A semiotic semi-automatic annotation for movie audiovisual document

The huge need of users to obtain a better and reliable description highlights the growth of interest of movie description. But, the lack of semantic description always persists. This lacks is mainly due to the failure of the semantic descriptions use extracted from the content. In this context, we therefore concentrate efforts to propose a mechanism to overcome this problem which has not been well treated yet. In fact, the semiotic descriptions represent an important source of information of movie document. In the herein presented work, we achieve a description level that combines two types of analysis such as the statistic and essentially the semantic. It consists in using an open (manual) description in order to extract the semiotic description.