Академический Документы
Профессиональный Документы
Культура Документы
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Mark Rittman
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Who Are Endeca? Why Did Oracle Acquire Them?
Recent (2012) acquisition by Oracle, aimed at strengthening ERP and E-Commerce offerings
Endeca was best known for search and guided navigation features for e-commerce sites
HMV.com
Autotrader.com
Sony e-commerce sites
etc
Key feature is searching across arbitrary
selections of attributes, fast response,
supports discovery and exploration
Faceted Search
Enabled through a hybrid search/analytic
database called the Endeca Server
Endeca Faceted
Search feature
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
What is Oracle Endeca Information Discovery?
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Information Discovery vs. Reporting & Analysis
Data volume, variety and growth presents issues in finding answers to business questions
Unstructured data, social network data, call centre logs as well as regular structured data sets
Datasets change, dont always fit dimensional models, and arrive quickly
Users want self-service access to data with minimal setup time
Reporting and Analysis is great for accurate answers to known questions ...
... Data discovery provides fast answers to new questions
Guiding principle : Quickly explore all relevant data
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Search-First Interface
Guided
Navigation,
free-form filtering
across all
atttributes
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Data Discovery Scenarios
Online and TV-based retailers Sales reporting only covers what Combine unstructured social
youve sold. not why youve sold it networking feeds with sales data
E-commerce operations Consumer sentiment is found on Content acquisition from non-
B2C companies with vocal, blogs, Facebook and Twitter, not traditional sources
online customer base easily brought into BI datasets Analyze consumer sentiment
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Oracle Endeca Information Discovery Platform Overview
A suite of tools to capture, load and then analyze all types of data
Built around the Oracle Endeca Server,
a hybrid search/analytic database Oracle Endeca Information
Discovery Studio
Full ETL capabiities plus web-based
dashboard and data exploration tools
Additional tools can be licenced for
Oracle Endeca Server
content acquisition and text enrichment /
sentiment analysis
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Oracle Endeca Server : A Hybrid Search/Analytic Database
Key to these capabilities is the Oracle Endeca Server and its datastores (databases)
Proprietary database engine focused on search and analytics
Data organized as records, made up of attributes stored as key/value pairs
No over-arching schema,
no tables, self-describing attributes
Every record can have its own unique
set of attributes, with the overall data model
emerging over time as data is loaded
Endeca Server hallmarks:
Minimal upfront design
Support for jagged data
Administered via web service calls
No data left behind
Load and Go
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Inside an Endeca Server Datastore
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Endeca Server Search and Document Analysis Capabilities
Search and unstructured data support adds powerful document analysis capabilities
Keyword search, boolean search, parametric search, wildcard search, dimension search and
dimension filters
Dimension precedence rules
Numeric range, geospatial, date/time and security filters
Spell correction/suggestion,
and do you mean-type alternative presentation
Find similar, and 1 and 2-way synonyms
Stemming and lemmatisation
Keyword-in-context snippeting
Results clustering, relevance ranking, sorting and paging
Multi-language support
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Oracle EID Integrator and Studio
Data is loaded into Oracle Endeca Server datastores using Oracle EID Integrator
Data Integration (ETL) tool build on open-source CloverETL tool (Eclipse framework)
Oracle EID functionality provided through components that call Endeca Server web services
User Interface created and delivered using Oracle EID Studio, 100% web-based
Create dashboards made up of search, navigation and data analysis components
Also provides Endeca Server / Studio admin features
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Endeca Information Discovery Platform & Software Requirements
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Oracle EID and Oracle BI Foundation
BI Multi-dimensional
Foundation Multi-dimensional engine, what-if? Essbase
OLAP analysis, scenario modeling, planning
Suite
Endeca
Information Data Discovery Search and contextual navigation for Endeca Information
structured and unstructured data Discovery
Discovery
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
An OEID 2.3 Development Scenario
End-User Analytic/
Search Application
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
High-Level Oracle EID Development Process (Simplified)
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
The Oracle EID Integrator Development Environment
Buttons to
run graphs, run
in debug mode
etc
Project elements
(graphs, in/out
data, params etc)
Additional graph
components
Graph, containing a (readers, writers,
a single component joiners + OEID
(Web Service call to components)
create a new Endeca
Server datastore)
Properties,
metadata etc for
selected object
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Integrator Graphs, Components and Edges
Graph
Port
Join Load
Edge
Input
Component
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Information Discovery-Specific Components in Integrator
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating a New Endeca Server Datastore
The Endeca Server runs as a server process, with datastores being created, then attached
New datastores are either created using the endeca-cmd Java command-line utility
cd c:\Oracle\Endeca\Server\2.3.0\endeca-cmd
endeca-cmd create-ds bikestore
Or through a graph component that calls the same Endeca Server web service API
Uses a configuration file to obtain
server host name, datastore name, port etc
for the Endeca Server being called
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Loading the Initial Record Set
The first stage in creating the dataset is to load the initial set of records
In this example, loaded from a relational fact table export, contains measures and keys
Records require a spec(ification), a unique key that uniquely identifies each record
In the example below, the spec is created using a transformation component that concatenates
the SalesOrderID and SalesOrderLineID
Transform (Reformat)
Reader component, Writer component,
component,
reading fact table writing records to the
concatenating input
rows (records) from Endeca Server data
columns to create
a flat file export store
record spec (PK)
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Transformations and Expressions
Data flowing through graphs can be transformed at various stages, for example
As it flows through an edge
During a join
Expression syntax is CTL1/CTL2
Clover Transformation Language
Similar to Java expressions, source
and target-technology neutral
Original CTL1 version now being
superceded by CTL2
Expression used to
create the spec
column,
concatenation two
input columns
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Viewing the Initial Record Set
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Loading Additional Structured and Semi-Structured Sources
Additional sources can be loaded using the same graph, joining on common IDs
Unstructured/semi-structured sources with at least a matching ID can be brought in
Continue adding sources until full set of required data is loaded into the datastore
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Text Parsing, Whitelisting, Enrichment and Sentiment Analysis
Text attributes can be analyzed, enriched and parsed using standard and additional features
Build in features include text tagging component
and various CTL functions
Oracle Endeca Text Enrichment module
Text analysis capabilities for extraction
of people, pleaces, themes etc
Automatic summarized abstracts etc
Uses Salience Engine from Lexalytics, Inc.
Additional cost option
Oracle Endeca Text Enrichment with Whitelist of terms to
be tagged using the
Sentiment Analysis module Text Tagger - Whitelist
component
As above, but with ability to calculate
aggregate sentiment scores Whitelist text
Determine positive or negative sentiment tagger, using the
whitelist of terms in
in comments, surveys, feedback the
SurveyWhiteList file
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Demonstrations
Oracle EID Integrator
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating the Studio Interface
Once you have some data loaded, you can start to create the Studio web-based interface
Create the initial Studio pages early, use as part of the prototyping process
Over time, add more visualizations and data views to aid discovery
Once the model and attribute set are fairly stable, you can perform additional configuration
Rename and relabel attributes, arrange into groups, configure for searching
Studio Individual
application menu datastore record,
using default
attribute names
from file metadata
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Typical Studio Development Tasks
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating Pages, Configuring Layouts and Adding Components
Studio uses pages with layouts to organize content, similar to OBIEEs dashboards
Components are then added to the pages, to create the dashboard content
Guided Navigation
component being
dropped onto page
layout template
Preferences and
Configuration buttons
for the component
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Performing Value Searches (Type-ahead Searching of Records)
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating and Enabling Search Interfaces
Search interfaces create sets of attributes that are then available for record searches
When values are returned, they are used to filter the record set
Requires search interfaces to be set up and enabled using
an Endeca Server web service API and a configuration file/source Graph data flow that
configures the selected
attributes as searchable,
using an Endeca Server
WS API
Reader
component,
reading in the
Search Index
configuration file
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Selecting and Enabling Search Interfaces in Studio
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Performing a Record Search
Record searches filter the record set by the specified search term, within the attributes in the
selected search interface
Record search is only performed with the search button is pressed
Default search interface is used, unless another one is specifically selected
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Configuring Attribute Naming and Grouping through Studio
Attribute metadata can be updated to provide display names, change sort order etc
Attributes can also be placed into groups, hidden from guided navigation menus
Configurable through Studio, with the
ability to export the new configuration
for use in Integrator List of attribute groups,
and their current
members
Studio Control
Panel, including
menu item for
managing
datastore
Attribute Settings
List of
unassigned
attributes,
currently in the
Other group
Selecting an
attribute group to
add the selected
attributes to
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Configuring Attribute Naming and Grouping through a Graph
Attribute configuration is applied through another Endeca Server web service API
Exported configuration from Studio can then be used in a graph to programatically apply
configuration to other / reset data stores
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Results Tables, Cross Tabs and Data Explorer Components
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Defining the Data Source for a Cross Tab Component
Cross tab components (and optionally, table components) need to be given a data source
Datasources for these components are defined using EQL (Endeca Query Language)
Similar to SQL, assumes a single table of records, implied SELECT with GROUP BY
clauses, and other variations suited to the Endeca Server record / attribute data model
EQL Statement
defining Cross Tab
data set
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
EQL Statement Types and Examples
Return SalesTotal
AS
SELECT SUM(FactSales_SalesAmount) as TotalSales
GROUP BY DimDate_FiscalYear, DimDate_FiscalQuarter,
DimSalesTerritory_SalesTerritoryCountry
DEFINE statement, creates a temporary table with the result set, used to create views
DEFINE RegionTotals
AS
SELECT
SUM(Amount) AS Total
GROUP BY Region
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Configuration and Viewing of Cross Tab Component
Once EQL query has been defined, component can then be configured and deployed on page
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Charting and Other Data Visualization Components
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating EQL Views
Physical
Model Oracle Endeca Server
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Creating Views and Using with Chart Components
Views are typically created using the View Manager is Studio, then exported for use in graphs
Once defined, they can be used (amongst other things) as data sources for charts
Chart configuration
options
List of previously-
defined views, available
as datasets for the
component
The view definition,
expressed as an EQL
DEFINE statement
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
The Finished Studio Application
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
The Finished Studio Application
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
The Finished Studio Application
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Demonstrations
Oracle EID Studio
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Getting Started with Oracle Endeca Information Discovery
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
More Information
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12
Endeca Information Discovery for BI/DW Devs
Mark Rittman, Technical Director, Rittman Mead
Oracle Openworld 2012, San Francisco, October 2012
T : +44 (0) 8446 697 995 or (888) 631 1410 (USA) E : enquiries@rittmanmead.com W: www.rittmanmead.com
Thursday, October 4, 12