on the personalization of event-based systems

19
Speaker: Opher Etzion [email protected] Joint work with Fabiana Fournier from IBM On the personalization of event-based systems

Upload: opher-etzion

Post on 02-Jul-2015

316 views

Category:

Data & Analytics


1 download

DESCRIPTION

Talk given in ACM Multimedia conference on Human Centered Event Understanding from Multimedia

TRANSCRIPT

Page 1: On the personalization of event-based systems

Speaker: Opher [email protected]

Joint work with Fabiana Fournier from IBM

On the personalization of event-based systems

Page 2: On the personalization of event-based systems

2

Example:

Personalized aides for elderly to maintain independent life

Motion sensor

Door sensor

ChairSensor

Voice Sensor

Alert family member

Alerts example:Door was not locked within 2 minutes after entranceFalling event detectedVocal distress detectedNo motion for certain time period detected

While much technology exists, it is not widely used. It needs to be more personalized, more affordable, and much simpler…

The research required is multi-disciplinary:

Technology oriented, human oriented, economic oriented and particular domain oriented

Page 3: On the personalization of event-based systems

3

On Personalization

The industrial revolution opened the era of mass production, variety depends on the economy of scale.

Current technology such as Internet of Things provides the opportunity to enable everybody to create their own systems. This requires multi-disciplinary effort.

Page 4: On the personalization of event-based systems

4

The term “Internet of Things” was coined by Kevin Ashton in 1999.

His observation was that all the data on the Internet has been created by a human.

His vision was: “we need to empower computers with their own means of gathering information, so they can see, hear, and smell the world by themselves”.

Page 5: On the personalization of event-based systems

5

The value of sensors

Kevin Ashton: “track and count everything, and greatly reduce waste, loss, and cost. We could know when things needs replacing, repairing or recalling, and whether they were fresh or past their best”

The value is in the ability to know and react in a timely manner to situations that are detected by sensors

Page 6: On the personalization of event-based systems

6

Differences between the traditional Internet to the Internet of Everything

Topic Traditional Internet Internet of Everything

Who creates content? Human Machine

How is the content

consumed?

By request By pushing information

and triggering actions

How content is

combined?

Using explicitly defined

links

Through explicitly

defined operators

What is the value? Answer questions Action and timely

knowledge

What was done so far? Both content creation

(HTML…) and content

consumption (search

engines)

Mainly content creation

Page 7: On the personalization of event-based systems

7

“How does Event Processing get into the picture?”

While the weakest link is now considered the data integration issue – looking beyond that we can find event processing

Combining data from multi-sensors to get observations, alerts, and actions in real-time gets us to the issue of detecting patterns in event streams

However much of the IoT world has not realized it yet…

Page 8: On the personalization of event-based systems

8

A major difference between traditional Internet and the IoE – usability

The success of the Internet is attributed to its relative simplicity:

to connect to create contentto search

Imagine that any search in the Internet would have been done using SQL queries… How pervasive do you think the

Internet would have been?

Page 9: On the personalization of event-based systems

For situational awareness….Languages are actually more complex than SQL

9

// Large cash deposit

insert into LargeCashDeposit

select * from Cash deposit where amount > 100,000

// Frequent (At least three) large cash deposits

create context AccountID partition by accountId on Cash deposit;

Context AccountID

Insert into FrequentLargeCashDeposits select count(*) from LargeCashDeposit

having count(*)>3;

// Frequent cash deposits followed by transfer abroad

Context AccountID

insert into SuspiciousAccount select * from pattern [

every f=FrequentCashDeposit -> t=TransferAbroad where timer.within(10 days)]

Page 10: On the personalization of event-based systems

10

12 Hurdles Hampering The Internet of Things

Chris Curran, October 30, 2014https://www.linkedin.com/pulse/article/20141030181835-509139-12-hurdles-hampering-the-internet-of-things

1. Basic Infrastructure Immaturity

2. Few Standards

3. Security Immaturity

4. Physical Security Tampering

5. Privacy Pitfalls

6. Data Islands

7. Information, but Not Insights

8. Power Consumption and Batteries

9. New Platforms with New Languages and Technologies

10.Enterprise Network Incompatibility

11.Device Overload

12.New Communications and Data Architectures

Page 11: On the personalization of event-based systems

11

Democratization of use in Internet of Everything

Challenges:

Integration of sensors and actuators Personalization of situation detection Pervasive use

Page 12: On the personalization of event-based systems

12

Personalization of situation detection

Page 13: On the personalization of event-based systems

Eliminating noise from the model

Current models are close to the

implementation models – and from pure

logic view contain “noise”.

Bringing data from current state

Query EnrichmentInclusion in

events

Examples:

Determine what food-type

the container carries

Fetch the temperature

regulations for a specific

food type

Other noise : workarounds

13

For simplification we need to clean the noise

Page 14: On the personalization of event-based systems

14

The Event Model Research project developed by IBM Haifa Research Lab and Knowledge Partners International that dealt with simplification of event processing using model driven engineering approach

The Event Model design goals

Short video can be found in:https://www.youtube.com/watch?v=9zjy8wngy5Y&feature=youtu.be

Page 15: On the personalization of event-based systems

TEM Concepts

Facts

Actors

EventsStates

Event Derivation

Logic Transitions

Goals IT elements

Glossary Logic

Computation

Logic

Page 16: On the personalization of event-based systems

Simple example:Top down design of event model for suspicious account derivation

Bank transaction system

Compliance officerSuspicious Account

Frequent large cash

deposits

Frequent large cash

deposits

Large cash deposit

Large cash deposit

cash amount

<Cash deposit>

customer threshold

Page 17: On the personalization of event-based systems

Simple example: TEM Logic Specification for deriving Suspicious Account

Suspicious account Logic

Row #When

ExpressionWhen Start

When End

Partition by Filter on event Pattern Filter on pattern

Account ID Frequent large cash deposits

1always same is Detected

Frequent large cash deposits Logic

Row #When

ExpressionWhen Start

When End

Partition by Filter on event Pattern Filter on pattern

Account ID Count(Large cash deposit)

1every 10

dayssame > 3

Large cash deposit Logic

Row #When

ExpressionWhen Start

When End

Partition by Filter on event Pattern Filter on pattern

Customer ID cash amount <Cash deposit>

1always same >= customer

threshold

Page 18: On the personalization of event-based systems

Pattern on events

Pattern on events designates what the relationship between events is. In this case conditions C states that an event should occur before another.

Suspicious customer logic

Row # Context Conditions

When Partition by Event filter Pattern on events Filter on patterned events

Expression

Start End Customer ID Amount <Cash deposit>

Amount <Transfer Abroad>

Cash deposit Account <Cash Deposit>

1 Every week

same >= 150K >= 100K OCCURS BEFORE

Transfer Abroad

IS NOT

Account <Transfer Abroad>

A B C D

Page 19: On the personalization of event-based systems

19

My main motivation is to use the experience and

knowledge I have accumulated over the years to make a

better world