Provided is a multimedia data generation method which includes allocating one or more pieces of image object information to each of a plurality of nodes, generating a scene tree by linking nodes according to correlation between the object information allocated to each node, and adding an event execution level attribute to each node, wherein the event execution level attribute determines whether to execute an event by using the image object information allocated to each node, thereby generating multimedia data in which the added event execution level attribute is represented with a scene description language and recorded on an area of a scene descriptor.