GraphNews

📣 Our spicy ICML 2025 position paper: “Graph Learning Will Lose Relevance Due To Poor Benchmarks”. Graph learning is less trendy in the ML world than it was in 2020-2022. We believe the problem is in poor benchmarks that hold the field back - and suggest ways to fix it! We identified three problems: #️⃣ P1: No transformative real-world applications - while LLMs and geometric generative models become more powerful and solve complex tasks every generation (from reasoning to protein folding), how transformative could a GNN on Cora or OGB be? P1 Remedies: The community is overlooking many significant and transformative applications, including chip design and broader ML for systems, combinatorial optimization, and relational data (as highlighted by RelBench). Each of them offers $billions in potential outcomes. #️⃣ P2: While everything can be modeled as a graph, often it should not be. We made a simple experiment and probed a vanilla DeepSet w/o edges and a GNN on Cayley graphs (fixed edges for a certain number of nodes) on molecular datasets and the performance is quite competitive. #️⃣ P3: Bad benchmarking culture (this one hits hard) - it’s a mess :) Small datasets (don’t use Cora and MUTAG in 2025), no standard splits, and in many cases recent models are clearly worse than GCN / Sage from 2020. It gets worse when evaluating generative models. Remedies for P3: We need more holistic benchmarks which are harder to game and saturate - while it’s a common problem for all ML fields, standard graph learning benchmarks are egregiously old and rather irrelevant for the scale of problems doable in 2025. 💡 As a result, it’s hard to build a true foundation model for graphs. Instead of training each model on each dataset, we suggest using GNNs / GTs as processors in the “encoder-processor-decoder” blueprint, train them at scale, and only tune graph-specific encoders/decoders. For example, we pre-trained several models on PCQM4M-v2, COCO-SP, and MalNet Tiny, and fine-tuned them on PascalVOC, Peptides-struct, and Stargazers to find that graph transformers benefit from pre-training. --- The project started around NeurIPS 2024 when Christopher Morris gathered us to discuss the peeve points of graph learning and how to continue to do impactful research in this area. I believe the outcomes appear promising, and we can re-imagine graph learning in 2025 and beyond! Massive work with 12 authors (everybody actually contributed): Maya Bechler-Speicher, Ben Finkelshtein, Fabrizio Frasca, Luis Müller, Jan Tönshoff, Antoine Siraudin, Viktor Zaverkin, Michael Bronstein, Mathias Niepert, Bryan Perozzi, and Christopher Morris (Chris you should create a LinkedIn account finally ;)

Integrating Neo4j, DuckDB, and Kestra for Metadata-Driven Graph Design

How to leverage Large Language Models to create consistent Knowledge Graphs from multiple sources

Contribute to LMMApplication/RAKG development by creating an account on GitHub.

How AI is learning to build its own conceptual frameworks

Official Implementation of "Affordable AI Assistants with Knowledge Graph of Thoughts" - spcl/knowledge-graph-of-thoughts

Graph visualisation for UK Research and Innovation (UKRI) funding, including NetworkX, PyVis and LlamaIndex graph retrieval-augmented generation (RAG)

Knowledge graphs to teach LLMs how to reason like doctors! Many medical LLMs can give you the right answer, but not the right reasoning which is a problem for clinical trust. 𝗠𝗲𝗱𝗥𝗲𝗮𝘀𝗼𝗻 𝗶𝘀 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝗳𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆-𝗴𝘂𝗶𝗱𝗲𝗱 𝗱𝗮𝘁𝗮𝘀𝗲𝘁 𝘁𝗼 𝘁𝗲𝗮𝗰𝗵 𝗟𝗟𝗠𝘀 𝗰𝗹𝗶𝗻𝗶𝗰𝗮𝗹 𝗖𝗵𝗮𝗶𝗻-𝗼𝗳-𝗧𝗵𝗼𝘂𝗴𝗵𝘁 (𝗖𝗼𝗧) 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝘂𝘀𝗶𝗻𝗴 𝗺𝗲𝗱𝗶𝗰𝗮𝗹 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗴𝗿𝗮𝗽𝗵𝘀.  1. Created 32,682 clinically validated QA explanations by linking symptoms, findings, and diagnoses through PrimeKG.  2. Generated CoT reasoning paths using GPT-4o, but retained only those that produced correct answers during post-hoc verification.  3. Validated with physicians across 7 specialties, with expert preference for MedReason’s reasoning in 80–100% of cases.  4. Enabled interpretable, step-by-step answers like linking difficulty walking to medulloblastoma via ataxia, preserving clinical fidelity throughout. Couple thoughts:   • introducing dynamic KG updates (e.g., weekly ingests of new clinical trial data) could keep reasoning current with evolving medical knowledge.  • Could also integrating visual KGs derived from DICOM metadata help coherent reasoning across text and imaging inputs? We don't use DICOM metadata enough tbh  • Adding testing with adversarial probing (like edge‑case clinical scenarios) and continuous alignment checks against updated evidence‑based guidelines might benefit the model performance Here's the awesome work: https://lnkd.in/g42-PKMG Congrats to Juncheng Wu, Wenlong Deng, Xiaoxiao Li, Yuyin Zhou and co! I post my takes on the latest developments in health AI – 𝗰𝗼𝗻𝗻𝗲𝗰𝘁 𝘄𝗶𝘁𝗵 𝗺𝗲 𝘁𝗼 𝘀𝘁𝗮𝘆 𝘂𝗽𝗱𝗮𝘁𝗲𝗱! Also, check out my health AI blog here: https://lnkd.in/g3nrQFxW | 40 comments on LinkedIn

𝐑𝐞𝐚𝐝𝐲 𝐟𝐨𝐫 𝐚 "𝐖𝐎𝐖!" 𝐦𝐨𝐦𝐞𝐧𝐭? The 𝘿𝙄𝙏𝘼 𝙂𝙧𝙖𝙥𝙝 𝙍𝘼𝙂 project automates building your content corpus's knowledge graph (KG) from structured documents. What few know are the (literally) hundreds of ways you can use a KG on your structured documents in addition to RAG retrieval for AI, so I've compiled a compendium of 150 DITA graph queries, what each does, the SPARQL queries themselves, and the business value of each. These 150 are only a sampling. Try doing THAT with the likes of Markdown, AsciiDoc, RsT, and other presentation-oriented document formats! 90 packed pages! https://lnkd.in/eY2kHcBe | 22 comments on LinkedIn

🔎 Lessons Learned from Evaluating NodeRAG vs Other RAG Systems I recently dug into the NodeRAG paper (https://lnkd.in/gwaJHP94) and it was eye-opening not just for how it performed, but for what it revealed about the evolution of RAG (Retrieval-Augmented Generation) systems. Some key takeaways for me: 👉 NaiveRAG is stronger than you think. Brute-force retrieval using simple vector search sometimes beats graph-based methods, especially when graph structures are too coarse or noisy. 👉 GraphRAG was an important step, but not the final answer. While it introduced knowledge graphs and community-based retrieval, GraphRAG sometimes underperformed NaiveRAG because its communities could be too coarse, leading to irrelevant retrieval. 👉 LightRAG reduced token cost, but at the expense of accuracy. By focusing on retrieving just 1-hop neighbors instead of traversing globally, LightRAG made retrieval cheaper — but often missed important multi-hop reasoning paths, losing precision. 👉 NodeRAG shows what mature RAG looks like. NodeRAG redesigned the graph structure itself: Instead of homogeneous graphs, it uses heterogeneous graphs with fine-grained semantic units, entities, relationships, and high-level summaries — all as nodes. It combines dual search (exact match + semantic search) and shallow Personalized PageRank to precisely retrieve the most relevant context. The result? 🚀 Highest accuracy across multi-hop and open-ended benchmarks 🚀 Lowest token retrieval (i.e., lower inference costs) 🚀 Faster indexing and querying 🧠 Key takeaway: In the RAG world, it’s no longer about retrieving more — it’s about retrieving better. Fine-grained, explainable, efficient retrieval will define the next generation of RAG systems. If you’re working on RAG architectures, NodeRAG’s design principles are well worth studying! Would love to hear how others are thinking about the future of RAG systems. 🚀📚 #RAG #KnowledgeGraphs #AI #LLM #NodeRAG #GraphRAG #LightRAG #MachineLearning #GenAI #KnowledegGraphs

In the world of representing and storing knowledge, things or entities have long monopolized center stage. Rows in databases represent things, where columns are just used to describe them.

📣 Byte #21: For those of you who want to visualize their graphs inside Jupyter notebooks - we have an exciting development! We recently released an integration with yWorks, who extended their yFiles Jupyter Graphs widget to support Kuzu databases! ✅ Once a Kuzu graph is created, we can instantiate the yFiles Jupyter KuzuGraphWidget, and use the `show_cypher` method to display a subgraph using regular Cypher queries. ✅ There are numerous custom layouts in the yFiles widget (tree, hierarchical, orthogonal, etc.). Give them a try! Here's an example of the tree layout, which is great for visualizing data like this that has rich tree structures. We can see the two-degree mentors of Christian Christiansen, a Nobel prize-winning laureate, in this example. ✅ You can customize the appearance of the nodes in the widget through `add_node_configuration` method. This way, you can display what you're looking for as you iterate through your graph building process. ✅ The Kuzu-yFiles integration is open source and you can begin using it right away for your own interactive visualizations. Give it a try and share around with fellow graph enthusiasts! pip install yfiles-jupyter-graphs-for-kuzu Docs page: https://lnkd.in/g97uSKRe GitHub repo: https://lnkd.in/gjA6ZjiF

A little peek into our developments of RDF-specific functionality for VS Code: 1️⃣ The autocompletion and hover-help for RDF vocabularies. Some are stored within the VS Code plugin, the rest are queried from LOV, giving intellisense to the most prominent ontologies. 2️⃣ We can use the ontology of the vocabularies to show when something is not typed correctly 3️⃣ SHACL has a SHACL meta-model. As we built a SHACL engine into VS Code, we can use this meta model to hint if something is not done correctly (e.g., a string as part of a datatype). We plan to release the plugin to the marketplace in some time (However, we are still building more functionality). To not take too much credit: https://lnkd.in/eFB2wKdz delivers important features like most syntax highlighting and auto-import of the prefixes.

To all the knowledge graph enthusiasts who've felt for a while that "graphs are the way to go" when it comes to enabling "intelligence," it was interesting to read Anthropic's "Tracing the thoughts of a large language model" - if you believe that LLMs need graphs to reason, you are right and now you have evidence: Claude answers questions by building and traversing a graph (in latent space) before it translates it back to language: https://lnkd.in/eWFWwfN4 | 20 comments on LinkedIn

A second usecase Thomas wrote for Veronika Heimsbakk’s SHACL for the Practitioner upcoming book is about Sparna’s work for the European Parliament. From validation of the data in the knowledge graph to further projects of data integration and dissemination, many different usages of SHACL specifications were explored… … and more exploratory usages of SHACL are foreseen ! “…

Large Language Models (LLMs) are revolutionizing the development of AI assistants capable of performing diverse tasks across domains. However, current state-of-the-art LLM-driven agents face...

What if your LLM is… a graph? A few days ago, Petar Veličković from Google DeepMind gave one of the most interesting and thought provoking conference I've seen in a while, "Large Language Models as Graph Neural Networks". Once you start seeing LLM as graph neural network, many structural oddities suddenly falls into place. For instance, OpenAI currently recommends to put the instructions at the top of a long prompt. Why is that so? Because due to the geometry of attention graphs, LLM are counter-intuitively biased in favors of the first tokens: they travel constinously through each generation steps, are internally repeated a lot and end up "over-squashing" the latter ones. Models then use a variety of internal metrics/transforms like softmax to moderate this bias and better ponderate distribution, but this is a late patch that cannot solve long time attention deficiencies, even more so for long context. The most interesting aspect of the conference from an applied perspective: graph/geometric representations directly affect accuracy and robustness. As the generated sequence grow and deal with sequences of complex reasoning steps, cannot build solid expert system when attention graphs have single point of failures. Or at least, without extrapolating this information in the first place and providing more detailed accuracy metrics. I do believe LLM explainability research is largely underexploited right now, despite being accordingly a key component of LLM devops in big labs. If anything, this is literal "prompt engineering", seeing models as nearly physical structure under stress and providing the right feedback loops to make them more reliable. | 30 comments on LinkedIn

In case you missed the Spanner Graph session at Google Cloud Next’25, the recording is now available: • Introduction of the graph space at 00:00 (https://lnkd.in/gsBFuDbt) • Spanner Graph overview at 07:24 (https://lnkd.in/ggxrzFrU) • How Snapchat builds its Identity Graph at 20:32 (https://lnkd.in/gFauYj-9) • Quick demo of an recommendation engine at 26:27 (https://lnkd.in/gvH4AbRF) • Recent launches at 32:00 (https://lnkd.in/gyCPq97t) • Vision: unified Google Cloud Graph solution with BigQuery Graph at 35:09 (https://lnkd.in/gRdbSMeu) I hope you like it! You can get started with Spanner Graph today! https://lnkd.in/gkwbGFbS Pratibha Suryadevara, Spoorthi Ravi, Sailesh Krishnamurthy, Andi Gutmans, Christopher Taylor, Girish Baliga, Tomas Talius, Candice Chen, Yun Zhang, Weidong Yang, Matthieu Besozzi, Giulia Rotondo, Leo Meyerovich, Thomas Cook, Arthur Bigeard #googlecloud #googlecloudnext25 #graphdatabases #spannergraph

In this post, we’ll dive into Kuzu’s core query capabilities and demonstrate how it pairs nicely with G.V()’s rich visualization tools.

An open-source Python framework for real-time, graph-based knowledge in Neo4j

How LLMs can used in a knowledge extraction and visualization pipeline

"I'm Ukrainian and I'm wearing a suit, so no complaints about me from the Oval Office" - that's the start of my lecture about building Artificial Intelligence with Croissant ML in the Dataverse data platform, for the Bio x AI Hackathon kick-off event in Berlin. https://lnkd.in/ePYHCfJt * 750,000+ FAIR datasets across the world forcing the innovation of the whole data landscape. * A knowledge graph with 50M+ triples. * AI-ready metadata exports. * Qdrant as a vector storage, Google Meta Mistral AI as LLM model providers. * Adrian Gschwend Qlever as fastest triple store for Dataverse knowledge graphs Multilingual, machine-readable, queryable scientific data at scale. If you're interested, you can also apply for the 2-month #BioAgentHack online hackathon: • $125K+ prizes • Mentorship from Biotech and AI leaders • Build alongside top open-science researchers & devs More info: https://lnkd.in/eGhvaKdH

We’re thrilled to announce new Text2Cypher models and Google’s MCP Toolbox for Databases from the collaboration between Google Cloud and Neo4j.

The Model Context Protocol gives users a wide range of services they can connect to from the comfort of their AI client.

Choosing the Right Format: How Knowledge Graph Layouts Impact AI Reasoning ... 👉 Why This Matters Most AI systems blend knowledge graphs (structured data) with large language models (flexible reasoning). But there’s a hidden variable: "how" you translate the graph into text for the AI. Researchers discovered that the formatting choice alone can swing performance by up to "17.5%" on reasoning tasks. Imagine solving 1 in 5 more problems correctly just by adjusting how you present data. 👉 What They Built KG-LLM-Bench is a new benchmark to test how language models reason with knowledge graphs. It includes five tasks: - Triple verification (“Does this fact exist?”) - Shortest path finding (“How are two concepts connected?”) - Aggregation (“How many entities meet X condition?”) - Multi-hop reasoning (“Which entities linked to A also have property B?”) - Global analysis (“Which node is most central?”) The team tested seven models (Claude, GPT-4o, Gemini, Llama, Nova) with five ways to “textualize” graphs, from simple edge lists to structured JSON and semantic web formats like RDF Turtle. 👉 Key Insights 1. Format matters more than assumed:   - Structured JSON and edge lists performed best overall, but results varied by task.   - For example, JSON excels at aggregation tasks (data is grouped by entity), while edge lists help identify central nodes (repeated mentions highlight connections). 2. Models don’t cheat: Replacing real entity names with fake ones (e.g., “France” → “Verdania”) caused only a 0.2% performance drop, proving models rely on context, not memorized knowledge. 3. Token efficiency:   - Edge lists used ~2,600 tokens vs. JSON-LD’s ~13,500. Shorter formats free up context space for complex reasoning.   - But concise ≠ always better: structured formats improved accuracy for tasks requiring grouped data. 4. Models struggle with directionality:   Counting outgoing edges (e.g., “Which countries does France border?”) is easier than incoming ones (“Which countries border France?”), likely due to formatting biases. 👉 Practical Takeaways - Optimize for your task: Use JSON for aggregation, edge lists for centrality. - Test your model: The best format depends on the LLM—Claude thrived with RDF Turtle, while Gemini preferred edge lists. - Don’t fear pseudonyms: Masking real names minimally impacts performance, useful for sensitive data. The benchmark is openly available, inviting researchers to add new tasks, graphs, and models. As AI handles larger knowledge bases, choosing the right “data language” becomes as critical as the reasoning logic itself. Paper: [KG-LLM-Bench: A Scalable Benchmark for Evaluating LLM Reasoning on Textualized Knowledge Graphs] Authors: Elan Markowitz, Krupa Galiya, Greg Ver Steeg, Aram Galstyan

Transform UML into a formal OWL ontology and SHACL shapes - OP-TED/model2owl

Root Github Pages.

Our first attempts at mechanistic interpretability of Transformers from the perspective of network science and graph theory! Check out our preprint: arxiv.org/abs/2502.12352 A wonderful collaboration with superstar MPhil students Batu El, Deepro Choudhury, as well as Pietro Lio' as part of the Geometric Deep Learning class last year at University of Cambridge Department of Computer Science and Technology We were motivated by Demis Hassabis calling AlphaFold and other AI systems for scientific discovery as ‘engineering artifacts’. We need new tools to interpret the underlying mechanisms and advance our scientific understanding. Graph Transformers are a good place to start. The key ideas are: - Attention across multi-heads and layers can be seen as a heterogenous, dynamically evolving graph. - Attention graphs are complex systems represent information flow in Transformers. - We can use network science to extract mechanistic insights from them! More to come on the network science perspective to understanding LLMs next! | 13 comments on LinkedIn

🚀 Thrilled to share our latest work published in Nature Machine Intelligence! 📄 "A comprehensive large-scale biomedical knowledge graph for AI-powered data-driven biomedical research" In this study, we constructed iKraph, one of the most comprehensive biomedical knowledge graphs to date, using a human-level information extraction pipeline that won both the LitCoin NLP Challenge and the BioCreative Challenge. iKraph integrates insights from over 34 million PubMed abstracts and 40 public databases, enabling unprecedented scale and precision in automated knowledge discovery (AKD). 💡 What sets our work apart? We developed a causal knowledge graph and a probabilistic semantic reasoning (PSR) algorithm to infer indirect entity relationships, such as drug-disease relationships. This time-aware framework allowed us to retrospectively and prospectively validate drug repurposing and drug target predictions, something rarely done in prior work. ✅ For COVID-19, we predicted hundreds of drug candidates in real-time, one-third of which were later supported by clinical trials or publications. ✅ For cystic fibrosis, we demonstrated our predictions were often validated up to a decade later, suggesting our method could significantly accelerate the drug discovery pipeline. ✅ Across diverse diseases and common drugs, we achieved benchmark-setting recall and positive predictive rates, pushing the boundaries of what's possible in drug repurposing. We believe this study sets a new frontier in biomedical discovery and demonstrates the power of structured knowledge and interpretability in real-world applications. 📚 Read the full paper: https://lnkd.in/egYgbYT4? 📌 Access the platform: https://lnkd.in/ecxwHBK7 📂 Access the data and code: https://lnkd.in/eBp2GEnH LitCoin NLP Challenge: https://lnkd.in/e-cBc6eR Kudos to our incredible team and collaborators who made this possible! #DrugDiscovery #AI #KnowledgeGraph #Bioinformatics #MachineLearning #NatureMachineIntelligence #DrugRepurposing #LLM #BiomedicalAI #NLP #COVID19 #Insilicom #NIH #NCI #NSF #ARPA-H | 10 comments on LinkedIn

Uncover data's hidden connections using graph analytics in BigQuery. This session shows how to use BigQuery's scalable infrastructure for graph analysis directly in your data warehouse. Identify patterns, connections, and influences for fraud detection, drug discovery, social network analysis, and recommendation engines. Join us to explore the latest innovations in graphs and see real-world examples. Transform your data into actionable insights with BigQuery's powerful graph capabilities.