Insights Interview | Bonnie Griffin, Enterprise Knowledge

We interviewed Bonnie Griffin, a taxonomy consultant at Enterprise Knowledge. Prior to joining EK, Bonnie held positions in the field of taxonomy including PayPal and Indeed. She holds a PhD in French Literature from Vanderbilt University. 

For this Insights interview Bonnie discusses her new role with Enterprise Knowledge, the impact of AI for taxonomists and how the role of taxonomist has evolved. 

Tell us a little about you and your career evolution. What led from your PhD in French Literature to your first role as a Taxonomist?

BG: When applying for the program, I knew there wasn’t a massive demand for PhD in French Literature, but I've always valued learning for learning's sake, and I felt the time at Vanderbilt was worthwhile. Following graduation, I initially considered academic positions but also applied for a mix of roles trying to find anything that valued my research skills and analytical skills.

One of these roles I pursued was with Indeed. They were looking for a taxonomy analyst for the French market. Initially, they started out developing occupational taxonomies for the US and UK markets, but there was a need to improve recommendations for the job seeker experience. They discovered that one market is completely different from another; think of how healthcare or education can differ from one country to another. The company wanted people who could do more than simply translate; individuals who could undertake in-depth research, understand cultural nuances, and command the language beyond translation. 

Prior to my interviews, I didn’t know a thing about taxonomy beyond what you learn in biology. I did lots of reading and research around the subject to learn more, and this discipline ended up being a good fit because, for anyone with a humanities background, you believe language is important. The words, the manner in which they’re used, and the context. You want to understand how humans make sense of things and recognize our natural tendency to categorize and classify. The role sounded engaging, plus it offered the chance to work with people from different countries, building my cross-cultural understanding, diving into language and meaning. 

I worked in the taxonomy analyst role for the French market for approximately a year. Then, there were two different options I could pursue, evolving into a senior taxonomy analyst on the technical side or pursuing the people-managing side. It was the latter that appealed to me. I was promoted to taxonomy team leader and with this role managed seven different international markets including Belgium, Italy, Sweden, and Hong Kong. I was using less of my French language skills and focused on management skills and developing people. 

My next career step was with PayPal as a taxonomist working in financial services. The role was focused on customer operations and I worked closely with an ontologist, knowledge engineer, and knowledge managers. It was a new dynamic, and I was able to use my taxonomy skills to improve areas like their customer intent registry. PayPal wanted to understand why a customer was calling, identify the issue and provide a solution to reduce resources allocated to customer support. 

The registry PayPal used was like a taxonomy with categories and subcategories but lacking semantic content. At the time, it did not yet follow hierarchical rules, and lacked definitions or alternative labels. My initial vision was to turn it into a taxonomy to look at relationships and introduce context, but I quickly realized that my approach had to be scaled down to realistic goals. I focused on definitions. This was an excellent learning experience in balancing expectations of what could be achieved while demonstrating tangible value. I think this type of experience will resonate with lots of solo taxonomists, especially those working at larger organizations, where you have to balance best practices with what can reasonably be accomplished to start demonstrating value and continue to build momentum around buy-in. 

Tell us about your new role with Enterprise Knowledge. 

BG: My main focus is developing taxonomies for clients. It’s a mix of talking with stakeholders, leading focus groups and traditional taxonomy design activities. It entails asking questions like:

• How does an organization (or part of an organization) work? 
• What are their current needs? 
• What knowledge management initiatives do they already have in place? 
• What’s succeeded, what’s faltered and why? 

Consulting was new to me, but I always think it’s best to jump into the deep end. I liked the challenge of trying to get context into their current condition and explore how taxonomy can help them get into a better place, and then doing the actual design work. It’s this mix that appealed to me about EK. 

In previous roles, I would meet with different teams, perhaps talk with sales, product and marketing teams but with a true consultancy role you get to learn about completely different industries, organizations and what their needs are. It's a nice balance between working with a cohesive internal team, and getting to know new clients as projects evolve. 

What are the key challenges in developing and implementing taxonomies for large enterprises?

BG: I learned partly through trial and error and, sometimes, the hard way. With a previous role I had to think about who the right stakeholders are, who I need to talk with, and the dependencies involved. Just being able to design a good taxonomy is not always enough, but I didn't realize this immediately. Long before taxonomy implementation can be part of the discussion, there are enhancements required to a content management system, simultaneous projects, other teams involved, resources and funding to contend with first. 

I grew to appreciate that you need to demonstrate impact early on, even with next to nothing. There is an assumption that you should only prioritize working towards developing an enterprise level taxonomy, and that anything short of that doesn't matter. Realistically though, to help people understand and develop their interest in taxonomy, something they often haven’t heard about before, you need to cultivate interest in the subject, help them understand how the taxonomy can directly impact them: 

• How do I benefit? 
• Why should I care? 
• Does this make my team look great? 
• How does this help the client? 
• Does this save us money? Does it increase profits?
• How does it cut down on time wasted looking for information? 

These are some of the little ways that lead to a ripple effect. Demonstrating impact, whether that’s saving time, facilitating a content audit, or some other value generated from the taxonomy even before taxonomy is implemented with search. Once I figured out ways to do that, that's when more and more people asked questions, wanted to see how it works, how they can use this approach, and that built buy-in. 

How do you advocate for taxonomy-driven solutions within an organization?

BG: Understand what the big priority is. Imagine you have a content author spending time looking for an article they wrote but can’t find. They might need to recreate this content. That’s wasteful. Of course, this same organization might have multiple content authors, as well as hundreds of millions of active users. While helping the authors save time otherwise wasted recreating content, I realized it was more impactful to speak to the savings if those hundreds of millions of users are able to find the answers that they want without contacting customer support. That becomes the story. 

So, my advice would be to find out what is the biggest problem for the business, or the greatest opportunities around return on investment. I don't think taxonomists naturally think like that – at least, I didn’t initially – I was more focused on the scope of the project, the content I was representing, and the domain of knowledge I had to learn about – but that was only a small piece of the puzzle. 

Another aspect is making sure people understand taxonomy. Using the word “taxonomy” too much can alienate employees who aren’t familiar with it. We all tend to clam up when it's around something that we don't understand, because we feel intimidated. Take time to explain and show what you mean, especially in terms of how it will make their job easier and help them save time.

How do you see generative AI tools like ChatGPT impacting the field of taxonomy development and management?

BG: View GenAI tools as a co-pilot, not a manager. For example, it’s really helpful for brainstorming – “here are all these concepts that I know go in the same space, but I don't know what to name them.” Ask for some examples. It will be helpful even if you don't like any of the suggestions. It can be helpful to see what you don’t want to be able to articulate the approach you do want to take. 

It can be helpful for generating definitions. Say you're working on a taxonomy that's going to support a lot of technical documentation and it's a little out of your area of expertise. To get a bit more of a familiarity, ask for an entry-level explanation of how APIs connect to a particular tool. It's good for generating synonyms. For example, ask the tool to provide 10-20 different terms for “buyer protection.” You might get great answers. You might get terrible answers. You might have to keep working on prompts. 

It can assist with generating SPARQL queries, which can be helpful for refining a taxonomy. Ask for a SPARQL query to “show me all of the taxonomy concepts in this tool that don't currently have a definition or have duplicate alternative labels.” 

Finally, it can help with sorting and categorization. You can submit a collection of articles and request they be sorted into general categories. This shouldn’t replace classic approaches like conducting card sorting activities with live participants, but can provide helpful examples to compare against. 

These tools help you polish and brainstorm. 

What are the potential benefits and risks of using AI to assist in taxonomy creation?

BG: One area discussed in relation to GenAI at the recent Taxonomy Boot Camp is you don't get an answer; you get a response that looks like an answer. I would add that you can get a different result every time. For the recent Semantic Data 2024 conference, I put together a hypothetical use case, creating a taxonomy for a travel company. For instance, I asked a Gen AI tool to come up with 10 category names for tours that have sports like kayaking, hiking, and biking. I liked a few, asked again and it gave completely different answers. These were responses that I could use, but I still thought it was significant to acknowledge that the same prompt returned entirely different responses. 

You have to be comfortable with the fact that you can ask the exact same prompt several times and the results will not be identical. They will be different. People tend to know this, and many are also aware that GenAI is capable of hallucinations. 

You also need to be aware that these different tools can be biased in their answers. I had an experience where the definitions I was provided were biased; I then found myself arguing with ChatGPT. It's important to embrace that natural skepticism. Remember that your prompts might be misconstrued.

How has the role of taxonomists evolved with the rise of big data and AI technologies?

BG: About two years ago there was a proliferation of educational resources and posts on how to use generative AI, what it is, and how it will transform the industry. There was a sense that these jobs would cease. My first instinct was to panic but instead, I was encouraged that organizations like Synaptica and Enterprise Knowledge were quickly able to demonstrate that there are ways to intentionally engage with these tools - and that taxonomies have great value in the context of AI technology and can be key to obtaining more accurate, usable results.

It’s an evolution from initially seeing this as something that will interrupt and take away jobs to showing that anyone working with data will be valuable, especially if they are able to effectively articulate and determine how the work adds to structure, semantics and context.

At KMWorld last year everyone mentioned GenAI in some way or another. What stood out to me were the people who were able to provide practical takeaways and open our eyes in terms of what’s achievable. There was one talk related to video/computer games that was very approachable and easy to understand. It’s this type of energizing talk that I would love to see more of, and these are the presentations that I think about, especially if I leave with one small thing I could try and implement. 

How do you maintain taxonomies as content and organizational needs change? How do you balance immediate project needs with long-term strategy planning?

BG: One thing that I've learned to do more explicitly since joining EK is to identify risks and obstacles. It can be tempting to want to avoid alluding to the possibility of failure, because that can make people lose confidence in you. Consistently ask your stakeholders what they see as their biggest obstacles, plus be aware of knowledge management initiatives or taxonomy projects that have failed in the past. How can we sidestep any of those obvious pitfalls? How can we not stumble where other projects have stumbled?

Documentation helps to ensure you can confidently explain any adjustments or exceptions that you have made, changes to the tool and resources. Having the ability to go back and review when and why is important; memory won't keep track. 

Don’t overestimate what people remember – always assume a tiny bit of amnesia. We all provide presentations and updates related to a project. We often believe everyone remembers whatever we last shared, or we assume they will read the documentation and agenda. The reality is most people have largely forgotten what was shared previously and are unaware of the current state of the project.

One of the reasons I was interested in joining EK is because they do this kind of knowledge sharing well; they'll be able to speak to multiple levels of a project at any given time. You can see the nitty gritty timelines that we're working on. We're working on many activities, but we are able to zoom out and see the time and scope of the project as well as the granular detail. It’s a fine balance. 

The first step is seeing what the client thinks they want and considering the resources, time, funding, subject matter expertise as well as systems required to support the project. It may be this isn't quite what that project is going to accomplish, but it will lay the foundation

What are some resources or strategies that you would recommend to other taxonomists who want to broaden their knowledge or deepen a skill set? 

BG: There are different groups online, communities of practice specific for taxonomy you can join and follow. These are great places to ask questions and share ideas. Finding those different groups, whether it's Women in KM on LinkedIn, groups associated with a conference, these are helpful to join to get a sense of who the thought leaders are, who contributes. With your own social media, I would try to be helpful by sharing summaries, pulling out impactful quotes or sharing posts from others that had a wider following.

Even as early as when I started working at Indeed, I found Enterprise Knowledge’s Knowledge Base to be an incredibly helpful resource. They have a talent for taking complicated topics and making them accessible, and providing actionable guidance, and I’ve referred to their blogs and case studies many times over the years for refreshers on best practices. 

Other helpful resources are Bright Talk and Content Wrangler (Scott Abel). They offer videos and blogs focused on taxonomy talks, generative AI and how it intersects with taxonomy in terms that are easy to understand, even if you are not an AI expert. 

Be comfortable with communities through Slack. I met Laura Rodriguez through a LinkedIn Community User Group and it led to an opportunity to chat, meet and talk through how people are using the different software tools. These are ways to build closer networks. For some people using LinkedIn can be daunting. The idea of putting yourself out there is not always comfortable or natural for all of us. But these smaller communities of practice allow you to connect and contact thought leaders in the sector. So don’t be afraid to start asking questions and finding answers. 

The first time I presented at Taxonomy Boot Camp was the first time I attended. Funding to attend wasn’t guaranteed so I submitted a talk and it was accepted. Take advantage of these opportunities. 

What do you think is around the corner for taxonomy and KM? What do we need to consider in the future? 

BG: One area emerging is the major problem of running out of training data. Running out of context. For instance, ChatGPT or other GenAI technologies are able to access content from everywhere from Reddit, from open-source content, and from content that isn’t open source. At some point there isn't going to be enough. As we continue to need more content for GenAI to provide readable human language answers, there isn't going to be enough content to continue that same trajectory of growth. 

One response is to use GenAI to generate context. But there's an obvious problem with that approach - it's a bit like copying the same piece of paper repeatedly. Or sending a fax again, and again - making a copy, then a copy of the copy, etc. Even if it’s just a simple picture or simple text, by the end of multiple uses it's going to be smudged, messy and unrecognizable. Basically, having GenAI to create training data will just result in this race being at the bottom. I've seen this referred to as Habsburg AI, it’s not my term but definitely something to consider and look for solutions to protect against. 

Synaptica Insights is our popular series of use cases sharing stories, news, and learning from our customers, partners, influencers, and colleagues. You can review the full list of Insight interviews online.