A regular edition video of a video title is split into shots or scenes with time codes, and provided information, which is the semantic evaluation of the story, is added to the respective scenes to organize the scene score. Necessary scenes for each purpose are extracted on the basis of the scene score and a threshold value (process Pr14). Video characteristic evaluation information is added for each shot constituting each of the extracted scenes so as to organize the shot score (process Pr16). Optimum shots for each of the extracted scenes are selected on the basis of a predetermined rule suitable for the purpose (process Pr16). The optimum shots are sequentially cut out from the regular edition video (process Pr21), thus automatically organizing a preview video suitable for the purpose (process Pr22).