Graphwise

Which genes are involved in a particular disease? And which biological processes and pathways do they participate in? Answering these questions often means searching through multiple biomedical databases and piecing together information manually. At the Oxford Drug Discovery Institute, researching a single therapeutic target for Alzheimer's disease could take several hours to one or two days. By bringing public and internal research data into one connected system, the team was able to evaluate and prioritize candidates much faster. This is one example of how we help #LifeSciences & #Pharma organizations work with fragmented biomedical data. See how #knowledgegraphs, harmonized data, and natural language queries help researchers move from scattered information to traceable answers, with practical examples from #drugdiscovery, pharmaceutical R&D, and regulatory workflows ⤵︎ https://lnkd.in/e_2asadu

A week after the #GraphwiseAISummit, the thing that stayed with us is how much the conversation has evolved compared with 10, 5, even 1 year ago. We are no longer talking about #semantics as an abstract data discipline or a niche concern for ontologists, knowledge engineers and #SemanticWeb people. We are talking about a technology with a long track record of solving some of the core problems AI is now running into at enterprise scale. That is, establishing shared meaning across fragmented systems, giving agents usable context, preserving provenance and governance, and making knowledge reusable across applications instead of rebuilding it for every new project. Those problems appeared in very different contexts - technical documentation at Avalara, FAIR and semantic governance at Roche and AstraZeneca, power-system data with Statnett SF and DNV, enterprise metadata at S&P Global, and manufacturing, maintenance, banking and #GraphRAG use cases across our partner network. The fact that the same architectural needs show up in such different environments is hard to ignore. The Summit also demonstrated that the strongest solutions are rarely built by one vendor or one discipline alone. They come from people who understand the domain, the data, the semantics and the technology, and who have spent enough time implementing these systems to know where they create real value and where the hard work still is. If you want to dig into the sessions, the recordings are now available 👉 https://lnkd.in/d8G-Y5UQ

🏆 Best Paper Award at Digital Heritage of European Conflicts Conference (DHECC) 2026! As the MEMORISE_EU project wraps up after three incredibly meaningful years, we're proud to look back on our work using digital technologies to preserve the stories of victims of Nazi persecution and Holocaust survivors, making their testimonies, diaries, and other historical materials accessible to new generations. And what better way to round off the project than with a Best Paper Award! Our joint paper, "Semantic Enrichment of Historical Archives: Semi-Automated Topical Annotation of Heritage of Nazi Persecution Collections," received the award at the DHECC, held last month in Odense, Denmark. The conference was chaired by Stefan Jänicke (conference chair) and Tobias Ebbrecht-Hartmann (program chair). What story does the paper tell? We built an #NLP framework designed specifically to process historical documents like testimonies and diaries related to the Heritage of Nazi Persecution. It handles the heavy lifting—like semantic tagging, translation, and extracting emotive phrases—while giving researchers a Data Curation Tool to easily review and refine the results. Our team at Graphwise led the Data Infrastructures work. We focused on building the pipeline that ties all these individual NLP models together into one shared, end-to-end platform, allowing researchers to bring in their own collections and process them seamlessly. A massive thank you to our brilliant co-authors at UFAL / Institute of Formal and Applied Linguistics, Charles University and the Syddansk Universitet - University of Southern Denmark - @Christopher Brückner, Nicklas Sindlev Andersen, Uladzislau Yorsh, Klára Tauchmanová, Adam Turčan, Pavel Pecina, and our own Artem Revenko. Thanks as well to the entire MEMORISE consortium for the collaboration. As we lose the firsthand voices of survivors, using technology to ensure that their stories remain findable, connected, and deeply understood is more critical than ever. We're honoured to have played a part in this. The paper will appear in this collection: https://lnkd.in/dfJH3iKE Read more about project: https://lnkd.in/dFycUeqh #MEMORISE #DigitalHeritage #NLP #KnowledgeGraphs #HolocaustRemembrance #HorizonEurope #DHECC2026

How exciting (& fitting) that the third Semantic Layer Symposium starting tomorrow is taking place in one of our hometowns - Vienna! Hosted by our partners Enterprise Knowledge, LLC, the event brings together practitioners working on #semanticlayers, #metadata, governance, AI readiness and #enterpriseknowledge, with hands-on training, real-world implementations and plenty of time for discussion. The carefully curated agenda this year will welcome: - Ole Olesen-Bagneux (Actian) talking about how metadata management is changing in the era of AI - Nicolas Seyot (Morgan Stanley) on the semantic layer as an enablement layer for trust in the AI-enabled enterprise - Rosanna Stephens, CPCC (Adobe) on building an HR knowledge strategy for AI worth trusting - Noz Urbina, Malcolm Hawker, Amy Duncan & Ben Clinch in a practitioner panel on semantic layers in action - Karthikeyan Palanisamy (ASML) on connecting human knowledge work with intelligent automation - Martin Romacker, Dr. (Roche) and Ilian Uzunov (Graphwise) on moving from strings to science with ontologies, FAIR vocabularies, metadata and GUPRIs for AI-ready data. If you are attending, join Andreas Blumauer, Martin Trajkov & Ilian Uzunov along with the rest of the speakers for the happy hour and informal conversations. Conference agenda & registration 👉 https://lnkd.in/dAj2gGtx Hear from the panelists ahead of the Symposium 👉 https://lnkd.in/d6Qwa23K

Last week at the #GraphwiseAISummit, we learned from several demonstrated use cases that for businesses whose product is organised knowledge, such as manufacturing, the #semanticlayer is not only an enabling infrastructure but it's the moat 💰📉🏆 Tomorrow, together with our partners PANTOPIX, exploring the case of a manufacturer of configurable industrial coffee machines, we’ll show how a semantic knowledge base can connect distributed product information, translate customer requirements into suitable configurations, and explain why a recommendation fits. Join us! https://lnkd.in/dBerFAG9

We look forward to meeting you at the Data Stack Conf tomorrow, October 13, at Inter Expo Center in Sofia! 🔗 https://lnkd.in/dCTzDQqu

Alan Morrison & Todd Blaschka closing the final day of the #GraphwiseAISUmmit with these takeaways that capture a lot of what we heard & saw: ✦ the #SemanticBackbone is becoming self-reflective; AI can help improve the backbone, and the backbone can improve the AI ✦ agentic workflows have moved into real operational use cases ✦ governance is what makes that scale ✦ the path from ontologies to #GraphRAG is practical; we saw it applied across real enterprise use cases. ✦ AI-ready data starts with metadata To everyone who prepped presentations, spoke, supported in the background, asked questions, joined live, we thank you! And if you were part of it, share with us what you took away from the summit. What presentation you liked best, what resonated? 🤓

Using the Chinese room thought experiment is a good metaphor for the role of semantics & meaning in enterprise AI. AI systems can manipulate symbols convincingly without understanding what they really mean. That's why enterprise data needs explicit meaning and that's how we make it AI-ready. The session of Szymon Klarman from S&P Global brings nicely together a thread running through pretty much all sessions - shared meaning, identifiers, provenance, governance and context are not peripheral to AI. And neither are the standards underneath them. #RDF is essentially that explicit meaning is far more useful when it is expressed in an interoperable, vendor-neutral form that can travel across systems rather than getting trapped in ona place. #GraphwiseAISummit

An angle we haven’t touched on just yet - what happens when AI starts writing back into enterprise knowledge? Agis Papantoniou from Cognizone says that we need governance in both directions. Human and AI-generated facts should pass through the same rules, carry provenance, and make it clear who - or what - asserted them. That turns governance from a brake on AI into the mechanism that lets AI safely become part of the knowledge-building process. #GraphwiseAISummit

The recipe of Gianluca Generali from BitBang for making #enterpriseAI useful is pretty simple - start with the meaning, not with the model. Define the shared domain language, connect the relevant sources, enrich them with context, build the #knowledgegraph, and only then activate that knowledge through #GraphRAG and AI applications. His showed us three use cases (marketing, industrial maintenance and banking) where that works and generates value. #GraphwiseAISummit