摘要 |
<P>PROBLEM TO BE SOLVED: To retrieve time-series data similar to attention data in a time-series data group while reducing calculation cost. <P>SOLUTION: Acceleration data composed of time-series data collected from a user terminal is grouped for each measurement period and stored in a sensor data storage unit 41. In this state, a symbolization processing unit 32 converts the sensor data into a symbol series represented in a lower dimension, and a representative character string extraction unit 33 obtains a reference segment character string in which a symbol appears most often on the basis of the symbol series and sets the reference segment character string as a representative character string. Under the control of a retrieval data setting unit 34, a similarity calculation unit 35 calculates similarity of the representative character string of each of other groups to the representative character string of a group selected by a user, and a retrieval result output unit 36 displays sensor data included in a group in which the similarity is equal to or less than a threshold on a display device 22. <P>COPYRIGHT: (C)2012,JPO&INPIT |