Whitepaper:
A Taxonomy Roadmap
Copyright 2002-2004 Gwen Thomas
Abstract:
Building a prototype taxonomy combines a classic pattern for software
development with steps used to assemble information products.
What happens if your organization does not have a Ta= xonomy Clearinghouse?
Consider the following functional groups and how the= y deal with a simple set of words and their meanings: the company’s products. Each functional group works with projects that result in documents. They all involve IT to some level.
Note that each silo project has similar steps, marke= d with a star, in its Design phase. In these steps, products must be named and described, and their relationships outlined. If the silo project uses only = that terminology familiar to them, the project is at risk of being out of alignm= ent with the rest of the enterprise.
<=
!--[if gte mso 9]>
The obvious answer is to amend those steps during pr= oject design phase to include validating product names, descriptions, and relatio= nships with all stakeholders. At this time, project teams could collect additional terms, map them against project terminology, determine whether they should = be included in project scope, and redefine project scope and effort as require= d.
Such an approach might help keep a silo project from= being out of alignment with the rest of the enterprise. However, it is totally impractical to consider that each project team will independently communica= te and align with all affected functional and project groups. Consider all the lines of communication required by such an approach:
<=
!--[if gte mso 9]>
A better approach is to designate a group within the enterprise to serve as a Taxonomy Clearinghouse. This group can collect and align taxonomies and vocabulary sets from across the organization and make = them available to all who need them.
<=
!--[if gte mso 9]>
The appropriate approach for building and maintainin= g the collection of taxonomies will vary from company to company. CIBER recommends that a company look for similar processes that are mature within the organization and have been accepted by users. Build a taxonomy management m= odel based upon those successes.
Many companies will find as they collect taxonomies = that they appear in dissimilar formats. It will be easiest to compare terms in different taxonomies if they have similar formats. And so, you may find that you want to follow the steps to convert informal collections of terms and relationships into formal taxonomies.
Building a taxonomy contains seven primary steps:
Ø Step one – plan
Ø Step two – design
Ø Step three – build the structure
Ø Step four – populate the taxonomy
Ø Step five – test
Ø Step six – implement
Ø Step seven – support
A best practice for taxonomy planning is to think globally, but work locally. If. like most taxonomy projects, you are starti= ng at a group or department level, you need to avoid the analysis paralysis th= at can come from considering enterprise factors over which you neither have control nor access to essential information. Instead, make assumptions about the needs of the enterprise or other groups. Document those assumptions and include revisiting them on a regular basis. Then build the taxonomy that wi= ll serve your needs, trying to avoid any known conflicts with enterprise or fu= ture needs.
Tasks to address during planning:
Ø define the purpose of the taxonomy you are building
Ø define a domain and formulate goals for taxo= nomy development and deployment;
Ø
develop a plan for analyzing content and user
needs
Ø define a development approach
Ø define success criteria and measurement
Ø define structural and implementation requirements
Ø conduct a domain analysis
Ø assemble a project team
Ø establish protocols, roles and responsibilities
Ø establish communication and collaboration sp= aces
Ø provide education as needed:
· business taxonomies purposes, characteristic= s
· taxonomy applications and functions
· centralized vs. specialized taxonomies
· the relationship between meta data and taxonomies
· the relationship between taxonomy governance= /stewardship and data governance/stewardship
Ø design a plan for evaluating the taxonomy
Ø design the prototype repository
· define objects and attributes
· design user interface
· design navigational structure
· design search function
Ø define standards for content
Ø define domain boundaries
Ø design staging areas for collected content
Ø design processes for moving content from staging areas to the prototype
Ø design and document a taxonomy governance policy and procedures
Ø build staging areas and movement processes
Ø build the prototype repository
Ø create navigation tools based on your taxono= my structure;
Ø assemble and mine materials for vocabulary terms
Ø build a basic vocabulary of terms
Ø build a hierarchy of categories
Ø refine controlled vocabulary lists<= /span>
Ø insert terms into the hierarchy of catego= ries
Ø establish relationships between ter= ms
Ø connect terms with sources= p>
Ø assemble the master thesaurus
Ø establish taxonomy governance
Ø test the taxonomy
Ø role out the taxonomy to potential users
Ø implement a mechanism for gathering feedback= and suggestions
Ø disseminate templates for project plans, software capabilities, and taxonomy governance policy
Ø work with the Project Office and Quality Groups to add taxonomy steps to Project Development Life Cycles and Quality Checks.
Ø implement a program to gauge user satisfaction
Ø implement a program for socializing the taxo= nomy across the enterprise
Ø continue to map benefits realized by the taxonomy to business benefits
&=
Oslash; =
continue
to refine taxonomy governance and stewardship prac=
tices
&=
Oslash; =
consider integrating taxonomies with other tools:=
· make taxonomies available to navigation tool= s
· incorporate taxonomies into business applications
· integrate taxonomies and search
Ø consider including taxonomy maintenance in editorial work flow
Ø establish ongoing processes for maintaining standards and collaboration
Ø consider the use of specialized software for future iterations of the taxonomy
· types of software
· selection tips
· concept extraction
· document summarization
· auto categorization