Classifying video scenes into semantic categories is a research problem of great interest. This process is usually handled using supervised classifiers based on features describing the global distribution of color, texture or edge information of the video image. Obviously, the speech of the video scene contains more detailed semantic information and the video transcript is now available through Closed-caption text. Furthermore, the analysis of the speech transcript of the video scenes is easier than video image analysis. Little work has been done using English closed caption text but Arabic closed-caption text is not even as fortunate. This paper proposes a framework for the classification of video scenes using Arabic Closed-Caption.