“In any given culture and at any given moment, there is always only one ‘episteme’ that defines the conditions of possibility of all knowledge, whether expressed in theory or silently invested in a practice.” – Michel Foucault, The Order of Things: An Archaeology of the Human Sciences

When the movie Johnny Mnemonic came out in 1995, I often heard people mispronounce the word “mnemonic” as “pneumonic”. I speculated at the time, rightly or wrongly, that more people knew the term “pneumonia” than “mnemonic” or maybe one was simply easier to say. Strangely enough, I’ve now associated the concepts of memory and sickness because of the similarity, and confusion, of those two concepts. Maybe a pneumomic device helps you remember something…or maybe it helps your lungs function properly. Are we concerned about another pneumonic plague or a pending mnemonic plague? Who can say?

The Mnemonic Plague

If we live in a knowledge zeitgeist, it may someday later be defined as the death of knowledge. Or, more accurately, the death of the belief in knowledge and expertise in favor of opinion and belief. The results of successful misinformation campaigns include a skepticism of expertise and knowledge, feelings that opinions are equal to or supersede knowable facts, or even the inability to know anything at all. Postmodern philosophy has concerned itself with the idea that there is no objective truth, either foreshadowing the imminent knowledge paradigm or driving it. Baudrillard has predicted, quite accurately, that simulations and simulacra will become so prevalent that all meaning will be meaningless.

Add to this another sociological trend, the current workplace generational turnover from a larger, knowledgeable, older generation to a younger generation buried in and swept up by these currents of skepticism. With reality changing so quickly, and so many people being distrustful of what reality is anyway, are we on the cusp of a mnemonic plague? A plague in which no one remembers anything? A plague in which we all doubt the ability to remember it correctly? A plague in which we are convinced by others that reality is not reality, facts are not facts, and that memory is susceptible to convincingly reality-sounding alternative histories?

Johnny Pneumonic

When we began the great shift from paper to electronic documentation, the speed at which we would be able to create, store, and access information grew exponentially. Imagine a library which could feasibly contain all knowledge from human history across all countries, languages, ethnicities, and religions. An electronic library that would make the Library of Alexandria, the Library of Congress, and the Bodleian Library look like corner news stands. One library to rule them all. That happened, sort of, with the expansion of the World Wide Web. Vast troves of information were digitized and made available online even as people continued to create informational web pages and create documents which were born digital, never seeing the pulpified remains of a tree. In parallel, all the other kinds of news, information, and entertainment were also being born digitally and, in the great democratization of the Internet, anyone anywhere could theoretically have access to the ability to view and create information assuming they had some kind of device and network or satellite infrastructure.

Imagine the possibilities! Imagine how knowledgeable we could all become! Imagine knowing anything, anywhere, at any time, at the moment you need to know it! And some of that is true. What is also true is that people create patently false information reflecting their personal beliefs and agendas. There are millions of mediums, channels, and formats available to anyone to tell us what magical healing herbs will make us live longer and with fewer wrinkles; entire video libraries dedicated to documenting the horrors of vaccines and Western medicine; travel logs with erroneous anecdotes and false facts with the only verification being the person who created the content; videos of lizard-people trafficking human children; videos of directors faking the moon landing. All of this available to anyone, anywhere who wants to be an audience.

Whether an individual or a state actor, whether the content creator believes their own content or not, whether we can decipher what is true and what is not, all of this information is there for us to consume and form our own opinions about. That is freedom. That is democracy. That is the democratization and platform of eight billion voices. We no longer have to be slaves to a single narrative; no, we can write our own narratives and gather our own followers, forming fragmented communities in a new-world Splinternet in which we can all, finally, show everyone else a picture of what we had for lunch.

The Codification of Memory

I could say things were better back in the good old days, whenever those were and for whomever they were good, but that’s also an opinion, not a fact. The genie is out of the bottle. The train has left the station. The die is cast. We have crossed the Rubicon. The horse has left the barn. We now live in a world of snowclones, memechés, and information of dubious provenance. Compounding the intersection of easily creatable and accessible information, skepticism of expertise, and generational turnover is the ever-improving use of AI tools to generate very impressive, lifelike, and believable images to support textual messages. Don’t believe we faked the moon landing? Here’s a photo! Didn’t that famous Hollywood actor appear in a porno? Here’s the video!

And now for something completely different. Or, rather, here is me now getting to the point for my target audience of information professionals: what do we do in the face of all this misinformation? Our jobs…the best we can.

I’ve been working on taxonomies for over 20 years. At various times–sometimes from one year to the next and other times one hour to the next–I have experienced that overwhelming sense of doom, frustration, or hopelessness in the face of shockingly ignorant misinformation and opinions. I have long since abandoned any hope for the Semantic Web’s promise that real, verified meaning could be captured in logical, formalized, human and machine-readable structures like taxonomies and ontologies. Despite the acknowledgement, both philosophical and practical, that there is no Truth with a capital “t” and that all facts are context-dependent, I still try to build semantic structures that codify the truth of my organization at the moment we are building it.

In the context of any moment, there are truths we can build and verify. We can create preferred concepts and connect them with human-readable, sensible relationships and build taxonomies into interconnected graphical knowledge bases applied to a variety of content vetted by subject matter experts. Verified and vetted content may no longer be true in a week or a year or a decade. The relationship between two concepts may quickly become outdated. A preferred concept may fall out of use or fashion. Despite all of this, codifying knowledge in taxonomies and ontologies is not an act of futility; it is an act of capturing truth and memory at a moment. We can document these changes over time and look back over the history to see what was true versus what is true now.

These semantic structures are a way we can document organizational knowledge and pass this knowledge on to the next employee, the next generation, the next iteration of a company. Creating good historical records in semantic models serves a knowledge management function in being one way we can enable knowledge handover from one person to the next, one team to the next, one project to the next.

The organizational book of knowledge as written in taxonomies is often edited and changed, but it is still a book to which we can refer with confidence that what we tried to build was accurate and true to the best of our abilities. Taxonomies are the new black. Ontologies are the mother of all semantic structures. In space, no one can hear you taxonomize…but your work is still valuable.

“Like a wave in the physical world, in the infinite ocean of the medium which pervades all, so in the world of organisms, in life, an impulse started proceeds onward, at times, may be, with the speed of light, at times, again, so slowly that for ages and ages it seems to stay, passing through processes of a complexity inconceivable to men, but in all its forms, in all its stages, its energy ever and ever integrally present.” – Nikola Tesla

Some of the most thorough, authoritative, and well-constructed controlled vocabularies have been built and curated over the course of decades. The NASA Thesaurus was first published in 1967. The Library of Congress Subject Headings can be traced back to its roots in 1898. The MLA Thesaurus was modernized in 1981 based on previous classification methods. The Getty Art & Architecture Thesaurus was started in the late 1970s. There are many more examples of well-established controlled vocabularies, some older and some newer.Maybe there’s something to the adage “with age comes wisdom” in the longevity and authority of these vocabularies.

Many of these longstanding vocabularies serve the academic system and do not necessarily “move with the speed of business”. In a business environment, the emphasis is typically on speed, agility, and efficiency. None of these qualities are in opposition to the typical development of good controlled vocabularies, but a fast-paced organizational environment can certainly be difficult to support with vocabulary development that can’t keep up.

In my experience, product vendors talk about how to speed up taxonomy and ontology development while semantic practitioners (librarians, taxonomists, ontologists, etc.) regularly have to strike a balance between maintaining quality semantic models and serving the business needs. The speed at which semantic models need to be developed in a business setting often leads to the extension of inappropriate taxonomy use cases in order to support and facilitate immediate business priorities. We need to step back and ask ourselves how we can as taxonomists effectively support the business while addressing what should be modeled in taxonomies and what should not.

In this blog, I’m going to focus on two closely related cases of data which I think are not appropriately modeled or maintained in a semantic model but are often brought up as examples of serving the business quickly.

Processes & Sequences

A business problem I’ve seen in my roles as a taxonomy management platform product manager for a vendor and as a taxonomy practitioner is the modeling of processes as taxonomies.

There are several difficulties with modeling processes as taxonomies, and why I tend to discourage their inclusion as part of the overall enterprise taxonomy and ontology semantic framework. First, steps in a process aren’t usually semantically hierarchical. For example, say you are asked to support a step-by-step process for submitting a ticket in a request tool such as those offered by ServiceNow or Atlassian in a centralized taxonomy. An IT service ticket can be opened, assigned, reassigned, resolved, closed, and reopened. Building these steps as a hierarchical taxonomy doesn’t convey their true relationship to each other. You can build these steps as a flat list, which addresses that problem, but then you have a sorting issue of making sure the steps are displayed in the proper order in the consuming system. Since the actual ticket requests live in a consuming ticketing system while the controlled values naming each step in the process live in a taxonomy management system, this may not be a problem at all. The ticket is an object that is simply tagged with a new value at each stage in its lifecycle, regardless of the flat or hierarchical structure in the taxonomy. So, the first problem may be solvable, but it needs to be addressed.

The second issue is that processes are rarely progressive steps from beginning to end. The stages of a ticket may move up and down a flat list or hierarchy, often skipping steps in between and moving back to a previous step. Again, if the ticket is tagged from a flat list which is properly ordered for display, this may not be an issue. If the consuming system must follow the flat list or tree order, however, there may be challenges in changing the value to a previous step. Where this gets complicated, however, are processes in which the same steps are repeated by different groups in the organization. A contract, for example, can go through contract review by the vendor, by the business team representatives, by legal, and again by the vendor’s legal. Modeling a process like this taxonomically often means that these repeated processes are either appended with the team or entity conducting the process step (business review, legal review, etc.) or repeating the steps as children under organizational group headings, creating a polyhierarchical nightmare. 

The final and most difficult challenge with modeling a process is that they often change. In another example, modeling the customer journey from start to finish is much like stages in an IT ticketing process. The customer rarely moves neatly from each step to the next. More importantly, however, is that marketing processes frequently are overhauled with changing values anytime there are changes in business strategy. Even when the values don’t change, their ordering of the process does and these orders are often reflected as navigational structures represented as filters on the front end. Representing a customer journey based on filterable values is challenging because of the numerous ways a customer may enter the UX pipeline. In retail apparel, for example, they may start looking for products at Gender (Men’s, Women’s, Children), then apparel type, then size, then filtering by material or color. Or, the customer may start with color, then apparel type, then size. There is a process here, but one with multiple entry and end points. Trying to represent this as a taxonomy is incredibly difficult.

A similar problem I’ve come across in taxonomy modeling is using lists or hierarchies to indicate sequence. Most processes have steps that are in sequence, but in this case I’m talking about strictly fixed sequential order. Examples can include the order of books or films (by date or by narrative order), the list of U.S. Presidents in order, or events by date.

The fundamental issue for most taxonomy management systems, and for many systems relying on taxonomy data for that matter, is the default to alphabetical display in lists. For most taxonomy hierarchies, alphabetical is the preferred display order, with each cascading branch also alphabetized. On most front end websites, navigational taxonomies are ordered by use with the most prevalent ways to access information listed first. Front end website platforms are built for this type of information display because the hierarchies do not necessarily follow what I would call semantic or “is a” taxonomy practices; parent child relationships are based on filtered drilldowns, not by strict contextual meaning. Where this becomes an issue is when an organization, quite rightly, wants to consume centralized taxonomy concepts for the front end experience.

Even when using the larger graph underlying taxonomy and ontology structures, conveying order can be challenging without the right functionality to support it or the ability to leverage the model in consuming systems that can only handle flat lists or shallow hierarchies.

Modeling Options

Modeling sequences isn’t out of the question in taxonomy management systems assuming that 1) the taxonomy and ontology management system includes functionality supporting modeling options, or 2) downstream systems can effectively handle or transform concepts received from a centralized taxonomy.

To model steps in a process or to capture sequences, there are a few options. If you are using a tool supporting RDF and core SKOS elements, you can consider using skos:OrderedCollection. An ordered collection is exactly what it sounds like: a collection of concepts put in a specified order. Using an ordered collection for a list of concepts in a branch of taxonomy allows listing those items in the order desired. There may be no other indicators of why those concepts are in that order if stripped from their contextual parent, but it will force sort a group of concepts. This assumes, of course, that the consuming systems don’t simply revert to alphabetical order once received.

A more flexible and sustainable way to model a process is to model semantically using “is a” rules and then leveraging a true ontological structure and a graph to map the journey. This means modeling concepts in their one best location in one or more taxonomies as part of a larger domain model. Modeling this way leans into the strengths of an ontology by using associative relationships between entities to make a graphical representation of the order while also connecting the entities to their owners. So, for example, books or films may have relationships like is followed by / is preceded by and then can be connected to their authors, directors, and actors as part of a greater graph.

Another option is to include a property on each concept. The property could be a field which indicates a numerical value listing its placement in the list. While this metadata field could be useful in the taxonomy user experience, whether or not these property values could be used in consuming systems to order items is still problematic. Furthermore, it gets complicated if it’s necessary to order multiple sets, all of them including the same numbers as an ordering property.

In advanced graph-based taxonomy and ontology management systems, there may be an option to use reification or RDF* to support metadata on triples. In this way, the ordering is embedded on the edges themselves. For example, books and films could include relationships with a release date. This could look something like “James Bond film” has release date [20061117] “Casino Royale” in addition to a broader/narrower relationship between the two concepts. There are several modeling options to make use of associative relationships with added metadata on the relationship edge.

In sum, it’s not impossible to model processes and sequences in taxonomies, but it requires thoughtful modeling in the context of other existing business taxonomies likely sharing the same overarching business ontology. Moreover, thoughtful modeling may not move with the speed at which an organization wants to move, but slowing down and getting it right the first time can save a lot of painful rework later.

“Hello, is it me you’re looking for? / ‘Cause I wonder where you are / And I wonder what you do / Are you somewhere feeling lonely?” – Lionel Richie, Hello

One of the most challenging activities in taxonomy work is communicating the value of taxonomy to potential business stakeholders. With so many shiny, promising technologies and methodologies, it can be daunting for the taxonomy strategist to win over converts to taxonomy use. Taxonomies and their applications are often misunderstood or are narrowly focused on a few common use cases like navigation. While business users can clearly articulate their needs, they may not be able to connect those needs to how taxonomies can be applied in the business.

The taxonomy strategist must be able to communicate the value of taxonomy while expressing the complexity of semantic structures like ontologies and their supporting technologies simply and succinctly to a variety of business stakeholders. 

Communicating the Value through Examples

To gain taxonomy users, it’s essential to communicate the value of taxonomy. One way to start is to seek out areas taxonomy can directly address, find examples of the current state problems, provide taxonomy-based solutions, and then communicate the findings to the business owners. This process can be initiated by the taxonomy strategist or by the business owners themselves, assuming, of course, they know to contact the taxonomy team in an effort to answer their need.

One simple, powerful example is to review a search-dependent organizational website–which could be an internal intranet or external, public-facing website–and collect examples of navigational and search barriers causing confusion, poor search results, or revenue-losing scenarios. For each example, provide an explanation of how taxonomy might help. For navigational issues, the taxonomy solution may be category restructuring or improved, facet-based results filtering aligned to the typical user journey. For search retrieval issues, taxonomy may be used for typeahead search keyword matching or to improve search relevance to include more accurate or additional results through content tagging or keywording. Navigation and search are often close to time-saving or profit-driving activities, improving the efficiency and bottom line of the organization. Search examples and their potential taxonomy solutions, therefore, are closer to the source of organizational revenue and make convincing use cases.

As generative AI becomes more prevalent in organizations, finding examples of general or inaccurate results and how an enterprise, domain-specific taxonomy (and ontology) can act as foundational training data to improve those results can result in convincing proof of concept projects. Generative AI and machine learning models can seem like magic to the average user who may not know the amount of time and data it takes to train a model to produce accurate and useful results. Providing examples of poor machine learning model output can illuminate the need for clean, accurate foundational data. As an organizational source of truth, taxonomies can provide such semantic data.

To overcome user confusion about what taxonomy means or clarify what they think taxonomy means, try starting with the end result and work backwards. When assessing the value of a new bathroom faucet, someone will look at whether the fixtures look appealing and if hot and cold water comes out as expected. Initially, no one is interested in the pipes. Taxonomy, unflatteringly, is the pipeline infrastructure providing clean water to downstream consuming systems. First show excellent search results or machine learning outcomes and then explain how taxonomy is the basis for those results. If business stakeholders are interested in taxonomy, all the better for your work and evangelization. If they aren’t, let them be impressed by the final state and develop a process of working together to get to and maintain that final result.

Communicating the Value through Time and ROI

One potential stakeholder hesitation may be the time it takes to perform discovery, conduct the build, and put taxonomy values into production. This process can take time in the initial business stakeholder relationship. Once established, however, the speed at which business users can request concepts and see them live can move as quickly as your organizational systems can handle. People often believe they need to “move at the speed of business”, which, ironically, they think is fast but is more often cumbersome, manual, and slow. What they want is the magical now in which thought is converted to action faster than Captain Kirk can have his shirt ripped when first confronting an alien species.

Machine learning techniques, once perfected, can offer the kind of rapid response business owners are looking for, but only after a lot of training. Specifically, a lot of training on assets and data tagged with taxonomy. Too often, the “magic” of artificial intelligence business users are sold isn’t artificial at all: it is thousands of hours of tagging content and training models to get the desired results. If done properly, there’s nothing wrong with using machine learning models to quickly react to trending topics or generate text on the fly. However, the slower growth of a taxonomy, as I cover in my blog The Taxonomy Tortoise and the ML Hare, actually creates speed in other areas, saving time in responding to consumers’ direct search queries and tagging content to train and evolve machine learning models. Communicating the need for time investment up front to generate time-savings later can be compelling.

Communicating taxonomy ROI, which I covered a few years ago in my blog for Synaptica, Running a Successful Taxonomy Campaign, can be extremely difficult. How do you explain how words become money? Again, show the examples. Mining successful and failed search results and mapping these to taxonomy as metadata tagged to assets can show a direct line between creating taxonomy concepts, applying them to content, and successful search results that end in a product purchase. Going back to time, time is money: time employees spend manually creating, tagging, and manipulating content which drives sales; time spent training machine learning models; time spent seeking information which has not been tagged with metadata. Ramping up taxonomy processes to more quickly tag content and put words into production will result in quicker time to money and realized ROI. While starting taxonomies can be slow at first, the more success the taxonomy strategist has in engaging business users, the more quickly the taxonomy is built out and covers the breadth needed to tag assets and express important concepts users are seeking.

Communicating Complexity

Communicating the nuance and complexity of taxonomies and ontologies may be necessary as the details of a pending or ongoing project develop. Few business contacts need to know the difference between a flat list, taxonomy, thesauri, or ontology. In fact, I find there are disagreements about the differences even among practitioners. That said, users can come to the discussion believing that taxonomies are only hierarchical lists of terms. For most practical discussions, I use the term “taxonomy” to include flat and hierarchical lists of terms, properties, and hierarchical and associative relationships. I rarely bother with ontology concepts like classes unless they are necessary to meet the project objectives.

If these terms do need clarification, however, I often clarify with simplicity. Taxonomies are concepts (preferred labels) that include synonyms (alternative labels) and other metadata attributes (properties) and these concepts can be related hierarchically and through custom relationships (associative relationships). When discussing ontology, I usually state that taxonomies are the words you want to use and the ontology includes the rules for the words you want to use. For example, how concepts are grouped (classes), how they can be related to each other (domain and range constrained by classes), and whether certain properties can be made available for a use case (properties constrained by classes). That’s often all a user needs to know.

There are more advanced use cases, like machine learning, which is, in my experience, more of a mapping of ideas than an education. Data scientists usually use all the same concepts as taxonomies and ontologies but may use different terms to express them. After one or two conversations, the mappings are understood and the complexity is simplified. It’s not often a data scientist needs convincing to leverage taxonomies, but getting on the same page with conceptual ideas is a good way to make taxonomy value clear.

In large organizations, there is usually information architecture complexity as well. Because of this, taxonomy can often become necessarily complex as values are consumed by and flow through various systems. Understanding this workflow is not always a prerequisite for understanding the value of taxonomy, however, and does not need to weigh down conversations with potential business stakeholders. If it does become necessary, simplify those information architecture diagrams into simple flowcharts between systems, showing at a high level how taxonomy concepts move from system to system and what they do in each.

Being a taxonomy strategist is challenging, but is a necessary part of the job for taxonomy to show and prove its value in the organization.