KGF22: Knowledge Graphs and The Not So Quiet Cognitive Revolution

Knowledge graphs seem to be the obvious solution for the complex world we live and do business in. Yet, knowledge graph projects are hard to manage not only because graph databases still aren’t the first thing enterprises consider when seeking a solution to their data problem, but also because the understanding about what a knowledge graph can and cannot do is still unclear.

Day 2, 3 and 4 of KGF22 brought clarity with presentations about specific use cases, challenges and shared best practices. With plenty of advice and practical steps towards conceptualizing a knowledge graph project, designing the solution and working towards a shift in the modeling perspective.

Below are some of the highlights and if you want to watch any of the sessions, you can register before the end of October to browse all the recorded talks and discussions.

Day 2: Amalgamating Knowledge and the Industry Uptake of Knowledge Graphs

Day 2 started with concrete examples of the ins-and-outs of specific industry approaches towards knowledge graph building. This second day was dedicated to talks covering use cases and best practices from the Industry domain, including manufacturing, aerospace, architecture, engineering and construction industries. We saw how knowledge graphs revolutionize these sectors and what the practical implications of designing knowledge graphs are.

Stories About Semantic Technologies, Automation And Beyond

Nikolay Krustev, Sales Director Manufacturing and Public Sector, talked about how knowledge graphs are bridging the gap between industries. Then, Vladimir Alexiev, Chief Data Architect at Ontotext, showcased the value of semantic technologies and a knowledge graph approach towards the energy sector data. He explained how Ontotext’s Transparency Energy KG (TEKG) project converted part of ENTSO-E electricity market transparency data to a semantic knowledge graph and complemented it with external data sources.

Knowledge Graphs As Instrumental for the Software and Automation Industry

We also had Chris Brockmann, CEO of eccenca, talking about process automation and its direct link to knowledge management and knowledge graphs. He shared his talk “Beyond Ro-Bots – Total Process Automation, An Edge Device Management Story” with Georg Geiger, Product Owner SW Entitlement Management and DevOPS Delivery Orchestration at Nokia.

Georg Geiger of Nokia (on the left) answering questions from Nikolay Krustev of Ontotext and Chris Brockmann of eccenca regarding the processes Nokia’s Knowledge Graph project went through.

Chris Brockman’s part, brought home the point about knowledge graphs being instrumental for the automation industry together with their much needed interplay with AI.

What is a curse for a human (to process that much information) can be a blessing for a machine (machine learning) - Chris Brockmann, eccenca Click To Tweet.

Then Georg Geider showed us how his company built a knowledge graph solution to support their digital software delivery processes. We saw that Nokia continues the line of connecting people by adding elements to the connection processes such as industries, automotives and software products. Georg outlined three main pain points of Nokia’s software business and explained how they’ve turned them into opportunities for more connectivity by the knowledge graph they worked on.

Knowledge Graphs and Wittgenstein Again

Having seen beyond the robot automation process, we then had a sneak peek into the future (and, for some industries, the now) of technical communication. In his presentation “Knowledge Graphs as Game Changer in Technical Communication”, Karsten Schrempp, Founder & Managing Director of Pantopix, talked about the application of knowledge graphs in manufacturing. He showed us how they can be used for creating a technical representation of products and for integrating the resulting information with content from technical documentation in operational and service processes.

This is where Wittgenstein came up again (Juan Sequeda had also brought the philosopher up on the first day of KGF22). Karsten reminded us of the philosopher’s early thinking about the world consisting of statements and sentences. It was, as Karsten framed it, as if the whole world was a knowledge graph.

The theoretical foundations were further backed with practical examples about technical documentation enhanced by a knowledge graph.

We also saw how a taxonomy combined with an ontology can be used to substantially improve “related content” experience.

Karsten Schrempp of Pantopix showcases the taxonomies and ontologies behind a solution that serves connected content to users.

Knowledge Graphs for Intelligent Data Flows Across Infrastructures

There were other engaging presentations from this day in the Industry vertical. In “Smart Flow of Information for the Transport Infrastructure”, Klas Westberg and Lars Wikström, Senior Consultants at our client Triona AB talked about their project. It deals with intelligent data flows from infrastructure projects to asset information systems at infrastructure owners using linked data technologies.

In “Next-Generation CPQ Tooling for Sales Enablement” Sebastian Schmidt, CEO at metaphacts, talked about an innovative solution that supports sales employees in configuring, pricing and quoting complex products tailored to meet specific customer requirements.

We also have to mention Jürgen Umbrich’s presentation about “Potentials & Advantages of Knowledge Graph Based Conversational AI in Manufacturing & Industry – Examples, Use Cases, Best Practices”. Jürgen is a Senior Data Scientist and Head of Data at Onlim.

And one more spotlight of the day was “Decentralized Semantic Data Management & Querying” presented by Jan Jelle Sijbesma (Conceptual Data Architect – Project Systems & Data department at Fluor), Jürgen Jakobitsch and Lutz Krüger from Semantic Web Company.

Knowledge Graph Use Cases

The day ended with a Panel Discussion about the “Top Knowledge Graph Use Cases in Industry” with insights from Nikolay Krustev, Vladimir Alexiev, Chris Brockmann, Lars Wikström and Karsten Schrempp.

Panelists from the “Top Knowledge Graph Use Cases in Industry” discussing industry standards and the way forward for knowledge graphs in various industry domains. Top (from left to right) Chris Brockmann, Vladimir Alexiev, Karsten Schrempp. Bottom (from left to right): Lars Wikström and Nikolay Krustev.

Brockman shared a passionate call to action about the knowledge graph value proposition and project initiation.

While small is fine for technical proof of concept, the thinking behind an enterprise knowledge graph should be big - Chris Brockmann Click To Tweet.

The real value, Chris continued, lies in their proven potential to save big bucks and add big value.

Register to browse all the recorded talks and discussions until the end of October!

Day 3: Knowledge Graphs as Connective Tissue and a Renewable Source of Insight

The third day of the KGF22 was streamed live from Basel, Switzerland, and it was dedicated to knowledge graph applications in the Healthcare & Life Sciences domain.

We had Novartis, IQVIa, Roche, The Swiss Personalized Health Network (SPHN) and more. They presented cutting edge solutions in the medical domain, highlighting that these solutions might well fit into other domains. Among the main topics were FAIR data, AI, graph analytics techniques and Linked Data approaches.

Knowledge Graphs and the Power of the Network for Propagating Messages

Zeshan Ghory, Product director AI and graph analytics at IQVIA, presented how they constructed a knowledge graph representing a network of relationships based on a combination of public and proprietary data sources. Their aim was to identify the most influential people around a given subject and further propagate messages over their network.

Zeshan Ghory, Product director AI and graph analytics at IQVIA, explains how the company made a graph from connecting public and proprietary data sources to identify KOLs

Zeshan started with a quick recap on the traditional methods of identifying key opinion leaders (KOL) in a given domain, which are labor intensive and slow. Then he led us to how IQVIA got to exploring the idea of redefining the search for KOL by connecting public and proprietary data sources and loading them in a knowledge graph.

Content Delivery Through A Knowledge Graph

Gunjan Aggarwal, Digital Data Products and Martech Strategy at Novartis, also demonstrated the building of a graph. This time it was constructed with a view to serving customers better with the right content at the right place.

The interesting perspective here was that this construction started with a deconstruction. Gunjan deconstructed the data deluge by identifying the sources of content and data that might potentially become part of a knowledge graph aiming to enhance customer experience.

The Martech data that needs to be integrated for excellent customer service and efficient marketing communication

Ontotext’s LinkedLifeData Inventory for FAIR Data

We also heard from Ontotext’s Todor Primov and Ilian Uzunov about how LinkedLifeData inventory is an accelerator for building comprehensive FAIR knowledge graphs. They explained that leveraging highly interlinked information across various data sources can make knowledge graph construction times faster and more efficient.

Todor Primov in the talk “Ontotext LinkedLifeData inventory – Accelerator for Building Comprehensive FAIR Knowledge Graphs”

Worth highlighting is Todor’s observation that the approach used in Ontotext’s LinkedLifeData Inventory is applicable for every single business domain, given there are industry acknowledged reference data sources to apply the normalization. Todor gave as an example from the Pharma industry, where even before the semantic boom, there were substantial efforts and investments in defining stable identifiers, which helps a lot, as most of the objects are really well structured and referenced.

Knowledge Graphs, R&D Data Discovery and FAIR-ification of Data

The day continued with a knowledge graph story from Martin Romacker, Senior Principal Scientist in Scientific Solution Engineering and Architecture at Roche. Their solution offered a modular approach to R&D data discovery and enabled end users to interact with huge volumes of data.

We also heard about FAIR-ification of data from The Swiss Personalized Health Network (SPHN) – a national initiative under the leadership of the Swiss Academy of Medical Sciences. SPHN talked about how they developed a national framework for standardizing the semantic representation of health data. The presentation focused on how semantically enriched Linked Data enables researchers to easily integrate disparate datasets and leverage clinical knowledge from ontologies for their research.

And before we move on, here are a couple more compelling presentations about applying knowledge graphs in Healthcare and Life Sciences: “Re-Imagining PV with Cutting Edge AI based Solution “Talosafe”” by Rajesh Singh, Gaurav Satsangi and Nilesh Doshi from our partner Wipro. And “Preclinical Discovery Knowledge Graph” by Kai Preuss from metaphacts, Felix Schwagereit from Roche and Todor Primov from Ontotext.

Exemplary Strategies For Drug Target Discovery

Martina Markova, Business analyst at Ontotext, shared her knowledge and expertise in integrating heterogeneous data sources and mapping and managing them in a central knowledge graph. Throughout the talk Martina showed how relevant information and insights can be unlocked in both structured and unstructured reference Pharma and Biotech data sources, equipping participants with exemplary strategies for Drug Target Discovery.

The day ended with another insightful panel discussion led by Martin Romacker and Chris Brockmann.

Another insightful panel discussion led by Martin Romacker and Chris Brockmann.

Register to browse all the recorded talks and discussions until the end of October!

Day 4: Lessons learned, Paths Triablazed and The Modeling Jump

Day 4 was dedicated to use cases and good practices in building knowledge graphs and solutions in the Financial Services and Publishing sectors.

And it started big.

With a simple yet powerful observation: It is always about the use cases. And they all boil down to: “How can I make decisions faster?” This came from Ragini Okhandiar and Krishna Potluri from JPMorgan Chase & Co. They shared their expertise in building an in-house knowledge graph.

Ragini Okhandiar answering questions from the audience after her talk about the way JPMorgan Chase & Co. built their knowledge graph.

The Value of Knowledge Graphs and Their Proper Project Management

After Ragini Okhandiar, we heard from Michael Atkin, Managing Director, Content Strategies LLC talking about “The Business Case for Data Management” and Tariq Farwana, Managing Director at StrateSphere, about “Leveraging Value Chain Mapping Methodologies Across Multiple Industries”.

The day continued with Lulit Tesfaye and Sara Nash from Enterprise Knowledge and their presentation about the worthlessness of knowledge graphs, when the use case is not set right.

In their talk “Knowledge Graphs are Worthless, Knowledge Graph Use Cases are Priceless” they continued the topic on value and the ways we can get the most out of any knowledge graph project. Sara and Lulit equipped the audience with caveats about why knowledge graph project might fail.

Slides from Lulit Tesfaye and Sara Nash from Enterprise Knowledge on the most frequent reasons why do graph projects fail.

They also provided a roadmap with vital elements that need to be sorted out before a project begins so it can be successful.

Not All Knowledge Graphs Are Created Equal

What followed was also about expected failure in some knowledge graph projects using property graphs. Borislav Iordanov, Knowledge Engineering Consultant at Semantic Arts, summarized some of the reasons why property graphs fail in the enterprise world.

Borislav Iordanov’s take on why do knowledge graph projects using property graphs fail?

Borislav also walked us through the whys and hows of turning a property graph into a knowledge graph, showing the actual steps needed to turn an LPG into a knowledge graph. You can read about them in his white paper: Turning Your Property Graph into a Robust Knowledge Graph.

Knowledge Graphs In Publishing and Finance

The day also featured a few more talks showcasing knowledge graphs in Publishing and Finance. “Sentinel One – Your Tool for AML in the Crypto World” was presented by Boris Khazin from EPAM Systems. Polly Alexander from WebMD Provider Services talked about “Harmonizing Content Publishing Practices Using a Semantic Approach: A WebMD Provider Services Use Case”. Joseph Hilger, COO at Enterprise Knowledge explained about “Making Knowledge Management Clickable with Knowledge Management Portals”. And Peio Popov, Financial Services Lead at Ontotext shared “Success Stories – Applications and Benefits of Knowledge Graphs in the FSI”.

The final panel was “Top Knowledge Graph Use Cases in Financial Services” and provided an enlightening discussion between Michael Atkin, Boris Khazin, Joseph Hilger and Peio Popov.

Panelists in the final panel “Top Knowledge Graph Use Cases in Financial Services” (from left to right): Peio Popov, Boris Khazin, Joseph Hilger, Michael Atkin, and Atanas Kiryakov.

Epilogue

We did have a wonderful Knowledge Graph Forum!

We saw that on the look out for quick clickable decision-making and data integration tools, there are different paths to take, all of them tangent to the knowledge management practice and the Semantic Web stack. Our gratitude goes to all our attendees and to all our partners and customers who shared their insight and experience in building knowledge graphs and designing enterprise-grade solutions with semantic technologies.

If you missed the party, it’s not too late.

Register to browse all the recorded talks and discussions until the end of October!