A context of media content is represented by context description data having a hierarchical stratum. The highest hierarchical layer is formed from a single element representing content. The lowest hierarchical layer is formed from an element representing a segment of media content which corresponds to a change between scenes or audible tones. The remaining hierarchical layers are formed from an element representing a scene or a collection of scenes. A score corresponding to the context of a scene of interest is appended, as an attribute, to each of the remaining hierarchical layers. A score relating to time information and a context is appended, as an attribute, to individual elements in the lowest hierarchical layer. In a selection step, the context of the media content is expressed, and one or more scenes are selected based on the score. In an extraction step, only data pertaining to the selected scenes are extracted.