Writing

Friday, 14th of October, 2022

Categories Or Tagging? Differences In Taxonomy And IA

Vinyl records are sorted by genre, as shown here, in a classic record store
Vinyl records are sorted by genre, as shown here, in a classic record store

When you’re building a house, the person you trust with sorting out the rooms and amenities is an architect. They learn what the environmental requirements are, what your intentions are, and how you want to live in it. Then they arrange floors and rooms in a way that makes most sense, so everything becomes accessible and usable, and the pipework for water, gas or electricity, is organised in a way that combines life comfort with efficiency and structural logic.

Information Architecture

Information in applications for web, on desktop computers and mobile devices, is also organised in a way that makes most sense. Ideally, it’s organised to be accessible, useable and efficient to find, and there’s an underlying structural logic, which will help users to navigate through information. The work done to achieve this used to be a profession in itself, called Information Architect. Today, Information Architects still exist, but it is rare as a job description. Applying IA, however, is a firm element of the UX development for every industry.

Information Architecture has roots in Information Science and psychology. It combines perception with semantics and ontologies with taxonomy and way-finding. IA isn’t just for apps or the web: airports, hospitals, university campuses, use way-finding too, which requires organisation and representation using an IA.

Familiarity, Learned Behaviour And Orientation

We humans like familiarity and context. We feel comfortable when we find our ways around a new situation and we are distressed when we feel lost and disoriented.

Our brains love recognising things we’re familiar with and they reward us with small doses of dopamine every time we successfully solved a problem. It is this mechanism that lets us experience something as satisfying or dissatisfying. Finding your way in an unfamiliar environment has been an evolutionary necessary treat. It’s no wonder we were equipped with a biochemical reward system for great orientation and successful hunting.

Making everything findable through context isn’t easy if there is no sense of order and organisation. It becomes easier to navigate when a taxonomy leads our understanding. Taxonomies are human made constructs, “the practice and science of categorisation and classification”, as Wikipedia puts it.

Familiarity and context rely on our ability to understand how something is related, or what it is for. No one needs to explain how a chair works, because we grew up with them and early on, we’ve learned how to use them. This familiarity with the object chair allows us to translate the same expectations for a function to sofas, bar stools and even tree trunks in the landscape. We’ve learned how to sit, and we know what can be used for sitting just by seeing or touching it.

When you’re entering a supermarket, you can assume that the milk is located somewhere where the rows of refrigerators are. It’s a learned behaviour of context, something we can rely on even without actively thinking about it. We are even able to connect seemingly detached context, like synonyms for the thing that’s on our mind.

If I’m looking for a pair of jeans and I am presented the word “trousers”, I recognise it’s in the same category, as trousers is a synonym for jeans. Likewise, antonyms – the opposite of what I’m looking for – are also helpful for finding orientation. It’s important and a good idea, when working on IA, to look at the meaning of words and differentiating things clearly from the opposite meaning.

Once we’ve learned one behaviour, we are able to extrapolate its meaning, not just with other objects, but also in other environments. Once you’ve used a bus in public transportation in one city, you’re able to apply the same learnings to the city transportation systems in other places. They all work the same and they all have a time schedule for each stop.

Nearly every environment that requires orientation has some form of taxonomy built into it, otherwise we would find it quite confusing and it would be rendered useless.

Taxonomy And Ontology

How do taxonomies work and what makes them useful? Here is Wikipedia’s definition:

A taxonomy (or taxonomical classification) is a scheme of classification, especially a hierarchical classification, in which things are organised into groups or types. Among other things, a taxonomy can be used to organise and index knowledge (stored as documents, articles, videos, etc.), such as in the form of a library classification system, or a search engine taxonomy, so that users can more easily find the information they are searching for. Many taxonomies are hierarchies (and thus, have an intrinsic tree structure), but not all are.

Taxonomy assigns meaning from the general towards the specific.

In biology, taxonomy (from Ancient Greek for arrangement and naming) is the scientific study of naming, defining and classifying groups of biological organisms based on shared characteristics. Organisms are grouped into taxa (singular: taxon) and these groups are given a taxonomic rank. Groups of a given rank can be aggregated to form a more inclusive group of higher rank, thus creating a taxonomic hierarchy (Source: Wikipedia).

Not all taxonomies are organised in hierarchies. Books, movies and music are traditionally sorted by genre, which is a rather flat hierarchy with grouping based on attribute. Of course this kind of categorisation runs into problems as soon you need to introduce sub-categories, or when a medium has more than one fitting attribute.

Historically, taxonomies have always played a big role in computing. The Microsoft Windows predecessor DOS had a strong hierarchical file structure. If you were deep down in one directory, you couldn’t jump directly to a different place, and you wouldn’t know what was in that other directory without going there and looking it up.

Ontology stands for a method of showing the properties of a subject area and how they are related, by defining a set of concepts and categories that represent the subject.

What ontologies in both information science and philosophy have in common is the attempt to represent entities, ideas and events, with all their interdependent properties and relations, according to a system of categories. (Source: Wikipedia)

Artificial Intelligence models also apply ontology in natural language processing within machine translation and knowledge representation.

Categories

In many offices, particularly those with a lot of documents stored in traditional folders, you’ll still find filing cabinets: heavy drawers, made of metal, holding hundreds of files in each drawer, usually sorted with tabs using alphabetical or alphanumerical categorisation.

The image of folders in a drawer is a metaphor that’s still in use today. Virtually all major desktop computer operating systems are using the file metaphor, with icons depicting folders and files. The term desktop itself is used metaphorically for the surface we use to navigate between hard drives, folders and files and the waste bin.

Cherries, bananas, pears and grapes belong to the same category: fruit
Cherries, bananas, pears and grapes belong to the same category: fruit

More modern UIs used for mobile operating systems were initially trying to omit these old metaphors. But with more recent versions, even iOS and Android cannot completely escape the underlying mind model. Both, iOS and Android, now let you access files through symbolic folders on the device or in the cloud.

Virtual folders – just like their physical counterparts – can be filled with anything we put inside. Categories, on the other hand, are very canonical. They are trees with trunks and branches that follow a certain logic, inherited from the parental branch. The canonic rule is that a subject cannot be in multiple categories simultaneously. This would lead to multiplication of the object and the resulting redundancy would introduce chaos to the system.

So what do you do if you have something that can be in one category, but also in another? This is where tags come in handy.

Tags And Metadata

Tags are metadata that is attached to subjects. So you’re not assigning subjects to tags, but rather tags to subjects. A tag adds relevance in context, which helps us associate the subject with others alike.

Tags enrich content with context. They are often described as keywords, and because search engines, like that of Google, are relying on context, those keywords become important for how the search engine informs itself about a given subject. But whilst keywords can be placed inside text (creating new context—useful for machine learning, for example), tags work more like a label that helps recognising a certain quality of the content it’s assigned with.

Tags add semantic meaning
Tags add semantic meaning

When you’re assigning tags to a data point, you can attach more meaning to it, which helps with finding it again, but also with connecting other, related information. Tags work more like filters than definitive grouping. Tags also freely follow multiple semantic expressions: they can describe things like qualities, attributes or similarity.

To narrow a search result or choose attributes within a category, tags can be used as filters to help us narrow down the specifics of what we’re looking for.

The benefits of tag-enriched content with metadata are clear. But who adds the metadata?

Google, Meta and Twitter use algorithms that help sorting and connecting metadata with potentially relevant content. However, tagging is usually in the hands of users. There lies another problem, because metadata becomes useless, or even counterproductive, the more diluted or misleading it becomes. This has the power to mislead even the smartest algorithms, and until today, no one came up with a final solution for this growing problem.

Hashtags

Hashtags entered our perception in 2007, when Chris Messina, a Twitter employee at the time, proposed his idea to use the pound sign (#) to designate a keyword. Twitter founder Evan Williams thought hashtags were “too nerdy” to go mainstream, because they originated in the older Internet Relay Chats (IRCs). In an IRC, the pound sign is used to note the topic of a specific channel, and anyone can join a channel by using the pound sign at the front of a channel name.

Two years later, in 2009, Twitter embraced hashtags as hot links connecting content with the same keyword.

Hashtags had positive, but also negative effects on the Twitter user experience. For one, they proved useful to organise content and people used them to find each other. On the other hand, hashtags on Twitter can be easily hijacked and abused for unrelated context. That makes them prone to spamming.

Stowe Boyd, a coworker of Messina, wrote in a blog posts:

“My sense is that tags in Twitter, as elsewhere, define shared experience of some kind, involving all those using the tag. And the use can be either actively putting a hash tag (like “#hashtag”) into a tweet, or more passively opting to follow a stream of tweets related to a tagged theme.”

Early on, Boyd recognised the social potential that comes with hashtags, beyond their quality as metadata. He was right: hashtags helped people to organise, like with the #MeToo movement, or earlier, when tags helped spark the 2011 Egyptian Revolution.

Virtual Folders

Historically, these folders were called Virtual Folders. A common term nowadays (thanks to Apple) is Smart Folders. It’s related in concept to several other topics in computer science, such as saved search, saved query, and filtering.

Virtual, or Smart Folders are basically “saved search results” related to one or many behaviours. All files matching the criteria are dynamically aggregated into the virtual folder. A Smart Folder in Apple Mail will behave according to the rules you assigned to it. It might sort out Junk Mail, for instance, given a blacklist of domain names.

When you click on a hashtag in Twitter, a new feed is shown with content that’s connected to this hashtag. In the same way, a Smart Folder shows a dynamic content list of items that fall under the Smart Folder’s rule, a filter often described with a keyword.

On the desktop, Smart Folders mimic folder behaviour. They act as if they’re showing the location of a certain file or directory. In fact, they show a virtual representation of those files, which are stored elsewhere, in a variety of folders.

Another form of virtual folders are shopping lists. You may know them from the Migros app, Digitec or Amazon. These online stores let you add items to a list, which is an equivalent to a bookmark list. Like Smart Folders, shopping lists provide a directory of things to be available and accessible for the user’s convenience.

Forget Everything

What’s the point of categories, tags, hashtags or smart folders, if we can find anything with search?

Although search and filtering have become a standard in nearly every UI today, we still use categories and tagging to navigate information. Why? Because related information, more context, can help us make sense of any given subject. And sometimes it even helps you to be inspired, a feeling you may recognise from when you were looking for a pair of jeans online and you ended up browsing shirts or shoes.

In our daily interaction with various devices, platforms and systems, we have gotten used to navigating fluidly and adapt our behaviour to any given environment. We use categories, tags, folders, search and filters without even thinking about the underlying structure. That’s how good UIs have become today.

These advances have set the bar higher. Because we’ve been spoiled by Google and Wikipedia, we tend to demand the same quality of service from every other service. If the architecture of organised information can’t deliver on these expectations, it ultimately leads to frustration and decreased usability.

Probably the best approach to handle access and navigation of information, especially with large datasets, is to provide multiple ways to access it. If you can, let people search, filter, find things and use tags and categories to make sense of it. Let them decide what works best for them. Ultimately, you’re enriching the experience and remove barriers that could hinder users from finding what they’re looking for.