Thomas Bergwinkl

@bergi
268 Followers
399 Following
128 Posts

I like wombats and trees. Creator of #rdfext and many other #rdfjs libraries.

#3dprinting #ai #arduino #iot #javascript #linkeddata #rdf #semweb

Homepagehttps://www.bergnet.org/
GitHubhttps://github.com/bergos/
Most data comes from closed-world systems — forms, sensors, logs — with known schemas. Yet the RDF community often says: adopt the full semantic web stack or nothing.
My latest post explores practical RDF for closed-world systems:
- When closed-world makes sense
- Real examples: sensors, IoT, forms
- How Netflix uses RDF+SHACL at scale
- New SHACL 1.2 features
Read more: https://www.bergnet.org/2026/03/closed-world-systems/
#RDF #SHACL #JSON-LD #Ontop
Bringing Semantics to Closed-World Systems

The origin of most data in the world comes from closed-world systems. Forms, logging systems, and sensors all generate data in predefined structures where the schema is known in advance. Yet when peop

bergis universe of software, hardware and ideas
LLMs don't need your secrets.
A proposal for extending MCP to share sensitive data safely between servers — enabling agentic workflows without leaking context.
Beyond Agentic Commerce: Sharing Secrets Safely in MCP 🛡️
https://www.bergnet.org/2025/10/mcp-secret/
Beyond Agentic Commerce: Sharing Secrets Safely in MCP

OpenAI has recently introduced the Agentic Commerce Protocol (ACP), which enables shopping and purchasing directly within ChatGPT. That is a step forward for agentic applications. Still, it only scrat

bergis universe of software, hardware and ideas
I added support for some experimental SHACL 1.2. features in shacl-engine. See my blog post https://www.bergnet.org/2025/03/shacl-engine-experimental/ for more details, and try it out on the SHACL Playground https://playground.rdf-ext.org/shacl-experimental/ #shacl
shacl-engine: Experimental SHACL 1.2 Features

The experimental branch of shacl-engine is now introducing features related to the upcoming SHACL 1.2 specifications, along with some proposals still under discussion. These additions are in an early

bergis universe of software, hardware and ideas

📢 Election results for the @tag! @w3c members elected @hadleybeeman, @marcosc, @csarven and Xiaocheng Hu. Congrats! 👏 👏

They join @tag's emeritus chair @timbl and continuing participants
@torgo, Matthew Atkinson, Dapeng (Max) Liu, @nitot, Martin Thomson and @jyasskin. @wz43rtx remains the staff contact.

The W3C Technical Architecture Group documents the Web's architecture and provides design reviews for Web-related specifications

Find out how to work with the @tag: https://tag.w3.org/workmode/

Working with the TAG

The W3C Technical Architecture Group (TAG) election for 2024 is happening now!

If you can vote, I highly recommend supporting Sarven! Read his nomination statement here: https://www.w3.org/2024/10/tag-nominations#sc

I’ve worked with Sarven in W3C groups and open-source projects—it’s always been a great experience. He listens to everyone, treats people fairly, and works hard to find good compromises.

Sarven knows a lot about W3C topics and has a strong background in RDF, crucial for the Web and enterprise market.

Nominations and Statements for W3C TAG 2024 Election

Many blog articles and LinkedIn posts claim that Knowledge Graphs and Language Models are a good match. However, it needs more than theoretical claims to convince people. That's why I wrote a blog post with easy-to-understand examples that will work in the next release of TopBraid EDG and only require a basic understanding of SPARQL, JS, and SHACL. If I haven't sparked your interest yet, let me tell you that everything will be explained based on examples with wombats!

https://www.bergnet.org/2024/09/llm-kg-wombat/

Language Models Meet Knowledge Graphs - A Powerful Duo (plus a Wombat)

Many blog articles and LinkedIn posts claim that Knowledge Graphs and Language Models are a good match and that potential future products that combine both worlds will be even more powerful. I work in

bergis universe of software, hardware and ideas

Ontop 5.2.0 is out! Among all, it features fine-grained access control through special SQL functions and new optimizations for dealing with denormalized data. A big thanks to all the contributors, especially new ones!

https://ontop-vkg.org/guide/releases.html#_5-2-0-august-16-2024

Release notes | Ontop

Thanks to everyone who gave feedback, reported bugs, and helped get this package to its current state. Special thanks to Ruben Taelman and the rest of the Comunica team!

Try it out on the playground, which runs 100% client-side in the browser [6]. Or, if you need something for the command line, try the rdf-ext-cli tool [7].

[6] https://playground.rdf-ext.org/shacl/
[7] https://github.com/rdf-ext/rdf-ext-cli

SPARQL is just one more unique feature in the JavaScript world of SHACL engines. The coverage report, which makes debugging much easier, is another feature exclusive to the shacl-engine. With a performance boost of factor 15-26x [4][5] compared to the alternative library, rdf-validate-shacl, the performance also becomes a standout feature.

[4] https://www.bergnet.org/2023/03/2023/shacl-engine/
[5] https://github.com/rdf-ext/shacl-engine/issues/12#issuecomment-1940875628

Implementing a 15x faster JavaScript SHACL Engine

One of my projects encountered a performance problem with the available SHACL engines. Besides, I needed to know which triples had been processed (coverage/fragments), so I implemented a new SHACL eng

bergis universe of software, hardware and ideas