What is the big deal with big data? Why is everyone talking about it? What, if anything, is anyone doing with it?
This session will discuss big data, starting with a definition of the 4 Vs and diving into the current and potential uses in personalized communication.
What is different from traditional data management and business intelligence is the sheer size of the datasets and the quality of sources of relevant data.
Each source has different structures, and the frequency of updates is faster than ever before. How can all of data from all facets of human activity be related? How can they be combined and analyzed to help us understand individuals and how they want to be communicated to individually?
2. ● What is Big Data and why is it important?
● How is Big Data being used for Marketing?
● Big Data is a driver of Artificial Intelligence?
● What is a Graph? Graph Database?
Accepting questions
goo.gl/slides/zzjzkj
3. ...or is it just ¯_(ツ)_/¯
❏ Big Data
❏ Semantics
❏ Patterns
❏ Paths
❏ Answers
❏ Insights
6. Beyond the Hype is Big Data Analytics
http://www.sciencedirect.com/science/article/pii/S0268401214001066
Text analytics
techniques that extract
information from textual
data.
● Information extraction
● Text summarization
● Question answering
● Sentiment analysis
Social Media
analytics
analysis of structured and
unstructured data from
social media channels.
● Community detection
● Social influence
analysis
● Link prediction
Predictive
analytics
techniques that predict
future outcomes based on
historical and current
data.
● Regression
techniques
● Machine learning
techniques
8. Why Big Data: Big Actionable Insights
Big Data NoSQL databases like MongoDB,
CounchDB, Cassandra, DynamoDB, MarkLogic,
and Neo4j.
Big Data processing tools such as Apache
Hadoop, HDFS, HBase, MapReduce , Spark...
“data mining,”
“data modeling”
“predictive modeling.”
9. ● Big Data often uses a different,
simpler, semantic data model
● Data is easily added and similar but
different data is relatable
● Powerful tools allow new
knowledge to be discovered and
explored
10. Semantics
əˈ
Semantic data models utilize Graph data structures to
link things to properties and to other things (think
things not strings).
With the form Object - RelationType - Object.
For example:
14. Big Data and Linked Data
● Semantic data models basis for Linked Data
● Open Datasets can extend LD objects
● Linked Open Data (LOD) repositories offer
50B+ triples with 10B in DBpedia alone
15. Big Data -> Artificial Intelligence
1. Big Data
2. Cheap parallel computation
3. Better algorithms
“Fueled by technology advancements (e.g. big
data processing power, advanced machine
learning, predictive analytics and natural
language processing) and by the marketing
engines of tech heavyweights, media are latching
onto AI as the next big technology trend.”
https://www.wired.com/2014/10/future-of-artificial-intelligence/
16. Artificial Intelligence Marketing Race
AI in common use
● Search
● Recommendation Systems
● Programmatic Advertising
● Marketing Forecasting
● Speech / Text Recognition
● Recommendations
● Fraud and data breaches
● Social semantics
● Website design
● Product pricing
● Predictive customer service
● Ad targeting
● Speech recognition
● Language recognition
● Customer Segmentation
● Sales forecasting
● Image recognition
● Content generation
● Bots, PAs and messengers
AI rapidly developing
● Image recognition
● Customer Segmentation
● Content Generation
● Personalization
● Personalize Content,
● Recommendations and
● Site Experiences
● Lifetime Value (LTV) Algorithms
● Whole Journey Optimize
● Personalized Recommendations
● A/B/N Testing to Create Unique,
Optimized Experiences
17. Will AIs want to use Electric Toasters?
“Blade Runner: Do Androids Dream of Electric Sheep? “
“AI is the new electricity,” he says. “Just as 100 years ago
electricity transformed industry after industry, AI will now do the
same.”
Why Deep Learning is Suddenly Changing Your Life
“AI is like electricity, and that when it was first incorporated into
appliances they were referred to by names such as “the electric
toaster.” Now it’s just a toaster. ”
Salesforce Einstein Proves that AI is Relative
21. Append Enhance Expand Infer
AI as a Service IBM
● IBM AlchemyLanguage
● IBM Conversation
● IBM Retrieve and Rank
● IBM Personality Insights
AI as a Service Google
● Prediction API
● Sentiment Analysis
● Purchase Prediction
● Spam Comment Detection
AI as a Service Microsoft
● Computer Vision API
● Emotion API
● Face API
● Bing Speech API
● Linguistic Analysis API
● Text Analytics API
● Recommendations API
AI as a Service Amazon
● Content Personalization
● Propensity Modeling
● Customer Churn Prediction
● Solution Recommendation
● Amazon Alexa
22. Personality Propensity?
● Analytics vendors user
Personality Profiles for
messaging / targeting
● Richer models helped
marketers to
understand and predict
behavior
● Use data that is
available in datasets
such as Acxiom and
Experian
● Leverage digital
content such as
individual writing
example or
self-improvement
surveys
23. Example: IBM Personality Insights
You are likely to...
● be sensitive to ownership cost
when buying automobiles
● have spent time volunteering
● prefer quality when buying clothes
You are unlikely to...
● prefer safety when buying
automobiles
● volunteer to learn about social
causes
● be influenced by brand names
when making product purchases
26. Relational databases cannot easily have
new varieties of data added
Similar but not exact data was difficult to
associate, align, understand
Richer semantic models can generate new
understanding, and questions
New questions generate more data, and
knowledge - processes increasing
autonomous
27. Answers
● Information Extraction
● Deep Learning
Knowledge Bases
● Pathfinding and Scoring
● Speech Recognition
● Natural Language
Processing
● Reasoners and Question
Answers
29. So what are the questions?
● How do marketers define successful customer experiences?
● How do customers define successful interactions with
brands?
● Does everyone want the same things?
● Isn’t the best price for the best product good enough?
● So many questions! Q&A conversations led to new
questions and to new insights about the nature of the
conversation.
32. If Answers are Easy...
A lesson of big data is that finding
answers to those questions is increasingly
trivial with AI based machines.
The challenge is to ask the right
questions.
As we'll see later the right question for
personalizing messaging are Who, What
and How?
33. Insights
What is the next best message
How can information be linked and analyzed to help
us understand individuals and how they want to be
communicated to individually?
How do I move from personalized communication to
individualized conversations?
34. Customize, Personalize, Emotionalize
7 Questions with suggestions for ...
● What are the intended outcomes for each
step?
● What data can we use as inputs to insight
generation?
● What AI / Big Data Tools that can be
considered?
35. Next Best Message 7 Questions
Why are we generating a message or conversation?
What do we start or continue a conversation about?
Who are we having a conversation with?
Where
is the best place to send message / have a
conversation?
When is the best time to send the next message?
With
individualized information do we communicate
personally?
How does an individual want to be talked with?
36. Why are we generating a message or conversation?
● Outcome
○ Triggering
○ Conditions
● Input
○ Campaign Map
○ Transaction History
○ Behavioral Event
● Services
○ IBM Conversation
○ Microsoft Bot Framework
○ Google DeepMind
○ Amazon Machine Learning
37. What do we start or continue a conversation about?
● Outcome
○ Campaign Trigger
○ Message Type
● Input
○ Segmentation Cluster
○ Campaign Persona
● Services
○ IBM Retrieve and Rank
○ Microsoft Text Analysis API
○ Google Purchase Prediction
○ Amazon Propensity Modeling
38. Example: Myers-Briggs Type Indicator
“THE ARCHITECT”
INTJ personality types think
strategically and see the big
picture.
Have original minds and great drive
for implementing their ideas and
achieving their goals. Quickly see
patterns in external events and
develop long-range explanatory
perspectives. When committed,
organize a job and carry it through.
Skeptical and independent, have high
standards of competence and
performance - for themselves and
others.
39. Who are we having a conversation with?
● Output
○ Segmenting
○ Audience
● Input
○ Campaign Recipients
○ Segment Candidates
○ GeoTargeted Customers
● Services
○ IBM AlchemyLanuage
○ Microsoft Linguistic Analysis
○ Google Prediction API
○ Amazon Churn Prediction
40. Example: PersonicX® Cluster Perspectives
Cluster #5: Active & Involved
Active & Involved households are wealthy
empty nesters. At a mean age of 60, they are
extremely well educated and still well
compensated in professional and managerial
white-collar jobs, as well as being active
investors. With a third having lived at their
residence for 6-14 years, and another third for
15+ years, these homeowners are well
established in their communities. They are likely
to own a recreation vehicle and enjoy travel to
Hawaii and to national parks. Their substantial
discretionary time and money are spent on
high-quality clothing, dining out, golf and live
theater. However, they are also community
activists, belonging to charitable, religious and
civic organizations.
41. Where is the best place to send message / have a conversation?
● Outcome
○ Channeling
○ Medium
● Input
○ GeoFencing
○ Device Preferences
○ Geography profile
● Services
○ IBM Conversation
○ Microsoft Entity Linking
○ Google Sentiment Analysis
○ Amazon Alexa
42. When is the best time to send the next message?
● Outcome
○ Customizing
○ Event Trigger
● Input
○ Campaign Map
○ TOD Best Practices
○ Preferences
○ Behavioral profile
● Services
○ IBM Conversation
○ Microsoft Entity Linking
○ Google Prediction API
○ Amazon Machine Learning
43. With which individualized information do we communicate personally?
● Outcome
○ Personalizing
○ Message Content
● Input
○ Cluster attributes
○ Demographic profile
○ Psychographic profile
○ Personality profile
● Services
○ Amazon Content Personalization
○ Microsoft Recommendation API
44. Example: DiSC Profile Comparison
Jeff Stewart John Leininger Eric Remington
Disc: Dci Disc: Isd Disc: Cdi
is fairly aggressive,
methodical, and
results-driven, but can be
approachable and
supportive of others.
thrives in an unstructured
environment, loves exploring
new ideas, and occasionally
makes gut-driven decisions
that might seem risky.
is analytical, inventive, and
craves tough problems to
solve, but you can bore him
easily with predictability.
Do: focus on a single,
clear message (ex: "I am
reaching out to get your
opinion.")
Do: use personal anecdotes
and information (ex: "I used
to work in the same industry
and want to get your
perspective")
Do: ask straightforward,
even yes or no questions
(ex: "Would you like to meet
about this?")
Don't: make any claims
that cannot be backed up
with proof (ex: "Our mutual
friend wanted us to
connect.")
Don't: be overly formal and
cold (ex: "I have 30 minutes
to review this information.")
Don't: use anecdotal
expressions (ex: "I thought
you might like this.")
45. How does an individual want to be talked with?
● Outcome
○ Emotionalizing
● Input
○ Psychographic profile
○ Temperament profile
● Services
○ IBM Personality Insights
○ Google Prediction API
○ CrystalKnows Profile
○ Traxion Customer Insights
46. Example: Traxion Temperament
Characteristics
● extroverted
● enthusiastic
● emotional
● sociable
● impulsive
● optimistic
You want to be the first to
experience something,
and never miss out on an
opportunity.
48. Personas Are Not Personal
Personas are analogies, useful but not personal.
What Is? Perse and meGraph
49. Perse Ontology
Πέρση əː ˈ ɪ
Perse is an ontology and set of classes for
creating and publishing a personalization
profile with multiple facets or dimensions.
60. Right message at the right time in the right
place with the right tone
● Effective use of good data with advanced models and
techniques can provide the margin of victory.
● Semantic models and information enhancement and
discovery can help with understanding how people want
to be communicated with.
● The right message at the right time in the right place
with the right tone can motivate customers along their
customer journey path.
61. Take-a-Ways...
...can I have a ( ͡° ͜ʖ ͡°) ?
❖ What is and Why Big Data
❖ NoSQL and Graph Databases
❖ Big Blue and others Deliver Answers
❖ The Best One is the Next One
❖ Me Per Se
62. More Questions? Contact me @
https://www.linkedin.com/in/jeffreyastewart
Jeffrey Stewart
IT and Management Consultant
Asterius Media LLC
Email: jstewart@asteriusmedia.com
stewjeffrey@gmail.com
Twitter: JeffreyAStewart
LinkedIn: jeffreyastewart
SlideShare: stewtrekk