Knowledge Graphs 101: Common Challenges With Adoption
By Monika Mueller
The buzz about knowledge graphs is getting louder. Inna Tokarev Sela, CEO and founder of illumex, has declared that 2023 is the Era of Knowledge Graphs. And according to Gartner, graph technologies will be used in 80% of data and analytics innovations by 2025, up from 10% in 2021. Industries like big tech, the life sciences, healthcare and cybersecurity have found valuable use cases for knowledge graphs that have spurred adoption.
Yet the majority of industries are still lagging behind. Why?
We sat down with Michelle Yi, Senior Director of Applied Artificial Intelligence at RelationalAI, to get her take on the roadblocks that are hindering knowledge graph adoption. Michelle has a long history working with knowledge graphs, dating back to her time at IBM in the early 2000s when knowledge graphs were known primarily as a semantic technology.
Michelle points out challenges on both the business and tech sides of the equation, from inconsistencies to technical ambiguities. Most of the issues can be boiled down to: who, what, why and how?
Awareness is Lagging on the Business Side
Google may have put knowledge graphs on the map a decade ago by coining the term in the groundbreaking “things not strings” article, but the majority of organizations aren’t exactly sure what a knowledge graph is, or what it can do for them.
What is a Knowledge Graph?
According to Michelle, a knowledge graph is “a structure or way of organizing information based on different concepts or the relationships between them.” Yet depending on who you ask, or where you search, you’re likely to find countless different definitions. Needless to say, there’s a general lack of consensus throughout the industry on how to define what, exactly, a knowledge graph is. This inconsistency plays directly into a general lack of awareness.
Michelle also points out that many don’t realize how broad the definition of knowledge truly is in this context. While most of the conversation is focused on data, she points out that knowledge can include everything from relationships to business logic. “Data can be encoded,” she explains. “The code can be counted as knowledge. Like the formulas you put into Excel — that kind of business logic is knowledge.”
Why Use Knowledge Graphs?
Beyond the element of mystery surrounding what a knowledge graph does, many are also in the dark about the value they can offer their organization. While mainstream technology like Google search and Amazon’s recommendation engine have provided accessible examples of knowledge graphs in action, most organizations still don’t have a solid grasp on how the technology can benefit them directly.
On the business side, many struggle to articulate the value of a knowledge graph for what it is. And with a limited amount of knowledge graph use cases to reference, finding relevant value in knowledge graphs is a big hurdle. Those that do dip a toe into the knowledge graph waters often run into challenges with the data, the infrastructure, or finding engineers with the right skill set.
Ambiguous Tech Contributes to a Lack of Expertise
According to Michelle, knowledge graph technology is still somewhat ambiguous, with no consistent tech stack, which makes it difficult to understand – much less master. Even when an organization is on board, the question becomes: How can I implement knowledge graph technology?
Organizations that are ready to adopt knowledge graphs often face challenges integrating the technology, and fitting it into their ecosystem. And they may be hard pressed to find the engineers they need to do the job. As Michelle points out, “There are different tech stacks, and different paradigms. If there’s no consistent tech stack, how are you going to train people?”
Overcoming the Hurdles that Hinder Knowledge Graph Adoption
In order for knowledge graph adoption to accelerate, organizations must understand what knowledge graphs can do for them, see a clear path to implementation, and find the engineers they need to support the technology. This, of course, begins with awareness.
“I think we need to do a lot more work socializing what they are, and defining the technology consistently,” says Michelle. “And we need more applications that drive business value and help impact people’s lives.”
Michelle points to conferences and evangelism as two ways to increase awareness. She’d like to see more vendors and industry players that know about knowledge graphs share how they’re being used with a larger community. And while there are a handful of knowledge graph conferences, Michelle thinks they could be more commercialized, stressing the importance of teaching business people to “speak knowledge graphs.”
On the tech side, Michelle is encouraged to see so many new players come into the arena on the infrastructure level and provide more database technologies. She expects the technology to continue to become more easy to use, scalable, performant and accessible. And hopes that 2023 will bring more business applications, and more sharing about how different industries are using knowledge graphs.
Hear more from Michelle in a Leadership Series video interview we did on Bias in the World of Artificial Intelligence.