GraphNews

The Memory Graph Layer for Agents

Finnish schools send their middle school students to companies for a week-long internship to learn about working life. I did my internship…

Joe Reis created a phenomenal visualization for his recently published 2026 Practical Data Community State of Data Engineering report. It included links to the raw CSV data and SQL queries associated with various charts. That simple act — something the Linked Open Data community ethos has always been about — enabled me to build on that work by adding a Knowledge Graph, Ontology, SPARQL, and SPASQL (SPARQL inside SQL)–based rendition using a live Virtuoso database instance for high-performance access, connectivity, and reuse. See comments for links. #CDO #CDIO #CAIO #CIO #CTO #CxO #CMO #SemanticLayer #ContextGraph #KnowledgeGraphs #SemanticWeb #LinkedData #DataVisualization #SPARQL #SQL #SPASQL #VirtuosoRDBMS #HowTo #CSV #PracticalDataSurvey

Agent control is a graph composition problem. Each layer answers a necessary control question. None is sufficient alone. Composing across them is hard.

Connecting the layers of Semantic Architecture

What would it take to finally connect data and meaning?

What "semantics" actually is and what we should focus on

Solution-level siloes and object-level descriptions aren't cutting it

An Introduction to Semantic Model-Driven AI

Knwler — turn documents into a structured knowledge graph, in minutes. We built Knwler to solve a problem every compliance team, legal department, and research analyst knows too well: extracting structured insight from dense, complex documents takes a lot of manual work.

Geospatial data analysis systems are currently very relevant. Most such systems use either relational databases or graph databases. This paper presents the idea of using both approaches, taking into account the main features and advantages of each. A concrete example of a city transport network is used to experimentally examine the use of this hybrid approach. A special ETL procedure was developed to transform data from the corresponding graph database to a relational one, as well as the reverse process from the relational to the graph database. The results show which type of queries are better suited for relational databases, and which for graph databases. Additionally, for certain specific queries and applications, neither database type is capable of providing any results. Although this kind of hybrid architecture has issues with data duplication, the performance gains achieved are significant, making this approach highly efficient.

I am an AI realist, not a skeptic.

Bridging Semantics and Syntax with Annotation-Driven Generation

How an ontology-based knowledge graph boosts the multi-hop Q&A accuracy of one of the leading schemaless GraphRAG systems

Trust has stood out more than ever in the light of recent innovations. Some examples are advances in artificial intelligence that make machines more and more humanlike, and the introduction of...

I ran 30 context experiments to understand which piece of context has the most impact on analytics agents performance

A Graph Practitioner’s Perspective

From Property Graphs to OWL Inference Using a Hybrid Pipeline

Here is a very simple example of graph with AGE PostgreSQL extension. In this example, we will assume we are trying to match a donor who can potentially donate to a patient based on the matching blood group.

Why copying data for graph analytics doesn’t scale, and what you can do instead

2025 was a big year for Spanner on Google Cloud, with new capabilities to help it evolve from a passive data repository to an intelligent context hub.

The integration patterns that make LLMs more reliable and controllable

AI agents act under your authority. If you can’t trace their reasoning or enforce boundaries, you don’t have governance. You have risk. Here’s the evidence.

Misinformation spreads quickly—but fact-checking can happen just as fast.

High-performance GraphRAG inspired from LightRag written in Rust - raphaelmansuy/edgequake

Part 2 — Generative Knowledge Operators

How Terminology Pollution Is Killing Graph Architecture And Where GraphQL Actually Belongs

Thursday was eventful - I had 37 phone calls from recruiters looking for ontologists. One of them sent a questionnaire that I thought was brilliant, and I am reproducing here, because if you are an ontologist, you should probably be able to address these issues: 1. Have you designed or implemented a Knowledge Graph in an enterprise environment? (Yes + can describe use case = proceed) 2. Which semantic technologies have you worked with hands-on (RDF, OWL, SPARQL, SHACL)? (Should name at least 2 confidently) 3. Was your work more architecture-focused, hands-on, or hybrid? (Looking for senior ownership) Semantic Depth 4. Have you designed or managed ontologies or semantic models tied to business domains? 5. How did your Knowledge Graph integrate with relational or warehouse data? AI / RAG / KAG Exposure 6. Have you worked on platforms supporting LLMs, RAG, or AI systems using enterprise data? 7. Do you understand the difference between RAG and Knowledge-Augmented Generation (KAG)? (Conceptual understanding is acceptable) 8. Have you integrated Knowledge Graphs with embeddings or vector search? Enterprise & Governance 9. How did governance, lineage, or trust factor into your semantic or AI architecture? 10. Have you worked in regulated or large enterprise environments? Platform Alignment 11. What cloud platforms have you worked on? Any hands-on GCP data services? 12. Have you embedded semantic logic into data pipelines or ingestion flows? Seniority & Communication 13. Have you defined standards or reference architectures used by multiple teams? 14. How do you explain Knowledge Graphs to business stakeholders? | 42 comments on LinkedIn