Taxonomy 101: Definition, best practices, and how it complements other AI work (2023)

Summary:A taxonomy is a behind-the-scenes structure that complements visible navigation. Taxonomies support the retrieval of consistent information by creating formal metadata rules.

What is a taxonomy and why is it important for UX professionals? At our UX conferenceinformational architectureOf course, I am often asked what a taxonomy is, how to create one, and how it fits into the larger taxonomic picture.Information architecture (IA) job.

Taxonomies Call for Information Science Professionalsreviewed vocabulary— Planned and required methods for adding descriptive metadata to content so that it can be retrieved efficiently. The idea is that the taxonomy defines a limited set of terms to describe our content behind the scenes; Content creators must attach them to any new content without the ability to expand that vocabulary on an ad hoc basis.

Definition:A taxonomy is a closed list of acceptable terms that are arranged hierarchically and take some getting used to.to describeand sort the content.

Taxonomies are essentially controlled tag systems; although each piece of content is associated with a set of taxonomic terms, content creators cannot define their own terms. Anytime we tag content using a taxonomy, we need to follow some basic rules about which terms to use and how those terms relate to other terms.

It's important to note that a taxonomy is different from the navigation framework that users interact with and the underlying AI framework. More on that below.

4 types of organizational models in AI

It's true that information architecture can be a confusing discipline: there are many different types of abstract organizational models, all of which look quite similar but describe different things. The four main organizational models are listed below:

1. Navigation is a set of user interface elements (menus, links, breadcrumbs, and accordions) that show the user what page or screen they are currently viewing and where they can go. In itservice designslang, navigation isstage(i.e. visible to users).
2. The AI ​​framework (sometimes called a sitemap) is a map of all the main nodes of a website (such as pages or screens) and the relationships between them. The AI ​​is behind the scenes: it is not directly exposed to users, but is used by staff to design navigation or decide where content should go. The AI ​​is usuallysimilar to navigation, but much more advanced: it is the complete map of all the content of the site and not the small fragment that a user can see at a time. Imagine a huge office building: the signage and elevators provide navigation (but you can't see the whole building at once), while the architectural design shows all the underlying structure of the building and adjusts for the AI.

(Information architects and taxonomists often debate whether the AI ​​tree structure represents a taxonomy; one could argue that this is a very flexible form of taxonomy, but for our purposes we will distinguish it from a formal metadata taxonomy. The irony of metadata -The rating is not lost!)

3. of taxonomies(and other types of controlled vocabulary) is metadata we use to describe each piece of content (e.g., pages, text, data, products, help articles, files, etc.) and establish connections with other similar content.

While the AI ​​framework is a map of how our content is organized, a taxonomy is a concept map. we use to describe our contentand how all these concepts relate to each other. A taxonomy is often very different from the AI ​​structure or navigation, and is often more extensive and technical than either.

Since the taxonomy itself is not shown to users per se (although we may show users some of the topics or tags as a way of navigation, search suggestions, or enhancements), this is a place where we can do a very deep dive. complete and logical sorting background work. Typically, with a taxonomy we care about logical precision, whereas with a visible navigation structure we care about grouping things in the way that suits them.user mental models.

4. Content models (or data)They are also part of the backstage, used to describe the differentThe typeof content, what information it contains, how it links to other types of content, what metadata is applied to it, etc.

For example, a NN/gArticlesombreroAuthor(s),that link to the profile pages of these authors,affairs(this link to other content on the same topic and, if available, full topic overview pages) and so on.

Why create taxonomies and why should UXers be involved?

Taxonomies are not created for the love of classifying things (although some UX users like me love it). Taxonomies allow us to effectively retrieveNocontent related to a specific concept. They are often used in several ways:

1. Make connections between content.Related content widgets often retrieve content that is in the same taxonomic category, even if it is in a different categoryNavigationFor example, on the NN/g website, the related articles you may see below this article come from our internal topic taxonomy, which is much more detailed and specific than the general topics that may appear on our website (for example, example, this article belongs on the topicinformation architecture,Behind the scenes, however, we use the detailed ranking even further.Information architecture > Metadata > Controlled vocabularies > Taxonomy).
2. faceted navigation.facets, which allow users to apply multiple filters at the same time, are based on faceted taxonomies. Facets allow users to very precisely meet their information needs without tediously scrolling through adeep navigationBaum.

3. search suggestionsand refinements. As users enter their search queries, the system can query the taxonomy and retrieve results for related terms (that differ from what the user entered). You can also present categories to broaden the user's search.

Other types of controlled vocabularies: thesauri and ontologies

Another common area of ​​confusion concerns the difference between a taxonomy, a thesaurus, an ontology, and a knowledge graph. They are all forms of metadata used for classification. All of them can be represented as graphs in which the nodes are concepts and the edges are different types of relationships between these concepts.

taxonomies arethe simplest of these structures. A taxonomy can be hierarchical or faceted. A faceted taxonomy consists of several different hierarchical taxonomies that work together to describe different aspects of the same resource. Taxonomies revolve around parent-child relationships between concepts. The deeper you dig into a taxonomy, the more specific the concepts become (they can also be parts of a larger whole).

a thesaurus(confusingly not the dictionary type of thesaurus) is a data structure that contains not only parent-child relationships between concepts, but alsoassociationRelationships (“related term” that is not synonymous but is conceptually related) andequivalence Relations (Synonymous terms, one being the official preferred term and the other non-preferred terms.) A thesaurus allows for control and consistency of synonyms, which is extremely important for complete information retrieval.

For example, if you have labels for your company's intranetsuave,Suggestion,opinion Von Work, miTomás, but informally they are all used interchangeably, so each of these tags is only assignedsomeof relevant content and therefore you will get incomplete results if you search for just one of the terms. A thesaurus can link all of these concepts as related or as synonyms (choose one as your preferred term). So in this case we probably would have done thisSuggestionmiTomásbe in an equivalence relation withSuggestionis the preferred term. These two concepts are in an associative relationship with each other.suavemiSpecifications, because they are not exactly synonymous, but conceptually very related. (Note: Most "taxonomies" created for digital products are actually thesauri, as taxonomy management software often includes thesaurus features, such as preferred terms and related terms.)

Ontologiststhey are the most flexible and complex of these types of metadata structures and are commonly used to map knowledge across complex technical domains. Ontologies support many different kinds of meaningful relationships (besides parent-child, associative, and equivalence relationships) that link concepts semantically.

For example, in Linnaeus' taxonomy of biology that many of us had to learn in school, a polar bear (Blueberry) is a specific species of bear (Ursidae), which is a specific type ofcarnivore(etc). We are limited to only one type of relationship: parent-child relationships (or class inclusion). In an ontology we can map other types of conceptual connections: not just onepolar boneKind ofCarry, but also youhabitatYPolar, That isstate of conservationYthreatened, etc. By building these relationships from what we can formalize about a domain of knowledge, we capture many more aspects in a formal structure that can then be used to build a database.

The flexibility of ontologies to represent many types of relationships between concepts allows a detailed multidimensional classification of information.

Ontologies are large, complex designs that are typically used to describe a specific domain of knowledge. These are typically not the kinds of projects that a single UXer does alone, as they are often the work of a team of ontologists, taxonomists, software developers, and domain experts. However, it is important to know them and their relationship with taxonomies. Especially when working in academic areas or Semantic Web projects, you are likely to come across ontologies, so it is helpful for a UX professional to understand what they are.

Construction of a taxonomy

Creating a taxonomy from scratch is quite a complex process, and the full details of how to do it are beyond the scope of this article, but there are some common themes on how to proceed regardless of project type. It's okay to start small, and a taxonomy doesn't have to be a multi-year project; simply having a small set of themes that you consistently map to your background content is a good start!

1. Inventory and review your contentto start. You cannot organize anything without knowing what you are going to organize.
2. See if there is a standard taxonomyAvailable first for your industry or domain. If there is an existing taxonomy, you can get a big advantage: you may need to adapt it to your specific context, but this can save you a lot of work. Note that the cost of existing taxonomies varies: some are free and open source, others are quite expensive and require a license. A good starting point is to checkBARTOC.orgto see what is available.
3. Identify theconceptsaround which you will build your taxonomy.They can be found in the content itself, any existing metadata (such as keywords or topics in your content management system), discussions with subject matter experts, and important internal business terms. The concepts should also come from existing user data, resulting for example from interviews, usability tests, etc.search logsto make sure you don't create taxonomy branches that don't benefit your users.
4. For each key concept you identify, collect information on other similar related concepts.or words and where the source of that concept was (for example, the URL on a website).
5. Evaluate the conditions of your candidate. Look for relevance to the user's needs and avoid "singularities": terms that aren't related to anything else or terms you have little knowledge of on the subject. Including concepts associated only with the content is probably a wasted effort.
6. Define preferred term and non-preferred variants for each concept. Since the full taxonomy is not displayed directly to users, you have a bit more freedom here.use internal business jargonthan you would in navigation, but it's still wise to choose preferred terms that are highsmell of informationfor users, as it can display some of the terms in topic links or search suggestions. It's better to choose a friendly term as your preferred term and associate internal business jargon or more technical language rather than non-preferred synonyms.
7. Build relationships between concepts..For a hierarchical taxonomy, this means defining the branches of its tree and deciding on parent-child relationships between concepts. This is where most of the work takes place, as you must decide the granularity of each level of the hierarchy. At this point, you'll also identify related concepts that aren't synonyms and associate them with relationships of related terms (so, technically, you'll also create your thesaurus).
8. Review and revise your taxonomywith your stakeholders, internal subject matter experts, and content strategists. This process is likely to involve iterative refinement, and it's up to the UXer to bring the user perspective into these conversations (especially in terms of choosing preferred terms).
9. Apply the taxonomy to your content.Depending on the amount of content you're dealing with, this is a significant effort and requires training the tagger to use the taxonomy (i.e. when to use one term over another). Of course, there are AI tools for automatic classification, but expect them to make mistakes and require manual editing and refinement.
10. Establish ongoing governance and maintenance.The long-term usefulness of the taxonomy depends on performing regular reviews to add, rename, merge, or delete terms, and also checking content markup examples to ensure the taxonomy is being used correctly.

Summary

Taxonomies are a powerful way to build content relationships between digital products and are the hidden and invisible organizational systems that fill in the gaps that user-centric navigation systems can leave behind. When properly defined and maintained, a taxonomy can provide better search suggestions and post-search refinements, faceted navigation, and automatic links to related content.

Fuentes

Dean Allemang, James Hendler. 2011.Semantic Web for the Work Ontologist (2nd ed.).Morgan Kauffman, Waltham, MA.

Heather Present. 2016.the accidental taxonomist(2nd ed.). Information today, Medford, New Jersey.

International Organization for Standardization. (2011)Information and documentation - Thesauri and interoperability with other vocabularies - Part 1: Thesauri for information retrieval(ISO 25964-1:2011 Standard) Obtained fromhttps://www.iso.org/standard/53657.html

Mary Whittaker and Kathryn Breininger. 2008. Development of taxonomies for knowledge management. In itWorld Library and Information Congress: 74th Ifla General Conference and Council,10.-14. August 2008, Quebec, Canadahttp://www.ifla.org/IV/ifla74/index.htm