Defining a taxonomy's domain and scope requires that several basic questions be asked:

• What is the subject domain that the taxonomy?
• How will the taxonomy be used?
• Who will use the taxonomy?
• What are the questions that can be answered if the taxonomy once constructed?

It is important to consider what others have done. Do relevant classification schemes or controlled vocabularies may already exist and should they be incorporated? There are a number of existing libraries:
American Society for Indexing
TaxoBank
Taxonomy Warehouse

The discovery, extraction and documentation of all terms/concepts (words, phrases, sentences) relevant to the subject domain is the central task of concept development. It is important to compile a comprehensive list in this step without concern for overlap between concepts or relationships between terms. Depending on the circumstances, this might be a manual or an automated process, or a combination of both.

Typically, the process starts by extracting terms and concepts or noun phrases from plain text using linguistic processors such as part-of-speech tagging and phrase chunking. Then statistical or symbolic techniques are used to extract relation signatures, often based on pattern-based or definition-based hypernym extraction techniques. During domain terminology extraction, domain-specific terms are extracted, which are used to derive concepts. Relevant terms can be determined by calculation of the TF/IDF values or by application of the C-value / NC-value method. The resulting list of terms has to be filtered by a domain expert. The most common methods is clustering and the application of statistical similarity measures.

In order to organize the concepts extracted and documented, it is necessary to identify and name a limited set of unifying dimensions or perspectives. Each of the concepts is then associated with one of these taxonomic dimensions. Concept organization requires the development of a controlled vocabulary set. The purpose of a controlled vocabulary is to organize information and to provide terminology to catalog and retrieve information. While capturing the richness of variant terms, controlled vocabularies also promote consistency in preferred terms and the assignment of the same terms to similar content. Constructing a controlled vocabulary requires deliberate development.

1. Identify terms
2. Pull together synonyms, disambiguate homographs
3. Identify relationships between terms
4. Make those relationships explicit in your metadata

The simplest kind of controlled vocabulary is a flat term list, sometimes called a ‘pick list’. Term lists are often utilized for administrative and structural metadata elements, such as a list of possible file formats, rights status or retention status. Term lists are also used in descriptive metadata elements, such as content type, language, department/source, etc. Controlled vocabularies of subject terms, however, may be too large and complex for simple term lists. Term lists are often displayed within drop-down boxes for a field, but could display as button or check-box items.

An authority file is a controlled vocabulary which includes synonyms or variants for each term which function as cross-references to guide the user from an ‘non-preferred term’ variant to the equivalent ‘preferred term’. In addition, authority files may provide a note for each term as to the authoritative source for the preferred term. The designation ‘authority file’ is used more often with named entities (proper nouns) only, and often authority files are simply called ‘controlled vocabularies’.

The classic meaning of a thesaurus is a kind of dictionary which contains synonyms or alternate expressions for each term and possibly even antonyms. An information/content retrieval thesaurus shares this characteristic of listing similar terms at each controlled vocabulary term entry. The difference is that in a dictionary-thesaurus all the associated terms might be used in place of the term entry depending upon the specific context, which the user needs to consider in each case. The content retrieval thesaurus, on the other hand, is designed for all contexts, regardless of a specific term usage or document. The synonyms or near-synonyms must therefore be suitably equivalent in all circumstances. A content retrieval thesaurus is also more structured than either a dictionary thesaurus or other types of controlled vocabularies, because it provides information about each term and its relationships to other terms within the same thesaurus. In addition to specifying which terms can be used as synonyms (labeled as ‘used from’), a thesaurus also indicates which terms are more specific (narrower terms), which are broader, and which are non-hierarchically related terms. In addition, some terms have scope note explanations, as needed.

One aspect of validation is the comparison of taxonomic dimensions against the knowledge gathered during the Domain and Scope Definition task. The validation step reviews all previous steps to insure that decisions still seem valid, that no important content has been missed, and that the level of detail within and across the dimensions is appropriate and consistent. There are two basic techniques for validating taxonomies.

Ensure a full exploration of the concepts that taxonomy employs describe the real world or the domain of interest. Focus on the established model. Encourage open discussion & brainstorming among the user group and taxonomist technicians to inform the structural design.

Examine the implementation of the existing design. Evaluate the workflow that implements the taxonomy. Gather feedback & critique to inform taxonomical design strategy.