SLIDE 1
Fakultät für Informatik
Technische Universität München
2
Outline
- YouTube Platform
- Motivation
- Dataset
- Modeling
- Conclusion
- Future work
- References
Popularity over time Analysis of Videos on YouTube Tizian Sarre - - PowerPoint PPT Presentation
Popularity over time Analysis of Videos on YouTube Tizian Sarre - - PowerPoint PPT Presentation
Fakultt fr Informatik Technische Universitt Mnchen Popularity over time Analysis of Videos on YouTube Tizian Sarre Advisor(s): Dr. Heiko Niedermayer Supervisor: Prof. Dr.-Ing. Georg Carle Chair of Network Architectures and Services
SLIDE 2
SLIDE 3
3
The Platform
- 2nd biggest website worldwide
- Over 1 billion users
- More than 4 billion daily views
- More than 300 hours of newly uploaded videos every minute
SLIDE 4
4
Motivation
Huge amounts of data
- Storage costs
- Networking costs
Video popularity analysis
- Better understanding user behavior
- Modeling video views
- (Network performance improvement)
SLIDE 5
5
Measured for 6 months (October 16th 2015 - April 14th 2016) YouTube metrics
- 59328 static video information
- 8909 unavailable videos
- 58594 measured videos
Facebook metrics
- About YouTube videos from dataset
Dataset Overview
SLIDE 6
6
YouTube metrics Static video information
- Title
- Duration
- Description
- Published
- …
Unavailable videos
- Time
Dynamic video measurements
- Views
- Likes
- Dislikes
- Comments
Dataset (YouTube)
SLIDE 7
7
YouTube Analysis -> Metrics Correlation
YouTube views strongly positively correlate with video likes
SLIDE 8
8
YouTube Analysis -> Search Ranking Benefits
Keywords, description, etc. barely correlate positively with views
SLIDE 9
9
YouTube Analysis -> Unavailable Videos
Most video taken down early Still, life time median surprisingly high (23)
SLIDE 10
10
Dataset (Facebook)
Facebook metrics Dynamic YouTube video measurements
- Likes
- Comments
- Shares
- Totals
SLIDE 11
11
Facebook Analysis: Influence on YouTube
Facebook shares correlates positively with YouTube video
views
Linear regression line “curved” due to logarithmic scale
SLIDE 12
12
Modeling -> Top Days
Certain videos more popular than others When and how often do the highest views gains (top days)
- ccur?
SLIDE 13
13
Modeling -> Event Days
When and how often do distinctly popular days (event days)
- ccur?
How exactly are event days defined? Desired properties:
- Independence of future data
- Outstanding popularity
- Independence of video age
- Event day occurrence independence
- Popularity only dependence
SLIDE 14
14
Modeling -> Event Days
Different event day definition attempts have been made
- Absolute median
- Popularity categories
- Power law-based
- Varying event models
SLIDE 15
15
Modeling -> Event Days -> Absolute Median
Idea: Calculate daily views gains medians of all videos in the dataset for each day in their lifetimes Use daily medians as event day decider
SLIDE 16
16
Modeling -> Event Days -> Absolute Median
Why not use the average? Too heavily influenced by higher values Represents reality less appropriately Daily views gains medians of the dataset:
SLIDE 17
17
Modeling -> Event Days -> Absolute Median
How do we use the daily views gains medians to derive event days? We could use the exact medians as decider
Better: Add arbitrary positive modifier to accomplish
- utstanding popularity
SLIDE 18
18
Modeling -> Event Days -> Absolute Median
Observation Lots of event days
SLIDE 19
19
Modeling -> Event Days -> Absolute Median
Problem Daily views gains medians no good decider…
- Values too small
Need more fine-grained decider
SLIDE 20
20
Modeling -> Event Days -> Popularity categories
Idea Classify videos dynamically according to popularity categories Smaller intervals for lower views gains (due to geometric views gains distribution) More realistic event day determination via interval median as decider
SLIDE 21
21
Problem Event day determination depends on interval choice
Seemingly random event days Popularity only dependence violated
Modeling -> Event Days -> Popularity categories
SLIDE 22
22
Modeling -> Event Days -> Power Law
Idea Calculate a power law based model on the dataset’s views gains Strong positive deviations are event days
SLIDE 23
23
Modeling -> Event Days -> Power Law
Results
Slightly fluctuating but overall decreasing event day occurrence Event days less likely than with absolute medians
SLIDE 24
24
Modeling -> Event Days -> Power Law
Positives:
- Relatively reasonable model
Negatives: Model changes are not considered Multiple models are not supported
SLIDE 25
25
Idea Adjust current power law model when strong deviations
Modeling -> Event Days -> Varying Event Models
SLIDE 26
26
Deviation weight using least squares? No good, high deviation between views gains too severely weighted
Modeling -> Event Days -> Varying Event Models
SLIDE 27
27
Modeling -> Event Days -> Varying Event Models
Better with relative measure:
Deviations weighted linearly
SLIDE 28
28
Modeling -> Event Days -> Varying Event Models
Decider for model adaption? More/Less than 50% of expected value
Further model adaptions less likely (consecutive event days less
likely)
SLIDE 29
29
Modeling -> Event Days -> Varying Event Models
Similar results compared to power law approach Still no multiple event models supported due to uncertainty
SLIDE 30
30
Conclusion
- We discussed various better and worse possible event day
definitions
- External popularity influence not considered due to uncertainty
- Vague models for predictions
SLIDE 31
31
Future Work
- Consider video recommendations for popularity analysis
- YouTube internal and external search engine rank/hits
- Consider YouTube channel subscriber base for video popularity
analysis
- Twitter/Snapchat/Instagram social media influence analysis
- Alternate event day definitions and multiple models detection
support (e.g. with CUSUM)
SLIDE 32
32
Thank you
SLIDE 33
33