Research On Content Structure Characteristics And Retrieval Of Network Flash Animation Learning Resources

Educational informatization is an important means to improve the quality of education,and also the basic condition for innovative application of education.Educational informatization cannot be separated from the construction of digital learning resources.As one of the types of digital learning resources,Flash animation is not only an important medium to express information content,but also an important network learning resource.Its content consists of text,graphics,images,audio,video,interaction,dynamic effects and other content.Because of its powerful multimedia interaction and performance ability,Flash animation is widely used in distance education,excellent course website,MOOC platform and other fields.A large number of Flash animation resources are accumulated on the network,which brings a lot of interference to the retrieval of animation demanders.How to get the Flash animation quickly and accurately is the problem that Flash search engine needs to solve.At present,Flash retrieval is generally based on key words,metadata features or web contexts,and the retrieval accuracy is not ideal.So people have carried out in-depth analysis and Research on the content characteristics of Flash animation.The topic of this study is based on the SWF file organization structure,which analyzes the content structure features of Flash animation,such as scene structure features,component element features and image emotion features.In this paper,according to the four-tier framework of semantic extraction of Flash animation(namely metadata,constituent elements,scene,emotion layer),several key technologies such as scene feature extraction,constituent element feature extraction,image emotion feature extraction are studied respectively.The significance of our study is to provide fast and accurate Flash animation retrieval engines for educators,network self-learners and Flash animation enthusiasts,so as to improve the efficiency of educational application of network Flash animation learning resources and give full play to its educational characteristics.Firstly,the definition of learning resources of network Flash animation is given,then thecontent structure feature description model and scene structure model of Flash animation are analyzed and established,and the segmentation algorithm of scene and the extraction process of scene features are given.Secondly,we complete the feature extraction of component elements.Thirdly,we build the emotion classification model of Flash animation,and use neural network to learn the mapping relationship between low-level visual features(mainly color and texture)and high-level emotion semantics,so as to complete the emotion classification of Flash animation,and compare the emotion classification model with SVM and deep learning.The research result of this paper is that the scene features,element features and emotion features extracted in the earlier stage are finally stored in the index database,and a content-based Flash retrieval system is established for the Flash animation retrieval of network users.Based on this database,through experiments,the researchers also use the gray correlation method to verify the relevance between the content structure features of Flash animation and learners' learning interests.The results show that the dynamic effect feature of Flash animation has the highest correlation with learning interest,which plays an important role in stimulating learners' interest and concentrating learners' attention.In the Flash animation of different stages and subjects,the visual features that play a key role in arousing learners' interest are inconsistent.The experimental results can provide theoretical guidance for Flash courseware creators to select visual features according to different stages and subjects.Based on the crawling program of network animation developed by previous researchers,we downloads a large number of Flash animation,and selects 4808 Flash animation learning resources which have obvious educational characteristics and can assist knowledge learning as the sample library of this study.Referring to education theory and literature review,this paper divides these 4808 samples according to subjects and learning stages,and analyzes the extracted visual scene,element features and emotional features according to different subjects and stages to obtain the characteristics of Flash animation in different stages and subjects,which provides guidance for the automatic classification of Flash animation in the later stage.The innovations of the paper are as follows.This paper established the content structure feature description model of Flash animation,and analyzed the content structure features ofnetwork Flash animation learning resources from three dimensions: learning stage,subject and teaching type;established a scene structure model and proposed a segmentation algorithm based on the combination of color histogram and edge density;established the emotion classification model,and completed the emotion semantic recognition of Flash animation based on BP neural network,SVM and CNN respectively;analyzed the relevance of the content structure features of network Flash animation learning resources to students' learning interests.