active insight behavioral targeting in the cloud
DESCRIPTION
TRANSCRIPT
www.activeinsight.net
Event Stream Processing in the Cloud
ACTIVE INSIGHT
Mike TelemBusiness Development
Table of Contents
> Background: The Digital Era
> Processing, Correlating and Aggregating Events
> Use Cases: From Behavioral Targeting to Electrical Smart Grids
> ESP in the Cloud
> Roadmap: Where is ActiveInsight headed
Our world is becoming digital…
Cell phones, web sites, GPS devices, cars, ads, Financial transactions,…
RFID, industrial eq., security sensors, border controls, medical eq.,…
Utilities, pipelines, meters, digital signage, home appliances, entertainment devices, cars, …
Applications, infrastructure, web-services, customer data,…
Markets, stocks, currencies, news, wiki’s, blogs, tweets,…
…
Multiple events share various perspectives
Event stream quantity and frequency will fluctuate
Effective time window for reactions is minimal
Reaction channels may vary Events should be correlated
with historical data
The Digital Era
Building blocks of Behavioral Targeting Event Stream Processing:
Processing application level events in a distributed environment
Event Correlation – Directing multiple event streams based on their context to the corresponding ESP containers
Complex Event Processing: Processing multiple events to detect
meaningful patterns using correlation, aggregation and time-frames
Pattern detection: Detecting specific event combinations and patterns in contexts
Cross-Context Correlation: Processing multiple streams into multiple contexts / perspectives (fraud / marketing)
Aggregation: Accumulating correlated events into time-based contexts, support for “event state machine” aggregation.
Data Integration: Caching data sources as “reference data” for processing
Reaction: Invoking an action after a successful event or pattern match
Different Use-cases > Similar Challenges Online Gaming : Real-time BI, money
laundering, local compliance, application offload
Online Advertisement: Behavioral targeting, multiple site click-stream correlation
Ecommerce : Identifying customer interests (up-sell/cross—sell) , Improving conversion rates, anonymous user hooking, campaign management
Online Self-Service : Identifying customer turnover or dissatisfaction, Monitor user experience and assist in transaction completion
Algo-Trading : performance and availability improvements and HW cost reduction
Auditing: Feeding “Who” did “What” and “When” to auditing and SIEM systems
Fraud detection: Fraudulent behavior pattern detection, Bot detection, alongside fraud detection systems
Electrical smart-grid: Detecting misuse, mal-functions, on-demand supply
Home Land Security: Enhance airport and border security, correlate multiple events, intelligence data and incoming alerts
Traffic management: Vehicle location management for Insurers, authorities and drivers
…Similar Challenges
React
Different Use-cases > Similar Challenges
Match
Correlate
Process
Aggregate
Behavioral Targeting in the Cloud Elastic
On-demand usage Scaling up and out to varying
event frequencies
… as a service Offloading event processing
Dynamic Stream Sources Dynamic event sources Handling remote event sources
SaaS Enabler Porting event-oriented
applications to the cloud
SaaS component Enhance SaaS applications Offload the core application Comply to regional regulations Provide SaaS Application
integration
IaaS/Hosting Value Added Services (Security,
Auditing, BI) Customer Experience
Management
ActiveInsight
Distributed Behavioral Targeting Platform
Real-time event processing
Multi-source event stream processing
Event correlation and aggregation
Pattern matching
Integrated data caching
Embeddable framework
Scalable, elastic cloud run-time
Process
Correlate
Aggregate
Match
React
Sample Architecture
AI Server Node
Distributed Cache
Reference Data
Context
Process Match React
Context
AI Server Node
Distributed Cache
Reference DataContext
Process Match React
Context
AI Server Node
Distributed Cache
Reference Data
Context
Process Match React
Context
ContextsMarketingSecurity
Web App
Mobile Device
Car GPS
Unique Value Proposition Embeddable, Real-time data stream processing
Flexible and dynamic pattern definition/detection
SpringSource development platform interoperability
Real-time, pattern-based logic invocation
Business driven behavior detection
User-centric actionable events
Real-time, value-based event feeds & user interactions
Non-intrusive deployment
Support for extreme transaction rates
“With ActiveInsight organizations can identify up-sell and cross-sell opportunities, react to potential customer
churn in time to prevent it, improve online self-service to customers and detect potential fraudulent activity in
real-time “
www.activeinsight.net