Alexis
Hi, everyone. Welcome to another episode of Data Dave Dives Deeper. I am Alexis, your host of this podcast and hopefully your favorite podcast, Talk Tech with Data Dave. I’m here today with Data Dave and Erik Lee to talk about some words that I don’t even understand. So, I’m really excited about that.
But before we get started, I love to remind our listeners, please submit your question to Data Dave by asking Data Dave talktech@d3clarity.com via email or hopping on the D3Clarity website to send a question that would be awesome. You can also volunteer to join us on one of these Data Dave Dives Deeper by emailing us that same email address, talktech@d3clarity.com.
Dave, how are you today?
Data Dave
I’m very well, Alexis. How are you? I’m excited. Good to have Erik with us today.
Alexis
I was just thinking about one of the previous episodes that you and I recorded, Dave. You mentioned the word taxonomy and you gave me, like, a quick definition, and then you were like, that’s for another podcast. And in the back of my mind I was like, it’s for the podcast with Erik Lee. I’m so excited.
Data Dave
That’s right, it is. So, Erik and I met at DGIQ, as you well know, and Erik is a professional taxonomist and information architect. I thought it would be better having mentioned taxonomy and mentioned talking about classification and other things in terms of machine learning and technology today quite frequently. So, I thought it was probably wise for us to have an expert to join us here to talk about taxonomies and on from that ontologies, and we’ll get into both of those.
So, Erik, welcome. Glad to have you here. Very excited to have you here. And you are a taxonomist and information architect. Let me ask you two questions right off the bat. One is, how does somebody become a taxonomist and information architect? And then what does that mean?
Erik Lee
All right, awesome. Well, I’m very glad to be here myself. Thank you so much. Usually there’s two routes by which somebody comes into the field. Usually, it’s either coming from a library background where you kind of learn about classification and organization. And a lot of classes that teach information architecture work or related work are in information schools where there are also library classes. So, usually they tend to get bundled together or at least have a little bit of overlap, or sometimes others will come in from the tech side, so they may be doing like it work or database administration and start learning more about how things are organized and become interested in that. For me, it was the former.
Data Dave
Okay, so tell us a little bit about that story, where did you start?
Erik Lee
I was always kind of a nerdy guy. So, all growing up, I’m like, “I’m going to work with books.” At some point in time, once I found out you could get a degree in something called library science, I was like, “Well, that’s me. Count me in.” Like, which is funny because I got my bachelor’s in film studies and then I still wanted to do library science. And eventually I decided, “You know what, I’m going to go for it. I’m going to go for my master’s in library science.” And then while I was taking my library science courses, one of the courses I took was an introduction to Information Architecture.
So, we went through a couple of the foundational texts in the field and talked about it and I’m like, “Oh, this is interesting. I am interested in this.” And at the time I was working for a retailer doing customer service work, and I had an question and I’m like, “Oh, this is a taxonomy question. I know what that is.” So I decided to look. I’m like, “I wonder if our company has a taxonomy team.” And they did. And also, they were hiring.
Data Dave
Excellent.
Erik Lee
Yeah. So, I decided to try it out, and I enjoyed it. And I’ve been working in the taxonomy field ever since. That would be about… since late 2015.
Data Dave
So taxonomy has nothing to do with tax?
Erik Lee
No, it has nothing to do with income tax. It has nothing to do with stuffing animals.
Data Dave
That was where I was going to go next.
Erik Lee
Oh, oh, somebody’s got jokes.
Alexis
I like that.
Erik Lee
Somebody actually asked me, “You’re leaving the company?” I’m like, “No, I’m just going to a new team.” And he said, “But taxidermy.” I’m like, “Oh, okay, okay.”
Alexis
So Dave, I think you’re going to ask this question too, but I’m going to ask it. You guys keep saying taxonomy. And Erik, I’ve heard you drop the word classification. One of my best friends has a master’s degree in library science, so I know a little bit about it. That’s what Dave said as far as a definition for taxonomy. But I need more information. Help me understand this word a little bit better. Hit me with some context.
Erik Lee
All right, so taxonomy, it’s a term borrowed from biology. So, you remember your kingdom of life. You have your five kingdoms, your animal, plant…My brain. So I don’t remember them.
Alexis
Animal, vegetable, mineral, that’s what I got.
Erik Lee
Animal, vegetable. I think like fungus something. A lot of single-celled organisms.
So biological taxonomy is like, “Where do these living things go in this kingdom of life? Like, how is it arranged? And what are the relationships?” Taxonomy, as far as knowledge management and information is kind of the same thing. Technically, a taxonomy is a hierarchical list of terms that have broader, narrower term relationships. So, like parent-child, so like car might have narrower relationships like coupe, sedan.
A lot of times in practice, taxonomy will also have relationships like related terms like dogs and dog food can be similar somewhat, but they’re not like a strict parent, child, for instance. And synonyms. So like car and automobile, these are the same thing. But if you’re, for instance, looking in a search engine, how does the search engine know it’s the same thing? Well, it has some metadata saying, “Hey, if somebody types in automobile, it’s the same thing as car.” So, whichever one you type in, it would get both. And that’s kind of how it relates to classification.
So, you might have a piece of data, a document, an object, and depending on your context and what you’re trying to do with it, you can use these taxonomies, these basically existing lists of values, to add the correct metadata to that, whatever you have it. That document, for instance, in a way where it’s already set up, so it has these synonyms, it has these related terms and stuff. And you don’t have to like add all this additional metadata every time you want to add an object.
Alexis
So, I’m going to dumb that way down. Tell me if I’m wrong. It’s the way we classify data and the way we explain data to a system.
Erik Lee
Yeah, it’s the terms we use to do that and how those terms relate to each other. So, you might have content type, like report, and you might say,”Well, a book report is a type of report.” So, you can have it set up as like, “This is a book report. Oh, I know it’s a report because this relationship between the terms already exists and it’s got it.” You can also use it to authorize lists too. So it’s not just somebody randomly typing in, like, “What is this thing?” It gives them classifiers, a guide as to what’s the appropriate metadata to add here.
Data Dave
It’s usually a fairly strict structure and set of terms. So, you alluded to the biological taxonomy earlier. Just for fun, I just looked it up just to put us on the right track, above the word species. So above the word species. So we haven’t gone, realistically, haven’t gone that deep. There are eight parent classifications from life, domain, kingdom, phylum, class, order, family, genus, and then species. This is the structure that starts to say, “I’ve got a life, I’ve got a crustacean, I’ve got a mammal.” And different things that fall into a very strict hierarchy that allows us to classify things, but with strict definitions.
Now, in our modern language, it’s correct to say that we don’t use our language as strictly as our language is necessarily defined. So, the idea of synonyms and ambiguity comes in, which is what I would imagine is a lot of, you know, when you say a company has a taxonomy department.
Let me ask this. Why would a company need a taxonomy department? That’s a little abstract.
Erik Lee
Let’s imagine that you’re like a retail company, for instance. You have a customer. You ask somebody from legal, “What’s a customer?” You ask somebody from marketing, “What’s a customer?” You ask somebody from logistics, shipping, “What’s a customer?” You’re going to get three different views of this concept of the customer. And there have, like, different sets of needs of what they need to know about the customer and how they understand it.
But when it comes time to do analytics, like, “Oh, we want to look at our marketing spend on customers.” Or legal, “Oh, we need to create a contract for a customer.” All this stuff they need to have across the board. Like, if you’re going to sell something to somebody, you have to know legally, can we sell it to them? Can we sell it to them in their location? Who is this person? If we want them to buy something, what target demographic are they in? What kind of work do they do all this?
So this shared understanding of this person, this customer, you need a way of saying all these little bits and pieces are all related to this one concept of the customer. So, here’s what a customer is. Here’s the metadata related to it. So that way system A and system B can say, “Oh, customer. Okay, it’s this thing, and this is the metadata we get,” and then we can put it together into an actual useful piece of information that you can analyze, that you can get information from, that you can make decisions on.
Data Dave
You just went into one of my favorite examples as well. That’s why Alexis was laughing. The definition of customer is something that’s near and dear to my heart. But it goes further than that, because the concept of a customer is a little bit of an abstract concept. But if I was to take a retailer and start to categorize the products that they sell, then that is an area where they don’t want any kind of ambiguity. It’s an area where this toaster is an appliance, where you kind of want to know that a toaster is an appliance.
Erik Lee
Right? And depending on what kind of retailer? Like, if you’re just a kitchen or bath retailer. Yes. It could be an appliance. If you’re doing whole home, you might need to say, “Hey, this is a kitchen appliance.”
Data Dave
Right.
Erik Lee
Because appliance, yeah, it could be all sorts of other things that you sell.
Data Dave
Right. So that’s interesting. Is there one taxonomy? Does an organization have one taxonomy? How does that work?
Erik Lee
Usually they’ll have multiple taxonomies that can be related in some ways. Because it would be a huge, giant mess to have a list of everything.
And depending on the context, not every system needs to know everything about everything. So. you might have taxonomies related to publishing your content. So, you know, content type outlet, author, creator. You might also have for marketing, target industry, job level, job type, things like that. For product. Product type. Even going into things like product capabilities or some kind of standards, you can have it. So, yeah, multiple taxonomies, one taxonomy would be so hard to manage. And then any time you’d save by having that list, you’d more than make up by trying to like, disentangle all of this every time you wanted to use it.
Data Dave
Because you might even have terms that cross taxonomies because terms are ambiguous and they might have different meanings to paring on where they are within the taxonomy.
Erik Lee
Exactly.
Data Dave
Would I be correct in saying that an address is a taxonomy where you say country has states, has towns or zip codes. Where they are constructed within that taxonomical view to give you a relationship?
Erik Lee
I would say it’s more of a collection of taxonomies put together in a particular way. You’d have like, your list of states, your list of zip codes, your list of street addresses, an optional thing for like a unit or building or other stuff. And then in the context, an address in Japan would look a lot different than an address in the US. So depending on what you’re using the address for, you might have different taxonomies for like, different country postal codes.
Data Dave
Yep. Okay.
Alexis
When we talked about taxonomy before, we brought up the conversation when we were talking about metadata and Erik, I’ve heard you say metadata a bunch of times. Based on what Dave and I talked about before about metadata and this conversation, it sounds like those two things are really, really similar, if not similar, very, very closely intertwined. Can you explain the difference or how they work together to me, just so I can understand that concept a little bit more?
Erik Lee
Basically, it’s in the way it’s used. So, the taxonomy is the list of terms. So, let’s say you have a content type taxonomy for publishing. And you have this list of different types of content. So, you might have a report, an email, a white paper.
Those lists can be used as metadata values. So, if someone’s going through and classifying it, they could use that term to add that as a metadata value. So, you know, white paper, this is a white paper content type, or author this person, and add that on and they have it as controlled values, basically. And then also a lot of taxonomy systems can have additional attributes that are in themselves metadata about the term. So, things like preferred label, for instance, or these synonym relationships. Sometimes the taxonomy system will do translations. That way it’ll show like term name and then like the alternate language term name say in Japanese. It will connect so you don’t have to hunt back through it. You can say, “Oh, I want this term name, but I want the Japanese version of it.” That metadata already exists. I can just add it on to my document or whatever
Alexis
Okay, so metadata being that which it is, and taxonomy specifically referring to the words that we’re using.
Erik Lee
Exactly.
Alexis
Got it.
Erik Lee
Yeah.
Alexis
It’s clicking now. That’s what I wanted to know. Thank you.
Data Dave
So, I’m going to go in a slightly different direction for a minute, which is you describe yourself as a taxonomist and an information architect. And we talked about metadata in terms of data when we talked about it before. You’re talking about metadata in terms of information and information architect. Can you, just for our audience, define a little bit the difference between data and information from your perspective as an information architect rather than data architect?
Erik Lee
So for me, data is basically the raw material that you have.
Data Dave
Yep.
Erik Lee
You might have like a sensor that provides a list of numbers for you.
Data Dave
Right.
Erik Lee
The information part of it is making that into something usable. “What are these numbers that I’m getting? And then how do I connect that to a system that can use those numbers to do something for me?”
Data Dave
It becomes the context. So you’ve added a level of context to the data that you’re receiving. You might be receiving a raw feed of numbers from a sensor, but you don’t know until you add the metadata to it that this data is actually a time series of temperatures.
Erik Lee
Right, Right, exactly.
Data Dave
It’s only when it becomes a series of temperatures that it becomes information that then you could decide to take action on. “My temperature is going up, therefore my information is then telling me that.” I’ve got knowledge that says I need to do something with this device because it’s getting too hot.
Erik Lee
Yep. And also the definition. So, what is temperature in this context.
Data Dave
Exactly what is too hot? What are those thresholds? Because it’s this sensor is on this piece of equipment, therefore it’s telling me something is wrong by getting too hot. And what is too hot for this piece of equipment versus that piece of equipment. That’s when this data, the raw data, your raw commodity, if you like, becomes information.
And so you’re in the information world, not that data world.
Erik Lee
Exactly. Yeah. They get the data and they’re like, “All right, we need a shared understanding of this and a definition of what this is so people can understand.”
Like, okay, you’re getting a data feed, a shared information. This is related to temperature. Here’s what temperature is. And then the systems are like, “We need to know stuff about temperature.” We’ll have that in places like, “Oh, this is marked as temperature. This is marked as temperature.” Like this is the one concept. Yeah, right.
Data Dave
And this is the unit that it’s in. It’s in whatever. And various other things.
So, you talked about coming to this from either a library science structure, which is what you did, or from the data perspective, the IT perspective, which is an interesting view. So, that’s kind of one coming up from data in a “What does this mean?” And the other one coming down from the knowledge view that says, “How do we pull this knowledge together?” Is that exactly a fair thought?
Erik Lee
I would say that. And it’s good when you have folks working together from those different backgrounds too, because they do bring that kind of bottom ups and top down view and it makes it a lot easier.
A lot of us taxonomists aren’t super technical folk, but it’s good to at least get a base understanding of kind of what the tech is and kind of. Right, yeah. What to expect from it.
Data Dave
But it’s nice as a technologist, I come from the technology background into data, and sometimes it’s nice to get the abstract view as well, from the higher level people who say, “So, what do you mean by that?” We tend to get pigeonholed into own cycles.
Erik Lee
It’s just nice to understand like what the stuff you’re doing is being used for. Like, “Oh, hey, I’m part of that.”
Data Dave
Yeah, exactly, exactly.
Alexis
What I’m hearing you guys say is that taxonomy- although it is obviously a word that we use in the data world- very often is part of all of our lives and we maybe don’t even realize it. Is that right?
Erik Lee
Yeah. Classifying it.
Alexis
We’re talking about libraries here and I’m thinking about a clothing store and how maybe Amazon suggests what I buy is based on the taxonomy of the stuff that I’ve been picking before.
Erik Lee
The retailer that I worked for, the online retailer, sold shoes and apparel. So, we even had to do things like, “What is a sandal?”, for instance. What are the types of shoe that there are so you can find it?
Data Dave
That’s an interesting question, isn’t it? That’s actually a fabulous question, because if I search for the word sandal, well, that’s an ambiguous term, but you want to present anything that I might buy. And my definition of sandal might be different to Alexis’s definition of sandal.
Erik Lee
Exactly. Yeah. So even in sandal, what types of sandals are there? Do you want a fisherman sandal, or do you want a healed sandal?
Data Dave
Exactly. Okay, so that’s now into the definition of terms that we see in everyday life and every structure. So, as we look at this, it becomes crucial, really, from an information point of view- and I’m using my words somewhat carefully- from an information point of view to really know what we’re talking about.
Erik Lee
Yeah. And then with multiple taxonomies, too, for, like, a product, you can have these multiple points of view that the customer can then be used as filters.
For instance, that metadata can be turned into filters. So, you might have a healed sandal. For instance, maybe you’re looking for a particular heel height, or maybe you’re looking for a particular color or, you know, a particular material. For instance, those different ways of describing that can be used with taxonomies. The taxonomy can be turned into metadata to classify that product and then used as filters on the user side to say, like, “All right, they want leather. Okay, here’s the stuff that’s classified as being made of leather.”
Alexis
Yeah. So, a lot more every day than I would have ever guessed.
Erik Lee
Yeah. And when you have that kind of multiple together, that’s called a faceted structure. So, you would have your materials facet, your heel height facet, that set of filters for a particular part or property of the item.
Alexis
I like it. Once we started talking about taxonomy in the terms of shoes, things made a lot more sense to me. I’m all in it now.
So, thank you guys for putting it in Alexis’s terms for me. I really appreciate it. I feel like that was, like, a great place to end the podcast in a very simple definition that probably most people can understand. So, I really appreciate you bringing it down to my level for me, guys, or at least taking it to the shoe level for me.
Data Dave
Excellent.
Erik Lee
Yeah. That’s what I’m here for.
Alexis
Thank you so much for being with us, Erik, today. This has been lovely.
Data Dave
Yes, thank you, Erik. We’ve still got a couple of things that we should touch on, but we’ll have to wait for another time because we didn’t actually go into ontology, and we didn’t actually go into knowledge graphs either, which is where I was kind of trying to get to. So, I think we might have to come back to those Alexis at some point.
Alexis
Ontology for sure. Because we talked about ontology for a hot second and I was like, “Cancer?”
Erik Lee
Yeah.
Alexis
No, a T, not C. On-TOL-o-gy. So yes, I love that idea.
Erik Lee
It’s not taxidermy and it’s not oncology. So.
But yeah, yeah, I would love to talk about it with you.
Data Dave
Excellent.
Alexis
Well, perfect. Thank you both so much again for being with us today. For our listeners out there, if you are interested in joining us for one of these episodes of Data Dave Dives Deeper, please reach out to us at talktech@d3clarity.com.
Erik Lee
Thanks.
