Mark Foster

Cognitive Computing and IBM Watson

Senior Vice President, GBS
for Aerospace
Dr. Sanjay K Prasad, IBM
February 18, 2019
THREE ERAS OF COMPUTING

Cognitive
Systems Era

Programmable
Computer Intelligence

Systems Era

212543BR

Tabulating
Systems Era

Time
THREE ERAS OF COMPUTING

Cognitive
Man-computer symbiosis is an expected development
Systems Era in
cooperative interaction between men and electronic co
mputers. It will involve very close coupling between
Programmable
the human and the electronic members of the partnersh
Computer Intelligence

Systems Era
ip………………Preliminary analyses indicate that the symbio
tic partnership will perform intellectual operations
212543BR

Tabulating
muchSystems
moreEra effectively than man alone can perform them
.”

J.C.R. Licklider, “Man-Computer Symbiosis”

March 1960

Time
BIG DATA: WHY WE MUST MOVE TO A NEW ERA OF
COMPUTING
Dimensions of data growth

Terabytes to Streaming data,

exabytes of milliseconds to
existing data seconds to
to process respond
Volume Velocity

Structured, Variety
unstructured,
text, multimedia Veracity
Uncertainty
from inconsistency,
ambiguities, etc.
WHAT IS A COGNITIVE SYSTEM?
It is a System that can

UNDERSTAND REASON LEARN INTERACT

Text, Image, Voice, Analyze, Induce, Recognize Patterns in See, talk and hear –
Data, Streaming Data, Deduce, Infer, Correlate, past interactions, Learn similar to natural, human
Massive Data, Hypothesize, from feedback, and interaction
Seemingly Random Recommend Adapt
Data

IBM Watson is a key platform for building Cognitive Systems 5

COGNITIVE SYSTEMS RELY ON COLLECTIONS OF
DATA AND INFORMATION

Data, information, and expertise

create the foundation
Examples include: Video and image Libraries
Analyst reports Wikis – Knowledge base
RFP Documents Bug alerts
Industry Data - IEEE Blogs
Tweets Tribal knowledge
Patents and Registrations FAQs
Service Manuals International crime database
Service Requests IoT sensor data
E-mails Warranty notes
Texts Field Maintenance Reports
 COGNITIVE IS UNDERPINNED BY MANY ADVANCED
TECHNOLOGIES
Impact of Cognitive on Business
Cognitive Technologies Business Value
Artificial intelligence
Simulation of human intelligence processes

Create new ways of interacting with

Recommendation
engines
Robotics customers
Analyze data and Conception, design,
suggest something as manufacture, and operation
per user’s interest of robots
Reveal more powerful on-demand
Predictive Machine learning business insights through real-time
analytics systems access to data
Predicting outcomes Ability to learn and
using statistical improve without explicit
algorithms and machine instructions
learning
Enable business model – and
Deep Natural language ecosystem – transformation
learning processing
Machine learning with artificial neural Ability to understand human speech
network algorithms as it is spoken
COGNITIVE SYSTEMS APPLY RELEVANT INDUSTRY
INFORMATION
information Watson + New Cognitive Insight
Industry domain • Connections
• Facets
Field Data
Manuals & expertise • Facet Pairs
Instructions • Sentiment
Reports Step 1: Extract data • Deviation/trends
•…
Step 2: Map to facets /
understand data & language
Field Narrative,
Notes Case History & Step 3: Explore data trends,
Tech logs deviations

Step 4: Correlations with

predictive models

Bulletins,
Errata
WATSON USE CASES FOR AIRLINES
10

COGNITIVE DISCOVERY
 …
Ideas

Reading
literature

experiments, simulations,
writing reports/papers/patents,

submissions per month

2000
4000
6000
8000

0
10000
12000

1991-07
1992-08
1993-09
1994-10
1995-11
1996-12
1998-01
1999-02
2000-03
2001-04
2002-05
2003-06
2004-07
2005-08
2006-09
2007-10
2008-11
2009-12
2011-01
FROM SERENDIPITOUS TO DATA-DRIVEN DISCOVERY …

2012-02
2013-03
2014-04
2015-05
2016-06
2017-07
COGNITIVE DISCOVERY

Ideas

Knowledge
Graph

Scalable Knowledge Ingestion:

• Document processing
• Fact Extraction (NLP)
Generate Supporting
Data-Set Knowledge Engineering:

• Knowledge Representation (graphs)

• Knowledge Querying (NLQ)
 COGNITIVE DISCOVERY: KNOWLEDGE Scalable Knowledge Ingestion
• Complex documents
(literature, lab-books)
• Structured DBs
• Public and proprietary • Identify trends:
Run historical queries
• Discover gaps in the
days knowledge:
What is known v.s.
what is not known (or
reported)
• Explore “what if”
• Tables, Images, Formulas, Diagrams…
scenarios: Extract structured data Create technical facts Transform and link
• “What are all the properties of a chemical”
Suggest hypotheses from complex media and information from facts and sources into
– “List all chemical reactions of a certain
type. Who, what, where, when?” and databases extracted data knowledge space

Cognitive Discovery @ IBM POWER

Deep Search Explore relevant
results... AC922
• High-performance Knowledge
Cloud Scalable Knowledge Deep Customer Graph
Apps Ingestion Search Apps • Training models
LC922
• User Interface
Resource / IBM Cloud Private • Databases
WL managers Kubernetes Storage

Data Layer Spectrum Scale Object Store

Do a search... Customize search workflow

• inputs/outputs Hardware P9 + Volta IBM ESS
• logic
• graph connections
14

COGNITIVE AIRCRAFT
MAINTENANCE
IBM Cognitive Aircraft Health and Maintenance
Combine analytics, cognitive and optimization capabilities to declare what
will happen, recommended actions and when to best implement

Prediction Diagnostic Optimization Advisory

+ + +
• Aircraft sensor / system data • Analyze manuals, maintenance, • Optimize corrective actions • Link probable repair to
to provide early warning service bulletins to correlate and maintenance diagnostics procedures
system predictions with probable corrective
• Determine next best actions • Technician assist through
repairs
• Determine time horizon and for allocation of equipment, repair and diagnostics
probability of events • Identifying relevant correlations resources and location
• Connect maintenance
between sensor activity and system
technician with engineering
failures/warnings
assist
Aircraft Maintenance
NR No : 2950896

NR Summary : DURING TAXI, F/CTL FLAP SYS2 FAULT

APPEAR

AC Type / AC No. : 330 / AB8276

Malfunction ATA / Corrective Action ATA : 27-51, 27-51

Corrective Action : MAINT MSG:275134, ACCORDING TO TSM

27-51-00-810-880-A, REPLACED NO.2 SFCC AND THEN SYS
OP’ CHECKED NORMAL PER AMM TASK 27-51-34-400-801-A
AND REMOVED PLACARD

Message No. : 275134

Request Date : 2014-05-04

16
Cognitive Application, Data Sources & API Interactions to support
aircraft health
System Drawings
[structured data]
Cognitive Maintenance Advisory
Maintenance Workflow Process & (Field or Non-Field)
Standards
[unstructured data]

Craft Tools &

Techniques
[unstructured data]

Maintenance
Information
[structured data]
Asset Mfg. Specs., Company
Asset Maintenance Standards
[unstructured data]

SCADA / Data
Historian
Regulatory Standards & Compliance
[structured data
[unstructured data]

Edge
Node
Fog
Node
Streamlining the Aircraft Maintenance Process
Augmented process
Data Upload: Failure Prediction: Diagnostics: Operations Scheduled Normal Flight
Sensor data At every cycle: Cognitive Analytics assesses Planning: Maintenance: Operations
uploaded for analytical models Prior work orders – provides Planes at risk Scheduled at Never
analytical risk score likelihood of proposed actions based prioritized – Best Location interrupted
scoring failure event before upon repair history Minimal Impact
the failure happens

Reliability Management
and planning + Advisor and
Optimization

Unscheduled
Maintenance
Normal Flight Scheduled Failure Disruptions: Normal Flight
Operations Maintenance Alert! • Aircraft On Ground Operations
• Flight Ops Resume
•Delayed flights
•Cancelled flights
•Unhappy customers
As-is process Spare Fleet
Technology creates possibilities and potential, but ultimately, the
future we get will depend on the choices we make. Technology is not
destiny. We shape our destiny.

— Erik Brynjolfsson, MIT