Big Data. New Physics. - Marist College

© 2016 IBM Corporation

Big Data. New Physics.And Geospatial Superfood

Jeff Jonas, IBM FellowChief Scientist, Context Computinghttp://www.twitter.com/jeffjonaswww.jeffjonas.typepad.com

http://www.twitter.com/jeffjonas

http://www.jeffjonas.typepad.com/


Jeff JonasIBM FellowChief Scientist, Context Computing

Founded Systems Research & Development (SRD) in 1985

Architected, designed, developed roughly 100 systems over the last three decades

– Defense, intelligence– Financial services– Gaming– Law enforcement

Acquired by IBM in 2005

Currently focused on Context Computing, Sensemaking and Privacy by Design (PbD)

2


”The data must find the data and the

relevance must find you.”

3


Trend: Organizations Are Getting Dumber

Time

Incr

easi

ng C

ompu

te P

ower

Sensemaking Algorithms

Available Observation

Space ContextEnterpriseAmnesia

Every two days now we create as much information as we did from the dawn of civilization up until 2003.”

~ Eric Schmidt, CEO Google

4


Trend: Organizations Are Getting Dumber

Time

Incr

easi

ng C

ompu

te P

ower

Sensemaking Algorithms

Available Observation

Space ContextWHY?

5


Algorithms at Dead End.

You Can’t Squeeze Knowledge

Out of a Pixel.

6


No Context

[email protected]

7


Context

“Better understanding something by taking into account the things around it.”

8


I ducked as the bat flew my way.

Another exciting baseball game.

9


In Context

VendorHigh ValueCustomer

Job Applicant

FormerEmployee Bad Guy

[email protected]

10


Context Accumulation

“The incremental process of integrating new observations with previous observations.”

11


Context Accumulating

ContextAccumulation

ContextualizedObservations

Observation(Any kind of data from

any kind of sensor)

12


Context Informs Decisioning

ContextAccumulation


ObservationIn Context

Decisioning

Act

Data Finds Data Relevance Finds YouThe Data is the Question

Observation(Any kind of data from

any kind of sensor)

13


The Puzzle Metaphor

Imagine an ever-growing pile of puzzle pieces of varying sizes, shapes, colors

What it represents is unknown – there is no picture on hand

Is it one puzzle, 15 puzzles, or 1,500 different puzzles?

Some pieces are duplicates, missing, incomplete or have errors

Some pieces may even be professionally fabricated lies

Until you take the pieces to the table, it is nearly impossible to assess the scene

14


Puzzling Images: Courtesy Ravensburger © 2011

270 pieces90%

200 pieces66%

150 pieces50%

6 pieces2%

30 pieces10% (duplicates)

15

© 2016 IBM Corporation16



First Discovery

18


More Data Finds Data

19


Duplicates in Front Of Your Eyes

20


First Duplicate Found Here

21



Incremental Context – Incremental Discovery

6:40pm START

22min “Hey, this one is a duplicate!”

35min “I think some pieces are missing.”

37min “Looks like a bunch of hillbillies on a porch.”

44min “Hillbillies, playing guitars, sitting on a porch, near a barber sign and a banjo!”

23


150 pieces50%

24


Incremental Context – Incremental Discovery

47min “We should take the sky and grass off the table.”

2hr “Let’s switch sides, and see if we can make sense of this fromdifferent perspectives.”

2hr10m “Wait, there are three … no, four puzzles.”

2hr18m “I think you threw in a few random pieces.”

25





PUZZLING EXPERIMENT #3ADULTS AT SIBOS 2011

29


SIBOS Conference 2011

100 executives, 10 teams 10 puzzles, 10 small tables Duplicate and missing pieces

Lessons:1. They learned federated

search bites.2. I watched as an early bias

misdirected their attention … but then over time new observations corrected this bias.

30

http://jeffjonas.typepad.com/jeff_jonas/2010/07/when-federated-search-bites.html


How Context Accumulates

With each new observation one asserts: 1) Un-associated; 2) near neighbors; or 3) connected

Must favor the false negative

New observations sometimes reverse earlier assertions

Some observations produce novel discovery

The emerging picture helps focus collection interests

As the working space expands, computational effort increases

Then given sufficient observations there comes a tipping point whereby decision certainty increases while compute effort decreases!

31


Overstated Population

Observations

Uni

que

Iden

titie

s

True Population

32


Counting Unique Identities is Difficult

File 1 File 2

Mark SmithDOB: 6/12/1978

SSN: 1234

Mark R Smith707.433.0000

DL: 5678

33


The Rise and Fall of a Population

Observations

Uni

que

Iden

titie

s

True Population

34


Context Accumulation

Mark SmithDOB: 6/12/1978

SSN: 1234

Mark R Smith707.433.0000

DL: 5678

File 1 File 2

New Record

Mark Randy SmithSSN: 1234DL: 5678

35


Data Triangulation

File 1

NAME DOB SSN DL PHONE

Mark Smith 6/12/1978 1234

Mark R Smith 5678 707.433.0000

Mark Randy Smith

1234 5678

36


Big Data [in context]. New Physics.

More data: better the predictions– Lower false positives– Lower false negatives

More data: bad data good– Suddenly glad your data is not perfect

More data: less compute

37


Big Data

Pile of ______ Information In Context38


One Essential Form of Context: “Entity Resolution”

Is it 5 people each with 1 account or is it 1 person with 5 accounts?

Is it 20 cases of SARS in 20 cities or one case reported 20 times?

If one cannot count, one cannot estimate vector or velocity (direction, speed).

Without vector and velocity prediction is nearly impossible.

39


Who is Fang Wong?

Fang WongTop 100 Customer

F A WongSeattle, DOB: 6/12/82

Former Customer

@FangWong2.5M Followers

[email protected] Subscriber

Fang [email protected] Department’s

Prospect List

40


Resolving the Fang Wong

Fang WongTop 100 Customer

F A WongSeattle, DOB: 6/12/82

Former Customer

@FangWong2.5M Followers

[email protected] Subscriber

Fang [email protected] Department’s

Prospect List

41


Resolving the Fang Wong

Fang WongTop 100 Customer2.5M Followers

Newsletter Subscriber

42


Graphing the (resolved) Fang Wong

Bill SmithMember of the Board

Employee

Customer

Customer

FraudsterFang Wong

Top 100 Customer2.5M Followers


43


Contextualizing Sandy MadenEntity Analytics + Events, Transactions, Space/Time, Etc.


Sandy MadenNew Customer

Employee

Lives With

Co-signer

FormerEmployee

(term no rehire)

Customer Customer

Customer

FraudsterFang Wong



44


“Entities”


Sandy MadenNew Customer

Employee

Lives With

Co-signer

FormerEmployee

(term no rehire)

Customer Customer

Customer

FraudsterFang Wong



Company

Boat

Plane

RouterCar

Asteroid

45


New Think About Entity Resolution

Exactly Same

Fuzzy

IncompatibleFeatures

Deceit

Bob Jones123455

Bob Jones123455

Bob Jones123455

Robert T Jonnes000123455

Bob Jones123455

Bob@TheCo

Bob Jones123455

Ken Wells550119

46


Key Features Enable Entity Resolution

Name License Plate No. Serial NumberAddress VIN MAC AddressDate of Birth Make IP AddressPhone Model MakePassport Year ModelNationality Color Firmware VersionBiometric Etc. Etc.Etc.

People Cars Router

47


Consider Lying Identical Twins

#123Sue3/3/84UberstanExp 2011

PASSPORT#123Sue3/3/84UberstanExp 2011

PASSPORT

Fingerprint

DNA Most TrustedAuthority

“Same person –trust me.”

Most TrustedAuthority

48


The same thing cannot be in two places … at the same time.

Two different things cannot occupy the same space … at the same time.

49


Space & Time Enables Absolute Disambiguation

Name License Plate No. Serial NumberAddress VIN MAC AddressDate of Birth Make IP AddressPhone Model MakePassport Year ModelNationality Color Firmware VersionBiometric Etc. Etc.Etc.

People Cars RouterWhen When WhenWhere Where Where

50


“Life Arcs” Are Also Telling

Bill Smith4/13/67

Salem, Oregon

Bill Smith4/13/67

Seattle, Washington

Address HistoryTampa, FL 2008-2015Biloxi, MS 2005-2008NY, NY 1996-2005Tampa, FL 1984-1996

Address HistorySan Diego, CA 2005-2015San Fran, CA 2005-2005Phoenix, AZ 1990-2005San Jose, CA 1982-1990

51


OMG

52


Space-Time-Travel

Cell phones are generating a staggering amount of geo-locational data – 600B transactions per day being created in the US alone

This data is being “de-identified” and shared with third parties – in volume and in real-time

Your movement quickly reveals where you spend your time

Re-identification (figuring out who is who) is somewhat trivial

And, oh so powerful predictions …

53


The 10 People I Spend the Most Time With(Not at Home and Not at Work)

Michelle Renee Peggy Erin Joshua Ivan Bob Amanda Dane Wesley

He must be following me!

54


Unfair Advantage?

The Uberstan intelligence service preempts the next mass protest in real-time

A political opponent is crushed and resigns two days after announcing their candidacy

55


Consequences

Space-time-travel data is the ultimate biometric

It will enable enormous opportunity

It will unravel one’s secrets

It will challenge existing notions of privacy

Adoption is now accelerating at a blistering pace

56


Introducing

Space-Time-Boxes (STB’s)

57


Using the Geohash Function

4 byte geohash w21z = +/- 20km6 byte geohash w21z4y = +/- 610m

58


4 Byte Geohashes (+/- 20km)

59


6 byte Geohashes (+/-610 meters)

60


8 Byte Geohashes (+/- 19 meters)

61


General Purpose Space-Time-Box (STB) Scale

Global Hierarchy Ever Yr Mon Day Hr 15min 5min 1min 5sec 500ms 100ms

2 char Geohash(630km)

X X X X X X X X X X X





5 char Geohash(2.4km)


6 char Geohash(610m)


7 char Geohash(75m)


8 char Geohash(19m)


9 char Geohash(2.83m)


10 char Geohash(59cm)


62


Example: STB’s for Banking Interactions in Singapore

Online banking (+/- 2.4km)Merchant transactions (+/- 610m)ATM (+/- 610m)Mobile services (+/- 610m)

“W21XX|2012”STB(1.192353, 103.421235, 2012-07-08 00:17:37)

“W21ZWQ|2011-12”STB(1.212363, 103.591156, 2011-12-29 10:56:09)

“W21Z73|2012-07”STB(1.2843622, 103.86103, 2012-07-04 15:08:12)

63


Toying with Publically Available Cell Phone Data

35,831 Call Data Records (CDRs)– 6 months: From 08-31-2009 through 02-27-2010

18,391 Total Number of Usable CDR’s– Excluded CDRs with missing latitude, longitude, time, flow, or accuracy>250 meters

2,444 Hangouts– Minimum of 2 events, spanning at least 15 minutes, in a 610m STB

The Pattern of Life– 130 Hangouts total– 64 Hangouts 3 or more times

Ummm … seems we are living in

habitrails!

64



Malte Spitz’s Hangouts

66


Getting to Know Malte Spitz

Six months of my life in 35,000 recordshttp://www.malte-spitz.de/blog/4103927.html

67

http://www.malte-spitz.de/blog/4103927.html


MARITIME DOMAIN AWARENESS

68


Data Sources

Vessel Registry

Real-time AIS

Other Sources

Historical AIS

Watch List

Sensemaking forMaritime Domain

Awareness

Arrival Notifications

AnnouncedCrew

BusinessListings

69


“Space-Time Box” Densities

Global Hierarchy Ever Yr Mon Day Hr 15min 5min 1min 5sec 500ms 100ms







5 char Geohash(2.4km)


6 char Geohash(610m)


7 char Geohash(75m)


8 char Geohash(19m)


9 char Geohash(2.83m)


10 char Geohash(59cm)


70


Vessels Hovering in Space-Time-Boxes

71


Example Insights

Insight Code Description ScoreWL_ENT Watch Listed Entity +85WL_RELATIONS Watch Listed Relationship +50IDENTITY_DECEIPT Identity Deceit +45UNFAMILIAR Unfamiliar Vessel +30CONFUSION Confused Entity +25HAZ_CARGO Hazardous Cargo +20POL_CHANGE Pattern of Life Change +10FAMILIAR Familiar Vessel -20TRUSTED_ENT Trusted Entity -40

72


Insights

Developing Actionable Intelligence

!

RelevantActionable

Insights accumulate

Enabling relevance detection

Prioritized with a max number of items

Hazardous CargoPattern of Life Change

Identity Deceit

73


Sensing and Responding

74


ASTEROID HUNTING

75


Single Detection

Image courtesy of: Eva Lilly, Institute of Astronomy, University of Hawaii76


From Orphans to Orbits

Backcasting

Single Detections(trash)

TrackletteTrack

OrbitForecasting

Named entity: S100ZUtza

Single Detection(orphan)

Anticipation

77

© 2016 IBM Corporationhttp://www.space.com/7854-slam-asteroids-suspected-space-collision.html

78


"We have directly observed a collision between asteroids for the first time, instead of having to infer that they happened from million-year-old remains."

Colin SnodgrassPlanetary Scientist

Max Planck Institute for Solar System Research

79

http://www.space.com/7854-slam-asteroids-suspected-space-collision.html


Geospatial Context via “Space Time Boxes”

80


Detecting Colocation

TIME1 day

1 hour

Determine encounter distance and time

Space Time Boxes

81


Asteroid vs. Asteroid Encounters

Encounter Distance Asteroid 1 Size Asteroid 2 SizeMay 1, 2032 299km 00A9170 2-4km 0008758 4-9km

Nov 24, 2016 449km 00P5634 1-2km 0055711 2-5km

Jan 11, 2018 449km K08E88J 530-1200m 00N0062 2-4km

82


Computing 600k Asteroid Interactions over 25 Years

4-5 orders of magnitude improvement

Initial Analysis

Adding 1 New Trajectory

Space-Time Box Method

2,880 CPU hours

15 CPU minutes

N-body Simulation Method

10,000,000 CPU hours

4,000 CPU hours

83


Collaborator Call-out

Mudhakar SrivatsaIBM Research

Raghu GantiIBM Research

Sexy Bald Guy

84


June 12th, 2015

Hi Jeff & the gang,

I have great news! On Tuesday I happened to observe a close encounter you guys predicted - one 1 km and the other one 2 km in diameter!

To my knowledge this is the first case ever of direct observation of a close encounter in the small main belt asteroids. The closest point of encounter unfortunately happened during bright daylight in Hawaii, so I missed it …

Cheers!Eva -

Image courtesy of Eva Lilly, Institute of Astronomy, University of Hawaii

86

© 2016 IBM CorporationImage courtesy of: Eva Lilly, Institute of Astronomy, University of Hawaii


[Theatrical Pause]

88


IN CLOSING

89


Action

Red Analytics

Green Analytics

Blue Analytics

ObservationSpace

Old School: Isolated Analytics

90


ObservationSpace

ActionInformationIn Context

Next: General Purpose Sensemaking

Data Finds Data Relevance Finds You

Sensemaking

91


ObservationSpace

ActionInformationIn Context


Helping Focusing Human Attention

Sensemaking

General Purpose• Threat & Fraud• Marketing• Asteroids

Simultaneously!

92


Sensemaking Architecture

Deep Reflection

DiscoveredPatterns

ContextAccumulation



Decisioning

ActObservation(Any kind of data from

any kind of sensor)


Data MiningMachine Learning

Feature Extraction Transformation

Scoring & Predictive ModelsEvent Processing

Entity ResolutionRelationship Graphing

93


The most competitive organizations

are going to make sense of what they are observing

fast enough to do something about it

while they are observing it.

94


Related Blog Postswww.jeffjonas.typepad.com

Data Finds Data

Puzzling: How Observations Are Accumulated Into Context

Big Data. New Physics.

Fantasy Analytics

G2 is 4

95


http://jeffjonas.typepad.com/jeff_jonas/2009/07/data-finds-data.html

http://jeffjonas.typepad.com/jeff_jonas/2012/11/fantasy-analytics.html

http://jeffjonas.typepad.com/jeff_jonas/2015/01/g2-is-4.html


“No one writes bomb on manifest!”

96


Email: [email protected]

Blog: www.jeffjonas.typepad.com

Twitter: http://www.twitter.com/jeffjonas

Questions?

mailto:[email protected]









WIDENING OBSERVATION SPACESA SNEAK PEEK INTO MY CURRENT WORK

99


Dealing with Probabilities

Deep Reflection

DiscoveredPatterns

ContextAccumulation



Decisioning


any kind of sensor)

Certainty6.25%

100


Additional DataOriginal Observations

Dealing with Probabilities

Mark Smith123 Main StreetSanta Rosa, CADOB: 5/12/1974

Mark SmithSanta Rosa, CA702.433.8871

Confirmed across 3 credit bureaus:Mark Smith123 Main StreetSanta Rosa, CADOB: 5/12/1974702.433.8871

Confirmed across two data aggregators:Mark Smith, Santa Rosa, 05/12/74- Only one observed123 Main Street, Santa Rosa, CA- No other Marks- No other Smiths702.433.8871- Exclusive to Mark Smith

(*) 16 Mark Smiths live in Santa Rosa, CA [ref: http://www.intelius.com/results.php?trackit=63&ReportType=1&qf=Mark&qmi=&qn=Smith&qs=CA&qc=Santa+Rosa]

Certainty6.25%*

Decision Certainty


Using Curiosity to Increase Decision Certainty

Deep Reflection

DiscoveredPatterns

ContextAccumulation



Decisioning


any kind of sensor)

SelectiveCuriosity

Figure Out Who to Ask If yes

Make Request(s)

Assembly of Responses into Observations

Certainty 6.25%

Is it worth being curious

about?

102


Before

Deep Reflection

DiscoveredPatterns

ContextAccumulation



Decisioning


any kind of sensor)

Certainty 6.25%

103


After

Deep Reflection

DiscoveredPatterns

ContextAccumulation



Decisioning


any kind of sensor)

Decision Certainty

104


Why Selective Curiosity MattersPatent US8620927

There are many domains where even 99% accuracy is not good enough e.g.,– Healthcare– Financial Services– National Security– Autonomous Vehicles

In the coming era of Internet of Things, robots, and cognitive computing “decision certainty” is going to make or break these advances.

Selective Curiosity will make this possible …

105


SELECTIVE CURIOSITY IN ACTIONA TRUE STORY


One Day …

A lady we will call “Jane Doe” friends me on Facebook. Not sure I know her, I check her About page and see she works for the US Government. But that’s it.

This peeks my curiosity.

Who is this and where does she work?


Next …

I notice we have a friend in common named “Stu Shea.”

I sat on Stu Shea’s Board of Directors at the United States Geospatial Intelligence Foundation (USGIF). He knows interesting people.

I find myself all that much more curious: Who is “Jane Doe?”


So …

I Google [“Jane Doe” and “Stu Shea”].

Ah Ha! She works for the National Geospatial Agency (NGA).

FYI: Job title, email address and more are easily found with one more search!


Email: [email protected]

Blog: www.jeffjonas.typepad.com

Twitter: http://www.twitter.com/jeffjonas

Questions?

mailto:[email protected]








Date post:	31-Dec-2021
Category:	Documents
Upload:	others
View:	3 times
Download:	0 times

Big Data. New Physics. - Marist College

Documents