In Episode 53 of Razorleaf’s Stay Sharp podcast, Character of the Data, co-hosts Jen Ferello and Jonathan Scott follow up on a topic that nearly always arises: data. Since data always comes up, the duo digs into why that is. Their first conclusion is that all data is definitely not created equal, and it’s definitely not all the same. From there, they discuss the differences between different types of data and they connect industry buzzwords with the real issues people face as they build out digital threads and start to develop product digital twins.
Why All Data Isn’t the Same
A foundational point that Jen makes to start the conversation is that a digital thread or digital twin is built with data that lives in more than one system, and different systems format, classify, and use data in different ways. The challenge isn’t about moving data from one system into another, in which case you could do a conversion process and be done. The issue is that the data required for the digital thread or twin must keep living in the other systems and doing its job there, while making sense in a digital thread with data from one, two, five, or even 10 other systems, Jonathan says.
Adding to that complication is the fact that data is also dynamic, not static. “The problem of how you connect it would be simple if it didn’t change,” Jonathan adds. “But when you think about using it in a digital thread or a digital twin, you have to decide what snapshot you want to use.” That need is further complicated by the fact that different systems manage data differently.
Data Differences = Data Character
Jen and Jonathan agree that successfully integrating systems—in order to create a digital thread or digital twin—requires understanding the character of data in each system, “not only understanding the underlying data itself, but also understanding how those systems represent that data in their data models. It’s understanding a lot of different pieces,” Jen says. Jonathan gives an example of an overloaded bill of materials (BOM) for products in a product line. Each product will likely use the same base items in the BOM, but with some variation in parts. The challenge is that if that BOM is shared to another system by people who don’t understand everything it’s representing, they might not understand that a single product doesn’t need all part variations.
Jen’s analogy summarizes the situation: “I’m going to make three different kinds of cookies—they all have the same basic ingredients [with variations]. If I came home, and my husband puts all of it in one bowl,” she adds. “That’s not going to be good.”
Jonathan points out that the character of the data is “going to dictate what you do with it, what you can do with it, what you should do with it, that sort of thing.” This leads the duo to a discussion of terminology, or how to characterize different types of data. Jonathan explains two basic ways to describe data in a simplified way as well as academic language. The first is traits or taxonomy. Data usually has observable traits, such as its size, its format, how it looks. Taxonomy, the science of classification, is simply how you sort items into categories based on traits.
The second way to describe data concerns its behavior or ontology. Behavior includes information like how data gets revised, how it matures, how it changes through time, how it’s released, and more. “But different revisions, different iterations of those have different meanings behind them,” Jonathan says. Ontology does some of the same things as taxonomy, in being about categorization, but an ontology also describes relationships, or how things relate to each other and work together. Referencing the BOM interpretation he described earlier, he says, “An ontology might go further and explain things, like it can also have a relationship known as an alternate or substitute or repair part. But there are all these meanings and an ontology lets you capture that.”
Jonathan gives us the bottom line for working with data: “If you don’t understand the nature or the character of your data, then you’re going to get it wrong. You’re going to apply the wrong rules, wrong logic when you move it over.”
Learn More About Data Character
The full podcast contains more discussion and details on a variety of topics, including the differences between graphical PMI and semantic PMI, why you need to be careful about how you handle data to keep it accurate, what to watch for when working with different sources of data, and more.
Check out the full conversation in Stay Sharp Episode 53: Character of the Data, and join us each week for a new podcast.
Follow the Razorleaf Podcast, Stay Sharp in Digital Engineering on:
