Many media forms can be stored easily at present. Photographs, for example, can be easily stored even though most of them have not been edited. This means they will gradually lose their value and become essentially unusable. To make better use of photographs, we tried to make use of information provided by viewers who had seen and commented on them. We felt that analyzing this information would enable us to make maximum use of photographic data. To do this, we defined a ''tag propagation'' model and relationships between photos. We also proposed a system that uses image processing to analyze viewers' handling of photos and how the photos are relevant to each other. We then validated our model by using it.