SlideShare a Scribd company logo
1 of 19
Download to read offline
R&D © BBC 2012
Automatic Mood Classification of TV
Programmes
Sam Davies, Jana Eggink, Denise Bland
BBC Research & Development
R&D © BBC MMVIII
British Broadcasting Corporation Archive
• BBC I&A
– Perivale, London
– >1,000,000 items
• ~ 650,000 TV
• ~ 350,000 Radio
• ~ 1.5 million hours
– Since 1922
• BBC Redux
– online
– 300,000 hours of TV and radio
– Since 2007
• BBC Written Archive
– 4 ½ miles of documents
– Caversham, Reading
R&D © BBC MMVIII
Current Programme Retrieval
• Infax
– > 1,500,000 programmes
• LonClasss
– > 52,000 concepts
• “pop music”, “Iraq”, “criticism
of growing plants for biofuels”,
“Dover Castle communications
centre”, “fake psychics”,
“posters of Ariel Sharon”.
– BBC Redux
R&D © BBC MMVIII
Current Programme Retrieval – BBC Internal
• Infax
– > 1,500,000 programmes
• LonClasss
– > 52,000 concepts
• “pop music”, “Iraq”, “criticism
of growing plants for biofuels”,
“Dover Castle communications
centre”, “fake psychics”,
“posters of Ariel Sharon”.
– BBC Redux
– BBC Snippets
R&D © BBC MMVIII
Current Programme Retrieval – Public facing
• BBC iPlayer
– Catch-up service
– Known item search
– Standard categorisation
R&D © BBC MMVIII
Current Programme Retrieval – Public facing
• BBC iPlayer
– Catch-up service
– Known item search
– Standard categorisation
• bbc.co.uk/programmes
– More episodes
– Editorially chosen similar
programmes
R&D © BBC MMVIII
Current Programme Retrieval – Public facing
• BBC iPlayer
– Catch-up service
– Known item search
– Standard categorisation
• bbc.co.uk/programmes
– More episodes
– Editorially chosen similar
programmes
• Link key contributors
– Editorially identified
– Automatically linked
R&D © BBC MMVIII
Mood Based Classification – System overview
Feature Extraction
• Video & Audio Analysis
– Colour histogram, motion
detection, brightness.
– Spectral audio components
• Object identification
– Faces, animals, objects
(Tardis)
– Gunshots, laughter, screaming
R&D © BBC MMVIII
Mood Based Classification: Ground truth collection
• Ground Truth Collection
– Video
• 200 members of public from varied
demographic
• 250 programmes
• Asked to classify programme clips
based around adjectives taken
from Affective Theory
R&D © BBC MMVIII
Mood Based Classification - GUI
R&D © BBC MMVIII
Mood Based Classification - GUI
R&D © BBC MMVIII
Other mood based features - Music
• Ground Truth Collection
– Video
• 200 members of public from
varied demographic
• 250 programmes
– Music
• MusicalMoods
– 20,000 members of
public
– 60 theme tunes
R&D © BBC MMVIII
Other mood based features - Text
• Identify mood of any text on three axis:
– Valence (positive/negative) e.g. triumphant, love, paradise
– Arousal (amount of emotion instilled) e.g. rage, thrill, explosion
– Dominance (power) e.g. winner, confident, admired
• Increases dimensionality of sentiment analysis
• Use on large datasets negates requirement for syntactical analysis
• Subtitles are more correct than derived metadata (automatic speech transcripts, machine
vision, machine listening)
• Fast, scalable
• Domain independent
R&D © BBC MMVIII
Other mood based features - Text
R&D © BBC MMVIII
Other mood based features: Text
Precision 0.95
Recall 0.91
F1 Score 0.93
R&D © BBC MMVIII
Future areas - Combination of Affect and Semantic
R&D © BBC MMVIII
Future areas - Highlights Generation
• Sports matches
– Audio based analysis
– Two stages
• Live match identification
• Interesting section detected
– Accuracy of 78%
– Looking currently to include
social media to increase
accuracy.
R&D © BBC MMVIII
Publications & more info
• Davies, S., Bland, D. & Grafton R (2010) “A Framework for Automatic Mood Classification of TV Programmes”
presented at SAMT 2010.
• Davies, S. (2010) “Interestingness Detection in Sports Audio Broadcasts” presented at IEEE ICMLA 2010
• Knoiusz, P. & Mikolajcyzk, K. (2011) “Soft Assignment Of Visual Words As Linear Coordinate Coding And
Optimisation Of Its Reconstruction Error” presented at ICIP 2011
• Knoiusz, P. & Mikolajcyzk, K. (2011) “Spatial Coordinate Coding To Reduce Histogram Representations, Dominant
Angle and Colour Pyramid Match” presented at ICIP 2011
• Davies, S. & Bland, D. (2011) “An Improved Framework for Affective Classification and Browsing of Large Scale
Broadcast Archives” presented at ACM SIGIR 2011
• Mann, M. & Cox, T. (2011) “Music Mood Classification of Television Theme Tunes” presented at ISMIR 2011
• Davies, S., Mann, M., Cox, T. & Allen, P. (2011) “Musical Moods: A Mass Participation Experiment for Affective Music
Classification” presented at ISMIR 2011
• Eggink, J. Allen, P. & Bland, D. (2011) “A Pilot Study for Mood-Based Classification of TV programmes” presented at
ACM SIGAC 2011
• Eggink, J. & Bland, D. (2012) “A Large Scale Experiment for Mood-Based Classification of TV Programmes”
presented at ICME 2012
• Available at http://www.bbc.co.uk/rd/publications/whitepapers.shtml
R&D © BBC MMVIII
Thank you
• Questions
• Contact;
– sam.davies@bbc.co.uk

More Related Content

Similar to Presentation 17 may morning casestudy 1 sam davies

COM 110: Chapter 8
COM 110: Chapter 8COM 110: Chapter 8
COM 110: Chapter 8
Val Bello
 
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 SlideshowNewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
Dr. William J. Ward
 
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
Jisc
 
Making bbc programmes discoverable
Making bbc programmes discoverableMaking bbc programmes discoverable
Making bbc programmes discoverable
Tom Scott
 
research into the client (BBC)
research into the client (BBC)research into the client (BBC)
research into the client (BBC)
JCKMRE
 

Similar to Presentation 17 may morning casestudy 1 sam davies (20)

Jisc MediaHub webinar
Jisc MediaHub webinarJisc MediaHub webinar
Jisc MediaHub webinar
 
Jisc MediaHub webinar
Jisc MediaHub webinarJisc MediaHub webinar
Jisc MediaHub webinar
 
Andrew Bevan, EDINA
Andrew Bevan, EDINAAndrew Bevan, EDINA
Andrew Bevan, EDINA
 
Discovering What You Can't Always Get From Google: Jisc MediaHub
Discovering What You Can't Always Get From Google: Jisc MediaHubDiscovering What You Can't Always Get From Google: Jisc MediaHub
Discovering What You Can't Always Get From Google: Jisc MediaHub
 
COM 110: Chapter 8
COM 110: Chapter 8COM 110: Chapter 8
COM 110: Chapter 8
 
101 ch8
101 ch8101 ch8
101 ch8
 
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 SlideshowNewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
NewhouseSU COM 107 Communications and Society #NH1074Ward - Ch. 4 Slideshow
 
Radio lessons 14 02 19
Radio lessons 14 02 19Radio lessons 14 02 19
Radio lessons 14 02 19
 
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
Discovering what you can't always get from Google - Andrew Bevan - Jisc Digit...
 
Jisc MediaHub webinar
Jisc MediaHub webinarJisc MediaHub webinar
Jisc MediaHub webinar
 
Making bbc programmes discoverable
Making bbc programmes discoverableMaking bbc programmes discoverable
Making bbc programmes discoverable
 
research into the client (BBC)
research into the client (BBC)research into the client (BBC)
research into the client (BBC)
 
Presentation 1
Presentation 1Presentation 1
Presentation 1
 
Research into the client (BBC
Research into the client (BBCResearch into the client (BBC
Research into the client (BBC
 
Radio lecture for Media and Communication Industries - QUT
Radio lecture for Media and Communication Industries - QUTRadio lecture for Media and Communication Industries - QUT
Radio lecture for Media and Communication Industries - QUT
 
Jisc MediaHub 2014/2015 Update
Jisc MediaHub 2014/2015 UpdateJisc MediaHub 2014/2015 Update
Jisc MediaHub 2014/2015 Update
 
Radio 1 Breakfast show - Revision ym jh
Radio 1 Breakfast show - Revision ym jhRadio 1 Breakfast show - Revision ym jh
Radio 1 Breakfast show - Revision ym jh
 
Music Objects to Social Machines
Music Objects to Social MachinesMusic Objects to Social Machines
Music Objects to Social Machines
 
Chapter 8
Chapter 8Chapter 8
Chapter 8
 
Jisc MediaHub webinar
Jisc MediaHub webinarJisc MediaHub webinar
Jisc MediaHub webinar
 

More from Nederlands Instituut voor Beeld en Geluid

More from Nederlands Instituut voor Beeld en Geluid (15)

TROVe Transmedia Observatory eindpresentatie
TROVe Transmedia Observatory eindpresentatieTROVe Transmedia Observatory eindpresentatie
TROVe Transmedia Observatory eindpresentatie
 
Presentation 17 may afternoon casestudy 1 yves raimond kopie
Presentation 17 may afternoon casestudy 1 yves raimond kopiePresentation 17 may afternoon casestudy 1 yves raimond kopie
Presentation 17 may afternoon casestudy 1 yves raimond kopie
 
Presentation 17 may morning case study 2 sarahhaye aziz
Presentation 17 may morning case study 2 sarahhaye azizPresentation 17 may morning case study 2 sarahhaye aziz
Presentation 17 may morning case study 2 sarahhaye aziz
 
Presentation 17 may keynote lara aroyo
Presentation 17 may keynote lara aroyoPresentation 17 may keynote lara aroyo
Presentation 17 may keynote lara aroyo
 
Presentation 17 may morning keynote cees snoek
Presentation 17 may morning keynote cees snoekPresentation 17 may morning keynote cees snoek
Presentation 17 may morning keynote cees snoek
 
Presentation 17 may afternoon casestudy 2 liam wylie
Presentation 17 may afternoon casestudy 2 liam wyliePresentation 17 may afternoon casestudy 2 liam wylie
Presentation 17 may afternoon casestudy 2 liam wylie
 
Presentation 16 may casestudy 2 evalisgreen kaisa unander
Presentation 16 may casestudy 2 evalisgreen kaisa unanderPresentation 16 may casestudy 2 evalisgreen kaisa unander
Presentation 16 may casestudy 2 evalisgreen kaisa unander
 
Presentation 16 may morning semantic linking rutger verhoeven
Presentation 16 may morning semantic linking rutger verhoevenPresentation 16 may morning semantic linking rutger verhoeven
Presentation 16 may morning semantic linking rutger verhoeven
 
Presentation 16 may morning casestudy 2 xavier jacques jourion
Presentation 16 may morning casestudy 2 xavier jacques jourionPresentation 16 may morning casestudy 2 xavier jacques jourion
Presentation 16 may morning casestudy 2 xavier jacques jourion
 
Presentation 16 may morning casestudy 1 maarten de rijke
Presentation 16 may morning casestudy 1 maarten de rijkePresentation 16 may morning casestudy 1 maarten de rijke
Presentation 16 may morning casestudy 1 maarten de rijke
 
Presentation 16 may morning keynote seth van hooland
Presentation 16 may morning keynote seth van hoolandPresentation 16 may morning keynote seth van hooland
Presentation 16 may morning keynote seth van hooland
 
Presentation 16 may keynote karin bredenberg
Presentation 16 may keynote karin bredenbergPresentation 16 may keynote karin bredenberg
Presentation 16 may keynote karin bredenberg
 
Presentation 16 may casestudy daniel steinmeier
Presentation 16 may casestudy daniel steinmeierPresentation 16 may casestudy daniel steinmeier
Presentation 16 may casestudy daniel steinmeier
 
Presentation 16 may casestudy 2 evalisgreen kaisa unander
Presentation 16 may casestudy 2 evalisgreen kaisa unanderPresentation 16 may casestudy 2 evalisgreen kaisa unander
Presentation 16 may casestudy 2 evalisgreen kaisa unander
 
Presentation 16 may archive achievements awards tom de smet
Presentation 16 may archive achievements awards tom de smetPresentation 16 may archive achievements awards tom de smet
Presentation 16 may archive achievements awards tom de smet
 

Recently uploaded

Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
SofiyaSharma5
 
Goa Call "Girls Service 9316020077 Call "Girls in Goa
Goa Call "Girls  Service   9316020077 Call "Girls in GoaGoa Call "Girls  Service   9316020077 Call "Girls in Goa
Goa Call "Girls Service 9316020077 Call "Girls in Goa
sexy call girls service in goa
 
Call Girls Agency In Goa 💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
Call Girls  Agency In Goa  💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...Call Girls  Agency In Goa  💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
Call Girls Agency In Goa 💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
russian goa call girl and escorts service
 
Call Girls In Goa 9316020077 Goa Call Girl By Indian Call Girls Goa
Call Girls In Goa  9316020077 Goa  Call Girl By Indian Call Girls GoaCall Girls In Goa  9316020077 Goa  Call Girl By Indian Call Girls Goa
Call Girls In Goa 9316020077 Goa Call Girl By Indian Call Girls Goa
sexy call girls service in goa
 
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
Apsara Of India
 

Recently uploaded (20)

Behala ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sexy Bhabi Ready ...
Behala ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sexy Bhabi Ready ...Behala ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sexy Bhabi Ready ...
Behala ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sexy Bhabi Ready ...
 
👙 Kolkata Call Girls Park Circus 💫💫7001035870 Model escorts Service
👙  Kolkata Call Girls Park Circus 💫💫7001035870 Model escorts Service👙  Kolkata Call Girls Park Circus 💫💫7001035870 Model escorts Service
👙 Kolkata Call Girls Park Circus 💫💫7001035870 Model escorts Service
 
Call Girl Service Belur - 7001035870 with real photos and phone numbers
Call Girl Service Belur - 7001035870 with real photos and phone numbersCall Girl Service Belur - 7001035870 with real photos and phone numbers
Call Girl Service Belur - 7001035870 with real photos and phone numbers
 
↑Top Model (Kolkata) Call Girls Howrah ⟟ 8250192130 ⟟ High Class Call Girl In...
↑Top Model (Kolkata) Call Girls Howrah ⟟ 8250192130 ⟟ High Class Call Girl In...↑Top Model (Kolkata) Call Girls Howrah ⟟ 8250192130 ⟟ High Class Call Girl In...
↑Top Model (Kolkata) Call Girls Howrah ⟟ 8250192130 ⟟ High Class Call Girl In...
 
Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
Low Rate Young Call Girls in Surajpur Greater Noida ✔️☆9289244007✔️☆ Female E...
 
👙 Kolkata Call Girls Sonagachi 💫💫7001035870 Model escorts Service
👙  Kolkata Call Girls Sonagachi 💫💫7001035870 Model escorts Service👙  Kolkata Call Girls Sonagachi 💫💫7001035870 Model escorts Service
👙 Kolkata Call Girls Sonagachi 💫💫7001035870 Model escorts Service
 
Science City Kolkata ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sex...
Science City Kolkata ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sex...Science City Kolkata ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sex...
Science City Kolkata ( Call Girls ) Kolkata ✔ 6297143586 ✔ Hot Model With Sex...
 
↑Top Model (Kolkata) Call Girls Rajpur ⟟ 8250192130 ⟟ High Class Call Girl In...
↑Top Model (Kolkata) Call Girls Rajpur ⟟ 8250192130 ⟟ High Class Call Girl In...↑Top Model (Kolkata) Call Girls Rajpur ⟟ 8250192130 ⟟ High Class Call Girl In...
↑Top Model (Kolkata) Call Girls Rajpur ⟟ 8250192130 ⟟ High Class Call Girl In...
 
Independent Joka Escorts ✔ 8250192130 ✔ Full Night With Room Online Booking 2...
Independent Joka Escorts ✔ 8250192130 ✔ Full Night With Room Online Booking 2...Independent Joka Escorts ✔ 8250192130 ✔ Full Night With Room Online Booking 2...
Independent Joka Escorts ✔ 8250192130 ✔ Full Night With Room Online Booking 2...
 
Call Girl Nashik Saloni 7001305949 Independent Escort Service Nashik
Call Girl Nashik Saloni 7001305949 Independent Escort Service NashikCall Girl Nashik Saloni 7001305949 Independent Escort Service Nashik
Call Girl Nashik Saloni 7001305949 Independent Escort Service Nashik
 
Call Girls Manjri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Manjri Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Manjri Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Manjri Call Me 7737669865 Budget Friendly No Advance Booking
 
2k Shot Call girls Laxmi Nagar Delhi 9205541914
2k Shot Call girls Laxmi Nagar Delhi 92055419142k Shot Call girls Laxmi Nagar Delhi 9205541914
2k Shot Call girls Laxmi Nagar Delhi 9205541914
 
Independent Hatiara Escorts ✔ 9332606886✔ Full Night With Room Online Booking...
Independent Hatiara Escorts ✔ 9332606886✔ Full Night With Room Online Booking...Independent Hatiara Escorts ✔ 9332606886✔ Full Night With Room Online Booking...
Independent Hatiara Escorts ✔ 9332606886✔ Full Night With Room Online Booking...
 
Top Rated Kolkata Call Girls Khardah ⟟ 6297143586 ⟟ Call Me For Genuine Sex S...
Top Rated Kolkata Call Girls Khardah ⟟ 6297143586 ⟟ Call Me For Genuine Sex S...Top Rated Kolkata Call Girls Khardah ⟟ 6297143586 ⟟ Call Me For Genuine Sex S...
Top Rated Kolkata Call Girls Khardah ⟟ 6297143586 ⟟ Call Me For Genuine Sex S...
 
Call Girl Nashik Amaira 7001305949 Independent Escort Service Nashik
Call Girl Nashik Amaira 7001305949 Independent Escort Service NashikCall Girl Nashik Amaira 7001305949 Independent Escort Service Nashik
Call Girl Nashik Amaira 7001305949 Independent Escort Service Nashik
 
Goa Call "Girls Service 9316020077 Call "Girls in Goa
Goa Call "Girls  Service   9316020077 Call "Girls in GoaGoa Call "Girls  Service   9316020077 Call "Girls in Goa
Goa Call "Girls Service 9316020077 Call "Girls in Goa
 
Call Girls in Barasat | 7001035870 At Low Cost Cash Payment Booking
Call Girls in Barasat | 7001035870 At Low Cost Cash Payment BookingCall Girls in Barasat | 7001035870 At Low Cost Cash Payment Booking
Call Girls in Barasat | 7001035870 At Low Cost Cash Payment Booking
 
Call Girls Agency In Goa 💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
Call Girls  Agency In Goa  💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...Call Girls  Agency In Goa  💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
Call Girls Agency In Goa 💚 9316020077 💚 Call Girl Goa By Russian Call Girl ...
 
Call Girls In Goa 9316020077 Goa Call Girl By Indian Call Girls Goa
Call Girls In Goa  9316020077 Goa  Call Girl By Indian Call Girls GoaCall Girls In Goa  9316020077 Goa  Call Girl By Indian Call Girls Goa
Call Girls In Goa 9316020077 Goa Call Girl By Indian Call Girls Goa
 
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
Karnal Call Girls 8860008073 Dyal Singh Colony Call Girls Service in Karnal E...
 

Presentation 17 may morning casestudy 1 sam davies

  • 1. R&D © BBC 2012 Automatic Mood Classification of TV Programmes Sam Davies, Jana Eggink, Denise Bland BBC Research & Development
  • 2. R&D © BBC MMVIII British Broadcasting Corporation Archive • BBC I&A – Perivale, London – >1,000,000 items • ~ 650,000 TV • ~ 350,000 Radio • ~ 1.5 million hours – Since 1922 • BBC Redux – online – 300,000 hours of TV and radio – Since 2007 • BBC Written Archive – 4 ½ miles of documents – Caversham, Reading
  • 3. R&D © BBC MMVIII Current Programme Retrieval • Infax – > 1,500,000 programmes • LonClasss – > 52,000 concepts • “pop music”, “Iraq”, “criticism of growing plants for biofuels”, “Dover Castle communications centre”, “fake psychics”, “posters of Ariel Sharon”. – BBC Redux
  • 4. R&D © BBC MMVIII Current Programme Retrieval – BBC Internal • Infax – > 1,500,000 programmes • LonClasss – > 52,000 concepts • “pop music”, “Iraq”, “criticism of growing plants for biofuels”, “Dover Castle communications centre”, “fake psychics”, “posters of Ariel Sharon”. – BBC Redux – BBC Snippets
  • 5. R&D © BBC MMVIII Current Programme Retrieval – Public facing • BBC iPlayer – Catch-up service – Known item search – Standard categorisation
  • 6. R&D © BBC MMVIII Current Programme Retrieval – Public facing • BBC iPlayer – Catch-up service – Known item search – Standard categorisation • bbc.co.uk/programmes – More episodes – Editorially chosen similar programmes
  • 7. R&D © BBC MMVIII Current Programme Retrieval – Public facing • BBC iPlayer – Catch-up service – Known item search – Standard categorisation • bbc.co.uk/programmes – More episodes – Editorially chosen similar programmes • Link key contributors – Editorially identified – Automatically linked
  • 8. R&D © BBC MMVIII Mood Based Classification – System overview Feature Extraction • Video & Audio Analysis – Colour histogram, motion detection, brightness. – Spectral audio components • Object identification – Faces, animals, objects (Tardis) – Gunshots, laughter, screaming
  • 9. R&D © BBC MMVIII Mood Based Classification: Ground truth collection • Ground Truth Collection – Video • 200 members of public from varied demographic • 250 programmes • Asked to classify programme clips based around adjectives taken from Affective Theory
  • 10. R&D © BBC MMVIII Mood Based Classification - GUI
  • 11. R&D © BBC MMVIII Mood Based Classification - GUI
  • 12. R&D © BBC MMVIII Other mood based features - Music • Ground Truth Collection – Video • 200 members of public from varied demographic • 250 programmes – Music • MusicalMoods – 20,000 members of public – 60 theme tunes
  • 13. R&D © BBC MMVIII Other mood based features - Text • Identify mood of any text on three axis: – Valence (positive/negative) e.g. triumphant, love, paradise – Arousal (amount of emotion instilled) e.g. rage, thrill, explosion – Dominance (power) e.g. winner, confident, admired • Increases dimensionality of sentiment analysis • Use on large datasets negates requirement for syntactical analysis • Subtitles are more correct than derived metadata (automatic speech transcripts, machine vision, machine listening) • Fast, scalable • Domain independent
  • 14. R&D © BBC MMVIII Other mood based features - Text
  • 15. R&D © BBC MMVIII Other mood based features: Text Precision 0.95 Recall 0.91 F1 Score 0.93
  • 16. R&D © BBC MMVIII Future areas - Combination of Affect and Semantic
  • 17. R&D © BBC MMVIII Future areas - Highlights Generation • Sports matches – Audio based analysis – Two stages • Live match identification • Interesting section detected – Accuracy of 78% – Looking currently to include social media to increase accuracy.
  • 18. R&D © BBC MMVIII Publications & more info • Davies, S., Bland, D. & Grafton R (2010) “A Framework for Automatic Mood Classification of TV Programmes” presented at SAMT 2010. • Davies, S. (2010) “Interestingness Detection in Sports Audio Broadcasts” presented at IEEE ICMLA 2010 • Knoiusz, P. & Mikolajcyzk, K. (2011) “Soft Assignment Of Visual Words As Linear Coordinate Coding And Optimisation Of Its Reconstruction Error” presented at ICIP 2011 • Knoiusz, P. & Mikolajcyzk, K. (2011) “Spatial Coordinate Coding To Reduce Histogram Representations, Dominant Angle and Colour Pyramid Match” presented at ICIP 2011 • Davies, S. & Bland, D. (2011) “An Improved Framework for Affective Classification and Browsing of Large Scale Broadcast Archives” presented at ACM SIGIR 2011 • Mann, M. & Cox, T. (2011) “Music Mood Classification of Television Theme Tunes” presented at ISMIR 2011 • Davies, S., Mann, M., Cox, T. & Allen, P. (2011) “Musical Moods: A Mass Participation Experiment for Affective Music Classification” presented at ISMIR 2011 • Eggink, J. Allen, P. & Bland, D. (2011) “A Pilot Study for Mood-Based Classification of TV programmes” presented at ACM SIGAC 2011 • Eggink, J. & Bland, D. (2012) “A Large Scale Experiment for Mood-Based Classification of TV Programmes” presented at ICME 2012 • Available at http://www.bbc.co.uk/rd/publications/whitepapers.shtml
  • 19. R&D © BBC MMVIII Thank you • Questions • Contact; – sam.davies@bbc.co.uk