In order to improve audience engagement, media companies must deal with vast amounts of raw data from web, social media, devices, catalogs, and back-channel sources. This session dives into predictive analytic solutions on AWS: We present architecture patterns for optimizing media delivery and tuning overall user experience based on representative data sources (video player clickstream, web logs, CDN, user profiles, social media sentiment, etc.). We dive into concrete implementations of cloud-based machine learning services and show how they can be leveraged for profiling audience demand, cueing content recommendations and prioritizing delivery of related media. Services covered include Amazon EC2, Amazon S3, Amazon CloudFront, and Amazon EMR.
4. • AWS Media Workloads
• AWS architecture for big data workloads
• Media Analytics – Use Cases
• AWS Media Partner Analytics Ecosystem
• Partner Speaker & Demo
• Pronam Chatterjee (CEO)
• Blue Pi Technologies
• Customer Speaker & Demo
• Retesh Gondal - Head Technology, APB News
• Anuj Sharma – Tech Lead, ABP News
Agenda
5. AWS Media Workloads
Content
Production
Content
Distribution &
Consumption
Processing &
Management
Content
Storage
Modelling
Rendering
Video editing
Post production
Broadcast signal
acquisition
Streaming of live
and VOD content
B2B distribution of
content
Insertion of Video
advertising for
live/on demand
content
High speed ingest
Library storage and
archiving
Tier management
Content/asset
management
En/Transcode
Packaging
Encryption,
watermarking
Digital Rights
Management
Consumer
Insight and
Analytics
Analytics,
reporting, log
analysis
Real-time
monitoring
Content discovery
Content
recommendations
Shared IT Services
NetworkSecurity OperationsInfrastructure
Partner Solutions
6. • Netflix: Over 75% of what people watch comes from recommendations [1]
• Nielsen: Social media activity [Tweets] drive higher broadcast TV ratings for
48% of shows [2]
• Google: 70% of the variation in box-office performance can be explained with
movie-related search volume [3]
Content discovery … and the conversation around it …
matter!
[1] http://www.slideshare.net/AmazonWebServices/maximizing-audience-engagement-in-media-delivery-med303-aws-reinvent-2013-28622676
[2] http://www.nielsen.com/content/corporate/us/en/press-room/2013/new-nielsen-research-indicates-two-way-causal-influence-between-.html
[3] http://www.google.com.au/think/research-studies/quantifying-movie-magic.html
Audience Engagement Signals
8. • Descriptive
• Retrospective
• What happened or is happening
• Simple aggregations and counters
• Predictive
• Statistical forecast
• Predict a value in a dataset
• Machine learning
• Prescriptive (emergent)
• What should I do about it?
Descriptive
Predictive
Prescriptive
The Analytics Spectrum
9. Common M&E Analytic Topics of Interest
Content
•Top Content
•Engagement
•Plays per
session
•Drop-off
• Referral path
• Recommendation
Audience
• Acquisitions
• Churn
• Where, when, who
• Segmentation
• Cohorts
Operations
• How much
buffering
• Best CDN paths
• Top devices
• Uniques per
platform
Monetization
• Monetization
• Ad Spend
• Social Media
• Mentions
• Sentiment
Analysis
10. Amazon S3
Amazon Kinesis
Amazon DynamoDB
Amazon RDS (Aurora)
AWS Lambda
Amazon
EMR
Amazon
Redshift
Amazon Machine
Learning
Collect Process Analyze
Store
Data Collection
and Storage
Data
Processing
Event
Processing
Data
Analysis
Data/
Audience
Engagement
Signals
Answers
Typical (Big Data Analytics) Pipeline
11. Typical (Big Data Analytics) Pipeline
Data/
Audience
Engagement
Signals
Answers
Collect Process AnalyzeCollect
Store
14. • Branch of Artificial Intelligence and Statistics
• Programming computers based on historical experience
• Focuses on prediction based on known properties learned from training data
Signals Predictions
Machine Learning for Predictive Analytics
20. • Business Impact:
• 20% traffic increase within 1 month
• 45% of story based traffic came after “related” content widget was placed
• Bounce rate reduced from 63.07% to 53.7%
• Pages per session increased to 3.08 Vs 2.53
• Average session duration increased to 4.09 mins from 3.05 mins
Case Study – Punjab Kesari Group
32. • Types of recommender systems
• Content-based recommendations
• Collaborative filtering recommendations
• User-user recommendations
• Item-item recommendations
• Applications
• Products and services you would like to buy
• People you might want to connect with
• Recommending movies, videos, songs, games,
and events you might like
Recommendations
36. Capture social media signals
Tweets, Comments on FB, IMDB, YouTube
Push through Sentiment Analyzer
Tokenize (words capture) create words from those streams
Classify (estimate) as Positive | Negative
Provide actionable insight
Improve sort order of recommendations
Alert / advise Digital Marketing team
Recommendations
42. • In addition to traditional ABR strategies, anticipate consumer demand for
media downloads
• Based on viewing patterns
• Time of day / season
• Initiate retrieval process ahead of time
• Restore from Amazon Glacier
• Retrieve from Amazon Simple Storage Service (S3) / Amazon CloudFront and
cache on device (STB, Mobile/Native Player App)
Other Applications: Minimize Buffering
57. INTRODUCTION
ABP News network
▸ ABP NEWS previously known as STAR NEWS
▸ Largest regional news network in India
▸ Available in 5 Indian languages Hindi, Punjabi, Marathi, Bengali,
Gujarati
▸ Weekly reach of 48 million users
58. INTRODUCTION
▸ Digital initiative by ABP NEWS
▸ News (abplive.in), Entertainment (FilmyMonkey.com), Cricket
(WahCricket.com)
▸ 300 million views a month across web and mobile
▸ 14 million monthly Video views
▸ Mobile Apps for iOS, Android, Windows
ABP Live
59. OTT AND VIDEO ANALYTICS
▸ In house developed OTT platform
▸ Manage, publish, measure, monetize video content on web and
mobile
▸ Deployed over AWS services
▸ EC2, S3, Elastic Transcoding, Elastic Search
ABP Live: OTT Platform
64. OTT AND VIDEO ANALYTICS
▸ In house developed Video Analytics system
▸ Measure reach, usage and Quality of Service across devices
▸ Deployed over AWS services
▸ API Gateway, Kineses, S3, Redshift, EMR
ABP Live: Video Analytics
69. OTT AND VIDEO ANALYTICS
▸ Quality of services across devices
▸ Measure daily reach over web and mobile devices
▸ Realtime analytics of Live and VoD users
▸ Realtime content categorisation of trending videos
Video Analytics
70. OTT AND VIDEO ANALYTICS
▸ Scalability
▸ Elasticity
▸ Realtime Video Encoding
▸ Easy deployment
▸ Cost effective: Pay only for what you use
Why AWS