The Study Of Content Recognition Based On Clustering Algorithm And The Realization Of Video Teaching System

This thesis built a video teaching system by meeting schools' needs in improving students' learning abilities in practical classes.The system was designed based on streaming media techniques,improved the existing K-MEANS clustering algorithm and designed the video-mining module.The system also helps students choose the corresponding internet teaching content that's are related,get the desired video list,and manage the data stream mining of sensitive teaching contents.Regarding the research methods,this article intro duced the content identification techniques that are based on clustering algorithm,which enriched the streaming media internet video teaching system and provided a more effective algorithm for illegal video identification system designer.The current mult imedia search and identification method has gradually developed from the keyword search to the object search,via inserting a part of suitable video content into a bulk of data information,and then find out the video file that the user wants or alike.Traditional search methods are flawed in many ways,such as the searching unit only based on traditional visual words,lack of correlation in time series of video frames,and not able to find the desired content.Therefore,when the author studied the school log data,some new monitoring rules about the video teaching contents in the website was found,based on which in this article the author presents some improvements in the existing K-MEANS clustering algorithm(named K-MEANS algorithm)based on her new findings.Compared with the improved K-MEANS clustering algorithm in theory,the old one has some flaws in sensitive teaching content recognition efficiency and speed when processing large data under the environment of K-MEANS clustering algorithm in data stream video website content.The advantage of the improved K-MEANS clustering algorithm is that it takes time series of video frames into consideration and maintains an acceptable search speed.And then,the thesis conducted further empirical study on the proposed innovative methods,delving into the given database,and designed three searching systems: the internet teaching content mining,the mining management module of sensitive teaching content video data stream and recognition module of highly suspected students of misdeed.At last,the thesis quantitatively analyzed the acquired results,tested the designed system,and drew an conclusion.The inherent law of streaming media content deep-mining found in the article is of great significance for schools in formulating and promoting the existing mining strategy.As the test results show in the article,the proposed algorithm and solutions are feasible and effective,and can achieve the desired effect.