 및 그 동작방법Customized intelligent systems and how they work
본 발명은 맞춤형 지능시스템에 관한 것으로, 보다 구체적으로는 사용자 장치 별 맞춤형 지능모델을 이용하여 사용자에게 즉각적이고 능동적인 지능서비스를 제공할 수 있는 맞춤형 지능시스템 및 그 동작방법에 관한 것이다.The present invention relates to a customized intelligent system, and more particularly, to a customized intelligent system capable of providing an immediate and active intelligent service to a user by using a customized intelligent model for each user apparatus and an operation method thereof.
최근 인공지능(Artificial Intelligence: AI) 기술이 발전하면서 다양한 AI 응용기술에 대한 연구개발이 활발하게 진행되고 있으며, 대표적인 AI 응용기술로서 머신러닝(Machine Learning)이 있다.In recent years, research and development on various AI application technologies have progressed actively with the development of artificial intelligence (AI) technology, and machine learning is a typical AI application technology.
머신러닝이란 컴퓨터가 스스로 데이터를 학습하고, 그 학습 결과를 바탕으로 예측 등을 수행하여 유용한 정보를 도출해내는 기술을 말한다. 머신러닝은 일반적으로 학습 데이터의 양과 질이 풍부할수록 예측 정확도가 높아진다. 따라서, 머신러닝은 대규모 데이터를 처리하는 기술인 빅데이터(Big Data)와 밀접한 관계를 갖게 되며, 그 대표적인 예로서 빅데이터를 이용한 머신러닝 기반의 지능서비스(personalized service)를 들 수 있다.Machine learning is a technique in which a computer learns data on its own and performs predictions based on the learning results to derive useful information. Machine learning generally increases the accuracy of prediction as the quantity and quality of learning data is richer. Therefore, machine learning has a close relationship with Big Data, which is a technology for processing large data, and a typical example is machine learning based personalized service using Big Data.
지능서비스란 사용자의 개인정보에 따라 맞춤형 컨텐츠를 제공하는 서비스를 말한다. 예를 들어, 포털 사이트 사용자들의 사용패턴을 분석하여 사용자 별 맞춤형 초기화면을 구성하거나, 검색패턴을 분석하여 다른 사용자들에게 추천 검색어를 제안하는 서비스 등을 들 수 있다. 이와 같은 지능서비스는 지능시스템(intelligent system)에 의해 사용자에게 제공되는바, 도 1을 참조하여 종래기술에 따른 범용(general) 지능시스템에 대해 살펴보기로 한다.An intelligent service is a service that provides customized content according to the user's personal information. For example, a service that constructs a customized initial screen for each user by analyzing usage patterns of portal site users, or analyzes a search pattern and suggests a recommendation word to other users. Such an intelligent service is provided to a user by an intelligent system. Referring to FIG. 1, a general intelligent system according to the prior art will be described.
도 1은 종래기술에 따른 범용 지능시스템을 개략적으로 나타내는 도면이다.1 is a schematic view of a general intelligent system according to the prior art.
도 1을 참조하면, 범용 지능시스템(100)은 사용자 장치(110)와 지능서버(130)를 포함할 수 있다.Referring to FIG. 1, the general purpose intelligent system 100 may include a user device 110 and an intelligent server 130.
사용자 장치(110)는, 사용자에게 지능서비스를 제공하기 위한 매개체로서, 스마트 폰(smart phone)이나 태블릿 PC(tablet PC) 등을 포함할 수 있다.The user device 110 may include a smart phone or a tablet PC as an agent for providing an intelligent service to a user.
지능서버(130)는, 적어도 하나의 사용자 장치(110)에 저장된 대량의 사용자 데이터, 예컨대 서비스 사용 이력 등을 수집하고, 빅데이터 기반의 머신러닝을 수행하여 범용 지능모델을 생성한다. 본 명세서에서, 지능모델이란 빅데이터에 대한 학습을 통해 생성된 예측함수 또는 패턴정보를 포함하는 소프트웨어 플랫폼을 말한다.The intelligence server 130 collects a large amount of user data stored in at least one user equipment 110, for example, service usage histories, and performs machine learning based on the big data to generate a general intelligence model. In this specification, an intelligent model refers to a software platform that includes a prediction function or pattern information generated through learning about big data.
지능서버(130)는 범용 지능모델을 이용하여 예측 데이터, 예컨대 잠재적인 사용자 의도 등을 예측한 정보를 사용자 장치(110)에 배포한다. 그리고, 배포된 예측 데이터는 사용자 장치(110)를 통해 사용자에게 제공된다.The intelligent server 130 distributes the predicted data, for example, information predictive of a potential user's intention, to the user device 110 using the general intelligence model. The distributed prediction data is provided to the user through the user device 110. [
이러한 범용 지능시스템(100)은, 예측 데이터의 생성이 지능서버(130)에 의해 수행되므로, 사용자 접속 트래픽 증가 또는 네트워크 장애로 인하여 실시간 서비스 제공이 지연되고, 서비스가 단절될 수 있다는 문제가 있다.In general-purpose intelligent system 100, generation of predicted data is performed by intelligent server 130, so that real-time service provision may be delayed due to increase in user connection traffic or network failure, and service may be disconnected.
또한, 범용 지능시스템(100)은, 사용자의 요청(즉, 사용자 데이터 전송)이 있는 경우에만 예측 데이터를 생성하여 제공하게 되므로, 즉각적인 서비스 제공이 어렵다는 문제가 있다.In addition, since the general-purpose intelligent system 100 generates and provides the predictive data only when there is a request from the user (i.e., user data transmission), there is a problem that it is difficult to provide the instantaneous service.
본 실시예는 상술한 문제점을 해결하기 위해 창출된 것으로서, 보다 상세하게는 사용자 장치가 지능서버에 의해 생성 및 배포된 맞춤형 지능모델을 이용하여 스몰데이터 기반의 즉각적이고 능동적인 지능서비스를 제공할 수 있게 하는 지능시스템을 제공하고자 한다.The embodiments of the present invention have been made to solve the above-described problems, and in particular, the present invention has been made to solve the above-mentioned problems, and it is an object of the present invention to provide a user data management system capable of providing an immediate and active intelligent service based on small data using a customized intelligent model created and distributed by an intelligent server Intelligent systems that allow them to do so.
본 실시예의 일 측면에 의하면, 스몰데이터(Small Data) 기반의 지능서비스를 제공하는 지능시스템에 있어서, 빅데이터(Big Data)로부터 지능모델을 생성하여 배포하는 지능서버; 및 상기 배포된 지능모델을 이용하여 사용자 데이터로부터 예측 데이터를 생성하고, 상기 예측 데이터에 기초한 의사 결정을 통해 사용자에게 지능서비스를 제공하는 적어도 하나의 사용자 장치를 포함하되, 상기 지능서버는, 상기 사용자 장치를 포함하는 빅데이터 소스로부터 빅데이터를 수집하는 빅데이터 수집부; 상기 수집된 빅데이터에 대해 전처리 및 학습 알고리즘에 의한 학습을 수행하여 예측함수 또는 패턴정보를 생성하고, 상기 예측함수 또는 상기 패턴정보를 분석하여 범용 지능모델을 생성하는 범용 지능모델 생성부; 및 상기 사용자 장치 각각의 프로파일 데이터를 이용하여 상기 범용 지능모델로부터 맞춤형 지능모델을 추출하는 맞춤형 지능모델 생성부를 포함하는 지능시스템을 제공한다.According to an aspect of the present invention, there is provided an intelligent system for providing an intelligent service based on Small Data, comprising: an intelligent server for generating and distributing intelligent models from Big Data; And at least one user device for generating predictive data from the user data using the distributed intelligence model and providing intelligent services to the user through a decision based on the predictive data, A big data collection unit for collecting big data from a big data source including a device; A general intelligence model generation unit for generating a predictive function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data and analyzing the predictive function or the pattern information to generate a general intelligence model; And a customized intelligent model generation unit for extracting a customized intelligent model from the general intelligent model using profile data of each of the user devices.
본 실시예의 다른 측면에 의하면, 지능모델을 생성하여 배포하는 지능서버 및 스몰데이터(Small Data) 기반의 지능서비스를 제공하는 적어도 하나의 사용자 장치를 포함하는 지능시스템의 동작방법에 있어서, 상기 지능서버가, 상기 사용자 장치를 포함하는 빅데이터(Big Data) 소스로부터 빅데이터를 수집하는 단계; 상기 수집된 빅데이터에 대해 전처리 및 학습 알고리즘에 의한 학습을 수행하여 예측함수 또는 패턴정보를 생성하고, 상기 예측함수 또는 패턴정보를 분석하여 범용 지능모델을 생성하는 단계; 상기 사용자 장치 각각의 프로파일 데이터를 이용하여 상기 범용 지능모델로부터 맞춤형 지능모델을 생성하는 단계; 상기 맞춤형 지능모델을 상기 사용자 장치 각각에 배포하는 단계; 및 상기 사용자 장치 각각이, 상기 배포된 맞춤형 지능모델을 이용하여 사용자 데이터로부터 예측 데이터를 생성하고, 상기 예측 데이터에 기초한 의사 결정을 통해 사용자에게 지능서비스를 제공하는 단계를 포함하는 지능시스템의 동작방법을 제공한다.According to another aspect of the present invention, there is provided an operation method of an intelligent system including an intelligent server for generating and distributing an intelligent model and at least one user device for providing an intelligent service based on Small Data, Collecting big data from a Big Data source including the user device; Generating a general function intelligence model by analyzing the prediction function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data to generate a prediction function or pattern information; Generating a customized intelligence model from the general intelligence model using profile data of each of the user devices; Distributing the customized intelligence model to each of the user devices; And each of the user devices generates predictive data from the user data using the distributed customized intelligent model and provides intelligent services to the user through a decision based on the predictive data. .
본 실시예에 따른 지능시스템은, 사용자 장치가 지능서버에 의해 생성 및 배포된 맞춤형 지능모델을 이용하여 스몰데이터 기반의 즉각적이고 능동적인 지능서비스를 제공할 수 있게 하는 효과가 있다.The intelligent system according to the present embodiment is effective in allowing the user device to provide an instant and active intelligent service based on small data using a customized intelligent model generated and distributed by the intelligent server.
도 1은 종래기술에 따른 범용 지능시스템을 개략적으로 나타내는 도면이다.1 is a schematic view of a general intelligent system according to the prior art.
도 2는 본 실시예에 따른 맞춤형 지능시스템을 개략적으로 나타내는 도면이다.2 schematically shows a customized intelligent system according to the present embodiment.
도 3은 본 실시예에 따른 사용자 장치의 구성을 나타내는 도면이다.3 is a diagram showing a configuration of a user apparatus according to the present embodiment.
도 4는 본 실시예에 따른 지능서버의 구성을 나타내는 도면이다.4 is a diagram showing a configuration of an intelligent server according to the present embodiment.
도 5는 도 4에 도시된 지능서버가 맞춤형 지능모델을 생성하여 배포하는 과정을 나타내는 도면이다.5 is a diagram illustrating a process of generating and distributing a customized intelligent model by the intelligent server shown in FIG.
도 6은 도 3에 도시된 사용자 장치가 사용자에게 지능서비스를 제공하는 과정을 나타내는 도면이다.6 is a diagram illustrating a process in which the user equipment shown in FIG. 3 provides an intelligent service to a user.
이하, 본 발명의 일부 실시예들을 예시적인 도면을 통해 상세하게 설명한다. 각 도면의 구성요소들에 참조부호를 부가함에 있어서, 동일한 구성요소들에 대해서는 비록 다른 도면상에 표시되더라도 가능한 한 동일한 부호를 가지도록 하고 있음에 유의해야 한다. 또한, 본 발명의 실시예를 설명함에 있어, 관련된 공지 구성 또는 기능에 대한 구체적인 설명이 본 발명의 요지를 흐릴 수 있다고 판단되는 경우에는 그 상세한 설명은 생략한다.Hereinafter, some embodiments of the present invention will be described in detail with reference to exemplary drawings. It should be noted that, in adding reference numerals to the constituent elements of the drawings, the same constituent elements are denoted by the same reference symbols as possible even if they are shown in different drawings. In the following description of the embodiments of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear.
또한, 본 발명의 구성 요소를 설명하는 데 있어서, 제 1, 제 2, A, B, (a), (b) 등의 용어를 사용할 수 있다. 이러한 용어는 그 구성 요소를 다른 구성 요소와 구별하기 위한 것일 뿐, 그 용어에 의해 해당 구성 요소의 본질이나 차례 또는 순서 등이 한정되지 않는다. 명세서 전체에서, 어떤 부분이 어떤 구성요소를 '포함', '구비'한다고 할 때, 이는 특별히 반대되는 기재가 없는 한 다른 구성요소를 제외하는 것이 아니라 다른 구성요소를 더 포함할 수 있는 것을 의미한다. 또한, 명세서에 기재된 '…부,' '모듈' 등의 용어는 적어도 하나의 기능이나 동작을 처리하는 단위를 의미하며, 이는 하드웨어나 소프트웨어 또는 하드웨어 및 소프트웨어의 결합으로 구현될 수 있다.In describing the components of the present invention, terms such as first, second, A, B, (a), and (b) may be used. These terms are intended to distinguish the constituent elements from other constituent elements, and the terms do not limit the nature, order or order of the constituent elements. Throughout the specification, when an element is referred to as being "comprising" or "comprising", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise . In addition, '... The term "module" refers to a unit that processes at least one function or operation, which may be implemented in hardware, software, or a combination of hardware and software.
이하, 첨부된 도면들을 참조하여 본 발명의 일 실시예들에 대해서 상세하게 설명하기로 한다.Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
도 2는 본 실시예에 따른 맞춤형 지능시스템을 개략적으로 나타내는 도면이다.2 schematically shows a customized intelligent system according to the present embodiment.
도 2를 참조하면, 맞춤형 지능시스템(200)은 사용자 장치(210)와 지능서버(210)를 포함할 수 있다.Referring to FIG. 2, the customized intelligent system 200 may include a user device 210 and an intelligent server 210.
사용자 장치(210)는 지능서버(210)로부터 생성 및 배포된 맞춤형 지능모델을 이용하여 사용자에게 자체적인 지능서비스를 제공한다.The user device 210 provides the user with his / her own intelligence service using the customized intelligence model generated and distributed by the intelligence server 210.
사용자 장치(210)는 사물 인터넷(IOT) 기기, 사용자(110)의 생체 정보(예를 들어, 심박수, 혈압 등)를 측정하기 위한 다양한 종류의 센서를 구비한 전자장치, 스마트 폰(smart phone), 스마트 워치(smart watch), 태블릿 PC(tablet PC) 등과 같은 이동식 디지털 기기, 스마트 TV 등과 같은 고정식 디지털 기기, 스마트 카(smart car), 드론(drone) 등 다양한 형태로 구현될 수 있다.The user device 210 may be an electronic device having various kinds of sensors for measuring the biometric information (e.g., heart rate, blood pressure) of the user IOT device, the user 110, a smart phone, A mobile digital device such as a smart watch, a tablet PC and the like, a fixed digital device such as a smart TV, a smart car, a drone, and the like.
사용자 장치(210)는 빅데이터 소스(source)로서, 사용자 장치(210)에 저장된 데이터는 지능서버(230)에 의해 수집되어 지능모델 생성에 이용된다. User device 210 is a big data source and data stored in user device 210 is collected by intelligence server 230 and used for intelligent model generation.
사용자 장치(210)에 저장된 데이터는, 사용자 데이터 및 사용자 장치의 프로파일 데이터를 포함할 수 있다. 여기서, 사용자 프로파일 데이터는, 사용자 장치(210)의 서비스 목적 및 시기, S/W 성능분석 결과(예: 운영체제 프레임워크나 라이브러리 등의 종류 및 버전) 및 H/W 성능분석 결과(예: CPU, RAM, 저장장치의 용량) 등을 포함할 수 있다.The data stored in the user device 210 may include user data and profile data of the user device. Here, the user profile data includes the service purpose and timing of the user device 210, the S / W performance analysis result (for example, the type and version of the operating system framework or library) and the H / W performance analysis result RAM, capacity of the storage device), and the like.
지능모델은 지능서버(230)에 의해 생성되어 사용자 장치(210)에 배포되며, 사용자 데이터가 추가됨에 따라 사용자 장치(210) 또는 지능서버(230)에 의해 갱신될 수 있다.The intelligent model is generated by the intelligence server 230 and distributed to the user device 210 and may be updated by the user device 210 or the intelligence server 230 as the user data is added.
지능서버(230)에 제공되는 사용자 데이터 및 사용자 장치의 프로파일 데이터는 다양한 암호화 알고리즘(예: AES(Advanced Enctyption Standarad), CMAC(Cipher-based Message Authentication Code) 등)을 이용하여 암호화 및 복호화됨으로써, 보안성이 강화될 수 있다.The user data provided to the intelligent server 230 and the profile data of the user device are encrypted and decrypted using various encryption algorithms (e.g., Advanced Encryption Standard (AES), Cipher-based Message Authentication Code (CMAC) Sex can be strengthened.
또한, 사용자 장치(210)는 메모리(미도시)에 저장된 학습 알고리즘을 지능서버(230)에 전송할 수 있다. 여기서, 메모리는 비휘발성 메모리(예: 낸드 플래쉬 등) 및/또는 휘발성 메모리(예: DRAM 등)를 포함할 수 있다. 이 때, 전송된 학습 알고리즘은 지능서버(230)가 빅데이터를 학습하여 지능모델을 생성하는 데 이용될 수 있다.The user device 210 may also send a learning algorithm stored in a memory (not shown) to the intelligence server 230. Here, the memory may include non-volatile memory (e.g., NAND flash) and / or volatile memory (e.g., DRAM, etc.). At this time, the transmitted learning algorithm can be used by the intelligence server 230 to learn the big data and generate the intelligence model.
지능서버(130)는, 적어도 하나의 사용자 장치(110)에 저장된 대량의 사용자 데이터, 예컨대 서비스 사용 이력 등을 수집하고, 빅데이터 기반의 머신러닝을 수행하여 범용 지능모델을 생성한다.The intelligence server 130 collects a large amount of user data stored in at least one user equipment 110, for example, service usage histories, and performs machine learning based on the big data to generate a general intelligence model.
다음으로, 지능서버(230)는, 사용자 장치(210)로부터 수신된 사용자 프로파일 데이터를 범용 지능모델을 이용하여 전처리, 학습 및 분석함으로써, 사용자 장치(210)별 맞춤형 지능모델을 생성할 수 있다.Next, the intelligence server 230 can generate a customized intelligence model for each user device 210 by preprocessing, learning, and analyzing the user profile data received from the user device 210 using the general purpose intelligence model.
사용자 장치(210)는 지능서버(230)에 비해 통상 하드웨어 및 소프트웨어 성능이 낮으므로, 빅데이터 기반 범용 지능모델을 사용자 장치(210)에 그대로 적용할 경우, 예측 속도가 느려지는 등의 성능 열화(degradation)가 발생할 수 있다. 또한, 사용자 장치(210)에서 제공하는 지능서비스는 스몰데이터(Small Data)를 기반으로 하므로, 범용 지능모델을 그대로 적용할 경우, 사용자 장치(210)의 자원(예: CPU, 메모리 용량 등)이 필요 이상으로 낭비될 수 있다.Since the user device 210 has a lower hardware and software performance than the intelligent server 230, performance degradation, such as slowing down the predicted speed, may occur when the big data-based general purpose intelligent model is directly applied to the user device 210 degradation may occur. Since the intelligent service provided by the user device 210 is based on Small Data, when the general-purpose intelligent model is directly applied, resources (e.g., CPU, memory capacity, etc.) It can be wasted more than necessary.
따라서, 본 실시예에서, 지능서버(230)는 사용자 장치(210) 각각의 기기 특성, 성능 및 서비스 정보 등(즉, 사용자 프로파일 데이터)을 기초로, 범용 지능모델로부터 사용자 장치(210)별 맞춤형 지능모델을 생성하여 배포한다.Therefore, in the present embodiment, the intelligent server 230 acquires, from the general intelligence model, personalized intelligent models based on the device characteristics, performance, and service information of each of the user devices 210 Create and distribute intelligence models.
그리고, 사용자 장치(210)는 배포된 맞춤형 지능모델을 이용하여 지능서비스, 즉 스몰데이터에 대한 예측 데이터를 생성하여 사용자에게 제공하게 된다.Then, the user device 210 generates prediction data for the intelligent service, that is, the small data, using the distributed customized intelligent model, and provides it to the user.
이상을 정리하면, 본 실시예에 따른 맞춤형 지능시스템(200)은, 종래기술에 따른 범용 지능시스템(100)과 달리, 맞춤형 지능모델을 자체적으로 이용하여 사용자에게 즉각적인 능동형 지능서비스를 제공할 수 있다.As described above, the customized intelligent system 200 according to the present embodiment can provide an active active intelligent service to the user by using the customized intelligent model itself, unlike the general intelligent system 100 according to the related art .
다음으로, 도 3을 참조하여, 맞춤형 지능시스템(200)에 포함되는 지능서버의 구성에 대해 살펴보기로 한다.Next, the configuration of the intelligent server included in the customized intelligent system 200 will be described with reference to FIG.
도 3은 본 실시예에 따른 사용자 장치의 구성을 나타내는 도면이다.3 is a diagram showing a configuration of a user apparatus according to the present embodiment.
도 3을 참조하면, 사용자 장치(210)는 맞춤형 지능모델을 이용하여 사용자에게 지능서비스를 제공하기 위해 필요한 구성요소로서, 지능서비스부(310), 통신부(330) 및 메모리(350)를 포함할 수 있다.3, the user device 210 includes an intelligent service unit 310, a communication unit 330, and a memory 350 as components necessary for providing an intelligent service to a user using a customized intelligent model .
지능서비스부(310)The intelligent service unit 310,
지능서비스부(310)는 지능서버(230)로부터 배포된 맞춤형 지능모델을 이용하여, 사용자 데이터로부터 예측 데이터를 생성하여 사용자에게 지능서비스를 제공할 수 있다. 구체적으로, 지능서비스부(310)는 지능서버(230)로부터 배포된 맞춤형 지능모델을 통신부(330)를 통해 수신한다.The intelligent service unit 310 can generate the predictive data from the user data using the customized intelligent model distributed from the intelligent server 230 and provide the intelligent service to the user. Specifically, the intelligent service unit 310 receives the customized intelligence model distributed from the intelligence server 230 through the communication unit 330. [
지능서비스부(310)는 사용자로부터 추가적으로 수신된 사용자 데이터를 전처리, 학습 및 분석하여 지능서버(230)로부터 배포된 맞춤형 지능모델을 갱신할 수 있다. 여기서, 전처리, 학습 및 분석 동작의 구체적인 내용은 도 4를 참조하여 상세하게 후술하기로 한다.The intelligent service unit 310 may update the customized intelligent model distributed from the intelligent server 230 by preprocessing, learning and analyzing the user data received from the user. Here, the detailed contents of the preprocessing, learning and analysis operations will be described later in detail with reference to FIG.
지능서비스부(310)는 사용자로부터 수신된 새로운 사용자 데이터에 대해 맞춤형 지능모델을 이용하여 예측을 수행하여, 소정의 예측 데이터를 생성한다. 그리고, 지능서비스부(310)는 생성된 예측 데이터에 기반한 의사결정 과정을 거쳐, 사용자에게 지능서비스를 제공한다.The intelligent service unit 310 performs prediction using the customized intelligence model for new user data received from the user, and generates predetermined prediction data. Then, the intelligent service unit 310 provides the intelligent service to the user through the decision process based on the generated predictive data.
지능서비스부(310)는 적어도 하나의 프로세서로 구현될 수 있으며, 특히, 지능서비스의 실시간성(real-time cast) 향상을 위해 병렬 프로세싱(parallel processing)을 수행하는 복수의 GPU(Graphics Processing Unit)로 구현될 수 있다.The intelligent service unit 310 may be implemented by at least one processor. In particular, the intelligent service unit 310 may include a plurality of GPUs (Graphics Processing Units) that perform parallel processing to improve the real- . ≪ / RTI >
통신부(330)The communication unit 330,
통신부(330)는 유/무선 통신연결을 통해 지능서버(230)와 데이터를 교환하기 위해 필요한 구성요소로서, 무선통신 모듈(LTE, Wi-Fi 등), 근거리 통신 모듈(블루투스(Bluetooth), 지그비(ZigBee), NFC(Near Field Communication) 등)을 포함할 수 있다. 다만, 이는 예시적인 것이고, 본 실시예가 이에 한정되는 것은 아니다.The communication unit 330 includes a wireless communication module (LTE, Wi-Fi, etc.), a short-range communication module (such as Bluetooth, ZigBee, etc.) as components necessary for exchanging data with the intelligent server 230 via a wired / (ZigBee), Near Field Communication (NFC), etc.). However, this is merely an example, and the present embodiment is not limited thereto.
통신부(330)는 지능서버(230)에 빅데이터를 제공할 수 있도록 지능서버(230)와의 통신연결을 지원하는 기능을 수행한다. 그리고, 통신부(330)는 지능서버(230)로부터 배포된 맞춤형 지능모델을 수신하는 기능을 수행한다.The communication unit 330 functions to support communication connection with the intelligent server 230 so as to provide the intelligent server 230 with the big data. The communication unit 330 performs a function of receiving the customized intelligence model distributed from the intelligence server 230. [
메모리(350)In memory 350,
메모리(350)는 지능모델 생성을 위해 필요한 학습 알고리즘을 저장하기 위하여 필요한 구성요소로서, 비휘발성 메모리(예: 낸드 플래쉬 등) 및/또는 휘발성 메모리(예: DRAM 등)를 포함할 수 있다. 여기서, 메모리에 저장된 학습 알고리즘은 통신부(330)를 통해 지능서버(230)로 전송되어, 지능모델 생성을 위한 빅데이터 학습에 이용될 수 있다.The memory 350 may include non-volatile memory (e.g., NAND flash) and / or volatile memory (e.g., DRAM, etc.) as components necessary to store learning algorithms required for intelligent model generation. Here, the learning algorithm stored in the memory is transmitted to the intelligence server 230 through the communication unit 330, and can be used for big data learning for intelligent model generation.
도 4는 본 실시예에 따른 지능서버의 구성을 나타내는 도면이다.4 is a diagram showing a configuration of an intelligent server according to the present embodiment.
도 4를 참조하면, 지능서버(230)는 사용자 장치(210)에 맞춤형 지능모델을 배포하기 위해 필요한 구성요소로서, 빅데이터 수집부(410), 범용 지능모델 생성부(430), 맞춤형 지능모델 생성부(450), 통신부(470) 및 저장부(490)를 포함할 수 있다.4, the intelligent server 230 includes components such as a big data collection unit 410, a general intelligent model generation unit 430, a customized intelligent model A generating unit 450, a communication unit 470, and a storage unit 490. [
빅데이터 수집부(410)The big data collection unit 410,
빅데이터 수집부(410)는 통신부(470)를 통해 연결된 빅데이터 소스로부터 능동적으로 또는 수동적으로 빅데이터를 수집한다.The big data collecting unit 410 collects big data actively or passively from a big data source connected through the communication unit 470.
여기서, 빅데이터는 데이터 베이스(database), 스프레드시트(spreadsheet) 등의 정형 데이터(structured data), XML, HTML 등의 반정형 데이터(semi-structured data) 및 텍스트 문서, 음성, 영상 등의 비정형 데이터(unstructured data)를 모두 포함할 수 있다.Here, the big data includes structured data such as a database, a spreadsheet, semi-structured data such as XML and HTML, and unstructured data such as text documents, (unstructured data).
그리고, 빅데이터 소스는 지능서버(230) 내부에 존재하는 내부 소스(예: 내부 데이터베이스 관리시스템(DBMS) 등)와 지능서버(230) 외부에 존재하는 외부 소스(예: 사용자 장치, 인터넷 등)를 포함할 수 있다.The big data source may be an internal source (e.g., an internal database management system (DBMS) or the like) existing in the intelligence server 230 and an external source (e.g., a user device, the Internet, etc.) . ≪ / RTI >
빅데이터 수집부(410)는 수집된 빅데이터를 저장부(490) 또는 별도의 외부 저장소(미도시)에 저장할 수 있다.The big data collection unit 410 may store the collected big data in the storage unit 490 or a separate external storage (not shown).
범용 지능모델 생성부(430)The general intelligence model generation unit 430
[전처리(pre-processing)][Pre-processing]
범용 지능모델 생성부(430)는 수집된 빅데이터를 용이하게 학습 및 분석하기 위하여 데이터를 정제(cleaning)하는 전처리 동작을 실시간으로 또는 사용자 장치(210)의 요청에 따라 수행한다. 여기서, 빅데이터 정제는 수집된 데이터의 불일치성을 교정하기 위하여 누락 값(missing value) 또는 오류가 존재하는 데이터를 검출하여 정정하는 동작을 의미한다.The general intelligence model generation unit 430 performs a preprocessing operation for cleaning data in order to easily learn and analyze the collected big data in real time or at the request of the user device 210. [ Here, the big data refinement means an operation of detecting and correcting data where a missing value or an error exists in order to correct the inconsistency of the collected data.
빅데이터 전처리 동작은 필터링(filtering) 동작, 데이터 변환(transformation) 동작 및 데이터 통합(integration) 동작을 포함할 수 있다.The big data preprocessing operation may include a filtering operation, a data transformation operation, and a data integration operation.
구체적으로, 필터링 동작은, 학습시간을 단축하고 저장공간을 효율적으로 활용하기 위하여, 빅데이터 수집목적에 맞지 않는 데이터를 제거하는 동작을 의미한다. 또한, 필터링 동작은 비정형 데이터의 오류나 중복을 제거하는 동작(즉, 데이터 마이닝)을 의미한다.Specifically, the filtering operation means an operation of removing data that does not fit the purpose of collecting big data in order to shorten learning time and efficiently utilize the storage space. In addition, the filtering operation means an operation (i.e., data mining) for eliminating errors or duplication of irregular data.
데이터 변환 동작은, 빅데이터 학습의 전제로서, 비정형 데이터를 구조적 형태로 전환하거나, 데이터를 정규화(normalization)하는 등의 동작을 의미한다.The data conversion operation refers to operations such as switching the unstructured data into a structured form or normalizing the data as a premise of the big data learning.
데이터 통합 동작은 여러 소스의 데이터를 병합하는 동작을 의미한다.The data integration operation refers to the operation of merging data from multiple sources.
빅데이터 수집부(410)는 전처리된 빅데이터를 저장부(490) 또는 별도의 외부 저장소(미도시)에 저장할 수 있다.The big data collecting unit 410 may store the preprocessed big data in the storage unit 490 or a separate external storage (not shown).
[학습(learning)][Learning]
범용 지능모델 생성부(430)는 빅데이터로부터 지능모델을 생성하기 위하여, 빅데이터를 학습 및 분석하는 빅데이터 처리 동작을 수행한다.The general intelligence model generation unit 430 performs a big data processing operation for learning and analyzing the big data in order to generate an intelligent model from the big data.
구체적으로, 범용 지능모델 생성부(430)는, 빅데이터로부터 예측함수를 생성하거나 패턴정보를 추론하기 위하여, 다양한 학습 알고리즘을 이용하여 빅데이터를 학습한다. 여기서, 학습 알고리즘은 지능서버(230)의 메모리(미도시)에 저장된 알고리즘 또는 사용자 장치(210)로부터 수신된 알고리즘일 수 있다. 또한, 학습 알고리즘은 감독 학습(supervised learning) 알고리즘(예: 다층신경망(퍼셉트론), 지지벡터머신(Support Vector Machine: SVM), 커널 머신, 결정 트리 등), 비감독 학습(unsupervised learning) 알고리즘(예: K-Means, 계층적 군집화, 자기조직지도(SOM) 등) 및 강화 학습(reinforcement learning) 알고리즘을 포함할 수 있다. 다만, 이는 예시적인 것이고, 본 실시예가 이에 한정되는 것은 아니다.Specifically, the general intelligence model generation unit 430 learns big data using various learning algorithms to generate a prediction function from big data or infer pattern information. Here, the learning algorithm may be an algorithm stored in a memory (not shown) of intelligence server 230 or an algorithm received from user device 210. In addition, the learning algorithms can be classified into supervised learning algorithms (eg, multilayer neural networks (Perceptron), support vector machines (SVM), kernel machines, decision trees, etc.), unsupervised learning algorithms : K-Means, hierarchical clustering, self-organizing maps (SOM), etc.) and reinforcement learning algorithms. However, this is merely an example, and the present embodiment is not limited thereto.
범용 지능모델 생성부(430)는 지능모델의 예측성능을 향상시키기 위하여, 2개 이상의 학습 알고리즘을 이용하여 빅데이터를 학습(즉, 앙상블(ensemble) 학습)할 수 있다.The general intelligent model generation unit 430 may learn (i.e., perform ensemble learning) of the big data using two or more learning algorithms in order to improve the prediction performance of the intelligent model.
[오류 분석(error analysis)][Error analysis]
범용 지능모델 생성부(430)는 지능모델의 예측성능을 향상시키기 위하여, 빅데이터 학습결과 생성된 예측함수 또는 패턴정보의 오류를 분석한다. 예를 들어, 범용 지능모델 생성부(430)는 학습결과의 과적합(overfitting) 문제를 해결하기 위하여 교차검증(cross-validation)을 수행할 수 있다.In order to improve the prediction performance of the intelligent model, the general intelligent model generation unit 430 analyzes the error of the prediction function or the pattern information generated as a result of the big data learning. For example, the general intelligence model generation unit 430 may perform cross-validation to solve an overfitting problem of learning results.
[범용 지능모델 생성][GENERAL INTELLIGENCE MODEL GENERATION]
범용 지능모델 생성부(430)는 오류 분석된 예측함수 또는 패턴정보를 기초로 범용(general) 지능모델을 생성한다. 여기서, 범용 지능모델은 새로운 사용자 데이터를 기초로 예측 데이터를 생성하기 위한 예측모델로서, 사용자 장치(210)에 배포될 수 있다.The general intelligence model generation unit 430 generates a general intelligence model based on the error function or the pattern information. Here, the general purpose intelligent model may be distributed to the user device 210 as a prediction model for generating predictive data based on new user data.
맞춤형 지능모델 생성부(450)The customized intelligent model generation unit 450 generates,
맞춤형 지능모델 생성부(450)는, 사용자 장치(210)로부터 수신된 사용자 프로파일 데이터를 범용 지능모델을 이용하여 전처리, 학습 및 분석함으로써, 사용자 장치 별 맞춤형 지능모델을 생성한다. 여기서, 사용자 프로파일 데이터는, 사용자 장치(210)의 서비스 목적 및 시기, S/W 성능분석 결과(예: 운영체제 프레임워크나 라이브러리 등의 종류 및 버전) 및 H/W 성능분석 결과(예: CPU, RAM, 저장장치의 용량) 등을 포함할 수 있다.The customized intelligence model generation unit 450 prepares, learns, and analyzes the user profile data received from the user device 210 using the general-purpose intelligence model to generate a customized intelligence model for each user device. Here, the user profile data includes the service purpose and timing of the user device 210, the S / W performance analysis result (for example, the type and version of the operating system framework or library) and the H / W performance analysis result RAM, capacity of the storage device), and the like.
사용자 장치(210)는 지능서버(230)에 비해 상대적으로 컴퓨팅 성능이 낮으므로, 빅데이터 기반 범용 지능모델을 사용자 장치(210)에 그대로 적용할 경우, 예측 속도가 느려지는 등의 성능 열화(degradation)가 발생한다. 또한, 사용자 장치(210)에서 제공하는 지능서비스는 스몰데이터(Small Data)를 기반으로 하므로, 범용 지능모델을 그대로 적용할 경우, 사용자 장치(210)의 자원(예: CPU, 메모리 용량 등)이 필요 이상으로 낭비된다.Since the user device 210 has a relatively low computing performance compared to the intelligent server 230, when the big data-based general purpose intelligent model is directly applied to the user device 210, degradation ). Since the intelligent service provided by the user device 210 is based on Small Data, when the general-purpose intelligent model is directly applied, resources (e.g., CPU, memory capacity, etc.) Waste more than necessary.
따라서, 본 실시예에서, 맞춤형 지능모델 생성부(450)는 사용자 장치(210) 각각의 기기 특성, 성능 및 서비스 정보 등(즉, 사용자 프로파일 데이터)을 기초로, 범용 지능모델로부터 사용자 장치(210)별 맞춤형 지능모델을 생성한다. 그리고, 사용자 장치(210)는 지능서버(230)로부터 배포된 맞춤형 지능모델을 이용하여 지능서비스, 즉 스몰데이터에 대한 예측 데이터를 생성하여 사용자에게 제공하게 된다.Therefore, in the present embodiment, the customized intelligent model generation unit 450 generates the customized intelligent model from the general intelligence model based on the device characteristics, performance, and service information (i.e., user profile data) of each of the user devices 210 ) Create customized intelligence models. Then, the user device 210 generates prediction data for the intelligent service, that is, the small data, using the customized intelligence model distributed from the intelligence server 230, and provides it to the user.
통신부(470)The communication unit 470
통신부(470)는 유/무선 통신연결을 통해 사용자 장치(210)와 데이터를 교환하기 위해 필요한 구성요소로서, 무선통신 모듈(LTE, Wi-Fi 등), 근거리 통신 모듈(블루투스(Bluetooth), 지그비(ZigBee), NFC(Near Field Communication) 등)을 포함할 수 있다. 다만, 이는 예시적인 것이고, 본 실시예가 이에 한정되는 것은 아니다.The communication unit 470 includes a wireless communication module (LTE, Wi-Fi, etc.), a short-range communication module (such as Bluetooth, ZigBee, etc.) as components necessary for exchanging data with the user device 210 via a wired / (ZigBee), Near Field Communication (NFC), etc.). However, this is merely an example, and the present embodiment is not limited thereto.
통신부(470)는 빅데이터 수집부(410)가 빅데이터를 수집할 수 있도록 빅데이터 소스와의 통신연결을 지원하는 기능을 수행한다. 그리고, 통신부(470)는 생성된 맞춤형 지능모델을 사용자 장치(210)에 배포하는 기능을 수행한다.The communication unit 470 functions to support a communication connection with a big data source so that the big data collection unit 410 can collect big data. The communication unit 470 then distributes the generated customized intelligence model to the user device 210. [
저장부(490)In the storage unit 490,
저장부(490)는 수집된 빅데이터 또는 전처리된 빅데이터를 저장하는 기능을 수행한다. 저장부(490)는 관계형 데이터베이스(relational Database, 예를 들어, MS-SQL 등), 비관계형 데이터베이스(non-relational Database, 예를 들어, NoSQL 등) 또는 분산형 파일 시스템(distributed file system, 예를 들어, 하둡(Hadoop) DFS 등) 등으로 구현될 수 있다.The storage unit 490 stores the collected big data or the preprocessed big data. The storage unit 490 may be a relational database (e.g., MS-SQL), a non-relational database (e.g., NoSQL, etc.), or a distributed file system For example, Hadoop DFS, etc.).
이하, 도 5를 참조하여, 맞춤형 지능모델 생성 및 배포과정을 살펴보기로 한다.Hereinafter, a process of creating and distributing a customized intelligent model will be described with reference to FIG.
도 5는 도 4에 도시된 지능서버가 맞춤형 지능모델을 생성 및 배포하는 과정을 나타내는 도면이다.5 is a diagram illustrating a process of generating and distributing a customized intelligent model by the intelligent server shown in FIG.
도 5를 참조하면, 단계 S510에서, 빅데이터 수집부(410)는 통신부(470)를 통해 연결된 빅데이터 소스로부터 능동적으로 또는 수동적으로 빅데이터를 수집한다. 그리고, 빅데이터 수집부(410)는 수집된 빅데이터를 저장부(490) 또는 별도의 외부 저장소(미도시)에 저장할 수 있다.Referring to FIG. 5, in step S510, the big data collection unit 410 collects big data actively or passively from a big data source connected through the communication unit 470. FIG. The big data collection unit 410 may store the collected big data in the storage unit 490 or a separate external storage (not shown).
단계 S520에서, 범용 지능모델 생성부(430)는 수집된 빅데이터를 용이하게 학습 및 분석하기 위하여 데이터를 정제(cleaning)하는 전처리 동작을 실시간으로 수행한다. 여기서, 빅데이터 전처리 동작은 필터링(filtering) 동작, 데이터 변환(transformation) 동작 및 데이터 통합(integration) 동작을 포함할 수 있다.In step S520, the general intelligent model generation unit 430 performs a pre-processing operation for cleaning data in order to easily learn and analyze the collected big data in real time. Here, the big data preprocessing operation may include a filtering operation, a data transformation operation, and a data integration operation.
단계 S530에서, 범용 지능모델 생성부(430)는, 빅데이터로부터 예측함수를 생성하거나 패턴정보를 추론하기 위하여, 다양한 학습 알고리즘을 이용하여 빅데이터를 실시간으로 학습한다. 여기서, 학습 알고리즘은 지능서버(230)의 저장부(490)에 저장된 알고리즘 또는 사용자 장치(210)로부터 수신된 알고리즘일 수 있다.In step S530, the general intelligence model generation unit 430 learns big data in real time using various learning algorithms to generate a prediction function from big data or infer pattern information. Here, the learning algorithm may be an algorithm stored in the storage unit 490 of the intelligence server 230 or an algorithm received from the user device 210. [
단계 S540에서, 범용 지능모델 생성부(430)는 지능모델의 예측성능을 향상시키기 위하여, 빅데이터 학습결과 생성된 예측함수 또는 패턴정보의 오류를 실시간으로 분석한다.In step S540, the general-purpose intelligent model generator 430 analyzes errors of the predictive function or pattern information generated as a result of the big data learning in real time in order to improve the prediction performance of the intelligent model.
단계 S550에서, 범용 지능모델 생성부(430)는 오류 분석된 예측함수 또는 패턴정보를 기초로 범용 지능모델을 실시간으로 생성한다.In step S550, the general-purpose intelligent model generation unit 430 generates the general intelligent model on the basis of the error function analyzed or the pattern information in real time.
단계 S560에서, 맞춤형 지능모델 생성부(450)는 사용자 장치(210) 각각의 기기 특성, 성능 및 서비스 정보 등(즉, 사용자 프로파일 데이터)을 기초로, 범용 지능모델로부터 사용자 장치(210)별 맞춤형 지능모델을 실시간으로 생성한다.In step S560, the customized intelligent model generation unit 450 generates a customized intelligent model based on the device characteristics, performance, and service information (i.e., user profile data) of each user device 210, Generate intelligent models in real time.
단계 S570에서, 통신부(470)는 유/무선 통신연결을 통해 사용자 장치(210)에 맞춤형 지능모델을 실시간으로 배포한다. 그리고, 사용자 장치(210)는 배포된 맞춤형 지능모델을 이용하여 지능서비스, 즉 스몰데이터에 대한 예측 데이터를 생성하여 사용자에게 제공하게 된다.In step S570, the communication unit 470 distributes the customized intelligent model to the user device 210 through the wired / wireless communication connection in real time. Then, the user device 210 generates prediction data for the intelligent service, that is, the small data, using the distributed customized intelligent model, and provides it to the user.
도 6은 도 3에 도시된 사용자 장치가 사용자에게 지능서비스를 제공하는 과정을 나타내는 도면이다.6 is a diagram illustrating a process in which the user equipment shown in FIG. 3 provides an intelligent service to a user.
도 6을 참조하면, 단계 S610에서, 지능서비스부(310)는 지능서버(230)로부터 배포된 맞춤형 지능모델을 통신부(350)를 통해 수신할 수 있다.6, in step S610, the intelligent service unit 310 can receive the customized intelligent model distributed from the intelligent server 230 through the communication unit 350. [
단계 S620에서, 지능서비스부(310)는 맞춤형 지능모델을 기반으로 사용자로부터 새롭게 수신된 사용자 데이터에 대해 예측을 수행하여, 소정의 예측 데이터를 생성할 수 있다.In step S620, the intelligent service unit 310 may generate predetermined prediction data by performing prediction on the user data newly received from the user based on the customized intelligence model.
단계 S630에서, 지능서비스부(310)는 생성된 예측 데이터를 기반으로 의사결정 과정을 거쳐, 사용자에게 지능서비스를 제공할 수 있다.In step S630, the intelligent service unit 310 may provide an intelligent service to the user through a decision process based on the generated predictive data.
이상 도 5 및 도 6에서는, 복수의 과정을 순차적으로 수행하는 것으로 기재하고 있으나, 이는 본 실시예의 기술 사상을 예시적으로 설명한 것에 불과한 것이다. 다시 말해, 본 실시예가 속하는 기술 분야에서 통상의 지식을 가진 자라면, 본 실시예의 본질적인 특성에서 벗어나지 않는 범위에서, 도 5 및 도 6에 기재된 순서를 변경하여 수행하거나 상기 복수의 과정 중 일부를 병렬적으로 수행하는 것으로 다양하게 수정 및 변경하여 적용 가능할 것이므로, 도 5 및 도 6은 시계열적인 순서로 한정되는 것은 아니다.Although FIGS. 5 and 6 illustrate that a plurality of processes are performed sequentially, this is merely illustrative of the technical idea of this embodiment. In other words, those skilled in the art will understand that the present invention can be implemented by changing the order described in FIGS. 5 and 6 or by changing some of the plurality of processes in parallel It should be understood that the present invention is not limited to the above-described embodiments.
한편, 도 5 및 도 6에 도시된 과정들은 컴퓨터로 읽을 수 있는 기록매체에 컴퓨터가 읽을 수 있는 코드로서 구현하는 것이 가능하다. 컴퓨터가 읽을 수 있는 기록매체는 컴퓨터 시스템에 의하여 읽혀질 수 있는 데이터가 저장되는 모든 종류의 기록장치를 포함한다. 즉, 컴퓨터가 읽을 수 있는 기록매체는 마그네틱 저장매체(예를 들면, 롬, 플로피 디스크, 하드디스크 등), 광학적 판독 매체(예를 들면, 시디롬, 디브이디 등) 및 캐리어 웨이브(예를 들면, 인터넷을 통한 전송)와 같은 저장매체를 포함한다. 또한 컴퓨터가 읽을 수 있는 기록매체는 네트워크로 연결된 컴퓨터 시스템에 분산되어 분산방식으로 컴퓨터가 읽을 수 있는 코드가 저장되고 실행될 수 있다.5 and 6 can be implemented as computer-readable codes on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. That is, a computer-readable recording medium includes a magnetic storage medium (e.g., ROM, floppy disk, hard disk, etc.), an optical reading medium (e.g., CD ROM, And the like). The computer-readable recording medium may also be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner.
이상의 설명은 본 실시예의 기술 사상을 예시적으로 설명한 것에 불과한 것으로서, 본 실시예가 속하는 기술 분야에서 통상의 지식을 가진 자라면 본 실시예의 본질적인 특성에서 벗어나지 않는 범위에서 다양한 수정 및 변형이 가능할 것이다. 따라서, 본 실시예들은 본 실시예의 기술 사상을 한정하기 위한 것이 아니라 설명하기 위한 것이고, 이러한 실시예에 의하여 본 실시예의 기술 사상의 범위가 한정되는 것은 아니다. 본 실시예의 보호 범위는 아래의 청구범위에 의하여 해석되어야 하며, 그와 동등한 범위 내에 있는 모든 기술 사상은 본 실시예의 권리범위에 포함되는 것으로 해석되어야 할 것이다.The foregoing description is merely illustrative of the technical idea of the present embodiment, and various modifications and changes may be made to those skilled in the art without departing from the essential characteristics of the embodiments. Therefore, the present embodiments are to be construed as illustrative rather than restrictive, and the scope of the technical idea of the present embodiment is not limited by these embodiments. The scope of protection of the present embodiment should be construed according to the following claims, and all technical ideas within the scope of equivalents thereof should be construed as being included in the scope of the present invention.
CROSS-REFERENCE TO RELATED APPLICATIONCROSS-REFERENCE TO RELATED APPLICATION
본 특허출원은 2017년 11월 23일 한국에 출원한 특허출원번호 제10-2017-0157067호에 대해 미국 특허법 119(a)조(35 U.S.C §119(a))에 따라 우선권을 주장하며, 그 모든 내용은 참고문헌으로 본 특허출원에 병합된다. 아울러, 본 특허출원은 미국 이외에 국가에 대해서도 위와 동일한 이유로 우선권을 주장하며 그 모든 내용은 참고문헌으로 본 특허출원에 병합된다.This patent application claims priority under 35 USC §119 (a) to US Patent Application No. 119 (a), US Patent Application No. 10-2017-0157067, filed on November 23, 2017, All content is incorporated herein by reference. In addition, this patent application claims priority to the countries other than the United States for the same reason as above, and the entire contents of which are incorporated herein by reference.

Claims (7)

  1. 스몰데이터(Small Data) 기반의 지능서비스를 제공하는 지능시스템에 있어서,In an intelligent system that provides an intelligent service based on Small Data,
    빅데이터(Big Data)로부터 지능모델을 생성하여 배포하는 지능서버; 및An intelligent server for generating and distributing intelligent models from Big Data; And
    상기 배포된 지능모델을 이용하여 사용자 데이터로부터 예측 데이터를 생성하고, 상기 예측 데이터에 기초한 의사 결정을 통해 상기 사용자에게 지능서비스를 제공하는 적어도 하나의 사용자 장치를 포함하되,At least one user device for generating predictive data from user data using the distributed intelligence model and providing intelligent services to the user through a decision based on the predictive data,
    상기 지능서버는,Wherein the intelligent server comprises:
    상기 사용자 장치를 포함하는 빅데이터 소스로부터 빅데이터를 수집하는 빅데이터 수집부;A big data collection unit for collecting big data from a big data source including the user device;
    상기 수집된 빅데이터에 대해 전처리 및 학습 알고리즘에 의한 학습을 수행하여 예측함수 또는 패턴정보를 생성하고, 상기 예측함수 또는 상기 패턴정보를 분석하여 범용 지능모델을 생성하는 범용 지능모델 생성부; 및A general intelligence model generation unit for generating a predictive function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data and analyzing the predictive function or the pattern information to generate a general intelligence model; And
    상기 사용자 장치 각각의 프로파일 데이터를 이용하여 상기 범용 지능모델로부터 맞춤형 지능모델을 추출하는 맞춤형 지능모델 생성부를 포함하는And a customized intelligent model generation unit for extracting a customized intelligent model from the general intelligent model using profile data of each of the user devices
    지능시스템.Intelligent system.
  2. 제 1항에 있어서,The method according to claim 1,
    상기 맞춤형 지능모델은 상기 사용자 장치에 의해 상기 사용자 프로파일 데이터와 연동하여 갱신되는Wherein the customized intelligence model is updated by the user device in association with the user profile data
    지능시스템.Intelligent system.
  3. 제 1항에 있어서,The method according to claim 1,
    상기 프로파일 데이터는 서비스 대상 정보와 상기 사용자 장치의 하드웨어 및 소프트웨어 정보를 포함하는Wherein the profile data includes service object information and hardware and software information of the user equipment
    지능시스템.Intelligent system.
  4. 제 1항에 있어서,The method according to claim 1,
    상기 학습 알고리즘은 감독 학습(supervised learning) 알고리즘, 비감독 학습(unsupervised learning) 알고리즘 및 강화(reinforcement) 학습 알고리즘 중 적어도 하나를 포함하고, 상기 사용자 장치 또는 상기 지능서버의 메모리에 저장되는Wherein the learning algorithm comprises at least one of a supervised learning algorithm, an unsupervised learning algorithm, and a reinforcement learning algorithm and is stored in the memory of the user device or the intelligent server
    지능시스템.Intelligent system.
  5. 제 1항에 있어서,The method according to claim 1,
    상기 사용자 장치와의 통신 연결을 지원하는 통신부; 및A communication unit for supporting a communication connection with the user apparatus; And
    상기 수집된 빅데이터, 상기 범용 지능모델 및 상기 맞춤형 지능모델을 저장하는 저장부를 더 포함하는Further comprising a storage unit for storing the collected big data, the general intelligence model, and the customized intelligence model
    지능시스템.Intelligent system.
  6. 제 5항에 있어서,6. The method of claim 5,
    상기 저장부는,Wherein,
    하둡 분산형 파일 시스템(Hadoop Distributed File System: HDFS)인The Hadoop Distributed File System (HDFS)
    지능시스템.Intelligent system.
  7. 지능모델을 생성하여 배포하는 지능서버 및 스몰데이터(Small Data) 기반의 지능서비스를 제공하는 적어도 하나의 사용자 장치를 포함하는 지능시스템의 동작방법에 있어서,A method of operating an intelligent system including an intelligent server for generating and distributing an intelligent model and at least one user device for providing an intelligent service based on Small Data,
    상기 지능서버가,Wherein the intelligent server comprises:
    상기 사용자 장치를 포함하는 빅데이터(Big Data) 소스로부터 빅데이터를 수집하는 단계;Collecting big data from a Big Data source including the user device;
    상기 수집된 빅데이터에 대해 전처리 및 학습 알고리즘에 의한 학습을 수행하여 예측함수 또는 패턴정보를 생성하고, 상기 예측함수 또는 패턴정보를 분석하여 범용 지능모델을 생성하는 단계;Generating a general function intelligence model by analyzing the prediction function or pattern information by performing learning by the preprocessing and learning algorithm on the collected big data to generate a prediction function or pattern information;
    상기 사용자 장치 각각의 프로파일 데이터를 이용하여 상기 범용 지능모델로부터 맞춤형 지능모델을 생성하는 단계;Generating a customized intelligence model from the general intelligence model using profile data of each of the user devices;
    상기 맞춤형 지능모델을 상기 사용자 장치 각각에 배포하는 단계; 및Distributing the customized intelligence model to each of the user devices; And
    상기 사용자 장치 각각이,Each of the user devices comprising:
    상기 배포된 맞춤형 지능모델을 이용하여 사용자 데이터로부터 예측 데이터를 생성하고, 상기 예측 데이터에 기초한 의사 결정을 통해 사용자에게 지능서비스를 제공하는 단계를 포함하는Generating predictive data from the user data using the distributed customized intelligent model and providing intelligent services to the user through a decision based on the predictive data;
    지능시스템의 동작방법.A method of operating an intelligent system.
PCT/KR2017/013509 2017-11-23 2017-11-24 Personalized intelligent system and method for operating same WO2019103199A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR10-2017-0157067 2017-11-23
KR1020170157067A KR20190060021A (en) 2017-11-23 2017-11-23 Customized intelligent system and operation method thereof

Publications (1)

Publication Number Publication Date
WO2019103199A1 true WO2019103199A1 (en) 2019-05-31

Family

ID=66630598

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2017/013509 WO2019103199A1 (en) 2017-11-23 2017-11-24 Personalized intelligent system and method for operating same

Country Status (2)

Country Link
KR (1) KR20190060021A (en)
WO (1) WO2019103199A1 (en)

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11797550B2 (en) * 2019-01-30 2023-10-24 Uptake Technologies, Inc. Data science platform
KR102241311B1 (en) * 2019-06-20 2021-04-16 경상국립대학교산학협력단 Autonomous moral judgment 0f artificial intelligence and its implementation system
KR102348174B1 (en) * 2019-08-09 2022-01-06 재단법인 아산사회복지재단 Method, program and system for learning in a volatile area based on big data
KR102230991B1 (en) * 2019-09-24 2021-03-23 허해연 Method for providing motion and interest patterns identifying service based on artificial intelligence and internet of things for customers behavior analysis and supply chain management
KR102345410B1 (en) * 2019-11-19 2021-12-30 주식회사 피씨엔 Big data intelligent collecting method and device
KR102283523B1 (en) * 2019-12-16 2021-07-28 박병훈 Method for providing artificial intelligence service
KR102366153B1 (en) * 2020-08-20 2022-02-23 한국전력공사 Fast ensembel inference in machine learning for real-time application and machine learning method accelerated ensembel inference

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20120033597A (en) * 2010-09-30 2012-04-09 성균관대학교산학협력단 Apparatus and method for recognizing user future context
KR20130035660A (en) * 2011-09-30 2013-04-09 주식회사 케이티 Recommendation system and method
JP2016048417A (en) * 2014-08-27 2016-04-07 石井 美恵子 Service providing system and program
KR20170009991A (en) * 2014-08-26 2017-01-25 구글 인코포레이티드 Localized learning from a global model
KR101787613B1 (en) * 2017-01-20 2017-11-15 주식회사 더디엔에이시스템 Artificial intelligence platform system by self-adaptive learning technique based on deep learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20120033597A (en) * 2010-09-30 2012-04-09 성균관대학교산학협력단 Apparatus and method for recognizing user future context
KR20130035660A (en) * 2011-09-30 2013-04-09 주식회사 케이티 Recommendation system and method
KR20170009991A (en) * 2014-08-26 2017-01-25 구글 인코포레이티드 Localized learning from a global model
JP2016048417A (en) * 2014-08-27 2016-04-07 石井 美恵子 Service providing system and program
KR101787613B1 (en) * 2017-01-20 2017-11-15 주식회사 더디엔에이시스템 Artificial intelligence platform system by self-adaptive learning technique based on deep learning

Also Published As

Publication number Publication date
KR20190060021A (en) 2019-06-03

Similar Documents

Publication Publication Date Title
WO2019103199A1 (en) Personalized intelligent system and method for operating same
Mehdipour et al. FOG-Engine: Towards big data analytics in the fog
Ahmed et al. 6G enabled federated learning for secure IoMT resource recommendation and propagation analysis
Ullah et al. FoG assisted secure De-duplicated data dissemination in smart healthcare IoT
WO2021054514A1 (en) User-customized question-answering system based on knowledge graph
Valerio et al. A communication efficient distributed learning framework for smart environments
WO2023068795A1 (en) Device and method for creating metaverse using image analysis
CN114817739B (en) Industrial big data processing system based on artificial intelligence algorithm
Esha et al. Trust IoHT: A trust management model for internet of healthcare things
Barceló-Armada et al. Amazon Alexa traffic traces
WO2021107444A1 (en) Knowledge graph-based marketing information analysis service provision method, and device therefor
Li et al. Research and application of AI in 5G network operation and maintenance
WO2021215551A1 (en) Blockchain-based electronic research note verification method and electronic research note management apparatus using same
WO2023106504A1 (en) Method, device, and computer-readable recording medium for machine learning-based observation level measurement using server system log, and for risk level calculation according to same measurement
US11704222B2 (en) Event log processing
Shih et al. Implementation and visualization of a netflow log data lake system for cyberattack detection using distributed deep learning
WO2021107446A1 (en) Apparatus and method for providing knowledge graph-based marketing analysis chatbot service
CN113691390A (en) Cloud-end-coordinated edge node alarm system and method
Maharajan et al. Membrane computing inspired protocol to enhance security in cloud network
WO2018216828A1 (en) Energy big data management system and method therefor
Tang et al. Federated learning of user mobility anomaly based on graph attention networks
Luksha et al. Method for filtering encrypted traffic using a neural network between an Industrial Internet of things system and Digital Twin
Benqdara et al. Ensemble of clustering algorithms for anomaly intrusion detection system
Senapati et al. A federated learning based connected vehicular framework for smart health care
Wang et al. Quality Control in the Clinical Medical Laboratory Based on Mobile Medical Edge Computing

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17932852

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17932852

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 17932852

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 200121)

122 Ep: pct application non-entry in european phase

Ref document number: 17932852

Country of ref document: EP

Kind code of ref document: A1