Transcript integrating search and business intelligence

Satish Ramanan
April 16, 2011
AGENDA
 Context
 Why
 How
 What
 Who
- Integrate Search with BI?
- do we get there?
- Tool Strategy
- is in it for me ?
- Outcomes & Benefits
- should I go to?
- Reference Vendors
Context
 Explosion of Web 2.0 Data
 Extended Scope of Enterprise Information Management
 Structured and Unstructured data
 Unified Information Access
 Guided Analytics
Growing
Search v/s BI
SEARCH
BI
Querying unstructured data
Querying structured databases
Humungous volumes of data that is
scattered all over
Size is usually known and manageable
Documents, web pages, emails, etc.
Tables and columns
Standard, Simple and brain-dead easy
Interface – Usually a search box or an
Advanced Search box
Complex formats taking time and effort
to build – Reports, Dashboards, Charts
& Graphs
Typically uses keywords as inputs
Selection criteria using one or more
fields as input
Usually a dump of results with drill
down capability
Aggregate and visualize data in tabular
or graphical formats with or w/o
complex calculations. Drill down
possible
Search with BI
 A system that gives u best of both worlds
 A user interface that can consume simple inputs but
produce complex outputs
 Ability to access and query information contained in
tables and columns in a database/warehouse as well as
web pages, documents, email
 Ability to predict or forecast future outcomes based on
such unified information
 Use innovation to succeed – linguistic technology,
converting simple text to SQL
Why Integrate Search with BI
 Information Explosion within organizations (and outside!)
 Gigabytes of structured data lying unused
 Terabytes of unstructured data lying unexplored and
unattended
 Unstructured data seen providing critical insights to
business rendering it almost indispensible
 True Predictive Analytics depends upon ability to
analyze outputs from unified information sources
Possible Approaches
 Categorized BI Search
 A Search engine indexes both metadata and data
generated
 A list of search results based on keyword is displayed in
the main body
 Categories / Sub Categories of results derived from
metadata are displayed on the left side
 Results contain drilldowns to records in the source
systems and reports that are created on the fly
 Eg. www.dice.com
 WebFocus Magnify
Possible Approaches….2
 Natural Language Processing (NLP) Search
 Uses Linguistic technology to decipher the meaning of keywords





used
Technology maps the meaning of keywords to metadata of
databases or documents
Users can query database using plain English (some other
languages as well) that will be converted to SQL automatically by
the technology
Technology actually uses a knowledgebase of concepts, business
rules, jargons & acronyms
The generated SQL is displayed alongwith the results, which can
then be converted into a table, chart, or dashboard.
Easyask, Semantra
Possible Approaches…….3
 Visual Search
 Not really a search, but uses advanced BI tools that runs
a visualization search directly against an analytic
platform
 The source is usually an in-memory columnar database
that helps returns results extremely fast
 Using a point and click paradigm, users can sort, filter,
group, drill, and visualize the results
 SAP Business Objects Explorer Accelerated, Endeca
Information Access
Typical Challenges
 Attempts to unify structured and unstructured data
facing major road blocks






Limited awareness
Lack of appropriate tools and technology
Data ownership lying with disparate groups
Huge time and costs involved in efforts to build a unified
information repository
ROI questionable and can be easily challenged
Big Bang approach coupled with Quest for the most economical
solution leading to major failures
Choosing the right tool
 Text Analytics
 Treatment of Data Relationships
 Query flexibility
 Information Availability
 Secure Access
 Ability to Build/Develop Powerful Interactive
Applications easily
 Enterprise Readiness
 Total Cost of Ownership
Outcomes and Benefits
 Increasing demand for High Performance DB Engines
 Teradata, Netezza, Exadata, Greenplum
 Increasing demand for Real Time Integration
 Need to accelerate Information access
 Providing TRUE ‘Single Source Of Truth’
 Analytics getting closer to ‘Predictive’ than ‘historical’
 Enhanced Knowledge Harvesting and Management
Vendors
 SAP - Business Objects Explorer
 ENDECA - Latitude
 ATTIVIO – Active Intelligence Engine
 INFORMATION BUILDERS – Webfocus Magnify
 FAST
 GOOGLE Analytics
THANK YOU
Satish Ramanan
[email protected]
+91 9769417381