Data Architecture Strategies: Artificial Intelligence - Real-World Applications for Your Organization

Artificial Intelligence: Real World
Applications for Your Organization
Donna Burbank, Managing Director
Global Data Strategy, Ltd.
June 28th, 2018
Follow on Twitter @donnaburbank
Twitter Event hashtag: #DAStrategies

Global Data Strategy, Ltd. 2018
Donna Burbank
Donna is a recognised industry expert in
information management with over 20 years
of experience in data strategy, information
management, data modeling, metadata
management, and enterprise architecture.
Her background is multi-faceted across
consulting, product development, product
management, brand strategy, marketing,
and business leadership.
She is currently the Managing Director at
Global Data Strategy, Ltd., an international
information management consulting
company that specializes in the alignment of
business drivers with data-centric
technology. In past roles, she has served in
key brand strategy and product
management roles at CA Technologies and
Embarcadero Technologies for several of the
leading data management products in the
market.
As an active contributor to the data
management community, she is a long time
DAMA International member, Past President
and Advisor to the DAMA Rocky Mountain
chapter, and was recently awarded the
Excellence in Data Management Award from
DAMA International in 2016.
Donna is also an analyst at the Boulder BI
Train Trust (BBBT) where she provides advice
and gains insight on the latest BI and
Analytics software in the market. She was on
several review committees for the Object
Management Group’s for key information
management and process modeling
notations.
She has worked with dozens of Fortune 500
companies worldwide in the Americas,
Europe, Asia, and Africa and speaks regularly
at industry conferences. She has co-
authored two books: Data Modeling for the
Business and Data Modeling Made Simple
with ERwin Data Modeler and is a regular
contributor to industry publications. She can
be reached at
donna.burbank@globaldatastrategy.com
Donna is based in Boulder, Colorado, USA.
2
Follow on Twitter @donnaburbank
Twitter Event hashtag: #DAStrategies

DATAVERSITY Data Architecture Strategies
• January - on demand Panel: Emerging Trends in Data Architecture – What’s the Next Big Thing?
• February - on demand Building an Enterprise Data Strategy – Where to Start?
• March - on demand Modern Metadata Strategies
• April - on demand The Rise of the Graph Database
• May - on demand Data Architecture Best Practices for Today’s Rapidly Changing Data Landscape
• June Artificial Intelligence: Real-World Applications for Your Organization
• July Panel: Data as a Profit Driver – Emerging Techniques to Monetize Data as a Strategic Asset
• August Data Lake Architecture – Modern Strategies & Approaches
• Sept Master Data Management: Practical Strategies for Integrating into Your Data Architecture
• October Business-Centric Data Modeling: Strategies for Maximizing Business Benefit
• December Panel: Self-Service Reporting and Data Prep – Benefits & Risks
3
This Year’s Line Up for 2018

Some Basic Definitions
• Data science produces insights1 based on human analysis, e.g.
• Descriptive or Exploratory: “75% of our customers are based in New England”
• Causal: “Customers in New England are less likely to purchase during a snowstorm”
• Machine learning produces predictions based on models & training data, e.g.
• Predict whether this patient is likely to go into remission
• “Predict” whether this image has a car in it.
• Artificial intelligence produces actions, e.g.
• Chat Bots
• Recommendation Engines
• Game algorithms
• Robotics
41 David Robinson, Data Scientist at Stack Overflow
What is in a name?
Some
overlap

Artificial Intelligence Usage Still Emerging
• According to DATAVERSITY’s 2017
Trends in Data Architecture survey,
only 18.9% of respondents
indicated Artificial Intelligence or
Machine Learning as a key driver.
5
AI lags behind other trends such as Data Analytics, BI, etc.
From Emerging Trends in Data Architecture, DATAVERSITY, by Donna Burbank & Charles Roe, October 2017

Survey: AI
• Are you currently implementing an Artificial Intelligence (AI) project in your
organization?
• Yes, currently in use
• Planning for future use
• No, not using or planning to use
6

Quality Data is the Foundation for AI
AI / Machine Learning Basics
Gather the Data
• What factors do I
want to focus on?
• Where will I source
the data to train
my model?
• What is the volume
of the data set?
• Etc.
7
Some common basic steps for AI/machine learning
Prepare the Data
• Analyze/Visualize
the data to
understand
patterns,
relationships, etc.
Is it a realistic mix
of factors?
• Randomize the
order.
• Etc.
Choose the Model
• What model is the
best fit for the
scenario at hand?,
e.g.
• Linear Regression
• Logistic Regression
• Naïve Bayes
• Random Forest
• Etc.
Train the Model
• Initialize parameter
values & run the
model with those
values.
• Compare model’s
predictions with
expected output
• Adjust the values
to have more
correct results.
Repeat
Evaluate & Tune
• Run the model
against data it has
never seen.
• Compare to
desired result and
tune parameters as
needed.

Machines Learn Like Humans Do
• Computer algorithms can “learn” just like humans do.
8
In many ways, computers learn the same way we do
This is a DOG, Johnny!
Look at the DOG!
Dog?

Use Case: Machine Learning & Metadata Discovery
• Machine Learning offers ways to automate
tedious tasks that may have been done
manually before:
• e.g. Data Mapping
• SSN -> Field1_SSN
• SSN -> Soc_Num
• Etc.
• Machine Learning Pattern Matching
• NNN-NN-NNNN -> Field_X follows this
pattern, it must be a SSN
9
Source kdnuggets.com
• There is a place for both methods:
• Sometimes you want to define specific mapping rules
• Sometimes you want a pattern-matching, discovery-
style approach.

Machine Learning & Metadata Discovery
10
This is a SSN, Johnny!
Look at the SSN!
SSN?
978-65-1239
097-27-9875
Note: All SSNs are fictitious and do not represent a known individual.
111-11-1111

• In many ways, we “learn” conditions responses to typical questions or situations.
11
What do you
say, Marco?
You say
THANK YOU!
You say
THANK YOU,
Marco!
Marco! Say
THANK YOU!
Mine?

• Most of us generally improve over time…
12
How are you?
I’m fine, and
you?

Chat Bots
• Chat bots are a common way to provide automated
answers to common questions.
13
Automating common questions
Eliza is still learning! Please let me know your experience with the
computer therapist, and anything you might want to see improved.
https://www.cyberpsych.org/eliza/

Quality Data is the Foundation for Chat Bots
Chat Bot Basics
Gather the Data
• Logs from Support
calls can be used.
• Training data sets
can be used.
14
Some common basic steps for building Chat bots
Prepare the Data
• Ensure that the
responses fit the
realistic use cases.
• Randomize the
order.
• Etc.
Choose the Model
• e.g. For Natural
language
processing,
Multinomial Naive
Bayes is often used.
Train the Model
• Train against
conversations.
• Models can learn
over time from real
customer input
Repeat
Evaluate & Tune
• Run the model
against data it has
never seen.
• Compare to
desired result and
tune parameters as
needed.
Sample Training set
class: greeting
“How are you”
“good morning”
“hi there”
Input sentence classification:
input: “How are you”
term: “how” (class: greeting)
Term: “are” (class: greeting)
term: “you” (class: greeting)
classification: greeting (score=3)
I’m fine, and
you?

Image Recognition
• By now, most of us have seen the Muffin or Chihuahua graphic
15
Identifying patterns

Image Recognition
• Labelled data sets can help with training algorithms.
16
• APIs are available to provide image tagging, e.g.
• Amazon’s Rekognition
• Google’s Vision API
• IBM Watson Vision
Photo from aws.amazon.com/rekognition/
Photo from www.image-net.org/

Real-World Use Cases for Image Recognition
Auto-Organizing your Image Library
17
Vacation Photos
Machine & Factory Maintenance
=
Part Number
PHY18374EU
Facial Recognition for Office Security
Entrance Allowed Etc! New use cases
constantly emerging.

Recommendation Engines – The North Face
18
• The North Face uses IBM’s Watson Artificial
Intelligence software to power its Expert Personal
Shopper
• Customized, Personalized Shopping Experience
• Integrates data from multiple sources
Weather data (external)
Location data (external)
Product Master Data (internal)

Artificial Intelligence & Data Quality
19
• Amazon.com’s Recommendation Engine uses Artificial Intelligence
• Based on analyzing data from shopping trends
• Is now available as an Open Source AI Platform - DSSTNE (pronounced “destiny”) - Deep Scalable
Sparse Tensor Network Engine
Product Master Data
Customer Purchasing Patterns

Artificial Intelligence & Data Quality
• Artificial Intelligence is based on evaluating data sets. If those data sets are faulty
or of poor quality, your AI results will be flawed.
• Especially if the data sets are small
20
AI is only as good as the underlying data

Don’t Forget the Business Value
21
Just because you “can” doesn’t mean it’s effective.

Governance & Metadata for Machine Learning/AI
• With Machine Learning (& Data Science), not only the data
needs to be governed with documented metadata, but the
models and algorithms themselves must be documented as
well.
• What data are we using and why?
• What algorithms are being used and what is the logic?
22
Source: David Robinson, Data Scientist at Stack Overflow

Ethics
• Ethics are a key consideration in the usage of Artificial Intelligence, i.e.
• Just because we can, does it mean we should?
• Some considerations
• Privacy – consideration of consumers’ rights
• Errors – how do we ensure a correct result (e.g. self-driving cars, decision algorithms)
• Job Loss – will this replace human staff? Is that a concern?
• Bias – do the training sets and algorithms promote inherent bias?
• Security – can data sets or algorithms be hacked by nefarious sources?
• Control – is there a risk of losing control over the algorithm and its results?
• The “Creep Factor” – perhaps it’s not illegal or doesn’t break official privacy rules, but does it “feel
right”? Would I want to be the consumer in this scenario?
• Etc.
23
Think before you code

Computers can “learn” Bias
24
Consider this fact in selecting your training data sets
Doctor Doctor

Data Governance is Critical for AI
Data Foundation
Quality Data Sets
Semantic Layer
Business Glossary, Data Models,
Labels & Meta Tags
Modeling & Analytical Layer
Modeling Techniques, Variables, Business
Understanding
The Crisp Methodology is one methodology for
governing analytical modelling.
Credit to Data Science Central
Governance is important at
a number of layers in the AI
ecosystem – from the data
to the algorithms.

When to use Artificial Intelligence
• Some guidelines:
• Is it useful in supporting my main business initiatives? – Only if YES
• Do I get to play with some cool technology? – NO, not as a main driver
• Is it ethical? – Only if YES!
• Do I have the right data sets to support it? – Only if YES
26
Can AI help your organization?

27
Business Driver
“Our customer satisfaction rating
are low. What can we do?”
Suggested Resolution
“Let’s implement a facial recognition
program that detects whether a customer
is smiling when they order online!”
• Is it useful in supporting my main business initiatives?
• Hmmm… NO, not really. Would this really help?
• Do I get to play with some cool technology?
• YES, but don’t waste my money and annoy my customers doing it.
• Is it ethical?
• Hmmm… seems sort of Creepy.
• Do I have the right data sets to support it?
• Hmmmm….NO …
• Is smiling when you are ordering the right indicator?
• Only 25% of our customers order online.
• Etc.

28
Business Driver
“A significant percentage of Students
who are accepted to College in the
Spring do not show up in the Fall.”
Suggested Resolution
“Let’s implement a Chat Bot to guide
students through the tough challenges like
financial aid, class registration, etc.”
• Is it useful in supporting my main business initiatives?
• Yes, this is a critical issue, and this could be useful to Students.
• Do I get to play with some cool technology?
• YES, and our target “customer” does, too. Students live on their cell phones.
• Is it ethical?
• Yes, students choose to interact, and we’re providing information that’s available at the
university, just in an easier way.
• Do I have the right data sets to support it?
• Yes, we have the data, we just need to create the right data sets and training to build it
into an intuitive app.
Georgia State University implemented this solution
and saw a significant decrease in “summer melt”.

Summary
• Artificial Intelligence and Machine Learning provide exciting opportunities.
• Image Recognition
• Recommendation Engines
• Chat Bots
• Etc.
• Quality Data is a core foundation for AI
• Ethics and Data Governance are critical
• Choose the right scenario for AI in your organization.

About Global Data Strategy, Ltd.
• Global Data Strategy is an international information management consulting company that specializes
in the alignment of business drivers with data-centric technology.
• Our passion is data, and helping organizations enrich their business opportunities through data and
information.
• Our core values center around providing solutions that are:
• Business-Driven: We put the needs of your business first, before we look at any technological solution.
• Clear & Relevant: We provide clear explanations using real-world examples, not technical jargon.
• Customized & Right-Sized: Our implementations are based on the unique needs of your organization’s
size, corporate culture, and geography.
• High Quality & Technically Precise: We pride ourselves in excellence of execution, and we attract high-
quality professionals with years of technical expertise in the industry.
30
Data-Driven Business Transformation
Business Strategy
Aligned With
Data Strategy
www.globaldatastrategy.com

White Paper: Trends in Data Architecture
31
Free Download
• Download from
www.globaldatastrategy.com
• Under ‘Resources/Whitepapers’

DATAVERSITY Data Architecture Strategies
• January - on demand Panel: Emerging Trends in Data Architecture – What’s the Next Big Thing?
• February - on demand Building an Enterprise Data Strategy – Where to Start?
• March - on demand Modern Metadata Strategies
• April - on demand The Rise of the Graph Database: Practical Use Cases & Approaches to Benefit your Business
• May - on demand Data Architecture Best Practices for Today’s Rapidly Changing Data Landscape
• June – soon on demand Artificial Intelligence: Real-World Applications for Your Organization
• July Panel: Data as a Profit Driver – Emerging Techniques to Monetize Data as a Strategic Asset
• August Data Lake Architecture – Modern Strategies & Approaches
• Sept Master Data Management: Practical Strategies for Integrating into Your Data Architecture
• October Business-Centric Data Modeling: Strategies for Maximizing Business Benefit
• December Panel: Self-Service Reporting and Data Prep – Benefits & Risks
32
This Year’s Line Up for 2018 – Join Us Next Month

Questions?
33
Thoughts? Ideas?

Data Architecture Strategies: Artificial Intelligence - Real-World Applications for Your Organization

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Data Architecture Strategies: Artificial Intelligence - Real-World Applications for Your Organization

Similar to Data Architecture Strategies: Artificial Intelligence - Real-World Applications for Your Organization (20)

More from DATAVERSITY

More from DATAVERSITY (20)

Recently uploaded

Recently uploaded (20)

Data Architecture Strategies: Artificial Intelligence - Real-World Applications for Your Organization