| With the rapid development of Internet, the network has gone deep into many areas oflives, become indispensable communication tools, information-accessed tools andentertainment tools to people. Flash Movies as a new kind of multimedia format, is a vitalcomponent of the Internet. With the advantages of strong artistic expression, simple producting,flexible interaction, small file, convenient communicate in network and many other advantages,it is widely used in the game, animation, MTV, advertising, instructional courseware, but also asa document, PPT and video containers. Its number is growing rapidly, and it is becoming theimportant multimedia information resources of network. Flash Movies resources play anincreasingly important role in the field of education, on the one hand, the development ofInternet technology makes publishing and sharing of these Flash resources are no longerconstraints to time and space, it is an important way for us to access those resources; but on theother hand, the huge rich and dynamic update of the network resources, so that find the FlashMovies resources we need has become increasingly difficult.Flash Movies as an important aspect of the construction of educational resources is theresearch field of educational technology. With the growing number of Flash Movies in thenetwork, how to manage to make people able to find the resources they want accurately andeasily is become a noticeable problem. The effective way to solve this problem is classification,category management and search can improve the efficiency of the managementgreatly.Therefore, by studying the content features of the FlashMovies and the classificationalgorithms, aims to establish a classification system based on the contents of Flash Movies,achieve to manage the Flash Movies classified and automatically. The classification systemselected 14 categories features to identify a type of Flash Movies, these features are file size,deformation number, the number of graphics, text number, voice number, number of buttons,the number of movie clips, script, frames, game matching, animation matching, MTV matching,courseware matching and ad matching, in which the last five are text features. This selectionmethod not only include content features of Flash Movies, but also turn text features incorporated into the classification system, it absorbs the advantages of both. The corealgorithm of this classification system uses BP neural network algorithm. BP neural network isa multilayer feedforward network that learning with a error back propagation algorithm, it isone of the most widely used neural network model. BP neural network is a simple imitation ofthe human brain, but it still has two key features: First, the artificial neural network with thehuman brain, are complicated network of highly connected by a number of computable units.Second, the connection between the unit and the mode of information processing determine thefunction of the network. BP neural network learning to get the parameters of the neurons,including the weights and threshold, by training samples, which is learning process of theneural network. Finally, put the value of the features into the BP neural network which afterlearning, and calculated its category.The work and research results of this study include the following aspects:(1) Search and download the Flash Movies from Internet, and build a resources database ofFlash Movies, analyzed the characteristic elements and attributes of the Flash Movies in thedatabase.(2) Extract the Keywords feature vectors of Flash Movies, computing similarity of FlashMovies with each type and bring similarity in the category characterization of Flash Movies.(3) Design the algorithm which suitable for the classification of the Flash Movies.(4) Research and analysis on the improvement of the BP-neural network algorithm.(5) On the basis of theoretical studies, we completed a classification system ofcontent-based Flash Movies, and applied this system to the resource management platform ofthe Flash Movies. |
PDF Full Text Request