Face-to-face detection by infrared-ray communication is effective in grasping interaction between persons. However the problem here is that infrared rays have a high directivity and detection fails unless the persons face each other right in front. Sensor signals having a high directivity and sensor signals having a low directivity are obtained from a sensor terminal (TR) carried by a person. Firstly, information on relative position is obtained with a sensor (TRIR) of infrared rays or the like having a high directivity and an initial group is formed at an application server (AS). A feature amount such as sound that has a low directivity and can sense surrounding environmental information is extracted from among the terminals (TRs) belonging to the initial group by personal feature extraction (ASIF), correlation with terminals (TRs) not belonging to a group is obtained, and thereby whether or not those terminals (TRs) belong to an identical group is judged.