GraphNews

Top modern AI use cases for ontology AI made enterprise knowledge feel instantly accessible — and exposed a deeper problem. Why ontology matters again? Ontology spent years in the background — …

Operating the Ontology Control Plane: A Practitioner’s Guide to the Last Mile The tips and tricks presented here are built on the work developed and shared by Tianxia Jin and Steve Mitchener in a …

Urban event management faces critical challenges in processing unstructured citizen complaints because of the limitations of traditional extraction models in handling colloquial text, static nature of conventional knowledge graphs that overlook spatiotemporal risk propagation, and brittleness of handcrafted symbolic rules that fail to generalize to novel scenarios. To address these issues, this study presents ST-CKG, a generalized spatiotemporal causal knowledge graph framework with Large Language Model (LLM)-enhanced reasoning. The framework integrates three tightly coupled components: a hybrid information extraction module that combines fine-tuned models with an LLM-based verifier to produce confidence-calibrated structured knowledge; a spatiotemporal causal hypergraph construction module that encodes spatial proximity and temporal causal hyperedges to model cascading urban risks; and a neuro-symbolic reasoning engine that distills expert-defined risk causal chain rules into a graph attention network, enabling interpretable and generalizable inference over incomplete or unseen event configurations. Extensive experiments conducted on over 2,000 real-world citizen complaints demonstrated that the hybrid extraction module outperformed standalone baselines by 3–5 percentage points in the F1-score, achieving 91% urgency assessment accuracy, significantly surpassing pure rule-based and pure graph neural network variants in rule-uncovered scenarios. Ablation studies have confirmed the indispensability of each component. The ST-CKG framework offers a replicable paradigm for bridging unstructured civic narratives with structured actionable governance decisions in smart cities.

Current LLM-based research agents have advanced through agent orchestration, yet largely overlook scientific knowledge orchestration. Existing works often reduce papers to abstracts, surface mentions, and flat \texttt{cites} edges, omitting key entities, claims, evidence, mechanisms, and method lineages essential for scientific reasoning. To this end, we introduce \textbf{Agents-K1}, an end-to-end knowledge orchestration pipeline that converts raw documents into agent-native scientific knowledge graphs. Agents-K1 integrates three components under a unifying theoretical foundation: a multimodal parser whose five-module schema captures entities, multimodal evidence, citations, and typed inter-entity relations across the full paper rather than abstracts alone; a 4B information-extraction backbone trained with GRPO under a rule-based reward; and a graphanything CLI, a tri-source agent interface that unifies web search, multimodal graph retrieval, and cross-document traversal. On top of this, we process 2.46 million scientific papers across six subjects to produce \textbf{Scholar-KG}, of which we release a one-million-paper subset, and the full Scholar-KG is accessible via the SCP link below. The same pipeline can be extended to general-domain corpora and to schema-conformant data synthesis. Extensive experiments demonstrate that Agents-K1 achieves superior performance in scientific information extraction, knowledge graph construction, and multi-hop scientific reasoning.

Cloud-computing relies on large-scale networks which are inherently complex systems. In this paper, we present a novel approach to root cause analysis (RCA) of cloud network incidents, leveraging graph-based causal discovery techniques. Our method addresses the limitations of rule-based automation by introducing a spatiotemporal grouping strategy and an automation ontology to reduce the dimensionality of the problem. We construct a causal graph from binary time series data using bivariate Granger causality and conditional independence tests. For inference, we introduce a probabilistic method that assigns edge-specific conditional probabilities as a function of time lag, allowing for interpretable, time-aware root cause scoring via causal graph traversal. We evaluated the system using a labeled dataset of 35 production incidents from a major cloud provider. The model successfully recalled the correct root cause in 85.7% of incidents and produced an exact match in 74.3%. In production, the deployed system has been used in over 800 real-world incidents, with positive qualitative feedback from network engineers. These results highlight the practicality of a data-driven, causal approach to RCA in dynamic and large-scale operational environments.

We propose KG-ER, a conceptual schema language for knowledge graphs that describes the structure of knowledge graphs independently of their representation (relational databases, property graphs, RDF) while helping to capture the semantics of the information stored in a knowledge graph.

Reframing Medical Reimbursement as a Temporal Knowledge Graph Abstract Most healthcare reimbursement systems operate on static documents. A chart is completed, coding begins, and reimbursement …

The Silicon Integration Initiative (Si2) AI/ML Schema/Ontology Working Group is introducing a public release of their AI for EDA Ontology. The goal of this public release of the Si2 AI/ML Schema Ontology Repository is to provide a strong example ontology and use cases to enable EDA agentic system experimentation for broader industry progress and collaboration.  With the rise of agentic AI systems,…

The Minimal Ontology Principle, Ontological Convergence, and What They Mean for Enterprise AI Platform Engineering

How better graph modelling reduces LLM context, improves retrieval precision, and makes GraphRAG easier to run

I've been hearing conversations around semantic layer. Everyone is talking about it. What the hell is a semantic layer actually? No one seems to explain that first. I hear people say that semantic layer doesn't increase accuracy but model docs does. Wait, isn't that exactly part of the semantic layer? This also came up in the roundtable that Shili Li from JetBrains organized with panelists André Baaij Baaij, Dylan Morrish, Evan Mullins and Jonathan Paquette for semantic layers this week. People seemed to have different ideas of what that exactly means. (It was a great discussion btw, thanks Shili for organising this!) So let's settle this properly. In fact, we need a universally accepted definition for this because there are now so many articles, research papers, etc. claiming semantic layer is useful or not without explaining what their semantic layer setup is and what it means for them. These articles are not even comparable as long as the semantic layer behind them is not comparable. ⁉️ What's your definition of a semantic layer? | 55 comments on LinkedIn

Why Taxonomies and Ontologies Are Not the Same Thing

Introducing Taxonomy 2.0(Agentic Version)

Tl;dr - Summary of the points. Graph databases are here for long time Adaption rates are low.

Last week I wrote about how "context for agents" isn't one problem: it's three. The response was way bigger than I expected: turns out a lot of people are quietly wrestling with the same question, which context problem am I actually solving, and who's out there solving it with me? So Miguel A. and I decided to do something about it. We consolidated the whole landscape into one map. The AI Context Layer Market Map lays out the entire space on a single page: → Bucket 1: the agent's context. Its own memory and operating manual, sorted by how much you model: pure RAG → hybrid → graph-native → full ontology. → Bucket 2: institutional knowledge. The scattered docs, chats and tickets your company already lives in. Surface-owners (who own where work happens) vs. neutral layers (who index across everything). → Bucket 3: systems of record. The hard operational data, split by who (or what) generates it: human-generated business records and machine-generated telemetry. Each side has its surface-owners and its neutral layers: centralized, federated, semantic. → Plus the two things that cut across all of it: governance & trust (the tax that scales with you), and the emerging public data context layer (verified facts and data from the world, so agents can query external knowledge). It's not exhaustive and it's a snapshot of a fast-moving space. We'll get things wrong and we'll miss players. That's kind of the point: tell us where we did, and we'll keep it honest. If you're building with agents and trying to figure out where you fit, this is meant to help you find your way. Map below. Curious what you'd move. 👇 | 33 comments on LinkedIn

I wonder, what is the reason this organization continuously excludes InfraNodus from all their “comprehensive” maps of the graph space? I think I’ll make a video on this topic for my YouTube channel because this is either a case of incompetence or gate keeping or maybe even both. State of the Graph as you will surely read this, please, leave your comment below, so I include your perspective in one of my videos at https://lnkd.in/d2wKHzSE There are 44K subscribers there who are very interested in graphs and I think they’ll be very curious to know why you continuously exclude leading tools (not only InfraNodus) from your reviews. Also a hint: you can also simply google GraphRAG API in a private browser session to break through the filter bubble…

How LLMs are changing ontology design: from machine-readable semantics to agent-digestible knowledge layers for retrieval, memory, validation, tools, governance, and autonomous AI workflows.

👉 Anthropic does not like Claude writing SQL directly against their database for their internal Self-Service Analytics! Instead, they require Claude querying a governed semantic layer as a function call (NOT SQL!) Their reasoning: the central problem is mapping a user's question to "specific and up-to-date entities in our data model" and knowing "the correct way of working with them." Get that wrong and the model grabs the wrong field, the wrong join, or a stale definition, and confidently returns a number that is subtly off. So for their key KPIs they want to have a single source of truth. At Veezoo, we built exactly that layer, our Knowledge Graph, already 10 years ago. It is two things at once: a store of governed business concepts, and very importantly, a query layer that turns them into answers. 👉  Where we differ (and why): For questions the semantic layer does not cover, Anthropic lets Claude fall back to raw SQL. If you have a technical team that understands the pitfalls and knows where to be careful with an answer, that is fine. But if your users are business people with limited technical know-how, or the number is headed for the CFO or board-level reporting, there is no room for a number that is subtly off. That is not a trade-off you are willing to make. Veezoo never writes SQL to answer a business user's question. It reasons over Knowledge Graph concepts and emits a query against the graph, which then compiles deterministically into SQL. And if the question cannot be answered with the Knowledge Graph, Veezoo says so, flagging it to both the business user and the Knowledge Graph admin. 🤝  I'm happy to see Anthropic arrive at an architecture so close to Veezoo's, while many companies and vendors still believe text-to-SQL, without a knowledge graph, is the way to go. And yes - you can also query the Veezoo Knowledge Graph using Claude 🙂! Source: How Anthropic enables self-service data analytics with Claude https://lnkd.in/euG2D99s | 20 comments on LinkedIn

There is an ongoing debate among semantic technology practitioners about when to use the Web Ontology Language (OWL) versus the Shapes Constraint Language (SHACL). Many people argue that OWL should be used for defining ontologies (classes, properties, logical axioms) while SHACL should be limited to

Kurt Cagle Charles Betz of Forrester Research published a piece in April titled "Context Graphs Are a Convergence, Not an Invention", and it deserves to be read widely. Not because it says something entirely new — those of us who have spent careers working at the intersection of knowledge graphs, en

Epistemic Decay: The Fourth Axis of Agent-Memory Architecture ⚙️ Graph retrieval ranks nodes by connectivity. Connectivity grows with age. So the best-connected node in your knowledge graph is often the one extracted two reorgs ago. That is a staleness bias built into the retrieval math. The diagnosis splits temporality into two different problems. 1️⃣ Event-driven invalidation is the solved half. Graphiti (26.9K stars, the memory layer behind Zep) stores bi-temporal validity windows. Memanto writes supersession links at admission, so each record knows what replaced it. APEX-MEM (Amazon AGI, ACL 2026) keeps an append-only store and resolves conflicts at query time with timestamp reasoning, beating resolved-store baselines by 11.6 points on LongMemEval. 2️⃣ Eventless decay is the unsolved half. Nothing replaced the fact. No supersession event fired. The domain just moved. A triple about territorial control decays in days; a triple about geology decays in centuries. Confidence here is a function of time, domain volatility, and source reliability, and no production system computes it along a traversal or returns it from the query API. The failure hides because verification checks the wrong axis. A hallucinated fact fails the audit. A stale fact passes it. Provenance confirms where a belief came from. Currency asks whether it is still alive. Most GraphRAG pipelines run the first check and skip the second, so the error lands on the model's reputation instead of the retrieval layer. And the bias compounds: the older the graph, the more centrality and staleness correlate. The maturity ladder, today: static certainty scalars per record (cheap, shipped). Validity windows plus supersession edges (production-grade). Query-time temporal reasoning over an append-only log (the benchmark champion). Stored decay functions parameterized per ontology class (experimental: one practitioner reports half-life decay with vitality pulses modulated by edge conductance). The open question sits on the last rung. A half-life per edge class is domain judgment, the same expert call as the ontology itself. A mis-set half-life silently demotes true facts, a failure harder to detect than the staleness it was meant to cure.

An interactive intelligent textbook on context graphs — enterprise graph data structures that get the right content into an LLM prompt in the fewest tokens.

An ontology is a blueprint of what matters to you. The data should not dictate the ontology, unless the data exactly reflects your focus. In this sense, most ontology designs start in the wrong place. You look at your documents, your database schemas, your existing data and you build an ontology that reflects what's there. It feels empirical and safe, but often it's also backwards. An ontology is not a description of your data, rather a statement of what you care about. It defines the concepts, relationships, and distinctions that are meaningful to you not the ones that happen to be easiest to extract. If you run a pharmaceutical company, your ontology should encode what matters to drug development: targets, mechanisms, adverse events, trial phases, regulatory pathways. Not whatever noun phrases your extraction pipeline surfaced most frequently. If you run a law firm, your ontology should reflect the structure of legal reasoning (jurisdiction, obligation, precedent, party) not the entity types your extraction model was trained on. Data follows intent. Ontology precedes data. I think people go from data to blueprint because it's easier to let engineering deal with ontologies, letting the data talk via extraction. On the other hand, defining what matters as a company (division) entails meetings, business goals, difficult questions, sometimes difficult people. Ontology design is a knowledge-elicitation exercise, not a data-mining exercise. The right people in the room are domain experts and decision-makers, not data engineers (or at least not only engineers). #Ontology #RDF #Semantics #KnowledgeGraphs | 12 comments on LinkedIn

Large Language Models (LLMs) have become great at sounding right — and frustratingly average at being right when precise, multi-hop facts…

Ok, Back to work.