Modeling Of Multimedia Files On The Web 2

Loading...

Flash Player 9 (or above) is needed to view presentations.
We have detected that you do not have it on your computer. To install it, go here.

0 comments

Post a comment

    Post a comment
    Embed Video
    Edit your comment Cancel

    Favorites, Groups & Events

    Modeling Of Multimedia Files On The Web 2 - Presentation Transcript

    1. MODELING OF MULTIMEDIA FILES ON THE WEB 2.0 陳泳宏
    2. Reference
      • Mojgan Soraya, Masood Zamani and Abdolreza Abhari
      • Electrical and Computer Engineering, 2008. CCECE 2008. Canadian on 4-7 May 2008 Page(s):001387 - 001392
    3. Outline
      • Introduction
      • Data Collection Strategy
      • Video File Characteristics
      • Zipf’s Analysis
      • Metadata Analysis
      • Conclusion
    4. Introduction
      • To understand the workload characteristics.
      • To obtain statistics from the popular videos of Youtube to exam the possibility of effective caching method.
      • Metadata may be used to make more effective caching strategies.
    5. Caching
      • 原理
      • 當 CPU 處理數據時,它會先到高速緩存中去尋找,如果數據因之前的操作已經讀取而被暫存其中,就不需要再從主內存中讀取數據,其目的是為了讓數據存取的速度適應 CPU 的處理速度。
      • 概念擴充
      • 凡是位於 速度相差較大 的兩種硬體之間的,用於協調兩者數據傳輸速度差異的結構,均可稱之為 Cache 。
    6. Data Collection Strategy
      • Method
      • Data Statistics
      • Analysis
    7. Method
      • Focus on Top 100 most viewed videos of the day and week.
      • 54 consecutive days(2007/9/13 ~ 2007/11/5)
      • Developed a software in java to retrieve the top 100 videos.
      • The result is an XML file include video ID, duration, rating average, rating count, view count…etc.
    8. Videos Statistics
    9. Analysis
      • The changes in daily list of the top 100 videos are quiet often.
      • The popular videos in both daily and weekly data set have high rating.
      • The COV (coefficient of variation) of the rating is low.
      • The popular videos have durations well below the Max.
      • of 10 minutes.
      • The file sizes are large, and not highly variable.
    10. Video File Characteristics
      • File Size
      • Video Duration
      • Rating of Videos
    11. File Size
      • 90% of the videos requested by users are less than 23.5 MB.
      • The files sizes are big. The more storage space are required.
      • Disk based caching should be used in this case.
    12. CDF of File Sizes
    13. Video Duration
      • 56% of daily videos and 64% of weekly videos are between 1 and 5 minutes long.
      • The popular videos with longer time popularity are shorter than others.
    14. Rating of Videos
      • Youtube rating system is used to rate the videos based on a scale of 0-5 “stars”.
      • Over 95% of the time the average rating is 3 or higher.
    15. Zipf’s Law(1/2)
      • 齊夫定律 可以表述為,在自然語言的語素庫里 , 一個單詞出現的 頻率與它在頻率表裡的排名成反比 。頻率最高的單詞出現的頻率大約是出現頻率第二位的單詞的 2 倍,而出現頻率第二位的單詞則是出現頻率第四位的單詞的 2 倍。
      • 是一個實驗定律,而非理論定律。
      • 齊夫定律很容易用 點陣圖 觀察,坐標為 log( 排名 ) 和 log( 頻率 ) 。如果所有的點接近 一條直線 ,那麼它就遵循齊夫定律。
    16. Zipf’s Law(2/2)
      • F ~ R − β F~1 / R
      • F : frequency of occurrence.
      • R : the rank of object.
      • β : a constant close to 1.
      • Based on a regression analysis , β = 0.80 and R 2 goodness of fit value is 0.98 .
    17. Goodness of Fit Test
      • 可以幫我們檢測母體比例是否為特定值,或是檢定母體分配是否為某一種特定分配。
      • 卡方檢定 ( the chi-square test )
      • 觀察次數和期望個數應相差無幾,這會使得的值很小。反之,若的值很大,則代表該樣本所對應的群體分布並不如期望。
    18. Rank Order VS Frequency(1/2)
    19. Rank Order VS Frequency(2/2)
      • There are not so many less popular videos at high rank position of data set.
    20. Metadata Analysis
      • r (correlation coefficient)= 0.2, between video duration and video rating average.
      • r = 0.08, for video duration and video view count.
      • r, for video rating average and video view count, is 0.18
    21. Conclusion
      • The huge growth of Web 2.0 creates scalability problem for its centralized resources and requires decentralized approaches like caching.
      • View count metadata can be a policy that use to design the caching system.
      • Exam effective caching algorithm for Web 2.0 multimedia sites such as Youtube.

    + Ian ChenIan Chen, 2 years ago

    custom

    332 views, 0 favs, 1 embeds more stats

    More info about this document

    © All Rights Reserved

    Go to text version

    • Total Views 332
      • 327 on SlideShare
      • 5 from embeds
    • Comments 0
    • Favorites 0
    • Downloads 2
    Most viewed embeds
    • 5 views on http://acmeian.blogspot.com

    more

    All embeds
    • 5 views on http://acmeian.blogspot.com

    less

    Flagged as inappropriate Flag as inappropriate
    Flag as inappropriate

    Select your reason for flagging this presentation as inappropriate. If needed, use the feedback form to let us know more details.

    Cancel
    File a copyright complaint
    Having problems? Go to our helpdesk?

    Categories