Integrating an Enterprise Taxonomy with Local Variations Tom Reamy Chief Knowledge Architect KAPS Group Program Chair – Text Analytics World Knowledge.

Integrating an Enterprise Taxonomy with Local Variations

Tom ReamyChief Knowledge Architect

KAPS Group

Program Chair – Text Analytics World

Knowledge Architecture Professional Services

http://www.kapsgroup.com

2

Agenda

Introduction – Information Environment Research Approach Integrated Solution – governance, technology – text analytics Conclusions

3

Introduction: KAPS Group

Knowledge Architecture Professional Services – Network of Consultants Applied Theory – Faceted & emotion taxonomies, natural categories

Services:– Strategy – IM & KM - Text Analytics, Social Media, Integration– Taxonomy/Text Analytics, Social Media development, consulting– Text Analytics Quick Start – Audit, Evaluation, Pilot

Partners – Smart Logic, Expert Systems, SAS, SAP, IBM, FAST, Concept Searching, Attensity, Clarabridge, Lexalytics

Clients: Genentech, Novartis, Northwestern Mutual Life, Financial Times, Hyatt, Home Depot, Harvard Business Library, British Parliament, Battelle, Amdocs, FDA, GAO, World Bank, Dept. of Transportation, etc.

Program Chair – Text Analytics World – March 29-April 1 - SF Presentations, Articles, White Papers – www.kapsgroup.com Current – Book – Text Analytics: How to Conquer Information Overload,

Get Real Value from Social Media, and Add Smart Text to Big Data

Information Environment

Multi-National Financial Institution-10,000+ Diversity - multiple languages, cultures, information needs

and behaviors, organizational cultures Initial Application – knowledge management networks

– Network definition – somewhat by subject area, but also political

Multiple applications – search, browse, web sites– Expertise location, Accounting-resource, analysis

Multiple audiences – internal and external, expert and non-expert (everyone a non-expert in something)

4

Approach

First step – research into variations– Use cases, levels of granularity– Common terms with different meanings

Interviews with multiple groups, roles, levels– Contextual interviews, information interviews– Taxonomy interviews – suggested terms and relationships

Analysis – taxonomies, search logs suggest facets, HR expertise descriptions, local web sites, keywords, clustering, new terms

Group sessions – representatives of multiple constituencies – talking out the differences

5

6

Current Environment Overview

Current form of Topics: Long and flat – 2 levels– Difficult to build on, desire for more specificity for experts and

content, usability issues, no place for new topics Multiple taxonomies – topics, organizational, Web site

browse, industry codes– Partial overlaps, conflicting – Political – Social Development & Gender

Variations – official term, relationships of terms– New terms mostly at lower levels and stable structure

Cross-cutting topics – Finance of Education, Poverty

7

Elements of the Solution

Taxonomy is only one part of the solution– Faceted metadata and text analytics– Enterprise taxonomy – death of?

Analysis of taxonomy – suitable for categorization & views– Structure – not too flat, not too large– Orthogonal categories – easier to tag and easier to map

variations Idea of Views – browse by local variations – map to official topics

– Supported by software – Pool Party– Role-based views, Activity-based views

Solution: integration of multiple components – two critical-Governance and Text Analytics

Mid-tier layer

Information Services/Semantic Layer

Enterprise Information Integration

Analysis & ReportingEnterprise search

engineData mining,

Text Analytics engine Statistical Analysis

Predictive Analytics

Front-end application

Dashboard Ad-hoc query Mobile Apps Portals/web LoB Client UI Mashups Enterprise Search UI

Metadata MgtWeb content EmailDocuments Business data

Structured Data

Statistical data

Structured Data

Shared drives

VocabulariesTaxonomies

Reference Data

Core Metadata

Corporate data Model

Reference ArchitectureG

over

nan

ce

Consolidated Repositories

Metadata Repository/

Registry

Operational Data Store

Data Warehouse

Master Data

Data marts

Unstructured data

Information Standards & Policy

Multilingual

Data SourcesUnstructured data Unstructured data Unstructured data

External

External Data

Data MovementETL Data Services ESB …

Data processing engine (e.g Hadoop)

e-publish, Day, Drupal, blogs etc

Wbdocs,Jolis

SAP, PeopleSoft, Finance etc

JiveSharepoint

Factiva NewsResearch dbs

DDP

9

Text Analytics – power and flexibility

Critical – Text Analytics tool – Same taxonomy term but different criteria, rules – Documents tagged for different uses, audiences

Education – for specialists– Deep complex rules, very fine granularity, specialists jargon-

acronyms Education – for generalists

– High level rules, general terms, simple Education within Social Development

– Generalist rules plus social development terms – birth weight

10

Proposed Model for a Taxonomy Eco-System

New Topic Taxonomy– Enhanced structure and coverage – deeper framework to

build on– Implemented in new software – flexible, solves old debates– Facets – remove complexity, increase coverage – 10 X 10.– Powered by auto-categorization - tagging, advanced

applications– Combine with current data –HR (Expertise), other

Taxonomy is part of an integrated information management – Search, Content management, IM Policy, External Web, etc.

Facets – Subject (topic), Industry, Program, Methods, Business Activities, Organization, Skills, Document Type, Project, Product, Geography

Governance Structure

Executive FunctionSponsorship

Tax Management Central

Tax Management (Anchors & Regions)

Users & SMEs

Taxonomy Management

IM focal Point, KM, web etc (Anchors & Regions, IMT)

Revise/approve tax structureRules for changesText Analytics/Research

Manage implementation Gather & make changesContent & tagging analysisCoordinate feedbackCommunication / Training

Provide feedback

Feedback loop

Set overall policy and strategies Drive direct acceptance

IT Systems Coordinate changes in dependent systems

Working Group

Prioritize changes, cross-cutting

Strategic LevelIntegration with existing Information Management

Operational

Resource DecisionsOrganizational structure

12

Critical Success Factors: Governance

Governance Policy & Process & Enforcement– Incorporate enforcement into publishing process / Hybrid Auto-cat

Taxonomy management is part of overall information management with additional taxonomy roles/functions

Best Practice: combination of central and distributed teams Taxonomy specific: Taxonomy Manager – Central & Networks

– Revise tax structure, rules for changes, manage implementation– Enforcement – combination of central & Networks

Feedback – metrics – identify need for new terms, remove old terms– Combination of user feedback in application & periodic analysis

Conclusion

Taxonomies are an enterprise resource Danger of monolithic over-riding local variations

– Less useful and/or ignored Danger of chaos of multiple variations losing ability to coordinate

and communicate Solution: Research into users, use cases, semantic resources Integrated solution – importance of distributed governance Integrated solution – text analytics to reflect local variations and

provide a means to integrate into unified solution Facets, text analytics and browse views solve 75%, rest is

manageable No one was entirely happy – must be doing something right

13

Questions? Tom Reamy

[email protected]

KAPS Group

Knowledge Architecture Professional Services

http://www.kapsgroup.com

www.TextAnalyticsWorld.com March 29-April 1, San Francisco