More Related Content Similar to Cognitive Internet of Things: Making Devices Intelligent (20) Cognitive Internet of Things: Making Devices Intelligent1. Session A3
Cognitive Internet of Things:
Making Devices Intelligent
Swami Chandrasekaran
Executive Architect - CTO Office
IBM Watson Innovations
swamchan@us.ibm.com
@swamichandra
2. Please Note
IBM’s statements regarding its plans, directions, and intent are subject to change or
withdrawal without notice at IBM’s sole discretion.
Information regarding potential future products is intended to outline our general product
direction and it should not be relied on in making a purchasing decision.
The information mentioned regarding potential future products is not a commitment,
promise, or legal obligation to deliver any material, code or functionality. Information
about potential future products may not be incorporated into any contract. The
development, release, and timing of any future features or functionality described for our
products remains at our sole discretion.
Performance is based on measurements and projections using standard IBM benchmarks
in a controlled environment. The actual throughput or performance that any user will
experience will vary depending upon many factors, including considerations such as the
amount of multiprogramming in the user’s job stream, the I/O configuration, the storage
configuration, and the workload processed. Therefore, no assurance can be given that
an individual user will achieve results similar to those stated here.
© 2014 IBM Corporation Mobility LIVE! 2014 2
3. Topics
• Preface
• Cognitive Computing
• Cognitive Computing & IoT Solutions/Apps
• What? – Why? – When?
– Observe
– Interpret & Evaluate
– Decide
© 2014 IBM Corporation Mobility LIVE! 2014 3
5. Consider the following questions for a moment
• How can I communicate and have a dialog with my connected
devices using natural language?
• Will the devices be able respond back to me with a degree of
confidence and contextually?
• Can these devices understand the end user at a deeper
psychographic segmentation level and alter the resonance of
the responses?
• How can I discover new insights from my customer / end user
interactions in a timely fashion?
• Will the wearables based apps learn from new data &
observations and get intelligent over time?
© 2014 IBM Corporation Mobility LIVE! 2014 5
7. Not …
Yet !
© 2014 IBM Corporation Mobility LIVE! 2014 7
8. Analytics Russian Doll
Cognitive
Tell me the best course of action
Prescriptive
What should I do to for the best outcome?
Predictive
What could happen?
Descriptive
What has happened?
Business Value
© 2014 IBM Corporation Mobility LIVE! 2014 8
9. This figure resembling a droid was purely unintentional J
So what are the
Characteristics of a
Cognitive System
Scale in
Proportion
Engage in a
Dialog
Generate &
Evaluate
Hypothesis
Understand
Natural
Language
Provide
Supporting
Evidence
Ingest Variety
of (Big) Data
Respond
with
Degree of
Confidence
Learn with
Every
Interaction
Offer
Contextual
Guidance &
Insights
Support for
Decision
Making
Understand
user at a
Deeper level
Relate
between
Terms &
Concepts
© 2014 IBM Corporation Mobility LIVE! 2014 9
10. Consider this Natural Language Question
A restaurant in
Chicago?
Several critics have raved about Zhivago and what a
masterpiece it was. Was it shown in Russia in 2001?
Are we talking about
Art or Sculpture or
Movie or Food?
Plain Number (or)
a Temporal
Reference?
Geographic
Reference?
Keyword search and expert systems are not able to recognize the subtleties,
idiosyncrasies, and ambiguities inherent in common human language
© 2014 IBM Corporation Mobility LIVE! 2014 10
11. This is how a Cognitive System like IBM Watson would
respond with movies related content ingested as Corpus
… and other
possible answers
… and other possible
answers
With a level of
confidence … and Evidence
© 2014 IBM Corporation Mobility LIVE! 2014 11
12. Cognitive systems enhance our abilities to observe, evaluate
and decide
Observe:
• Learns from a vast body of (unstructured) content
• Looks beyond the surface
• Understand Natural Language
Interpret & Evaluate:
• Generate & Evaluate hypotheses
• Finds relationships between terms and concepts
• Simplifies complex thinking
Decide:
• Understands with me at a deeper level
• Evaluates pros and cons. Helps me discover new ideas
• Lets me be myself & engages with me personally
Learning:
• Learns from every interaction
• Trains with experts and improves with feedback
Observe
Decide
Interpret
&
Evaluate
© 2014 IBM Corporation Mobility LIVE! 2014 12
14. Anatomy of an IoT Solution / Application - Setting Context
Systems Integration
Applications
Cognitive Services
Data at Rest Analytics
Focus of
the session
✔
Data Ingestion & Streaming Analytics
Connectivity Management
Network
Devices / Sensors
Platform &
Services
Connectivity
& Devices
Users of Things
Wizard’s
stitching the
perfect
Composable
Apps
Platform,
Services,
IoT Cloud
Providers of
Connectivity
Makers of Things
© 2014 IBM Corporation Mobility LIVE! 2014 14
15. Introducing Cognitive Internet of Things (IoT)
• Provide capabilities for IoT apps & solutions to have cognition
• Allows IoT apps & solutions to exhibit characteristics such as,
– Deep natural language understanding
– Accurate & evidence based decisions
– Relating & linguistic analysis
– Maps euphemisms or colloquial terms
– Deeper understanding of user intrinsic preferences / characteristics
– Communicating with resonance
– Knowledge & relationships discovery
– Continuous learning
© 2014 IBM Corporation Mobility LIVE! 2014 15
16. Cognitive Enabled IoT Apps / Solutions – Art of the Possible
Connected Car Digital Life Smarter Cities Smarter Care
API Management
Cognitive Services Platform
Observe Interpret & Evaluate Decide
Models | Annotations | Content | Tools
Orchestration
Mediation | Composition | Rules
Device Registration & Connectivity
Data services
Historian | File | Archive
Connectivity | Awareness | Security & Privacy | Asset mgmt
Big Data Analytics
Streaming | Batch Analytics
© 2014 IBM Corporation Mobility LIVE! 2014 16
17. Now let’s see some examples of a how cognitive services &
capabilities can make IoT apps / solutions intelligent …
© 2014 IBM Corporation Mobility LIVE! 2014 17
19. Question & Answer
• Allow end applications users to converse using natural language
• Understand a question in NL, generate and evaluate hypothesis
and respond with degree of confidence and evidence
• Interpret questions & answers user questions directly based on,
– unstructured content (PDF, Word, HTML, TXT)
– primary data sources (brochures, web pages, manuals, etc.)
– selected and gathered into a body of corpus
© 2014 IBM Corporation Mobility LIVE! 2014 19
20. Meet Rosy
• Recently purchased and
installed a smarter thermostat
• Very savvy smart phone user
• She has a question about
restricting only authorized
users to be able to configure
& access the thermostat
• Has a smart phone app that
allows her to pose questions in
natural language and have a
conversational dialog
I need to restrict the access to modify certain capabilities in the
thermostat. How can set it up?
Integrate via API
Q&A Dialog
Interact
using NL
Product Corpus
Manuals
© 2014 IBM Corporation Mobility LIVE! 2014 20
22. Concept Expansion
• Allows IoT apps greater insight across multitude of unstructured documents
• Map euphemisms or colloquial terms to more commonly understood
phrases
• Analyze text and interpret its meaning based on usage in other similar
contexts
• For e.g., a semantic class, such as “drugs” can be expanded to,
– start seed terms à motrin, aspirin, keflex
– post expansion à allegra, lisinopril, metformin, aspirin, equagesic, cimetidine, fiorinal,
vancomycin, avelox, protonix, glimepiride, protonix, verapamil, norco, inderal, hctz, advair
• Well suited for expanding where the unstructured source text does not
contain well formed language (e.g., social media data, email, helpdesk
reports, and other less formal communications)
© 2014 IBM Corporation Mobility LIVE! 2014 22
23. Meet Zhang
• Remotely monitored patient
• Can use an app to interact with
his healthcare providers
• English not first language
• Need for understanding
nuances in his less than formal
communications
• Concept Expansion service
returns a ranked list of
contextually similar terms
• Learned from the provided
'seed list' against the Zhang’s
interaction history
Interact
using NL
informally
Integrate via API
Concept Expansion
Interaction Seed List
History
© 2014 IBM Corporation Mobility LIVE! 2014 23
26. User Modeling
• Use linguistic analytics to extract personality and social traits,
including Big 5, Values, and Needs, from the way that a person
communicates.
• Analyze any digital footprint that the user makes available, such as
email, text messages, tweets, forum posts, and more.
• Leverage cognitive and social characteristics with their
corresponding percentile values as the basis for analyzing
personality and social traits.
• IoT apps can leverage this for targeted customer and end user
interaction and acquisition via personality-driven engagements
(offers, recommendations etc.)
© 2014 IBM Corporation Mobility LIVE! 2014 26
27. Meet Ravi
• Very vocal and maintains a strong
digital presence
• Has a long day at work !!
• Tweets with certain emotions and
walks to the car
• Car has done a psychographic
analysis of his digital footprint and
alters its response resonance
• Recommends or tunes to the
“comedy” radio channel
• Understands from past behavior,
the driver would want to go to the
gym
Walk to Car
Share Location &
User
Context
Modeling
Recommended
Infotainment
Prediction
Past
Interactions
© 2014 IBM Corporation Mobility LIVE! 2014 27
28. Meet Ravi … again
• Very vocal and maintains a
strong digital presence
• Seeks recommendation for
happy hour
• Based on Ravi’s location fetch
digital tweets, blogs, FB
updates about relevant places
• Extract Big 5, Values, and needs
• Provide as input to a a
classification ML model to
recommend apt place for Ravi
Seek
recommendation
Share Location &
User
Modeling
Context
Prediction
© 2014 IBM Corporation Mobility LIVE! 2014 28
31. Relationship Extraction
• Intelligently finds relationships between sentences components
(nouns, verbs, subjects, objects, etc.) in unstructured text
• Extract entities such as person, organizations, locations, devices,
events, etc., and relationships between them
• Relationships help easily understand the meaning & intent of
individual sentences and documents
• Enables automated processes to understand unstructured
content in healthcare, drug discovery, financial reports, news
and blog monitoring, etc.
© 2014 IBM Corporation Mobility LIVE! 2014 31
34. Message Resonance
• Communicate with people with a style & words that suits them
• Analyzes content and score how well it is likely to be received
by a specific target audience.
• Analysis is based on content that’s been written by the target
audience itself (team sports fans, new product users, etc)
• Add capabilities within the IoT apps (smart TV, personal wellness)
to maximize the resonance of your messaging
• Enables IoT apps to best engage with end users and facilitate
consistent messaging
© 2014 IBM Corporation Mobility LIVE! 2014 34
37. How Cognitive Computing will help IoT Solutions & Apps?
• Overcome Complexity [1]
– Ingestion of 4V (volume, variety, velocity, veracity) data esp., unstructured ones
– Uncover hidden relationships (Bad weather à Tweets à Extract Insights)
• Getting the Big Picture
– Extract & Correlate patterns, concepts & relationships across content
– Expansion of concepts
• Augment our Senses [1]
– Converse in natural language
– Deeper understanding of the individual & group
– Chip modeled on the brain
• Support Discovery & Decision Making
– Hypotheses based inferences
– Offer contextual insights
– Support at point of decision making
1 - Smart Machines: IBM’s Watson and the Era of Cognitive Computing - John E. Kelly III and Steve Hamm
© 2014 IBM Corporation Mobility LIVE! 2014 37
39. Some Interesting Links
• IBM Watson
– http://www.ibmwatson.com
• IBM Bluemix
– http://www.bluemix.net
• Node-RED
– http://nodered.org/
• Watson Ecosystem program
– http://www.ibm.com/smarterplanet/us/en/ibmwatson/ecosystem.html
• Growing social dialogue
– Twitter: @IBMWatson
• IBM Watson: A Platform for Innovation in the New Era of Computing
– http://asmarterplanet.com/blog/2014/04/ibm-watson-ecosystem-platform-innovation.html
• IBM unveils a computer that can argue
– http://finance.yahoo.com/blogs/the-exchange/ibm-unveils-a-computer-than-can-argue-181228620.html
© 2014 IBM Corporation Mobility LIVE! 2014 39