Design And Implementation Of Streaming Media System Based On Popularity

With the rapid development of Internet technology and the continuous enrichment of network applications,streaming media technology has received extensive attention.Different from the traditional media transmission and playback methods,the streaming media technology adopts the strategy of downloading while playing,which greatly reduces the user's waiting time.However,with the continuous improvement of the response time requirements of users,the adoption of streaming media technology alone cannot satisfy the needs of users.As a result,popularity prediction technology has emerged.By predicting the popularity of the video,multi-rate transcoding and buffering of the video of interest to the user in advance can further reduce the user access delay and improve the user experience.Therefore,based on the research of domestic and foreign video popularity prediction algorithms,this paper improves a popularity prediction method based on user preference,proposes a prediction model based on social and search data,and implements a popularity-based video distribution.system.The specific work is as follows:(1)Introduced the latest streaming media related transmission protocols,video publishing system technology framework,and popularity prediction technology theory,algorithms,etc.(2)Improving a popularity prediction method based on user preference and proposes BMF-Score and BMF-Link algorithm.This method first uses matrix decomposition algorithm to mine user's interest preference for video,and then statistically analyzes it to predict the on-demand popularity of video.Use the data set to verify the algorithm and compare it with related algorithms.The results show that the accuracy of BMF-Score and BMF-Link methods has improved..(3)Using Weibo data and search index to predict the popularity of the video.By crawling the real data in the analysis network,the relationship between the two different social networks of microblog data and search index and the popularity of the video is discovered,and a popularity prediction model SSPM that combines microblog and search data is proposed.The model uses correlation analysis to determine factors that are significantly related to popularity,and uses multiple linear regression algorithms to predict video popularity.Experiments show that the SSPM model can accurately predict the popularity of the video.(4)Design and implement a popularity-based video on demand system.The server uses Nginx and uses the FFmpeg tool to transcode video.Based on the SSH framework,it uses the My SQL database and uses the high-performance non-relational database Redis as the system cache.At the same time,the system uses the prediction model SSPM to predict the popularity of the movie,and periodically replaces the cached content according to the prediction result.By testing the system,the system can respond to user requests more quickly,further improve system performance and has better practical value.