Developing an Enterprise-wide, Multi-Lingual Taxonomy.: Information Architecture Summit Vancouver, Canada March, 2006 Marti Heyman, Senior Global Taxonomy Manager Deloitte Touche Tohmatsu Developing an Enterprise-wide, Multi-Lingual Taxonomy.


Setting the context for the story…: Setting the context for the story… Overview of Deloitte A professional services firm that has a global reach while offering local solutions 80+ member firms with offices in 148 countries Annual revenues of $US16.4 billion Approximately 115,000 employees Business offerings: Audit, Consulting, Enterprise Risk Services, Financial Advisory, and Tax Corporate culture: Autonomy is highly valued and autonomous actions are highly rewarded Consensus-driven, no top-down authority


Setting the context for the story…: Setting the context for the story… The “knowledge sharing” environment is actually a “knowledge hiding” environment : Cultural barriers to knowledge sharing Culture of autonomy = knowledge/content silos Unique knowledge = competitive advantage Functional silos = limited cross-fertilization Technological barriers to knowledge sharing Variety of content management systems Variety of document management systems Variety of search utilities Outcome: Ambiguous language, limited or no controlled vocabulary Too many hits, too few hits Lost revenue, wasted time searching Business decisions made with incomplete information


The Case for Taxonomy: The Case for Taxonomy 40% of corporate users…can not find the information they need to do their jobs on their intranets. Susan Feldman, “The High Cost of Not Finding Information,” KMWorld, March 2004 Program Objectives: Disambiguate language so as to facilitate effective, efficient knowledge discovery within a single knowledge base or across multiple knowledge bases (i.e., ROI for search technology) Facilitate balancing precision with comprehensiveness Increase practitioner productivity leading to increased revenues and increased profits Reduce exposure of the firm resulting from decisions being made with incomplete information


Development Strategy and Processes: Development Strategy and Processes Launched Version 1 of Global Taxonomy with twelve facets in 2004 after THREE years of development. Working group of 30 cross-functional and multi-regional participants Yes, consensus takes time! Multilingual: English French German Japanese Spanish External Authority File Thesaurus Taxonomy Internal Authority File Initial twelve facets: Business Topics Client Services Confidentiality Content Types Deloitte Locations Economic Communities Internal Services Format Types Industries ISO Countries Languages People Levels


Extensible and Flexible: Extensible and Flexible Global (Core) Taxonomy Business Unit Extension Member Firm Extension Increasing Granularity and Specificity Increasing Generality and Broadness 1 or more Languages


Implementation Strategy: Implementation Strategy Critical success factors for broad adoption: Flexible Core needs to be broad and shallow to maximize flexibility and adaptability Extensible Member firms and Business Units need to be able to extend the core to meet their specific business needs Granularity Specificity Preferred terms


Implementation Strategy: Implementation Strategy Critical success factors for broad adoption: All “local extensions” must stay synchronized with core Single term record shared amongst all taxonomies ANSI/NISO Z39.19 systematically enforced All extensions must be visible to prevent silos Transparency amongst taxonomies Edit rights definable down to term level Ability to adopt concept trees built by others either by cloning or by integrating into Global (core) taxonomy Critical success factors for broad adoption: Ability to effectively implement periodic revisions to the Global (core) taxonomy without negatively impacting local extensions Shared system with workflow controls, term statuses, and editor/taxonomist notes fields Ability to feed taxonomy date to downstream systems Variety of export formats (CSV, XML, HTML) Ability to easily take only portions of selected vocabularies


Synchronized and Visible: Synchronized and Visible Global (core) taxonomy – consumer business industry segment and sub-segments:


Synchronized and Visible: Synchronized and Visible Global (core) taxonomy with local extensions:


Implementation Requirements: Implementation Requirements In order for the Global Taxonomy to “feel” local, we needed to be able to: Apply local labels to global terms Add terminology specific to the business needs of the Member Firm or Business Unit and amalgamate it seamlessly into the global taxonomy Exclude extraneous global terms Manipulate global category hierarchies Manage the number of levels seen within categories Mirror English Taxonomy hierarchy in 4 other languages, and ensure the taxonomies stay synchronized


Taxonomy Storage - Object Classes: Taxonomy Storage - Object Classes Each taxonomy category is housed in a number of related object classes Separate object classes store global, local and language equivalent terms Relationship types establish the connection between terms within related object classes The combining of object classes provides the basis of a local category extension


Object Classes: Object Classes Object Classes used in the Canadian Industry Extension


Allowing Display of Local Preferred Terms (i.e. dominance within Local Systems): Allowing Display of Local Preferred Terms (i.e. dominance within Local Systems)


Handling Multi-lingual Aspect: Handling Multi-lingual Aspect Use of Language Equivalent (LEQ)


Micro-Thesauri: Micro-Thesauri Broader categories are sliced into smaller term sets, excluding extraneous global terms from a local extension Micro-thesauri tags are applied to terms within the related category object classes Tags can also be used to distinguish top level terms from narrower terms


Inferred Hierarchies: Inferred Hierarchies Used in conjunction with Micro-thesauri tagging Collapses the hierarchy between untagged teams, excluding extraneous global terms


Ensuring Sustainability – Devil in the Details: Ensuring Sustainability – Devil in the Details


Establishing Policies and Procedures: Establishing Policies and Procedures Change management and version control Established process for quarterly review and revision Broadly distributed request for proposed adds/deletes/modifications Global Taxonomy Review Committee approves after review and discussion Changes not implemented until Local Taxonomist assesses impact on local extension Changes implemented within Global Taxonomy Changes adopted/modified as needed at local level Keeping it multi-lingual Newly added or modified global terms placed in “needs translation” workflow


Staying Flexible: Staying Flexible Coping with nuances of scope note variances Creation of LPT/GPT even though term is identical Able to create new scope note for the LPT Requires monitoring that identical terms not used for different concepts Coping with differing hierarchical placement of concepts Significant complexity introduced


Future Vision: Future Vision Audience recognition on Global Intranet Display the right taxonomy variant for right audience Canadian User


Future Vision: Future Vision Audience recognition on Global Intranet Display the right taxonomy variant for right audience Australian User


Ultimate Goal: Ultimate Goal Effective and efficient cross-repository searches Enable Topic Maps: Powerful route to knowledge discovery