Research
At NeverBlink, we’re advancing neurosymbolic technologies toward real-world adoption by centering our research on scalability, reliability, and seamless integration. Discover our latest work and see how we’re bringing real-time reasoning into production.
LinkML-Scala: a Robust, Fast, and Portable Implementation of LinkML
Piotr Sowinski, Kacper Grzymkowski, Andriy Plokhotnyuk
We present LinkML-Scala, a fast, portable (JS, JVM, native) of the LinkML cross-language modeling framework that outperforms the Python implementation by 22.9–38.5x.
A Robust Decentralized Infrastructure for Trust-Aware Open Knowledge Sharing
Tobias Kuhn, Virginia Balseiro, Ashley Caselli, Ziroli Plutschow, Piotr Sowiński, Anastasiya Danilenka
Accepted as a research paper at the International Semantic Web Conference (ISWC) 2026
Led by Knowledge Pixels, this paper presents a decentralized, global ecosystem based on nanopublications and a novel trust propagation algorithm (IDEBT) to reliably publish and query semantically structured knowledge for both humans and AI agents.
It’s Time to Standardize RDF Messages
Pieter Colpaert, Piotr Sowiński
We propose the concept of an RDF Message as an RDF Dataset intended to be interpreted atomically as a single communicative act, laying the foundation for defining RDF Message Streams and RDF Message Logs.
Jelly-Patch: a Fast Format for Recording Changes in RDF Datasets
Piotr Sowiński, Kacper Grzymkowski, Anastasiya Danilenka
Presented as poster paper at the 24th International Semantic Web Conference (ISWC) 2025 – see poster
We introduce Jelly-Patch, a binary format for recording RDF dataset changes that achieves far better compression than existing patch formats.
Tackling inter-service RDF communication bottlenecks in the Nanopublication network with Jelly
Piotr Sowiński, Tobias Kuhn, Karolina Bogacka
Led by Knowledge Pixels, the Nanopublication Network overcame scalability issues with Jelly, a binary RDF format that cut retrieval times from over an hour to seconds. Nominated for Best Industry Paper at SEMANTiCS 2025 Vienna. Read full case study.
Jelly: a Fast and Convenient RDF Serialization Format
Piotr Sowiński, Karolina Bogacka, Anastasiya Danilenka, Nikita Kozlov
We introduce Jelly, a binary RDF format that supports batch and streaming use cases, offering high throughput, strong compression, and easy integration for the Semantic Web.
Representing and querying data tensors in RDF and SPARQL
Piotr Marciniak, Piotr Sowiński, Maria Ganzha
We present a method for representing tensors as RDF literals, supported by a SPARQL extension and open-source Jena implementation to integrate machine learning with knowledge graphs. Awarded with Best Poster Award and Dieter Fensel Visionary Award at ESWC 2025.