System for Recommendation of APP User Terminal and Method Therefor

Disclosed is an app recommendation service system including a user terminal connected to an app recommendation service server via a communication network. According to one embodiment of the present invention, a user terminal includes a sensor information processing unit for generating time-series data of sensor information vectors from sensor information provided by a plurality of sensors; a situation estimation unit for estimating a specified number of situations using an unsupervised learning method based on the time- series data of a sensor information vector, generating the apps executed in each of the specified number of situations and the frequency of the executed apps by an app recommendation table, and estimating a probability of a specific situation among the specified number of situations based on the time-series data of the sensor information vector; a time point estimation unit for learning an app recommendation time point by estimating the probability of a moment that a user wants to execute an app based on the time-series data of the sensor information vector or the time-series data of vector having a dimension corresponding to a specified number having a probability value of each situation and a preceding event before a specific app is executed, and determining as an app recommendation time point if the probability value of the moment that the estimated app is to be executed exceeds a predetermined threshold value, wherein the preceding event is configured to include a time point when an operation of putting up the user terminal is detected or a time point when the state that the screen of the user terminal is turned is detected, and a time point when the predetermined app is executed; and an app recommendation unit for detecting, as a specific situation among the specified number of situations is recognized, at least one execution candidate app corresponding to the recognized specific situation by referring to the app recommendation table, and outputting the at least one execution candidate app through a user interface when determined as the app recommendation time point. 본 기술의 일 실시예에 의한 앱 추천 서비스 시스템은 통신망을 통해 앱 추천 서비스 서버와 접속되는 사용자 단말기를 포함하는 앱 추천 서비스 시스템으로서, 사용자 단말기는, 복수의 센서, 복수의 센서에서 제공되는 센서정보로부터 센서정보벡터의 시계열 데이터를 생성하는 센서정보 처리부,센서정보벡터의 시계열 데이터를 입력으로 한 비지도 학습 방법을 이용하여 지정된 개수의 상황을 추정하며, 지정된 개수의 상황 각각에서 실행된 앱 및 실행된 앱의 빈도를 앱 추천 테이블로 생성하고, 센서정보벡터의 시계열 데이터에 기초하여 지정된 개수의 상황 중 특정 상황일 확률을 추정하는 상황 추정부, 센서정보벡터의 시계열 데이터 또는 각 상황의 확률값을 갖는 지정된 개수에 대응하는 차원을