Getting Started with Knowledge Graphs and Machine Learning: Part 2 Q&A with Sebastian Schmidt, CEO of metaphacts

Video produced by Steve Nathans-Kelly

Getting information to the people in an organization who need it when they need it—continues to be a widespread challenge.

In part 1 of this KMWorld Drill Down Video Interview with Sebastian Schmidt, CEO of metaphacts, he spoke about the promise of knowledge empowerment for more users and the challenges that have prevented that from happening. Here, Schmidt outlines the approach that metaphacts takes with customers to help make knowledge democratization a reality and explains the role that knowledge graphs play in the process.

Joyce Wells: You mentioned a four-step approach. Can you tell us more about that?

Sebastian Schmidt: Step 1 focuses on identifying the specific information need; that could be a specific source dataset to start out with. The idea behind that is to reduce the complexity of the knowledge democratization process by not trying to integrate all data sources or answer all end-user questions at once. We are starting off from something that's easy to capture. And then, because we are building on the knowledge graph, which allows us to easily expand our use case and to extend the semantic model in which we describe and reuse data, we can easily expand from this one use case to all others and, through that, achieve company-wide adoption.

In the second step, we then model the meaning of data. As I said earlier, this is a key step. This requires the involvement of the domain experts and business users throughout the enterprise—those users who are creating, using and reusing this domain-specific information. And it's crucial to really capture the semantic description of data together with the data itself and also defining and tying in controlled vocabularies or common terms to be used with your data. We commonly see customers reuse public and published semantic models and vocabularies to bootstrap this process. That's possible and works great because with knowledge graphs you're building on an open standard—a W3C-recognized standard which means that all of this can be easily shared with others.

In the past, this process of semantic knowledge modeling has also been very much a headache due to missing tooling, because you need a tool where you can just allow your business users and domain experts to participate in that process without them being full-fledged ontology engineers. And that's what we have enabled in our product metaphactory. And I even wrote a blog post about this specific topic.