Abstract:
PURPOSE: A real-time prediction method of a moving path of a user by using a smart terminal, and a system thereof are provided to predict a future moving path desired by the user in real time from the moving pattern of the user by using various sensing data collected from a smart terminal and learning about the moving path of the user. CONSTITUTION: A sensing data processing part (120) divides and classifies collected sensing data into routine data according to predetermined classification and stores the data time-serially. A major place extraction part (140) extracts multiple major places for the user of a smart terminal from the routine data by using the number and duration of visits. A major path extraction part (150) extracts major paths among the multiple major places or passing through at least one major place. An inference model construction part (170) constructs an inference graph model by reflecting the current location, and the hierarchical relation and time-serial relation between the multiple major places and the major paths. [Reference numerals] (110) Sensing part; (120) Sensing data processing part; (130) Log DB; (140) Major place extraction part; (150) Major path extraction part; (160) Behavior recognition unit; (170) Inference model construction part; (180) Learning unit
Abstract:
PURPOSE: A method and a system for supplying a condition recognition service are provided to recognize a current condition of a user and to supply recommendation information to a user. CONSTITUTION: A mobile terminal(110) includes a plurality of sensors. The mobile terminal offers sensing data collected from a plurality of sensors. A condition recognition server(120) generates a condition model based on the sensing data. The condition recognition server deduces a condition of the mobile terminal based on the condition model. The condition recognition server supplies recommendation information related to the deduced condition.
Abstract:
A method for estimating a moving path of a user in real time is performed in a real-time estimation system of the moving path of the user, which can be connected to a sensor capable of acquiring sensing data including GPS signals. The method for estimating the moving path of the user in real time comprises the steps: (a) generating and storing road information and behavior information from the sensing data collected by the sensor as time series; (b) extracting multiple key locations and key paths which can be expressed as a daily pattern of the user based on at least one among the collected sensing data, road information, and behavior information; (c) building a probability graph model including switches respectively associated with the extracted multiple key locations and key paths; and (d) learning the built probability graph model and estimating at least one among a moving location and a path of the user based on the learned model and the sensing data inputted from the sensor.
Abstract:
본 출원은 이동체의 이동 경로 예측 기술에 관한 것으로, 개시된 기술에 따른 이동 경로 실시간 예측 방법은 위치 정보를 획득할 수 있는 센서를 포함하는 스마트 단말에서 수행된다. 상기 이동 경로 실시간 예측 방법은 (a) 복수의 센서부에서 수집된 센싱 데이터의 적어도 일부를 일상 데이터로서 구분하고 시계열적으로 저장하는 단계, (b) 상기 시계열적으로 저장된 일상 데이터에 대하여, 사용자에게 일상적인 패턴으로서 표현될 수 있는 복수의 주요 장소들 및 주요 경로들을 추출하는 단계, (c) 현재 위치, 상기 복수의 주요 장소들 및 주요 경로들 사이의 계층적 연관 관계로 이루어지는 추론 모델을 구성하는 단계 및 (d) 상기 구성된 추론 모델의 적어도 일부 그래프에 대하여 학습을 수행하고, 학습된 데이터를 기초로 현재 일상 데이터로부터 예측되는 이동 경로 또는 이동 장소를 추출하는 단계를 포함한다. 본 출원의 개시된 기술에 따르면, 스마트 단말에서 수집되는 다양한 센싱 데이터 및 사용자의 이동 경로에 대한 학습을 이용하여 사용자의 이동 현황으로부터 사용자가 이동하려는 장래적 이동 경로를 실시간으로 예측할 수 있는 효과가 있다.