Read about the three core features that give knowledge graphs their capabilities and how they generate value for four key players.
The 17th edition of the annual SEMANTiCS conference – a knowledge hub and a vibrant arena for industry and academia dialogue – took place in Vienna from September 13th to 15th. Within three days, packed with tutorials, talks and insightful keynotes, technology professionals, industry experts and researchers shared their visions and learned about new research, technologies and good practices in the fields of Linked Data and knowledge graphs.
Organized by our partners at the Semantic Web Company, SEMANTiCS 2022 brought a total of 4 topical tracks, a dedicated DBPedia day, a poster park and a Data meetup. This allowed for a rich experience of exchange between people interested in semantic technologies.
SEMANTiCS 2022 opened with lots of coffee. Literally. The first keynote was held by Olaf Hartig of Linköping University in a packed hall and featured a fictitious company that would connect brewers, people and machines through federated queries.
Olaf’s “Towards Querying Heterogeneous Federations of Interlinked Knowledge Graphs” started with a fictitious company Jumpstart Brew. The vision was to connect their knowledge graphs, holding information about coffee brewers, researchers of coffee brews and recipes, data about vending machines delivering coffee, their parts, contracts, documentation.
To underpin that vision, Olaf shared techniques for querying federations of the interlinked knowledge graphs that would enable getting answers from different graphs, combined with one another. He talked about new types of data access interfaces for RDF and shared a bit about FedQPL – A Language for Logical Query Plans over Heterogeneous Federations of RDF Data Sources.
A great recap shot is available from Christoph Broun.
— Christoph Braun (@_uvdsl) September 13, 2022
The first day of SEMANTiCS also featured a tutorial on knowledge engineering, held by Heather Hedden, Data and Knowledge Engineer at Semantic Web Company. This was an insightful combination of a crash course on taxonomies, given through the perspective of ontologies and thesauri. Heather gave plenty of examples and a lot of useful comparisons between taxonomies, thesauri and ontologies and the use cases they are best suitable for.
This tutorial gave us the fundamentals for creating quality taxonomies with an emphasis on serving specific use cases, rather than linking and categorizing all the knowledge in the world. Grounding the tutorial in the sound basis of the practice of knowledge engineering, Heather systematically led us through different kinds of knowledge organization systems (taxonomies, thesauri, ontologies, etc.) and explained the differences between them.
The hall was packed with people and there were a lot of questions, all trying to find that bridge between taxonomies and ontologies. Is medicine the umbrella of the hospital, well yes and no. While I thought it was, the good taxonomies practice wouldn’t allow such a level of abstraction, as it would not serve the purpose of the user facing the information.
How are documents, documentation and information in general exchanged today? What is the fundamental change in delivering content? And what does the future of semantically tagged modular content look like? These were the questions Jan Benedictus of Fonto and Jörg Schmidt of RWS Group promised to answer in their tutorial “The Future of Documents: Semantics and Structured Content”.
And they delivered.
The session was full of examples about how authoring and publishing changes when a knowledge graph and a process of semantic tagging enter the picture. The vision about the end of documents and “disconnected boxes” can be underpinned by the use of knowledge graphs in content management, where content is managed and created at the level of data, not documents.
As a main takeaway from this tutorial, I would highlight the understanding about semantic tagging and having a system that brings all modular content together. It is such a system that sets documents (and authors) free and takes content management to another, knowledge-graph powered level.
SEMANTiCS 2022 also hosted a wonderful Posters park, where the seeds of all kinds of knowledge graph use cases were showcased.
The papers, accompanying this garden of ideas are available in the conference’s Poster’s Park.
It is worth mentioning Solid – the project led by Prof. Tim Berners-Lee, aiming to change the way Web applications work today, resulting in true data ownership as well as improved privacy. It was used in several posters, showcasing how its value can turn out to be a value in an enterprise setting. For example, the paper Solid Proof of Concept in an Enterprise Loan Request Use Case Solid, WebID, Enterprise Data Exchange, showed how Solid pods could be used to share data between companies in a tangible scenario.
The best award for paper went to Franz Krause, Tobias Wleer and Heiko Paulheim for their “On a generalized Framework for Time-Aware Knowledge Graphs”.
Do you like fabrics or leather sofas? Do you prefer a small or a large sofa? And how do I, as an IKEA team member, rearrange all the sofas in the IKEA hall to match your personal preferences while you are browsing the physical store? Of course, I can’t. But if I am a UX designer, a tiny, yet mighty feature as a checkbox on IKEA’s website, might help. It is the checkbox with which you as a user would choose personalized content, fabrics or leather sofas included.
And it is that checkbox that in 2019 put Adam Keresztes on the knowledge graph road.
In his keynote speech “UX Design & Knowledge Graphs – The Perfect Match” Adam explained the huge role knowledge graphs can and do play in bridging the gap between data and design. In this talk that simple truth became evident, Knowledge Graphs Are Not About Data, they are about diminishing the distance between people and data. By explaining the need for UX design to embrace the knowledge graph technology (and vice versa), Adam brought home the point that designing customer experiences for the digital environment creates extremely big dependency on data.
Yet, the main takeaway from this keynote came after the talk. “It is the knowledge we want to embed in our system that matters to me”, is what Adam told me in the coffee break. It is that sentence that truly rang a bell and further connected to the main thread running through the next two sessions I attended.
Following Adam’s talk was a session on Recommender systems, where our own Nikolay Krustev continued this narrative of serving users the right experiences through data.
In his presentation “Lowering operational complexity and building bridges between industries”, Nikolay showed how knowledge graphs (built with standardized data) can be leveraged to help businesses across various industries create synergies and optimize their current business models.
He walked us through our own everyday life where we grab a coffee, drive our car, go through the city’s infrastructure, stop by to shop and enter our office spaces. When seen through the perspective of knowledge graph technology, this is actually a story about knowledge graphs. And that story includes interoperable data across the automobile industry, retail industry, architecture, energy and more. 13 industries, all utilizing knowledge graphs to enhance their operations.
To walk the knowledge graph story with Nikolay across the 13 industries where Ontotext is working to help enterprises serve their users better – watch the recording.
One of the pinnacles of the conference was the panel of knowledge graph experts, practitioners and technology leaders, held on the second day of SEMANTiCS. Andreas Blumauer was a terrific host, managing to keep things grounded in the why of knowledge graphs, rather than the “how”.
At “The Global Knowledge Graph Market: State of Play and Upcoming Trends”, the panelists were our own Atanas Kiryakov, Joseph Hilger COO of Enterprise Knowledge, Jans Aasman, CEO at Franz Inc and Peter Crocker, co-founder and CEO of RDF FOX. It was moderated by Andreas Blumauer and discussed the relation of knowledge graphs and semantic technologies to sustainability and organizational change. They shared their views on the state of the knowledge graph market and the maturity of the requests coming from clients.
The panelists’ insight and high-level perspective helped us, the audience, understand more about current and future practices in the way enterprise knowledge graphs are conceived. Sharing their expertise in the actual creation of knowledge graphs across diverse industries, the panelists paved the way towards an informed way of deciding how to build sustainable solutions with semantic technology.
During the session Andreas Blumauer challenged them to a myth busting session.
KNOWLEDGE GRAPH MYTHS DEBUNKED
|Myth 1. Knowledge graphs are useful only for data integration.
|Knowledge graphs are primarily about models that support questions and understanding that underpins the success of any project.
|Myth 2: You have to have a complete, perfect model.
|Knowledge graphs should start small and be expanded later depending on what the technology allows, otherwise it’s a road to spending a lot of money.
|Myth 3: Knowledge graphs are such an effort!
|Knowledge graphs are not an effort if you start your project with smart people and do the thinking and modeling wisely and in a future-proof way.
|Myth 4: RDF is so complicated.
|Semantic technologies (including RDF) make it much easier and more intuitive to model business challenges in a mindmap manner. Modeling in a relational database is what should scare people away.
|Myth 5: To have a knowledge graph, you only need to take your taxonomy and apply some machine learning magic.
|The real value of knowledge graphs is in formal semantics, which helps you do proper machine-readable descriptions.
The main takeaway from the panel was about the maturing of the technology and the client’s expectations. Things have changed. Customer’s expectations have changed. People have started asking the business questions. They are now asking for a solution, not just a knowledge graph, because Google has it, but because they have a business challenge that they know would take a knowledge-graph based approach.
In other words, business leaders now know better that, as Jim Hendler famously put it, a little semantics goes a long way!
You can watch the recording of the entire panel.
For an intense three days, the event gathered industry, academia and people who live and walk the talk about connecting data. The conference delivered the perspectives needed to understand knowledge graphs and semantic technologies on various levels of abstraction. It gave a plethora of perspectives on the technology layer and the user-facing layer, where the real value of knowledge graphs and semantic technologies lies.
And that’s not all! There’s more coming.
Want to hear from world-leading experts about the state of knowledge graph technology?