Bob Kasenchak joins the Synaptica team as the Senior Manager for Client Solutions. For this Insights interview, we found out more about Bob’s experiences, his views on the future of taxonomy and a hint of his new role.
After early career stops as a philosopher, music theorist, and wine geek, Bob has spent the last eight years in the information industry developing taxonomy projects; research interests include linked data, graph databases, and all things categorization. Bob lives in Albuquerque with his wife and an elderly cat.
Tell us about you, your experience and interests.
BK: I grew up in Iowa and went to a liberal arts college in Santa Fe, New Mexico called St. John’s College. Their education curriculum focuses on books, classics with small class sizes. My degree is in liberal arts with strong philosophical overtones. The education involved lots of integral critical thinking and discussion. Following this, I stayed a few more years in Santa Fe during my early 20s. My first work experience with information science — although, at the time, I didn’t realize it — was when I joined a classical music magazine. The role involved cataloging classical music for a print publication. We collated different recordings by the conductor, orchestra, and performers to help people find and purchase the recording they wanted. This was in the 90s. The publication doesn’t exist anymore but exposed me to that first information-based experience; I also met my future wife as we worked in the same office.
Following this, I decided to return to education. I took a master’s degree in music theory at the New England Conservatory based in Boston. This meant I had a lot of catching up to do as I didn’t have years of undergraduate music or sight reading training, but I wasn’t looking to be a performer. I wanted to focus on the theory of music and write about it. Again, this was a wonderful small institution, and I was lucky enough to be taught by an inspirational professor who was fascinated by the intersection between math and music.
Studying music theory meant I was based in Boston for four years. I wrote a thesis on the inter-war avant-garde composers; Ives, Cage. My work explored experimental artists and delved deeper into my favorite composer, Ruth Crawford Seeger. Seeger was a fascinating subject; she had one foot in the radical political world, another in the classical music world and then during the 1930s became involved in folk music. Her legacy still exists today through her children, including her step-son Pete Seeger.
During 2003 I moved to Austin, Texas again for education, this time undertaking a Ph.D. After nearly six years in academia, the culture had changed. I was working and studying full-time plus teaching part-time. After some soul searching I decided to walk away and divert to another direction. It’s one of the reasons that today I like to describe myself as a recovering academic.
How did you move into the world of Taxonomy and Knowledge Management?
BK: I returned to New Mexico and initially worked as a freelance writer and editor. Purely by accident, I ended up working in taxonomy. I saw an intriguing ad on craigslist and responded. The role involved working with databases, building a large vocabulary indexing 900,000 scientific articles. I spent 7 years getting an in-depth education in information science.
The work was enjoyable and much of it was linked to academic and scientific publishing. My first project involved thousands of preferred terms. The work was demanding and involved a great deal of specialized content which was used by the sector researchers and experts. They know their subject matter and expect effective search. Over time we have seen fresh generations of researchers emerge. They want improved search. There is a need to see value from publishers and academic bodies. It’s crucial going forward that as a sector these organizations invest in their information infrastructure, taxonomies, as well as search and interface.
Many of us who work in this sector embrace words, grammar, punctuation, and language. However, “word nerds” can be shy about talking in public. I was asked to give a talk at a conference and I really enjoyed it. From then on, I was working less on taxonomy projects and working closely with customers. Building relationships, designing projects with clients, listening to their needs and solving their challenges. It’s great fun and personally rewarding.
Can you tell us about your involvement with events?
BK: I enjoy going to London for the Taxonomy Boot Camp in the UK. It’s a great place to meet established practitioners, novices, and people working in the space in Europe in industry and other organizations.
I also like Taxonomy Boot Camp’s USA-based event which is part of a larger KMWorld conference and has a very different feel. It’s co-located with lots of other events and draws a larger crowd. You can definitely find all of the major taxonomy vendors and consultants there every year.
It’s great that Synaptica is a premium sponsor of both events. This year I will be joining the rest of the Synaptica team who regularly speak and provide workshops at these events. It’s an invaluable opportunity to reconnect with our customers and colleagues as well as meet new people.
Perhaps my favorite Conference is the Information Architecture Conference (formerly IA Summit). This audience is a mix of information architects which includes UX experts, web designers, and strategists as well as metadata people and taxonomists. The conversations and tones are diverse but related. Taxonomy work is growing in this space and taxonomists bring expertise to this community. It’s an area to watch out for. I have found that when we talk about taxonomy it’s (ironically) not always the same thing. People who work in content strategy are much more into deep structures. They are concerned about what is displayable. Others might view taxonomy as a tool for web navigation.
As a way to address this, I often talk about the different kinds of taxonomies and their features. Explaining the differences between web navigation, e-commerce, and products or the larger information retrieval structures is interesting and relevant to the audience. They are similar but there are variables, and you need the right strategy to resolve each method and understand their properties.
Was there a specific reason you wanted to join the Synaptica team?
BK: I have met Dave Clarke, Jim Sweeney and the team at industry events for several years. The team is all very approachable. There is only a handful of companies who develop taxonomy tools and you need to understand what they do, don’t do and the direction each company is heading in. I feel strongly that this industry and the wider business world are moving in the direction of graph databases and ontologies. As a sector, we need to embrace this. There has been a recent Twitter debate about this issue and what it means. We need to be able to define relationships, not just build models, and be able to name the edges of graphs. It was very clear Synaptica had decided this was the direction the industry was going in and that they want to be not just part of it but leading. I believe that is smart.
Can you tell us more about your new role with Synaptica?
BK: I will be leading client solutions and business development. Client solutions starts with a deep understanding of a customer’s needs and business challenges, and then involves pulling together various complementary products and services to form an integrated solution. This isn’t limited to Synaptica products and services; a solution may also embrace consulting services and technologies provided by our partners. Synaptica already has a lot of really cool tools. The newest one, called Graphite, is a game-changer that will make it incredibly easy for people to build ontology-based knowledge organization systems and knowledge graphs, as well as leverage the enormously valuable resources of the Linked Open Data cloud. Another important aspect of my role is to ensure that what we learn about client needs and challenges feeds straight back into software enhancements and new product development.
It’s important to cultivate and nurture valuable relationships, not just with our clients but with our partners and consultants in the industry. We need to be clear about our enterprise solutions and confident about our business ethos.
Where do you see Linked Data fitting into the value proposition for enterprise taxonomies?
BK: It depends on what they are trying to do. Linked Data is extraordinarily important because you can’t have one taxonomy that everyone can use. There is no taxonomy tool for everything, it is not practical for information retrieval and no one would ever agree on it. Instead, we have an evolving “hub and spoke model” where there are these Linked Data repositories. DBpedia and Wikidata are the big general ones, but there are small specialized ones developing as well.
It’s powerful. It allows you to tie together and aggregate content that might be disparate or held in different places. Taking a half-hearted approach with data-links is not useful, and working with large vocabularies is a challenge. How would this work at the enterprise level? Your business may be in manufacturing with a taxonomy explaining your products, their parts, suppliers and different elements they need. All of this valuable information needs to be organized. You will want to link out to a variety of sources, some may be external, some internal. It makes sense to build it right but this can be expensive and time-consuming. You need a specific use case describing what value it’s going to add to you and your business. We see it in the publishing sector with ideas, concepts linked to an authoritative source directing you to read more about the topic or another collection.
Linked Data is a big external set of datasets you can use to point terms and relationships to equivalent Linked Data sources and describe connections. We need to realize the potential of what is achievable with Linked Data. There are tools that are going to emerge that we don’t even know about yet.
AI and machine learning are the hot topics, but others argue taxonomy plus rules-based auto-categorization systems are more appropriate. What’s your view?
BK: I am on the rules-based categorization side for several reasons:
- Any machine-learning-based document classification system is actually a rules-based system — with machine-generated rules based on the machine learning output. But they are almost always not human-readable or -editable. You need to return to your vendor to change the rules if they are not working. Such systems generate black boxes that are proprietary, so you can’t edit the rule engine.
- Inferential systems are extraordinarily expensive to train. If you have 10,000 terms in your taxonomy you are talking about 50-100 documents you will have to tag by hand and tag correctly to feed an accurate training set to the inferencer. It’s much faster to write rules than to build a 100,000-document machine learning system.
- I think the rule-writing process is fun and interesting. It makes you learn about your taxonomy and your content. There are ways we can improve them and it’s an area I would be keen for us at Synaptica to develop.
Do you have any views on ontology versus taxonomy?
BK: Taxonomies are becoming more and more complex; they are edging towards ontologies. Some think they want an ontology but really need a taxonomy. Similarly many want to build an ontology because they want data-driven semantics, they want to model the relationships. For example, dog food cannot be categorized as a dog, it’s not an example of a dog – you need to be able to describe the nature of the relationship between dogs and dog food. You need to be able to model this in your knowledge organization system and be able to surface the right information.
If we want to build smart content, we can’t just have a cluster of taxonomies, we need to model the associations between concepts. I think there’s a new generation of semantic tools about to emerge, the ideas for which are just developing. We need to be able to have semantic structures to drive these tools — to allow them to connect to the semantic web and enrich content for both readers and in a way that’s machine-readable..
I guess it depends on your use case. There is no ontology without taxonomy — I could argue that a thesaurus is just a lightweight ontology — the two are intertwined. Certainly any taxonomy can be expressed in ontology languages. But you can’t model a supply chain with a taxonomy, and you probably wouldn’t want to use a supply-chain model ontology to classify documents for retrieval; they have different purposes.
Let me clarify this a bit. The “Ontology v. Taxonomy” debate comes up regularly, as do claims that taxonomy is dead. This is unhelpful and misses the point. Taxonomies and ontologies do fundamentally different but highly complementary things. Throw in the recent rise in interest in knowledge graphs and you have a potentially (and actually) confusing mess.
Let’s think about SKOS for a minute to try to clear up some common misunderstandings and frame the relationship between these things. SKOS is, essentially, an ontology for building and specifying taxonomies. Let me say that again. SKOS is an ontology for building taxonomies. The ontology specifies the logical structure (the classes of objects and relationships between them), and the taxonomy populates that structure with a vocabulary of semantically defined things. Together the ontology and the taxonomy form a Knowledge Organization System (KOS) that can be expressed as a knowledge graph.
Taxonomies, thesauri, name authorities, classification schemes and glossaries have always had logical structure, and they have always been human-readable. What happens when we combine them with an ontology is they also become machine-readable, actually it’s more than just readable: it’s machine-intelligible. This is why this is such a cool field to be working in. Ontologies working in combination with taxonomies are integral to voice search, AI, and other important new fields.
What makes a good taxonomist?
BK: Beyond the passion for words, language, and information a good taxonomist needs to be prepared to be open-minded and a generalist. You need to have a sense of how to organize and embrace the minutia. Very few of us work in one sector all our lives. A taxonomist needs to be prepared to build a taxonomy in areas where you have no prior knowledge or expertise. You can read up on the subject and industry to figure out where it fits. Be prepared to pick things up quickly, recognize important concepts, and be prepared to bend and flex. Flexible critical thinking is essential.
What are you looking forward to about joining the Synaptica team?
BK: Beyond embracing new challenges, discovering new tools and technology, and working with a team of smart people, what I am most excited about is learning how to leverage Synaptica’s software tools to help people solve problems, I think this is at the heart of what client solutions is all about.
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.