SlideShare a Scribd company logo
1 of 27
Download to read offline
Mining TV-on-demand Services
EPSRC project
Dmytro Karamshuk
Users - 32 M/month
IP address – 20 M/month
Sessions - 1.9 Billion
May 2013 – Jan 2014
≈ 50% of population
Large-scale study of BBC iPlayer
UK Population – 64M
2  x  INFOCOM’2015,  ToN’2015,  JSAC’2016
Longitudinal View across ISPs
Fixed-line Internet market
(5 representative providers)
Mobile market is more dynamic than the fixed-line Internet market
Mobile Internet market
(5 representative providers)
Data caps decrease market share
All-you-can-eat data
(M1, M5)
Limited-cap data packages
(M2 – M4)
All-you-can-eat plans boost user consumption
Temporal Patterns in different ISPs
Fixed-line accesses (F1-F5) peaks
in the evening hours
Mobile users watch more
during commutes
FixedLined
ISPs
Mobile,limiteddata
caps
There is a problem…
Internet on trains in the UK is no good
A study shows that 23.2% 3G packets and 37.2% 4G packets on
the major train routes failed
A useful insight: users watch across networks
Users complete watching across different sessions and networks
Fixed-line ISPs Mobile ISPs
Per user completion ratio
Speculative Content Pre-fetching
Pre-fetch at home Watch during commutes
Speculative Content Pre-fetching
Not very efficient…
Per-user mobile savings with pre-fetching
Can we do better with predictive preloading?
Towards Predicting User Preferences
Featured content
Most Popular Content
How important are UI guidance?
For 20% of users > 60% of their access are from the Front Page
Content Types
11 channels
11 categories and 172 genres
thousands shows
1 channel 2 channels 3 channels
20%0% 40% 60% 100%
1 category 2 category 3 categories
30%0% 75%55% 100%
1 genre 2 gen. 3 gen.
15%0% 40% 50%30%
4 gen.
100%
1 sh. 2 sh. 3
10%0% 25%20%
4 sh.
100%35%
User Focus on Different Content Types
share of users with all their sessions from:
out of 11 channels
out of 171 genres
out of thousands shows
out of 11 categ.
importance
content category 0.038
content genre 0.063
category affinity 0.042
genre affinity 0.103
show affinity 0.179
channel affinity 0.043
content age 0.087
User Preferences
Total importance: 0.555
importance
featured content 0.061
featured position 0.061
content popularity rank 0.071
popularity position 0.008
featured probability 0.091
UI Guidance
Total importance: 0.292
importance
previously watched 0.066
completion ratio 0.081
probability of re-watching 0.007
Repeatedly Watched Content
Total importance: 0.154
Engineering Features
Supervised Learning
Problem: For a given user U and an episode E
predict whether U will watch E
Binary Classification Problem f(U,E) -> {0,1}
Random Forest: fast,
good performance on high dimensional data
Negative Examples: randomly sample from
what users did not watch
Predictions: Predict probability, rank all
episodes by probability
Accuracy of Personalized Predictions
For 50% of users over 70% chance
of fitting in Top-10 predictions
When do we do predictions?
Front Pages are updated over night…
When do we do predictions?
… and remain largely unchanged for 24h
How much traffic can be saved?
Predictive pre-fetching can potentially
save near 71% of mobile usage
We made mobile users happy!
How about the rest?
Access Patterns
Average per-user # sessions Correlation with Internet speed
Content Delivery for Home Broadband
Install more
distributed caches
May requires
significant investments
Any alternatives?
Problem: how to handle peak load from 32M users
Alternative: Peer-assisted Content Delivery
Content Servers
user
user user
user
user
user
average of 5K users online every sec in the first day after release
5K duplicates
every second!!!
Ask users for assistance
Elegant Theoretical Model for very Complex Behavior
around 88% of savings can be achieved
Data Analysis
TheoreticalModel
G c 1 e
c
Why it works?
Top-5% of the content
corpus accounts for 80% of traffic
Most of accesses happen in the
first day after release
Yes, it’s all about very popular content
Dmytro Karamshuk
King’s College London
“True genius resides in the capacity for evaluation of
uncertain, hazardous, and conflicting information” -
Winston Churchill

More Related Content

Similar to Take-away TV: Recharging Work Commutes with Greedy and Predictive Preloading of TV Content

Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...
Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...
Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...Citrix
 
Cloud Computing and Mobile VAS by E. Jay Saunders
Cloud Computing and Mobile VAS by E. Jay SaundersCloud Computing and Mobile VAS by E. Jay Saunders
Cloud Computing and Mobile VAS by E. Jay SaundersE. Jay Saunders
 
Zahid Hussain - Internet Tv Aug 2008 Poland
Zahid Hussain - Internet Tv Aug 2008 PolandZahid Hussain - Internet Tv Aug 2008 Poland
Zahid Hussain - Internet Tv Aug 2008 Polandguest4d4d00
 
Content for-all-the-potential-for-lte-broadcast-embms-white-paper
Content for-all-the-potential-for-lte-broadcast-embms-white-paperContent for-all-the-potential-for-lte-broadcast-embms-white-paper
Content for-all-the-potential-for-lte-broadcast-embms-white-paperKamal Kishor Pandey
 
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound Waves
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound WavesIRJET- Data Transmission for Proximity Devices using Ultrasonic Sound Waves
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound WavesIRJET Journal
 
Lecture 1 - Introduction to Course & Course outline.pptx
Lecture 1 - Introduction to Course & Course outline.pptxLecture 1 - Introduction to Course & Course outline.pptx
Lecture 1 - Introduction to Course & Course outline.pptxSameer Ali
 
Predicting whether users view dynamic content on the World Wide Web (and beyo...
Predicting whether users view dynamic content on the World Wide Web (and beyo...Predicting whether users view dynamic content on the World Wide Web (and beyo...
Predicting whether users view dynamic content on the World Wide Web (and beyo...Caroline Jay
 
The Impact of OTT on Media Consumption Habits
The Impact of OTT on Media Consumption HabitsThe Impact of OTT on Media Consumption Habits
The Impact of OTT on Media Consumption Habitssonalithakurvns1999
 
Vibemedia: Mobile Internet and Connected Services
Vibemedia: Mobile Internet and Connected ServicesVibemedia: Mobile Internet and Connected Services
Vibemedia: Mobile Internet and Connected ServicesGareth Capon
 
Latest trends in wireless technology
Latest trends in wireless technology Latest trends in wireless technology
Latest trends in wireless technology Dr. Mazlan Abbas
 
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...ijma
 
Tacconi PhD final exam
Tacconi PhD final examTacconi PhD final exam
Tacconi PhD final examCoRehab
 
Efficient multimedia query by-content from mobile devices
Efficient multimedia query by-content from mobile devicesEfficient multimedia query by-content from mobile devices
Efficient multimedia query by-content from mobile devicesBrohi Aijaz Ali
 
Using LTE to Boost ARPU
Using LTE to Boost ARPUUsing LTE to Boost ARPU
Using LTE to Boost ARPUeXplanoTech
 
Program for 2015 ieee international conference on consumer electronics taiw...
Program for 2015 ieee international conference on consumer electronics   taiw...Program for 2015 ieee international conference on consumer electronics   taiw...
Program for 2015 ieee international conference on consumer electronics taiw...supra_uny
 

Similar to Take-away TV: Recharging Work Commutes with Greedy and Predictive Preloading of TV Content (20)

Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...
Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...
Citrix Mobile Analytics Report September 2014: Mobile subscriber data usage t...
 
Cloud Computing and Mobile VAS by E. Jay Saunders
Cloud Computing and Mobile VAS by E. Jay SaundersCloud Computing and Mobile VAS by E. Jay Saunders
Cloud Computing and Mobile VAS by E. Jay Saunders
 
Zahid Hussain - Internet Tv Aug 2008 Poland
Zahid Hussain - Internet Tv Aug 2008 PolandZahid Hussain - Internet Tv Aug 2008 Poland
Zahid Hussain - Internet Tv Aug 2008 Poland
 
Content for-all-the-potential-for-lte-broadcast-embms-white-paper
Content for-all-the-potential-for-lte-broadcast-embms-white-paperContent for-all-the-potential-for-lte-broadcast-embms-white-paper
Content for-all-the-potential-for-lte-broadcast-embms-white-paper
 
Sdn nf v_cala_slides
Sdn nf v_cala_slidesSdn nf v_cala_slides
Sdn nf v_cala_slides
 
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound Waves
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound WavesIRJET- Data Transmission for Proximity Devices using Ultrasonic Sound Waves
IRJET- Data Transmission for Proximity Devices using Ultrasonic Sound Waves
 
Lecture 1 - Introduction to Course & Course outline.pptx
Lecture 1 - Introduction to Course & Course outline.pptxLecture 1 - Introduction to Course & Course outline.pptx
Lecture 1 - Introduction to Course & Course outline.pptx
 
PhD_Thesis
PhD_ThesisPhD_Thesis
PhD_Thesis
 
Predicting whether users view dynamic content on the World Wide Web (and beyo...
Predicting whether users view dynamic content on the World Wide Web (and beyo...Predicting whether users view dynamic content on the World Wide Web (and beyo...
Predicting whether users view dynamic content on the World Wide Web (and beyo...
 
The Impact of OTT on Media Consumption Habits
The Impact of OTT on Media Consumption HabitsThe Impact of OTT on Media Consumption Habits
The Impact of OTT on Media Consumption Habits
 
Vibemedia: Mobile Internet and Connected Services
Vibemedia: Mobile Internet and Connected ServicesVibemedia: Mobile Internet and Connected Services
Vibemedia: Mobile Internet and Connected Services
 
Latest trends in wireless technology
Latest trends in wireless technology Latest trends in wireless technology
Latest trends in wireless technology
 
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...
ANALYSIS AND MODELLING OF POWER CONSUMPTION IN IOT WITH VIDEO QUALITY COMMUNI...
 
Tacconi PhD final exam
Tacconi PhD final examTacconi PhD final exam
Tacconi PhD final exam
 
Distributed Systems, Mobile Computing and Security
Distributed Systems, Mobile Computing and SecurityDistributed Systems, Mobile Computing and Security
Distributed Systems, Mobile Computing and Security
 
Ijariie1186
Ijariie1186Ijariie1186
Ijariie1186
 
Efficient multimedia query by-content from mobile devices
Efficient multimedia query by-content from mobile devicesEfficient multimedia query by-content from mobile devices
Efficient multimedia query by-content from mobile devices
 
Using LTE to Boost ARPU
Using LTE to Boost ARPUUsing LTE to Boost ARPU
Using LTE to Boost ARPU
 
Program for 2015 ieee international conference on consumer electronics taiw...
Program for 2015 ieee international conference on consumer electronics   taiw...Program for 2015 ieee international conference on consumer electronics   taiw...
Program for 2015 ieee international conference on consumer electronics taiw...
 
Li fi
Li fiLi fi
Li fi
 

Recently uploaded

Film cover research (1).pptxsdasdasdasdasdasa
Film cover research (1).pptxsdasdasdasdasdasaFilm cover research (1).pptxsdasdasdasdasdasa
Film cover research (1).pptxsdasdasdasdasdasa494f574xmv
 
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一Fs
 
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书rnrncn29
 
Contact Rya Baby for Call Girls New Delhi
Contact Rya Baby for Call Girls New DelhiContact Rya Baby for Call Girls New Delhi
Contact Rya Baby for Call Girls New Delhimiss dipika
 
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一Fs
 
Magic exist by Marta Loveguard - presentation.pptx
Magic exist by Marta Loveguard - presentation.pptxMagic exist by Marta Loveguard - presentation.pptx
Magic exist by Marta Loveguard - presentation.pptxMartaLoveguard
 
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书rnrncn29
 
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书zdzoqco
 
PHP-based rendering of TYPO3 Documentation
PHP-based rendering of TYPO3 DocumentationPHP-based rendering of TYPO3 Documentation
PHP-based rendering of TYPO3 DocumentationLinaWolf1
 
Git and Github workshop GDSC MLRITM
Git and Github  workshop GDSC MLRITMGit and Github  workshop GDSC MLRITM
Git and Github workshop GDSC MLRITMgdsc13
 
Blepharitis inflammation of eyelid symptoms cause everything included along w...
Blepharitis inflammation of eyelid symptoms cause everything included along w...Blepharitis inflammation of eyelid symptoms cause everything included along w...
Blepharitis inflammation of eyelid symptoms cause everything included along w...Excelmac1
 
Elevate Your Business with Our IT Expertise in New Orleans
Elevate Your Business with Our IT Expertise in New OrleansElevate Your Business with Our IT Expertise in New Orleans
Elevate Your Business with Our IT Expertise in New Orleanscorenetworkseo
 
NSX-T and Service Interfaces presentation
NSX-T and Service Interfaces presentationNSX-T and Service Interfaces presentation
NSX-T and Service Interfaces presentationMarko4394
 
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一Fs
 
SCM Symposium PPT Format Customer loyalty is predi
SCM Symposium PPT Format Customer loyalty is prediSCM Symposium PPT Format Customer loyalty is predi
SCM Symposium PPT Format Customer loyalty is predieusebiomeyer
 
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170Sonam Pathan
 
Q4-1-Illustrating-Hypothesis-Testing.pptx
Q4-1-Illustrating-Hypothesis-Testing.pptxQ4-1-Illustrating-Hypothesis-Testing.pptx
Q4-1-Illustrating-Hypothesis-Testing.pptxeditsforyah
 
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)Christopher H Felton
 
Call Girls Near The Suryaa Hotel New Delhi 9873777170
Call Girls Near The Suryaa Hotel New Delhi 9873777170Call Girls Near The Suryaa Hotel New Delhi 9873777170
Call Girls Near The Suryaa Hotel New Delhi 9873777170Sonam Pathan
 

Recently uploaded (20)

Film cover research (1).pptxsdasdasdasdasdasa
Film cover research (1).pptxsdasdasdasdasdasaFilm cover research (1).pptxsdasdasdasdasdasa
Film cover research (1).pptxsdasdasdasdasdasa
 
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一
定制(AUT毕业证书)新西兰奥克兰理工大学毕业证成绩单原版一比一
 
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书
『澳洲文凭』买拉筹伯大学毕业证书成绩单办理澳洲LTU文凭学位证书
 
Contact Rya Baby for Call Girls New Delhi
Contact Rya Baby for Call Girls New DelhiContact Rya Baby for Call Girls New Delhi
Contact Rya Baby for Call Girls New Delhi
 
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一
定制(Management毕业证书)新加坡管理大学毕业证成绩单原版一比一
 
Magic exist by Marta Loveguard - presentation.pptx
Magic exist by Marta Loveguard - presentation.pptxMagic exist by Marta Loveguard - presentation.pptx
Magic exist by Marta Loveguard - presentation.pptx
 
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书
『澳洲文凭』买詹姆士库克大学毕业证书成绩单办理澳洲JCU文凭学位证书
 
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书
办理多伦多大学毕业证成绩单|购买加拿大UTSG文凭证书
 
PHP-based rendering of TYPO3 Documentation
PHP-based rendering of TYPO3 DocumentationPHP-based rendering of TYPO3 Documentation
PHP-based rendering of TYPO3 Documentation
 
Git and Github workshop GDSC MLRITM
Git and Github  workshop GDSC MLRITMGit and Github  workshop GDSC MLRITM
Git and Github workshop GDSC MLRITM
 
Blepharitis inflammation of eyelid symptoms cause everything included along w...
Blepharitis inflammation of eyelid symptoms cause everything included along w...Blepharitis inflammation of eyelid symptoms cause everything included along w...
Blepharitis inflammation of eyelid symptoms cause everything included along w...
 
Elevate Your Business with Our IT Expertise in New Orleans
Elevate Your Business with Our IT Expertise in New OrleansElevate Your Business with Our IT Expertise in New Orleans
Elevate Your Business with Our IT Expertise in New Orleans
 
NSX-T and Service Interfaces presentation
NSX-T and Service Interfaces presentationNSX-T and Service Interfaces presentation
NSX-T and Service Interfaces presentation
 
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一
定制(Lincoln毕业证书)新西兰林肯大学毕业证成绩单原版一比一
 
SCM Symposium PPT Format Customer loyalty is predi
SCM Symposium PPT Format Customer loyalty is prediSCM Symposium PPT Format Customer loyalty is predi
SCM Symposium PPT Format Customer loyalty is predi
 
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170
Call Girls In The Ocean Pearl Retreat Hotel New Delhi 9873777170
 
young call girls in Uttam Nagar🔝 9953056974 🔝 Delhi escort Service
young call girls in Uttam Nagar🔝 9953056974 🔝 Delhi escort Serviceyoung call girls in Uttam Nagar🔝 9953056974 🔝 Delhi escort Service
young call girls in Uttam Nagar🔝 9953056974 🔝 Delhi escort Service
 
Q4-1-Illustrating-Hypothesis-Testing.pptx
Q4-1-Illustrating-Hypothesis-Testing.pptxQ4-1-Illustrating-Hypothesis-Testing.pptx
Q4-1-Illustrating-Hypothesis-Testing.pptx
 
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)
A Good Girl's Guide to Murder (A Good Girl's Guide to Murder, #1)
 
Call Girls Near The Suryaa Hotel New Delhi 9873777170
Call Girls Near The Suryaa Hotel New Delhi 9873777170Call Girls Near The Suryaa Hotel New Delhi 9873777170
Call Girls Near The Suryaa Hotel New Delhi 9873777170
 

Take-away TV: Recharging Work Commutes with Greedy and Predictive Preloading of TV Content

  • 1. Mining TV-on-demand Services EPSRC project Dmytro Karamshuk
  • 2. Users - 32 M/month IP address – 20 M/month Sessions - 1.9 Billion May 2013 – Jan 2014 ≈ 50% of population Large-scale study of BBC iPlayer UK Population – 64M 2  x  INFOCOM’2015,  ToN’2015,  JSAC’2016
  • 3. Longitudinal View across ISPs Fixed-line Internet market (5 representative providers) Mobile market is more dynamic than the fixed-line Internet market Mobile Internet market (5 representative providers)
  • 4. Data caps decrease market share All-you-can-eat data (M1, M5) Limited-cap data packages (M2 – M4) All-you-can-eat plans boost user consumption
  • 5. Temporal Patterns in different ISPs Fixed-line accesses (F1-F5) peaks in the evening hours Mobile users watch more during commutes FixedLined ISPs Mobile,limiteddata caps
  • 6. There is a problem… Internet on trains in the UK is no good A study shows that 23.2% 3G packets and 37.2% 4G packets on the major train routes failed
  • 7. A useful insight: users watch across networks Users complete watching across different sessions and networks Fixed-line ISPs Mobile ISPs Per user completion ratio
  • 8. Speculative Content Pre-fetching Pre-fetch at home Watch during commutes
  • 9. Speculative Content Pre-fetching Not very efficient… Per-user mobile savings with pre-fetching
  • 10. Can we do better with predictive preloading?
  • 11. Towards Predicting User Preferences Featured content Most Popular Content
  • 12. How important are UI guidance? For 20% of users > 60% of their access are from the Front Page
  • 13. Content Types 11 channels 11 categories and 172 genres thousands shows
  • 14. 1 channel 2 channels 3 channels 20%0% 40% 60% 100% 1 category 2 category 3 categories 30%0% 75%55% 100% 1 genre 2 gen. 3 gen. 15%0% 40% 50%30% 4 gen. 100% 1 sh. 2 sh. 3 10%0% 25%20% 4 sh. 100%35% User Focus on Different Content Types share of users with all their sessions from: out of 11 channels out of 171 genres out of thousands shows out of 11 categ.
  • 15. importance content category 0.038 content genre 0.063 category affinity 0.042 genre affinity 0.103 show affinity 0.179 channel affinity 0.043 content age 0.087 User Preferences Total importance: 0.555 importance featured content 0.061 featured position 0.061 content popularity rank 0.071 popularity position 0.008 featured probability 0.091 UI Guidance Total importance: 0.292 importance previously watched 0.066 completion ratio 0.081 probability of re-watching 0.007 Repeatedly Watched Content Total importance: 0.154 Engineering Features
  • 16. Supervised Learning Problem: For a given user U and an episode E predict whether U will watch E Binary Classification Problem f(U,E) -> {0,1} Random Forest: fast, good performance on high dimensional data Negative Examples: randomly sample from what users did not watch Predictions: Predict probability, rank all episodes by probability
  • 17. Accuracy of Personalized Predictions For 50% of users over 70% chance of fitting in Top-10 predictions
  • 18. When do we do predictions? Front Pages are updated over night…
  • 19. When do we do predictions? … and remain largely unchanged for 24h
  • 20. How much traffic can be saved? Predictive pre-fetching can potentially save near 71% of mobile usage
  • 21. We made mobile users happy! How about the rest?
  • 22. Access Patterns Average per-user # sessions Correlation with Internet speed
  • 23. Content Delivery for Home Broadband Install more distributed caches May requires significant investments Any alternatives? Problem: how to handle peak load from 32M users
  • 24. Alternative: Peer-assisted Content Delivery Content Servers user user user user user user average of 5K users online every sec in the first day after release 5K duplicates every second!!! Ask users for assistance
  • 25. Elegant Theoretical Model for very Complex Behavior around 88% of savings can be achieved Data Analysis TheoreticalModel G c 1 e c
  • 26. Why it works? Top-5% of the content corpus accounts for 80% of traffic Most of accesses happen in the first day after release Yes, it’s all about very popular content
  • 27. Dmytro Karamshuk King’s College London “True genius resides in the capacity for evaluation of uncertain, hazardous, and conflicting information” - Winston Churchill