Tabular data is everywhere—from research datasets to business reports—but unlocking its full potential requires linking it to structured knowledge sources. That’s where s-elBat, developed by UNIMIB and Sintef as part of the enRichMyData toolbox, comes in. Designed to automate Semantic Table Interpretation (STI) tasks, s-elBat enables seamless entity linking, reconciliation, and tabular data transformation into knowledge graphs (KGs).
What is s-elBat?
s-elBat is a comprehensive STI framework with advanced algorithms for automatically annotating tabular data. By leveraging reference Knowledge Graphs (KGs) such as Wikidata and DBpedia, it enables users to:
Link table values to entities in KGs (entity linking)
Discover schema-level relationships between data points
Transform tabular data into graph format, ensuring compliance with predefined ontologies
Enrich datasets with additional knowledge from external sources
Alligator: A Next-Generation Reconciliation Service
One of s-elBat’s standout features is Alligator, a newly developed reconciliation service that follows the W3C reconciliation API standard. This ensures smooth interoperability with third-party tools while offering:
🔹 Customizable entity linking– Users can specify which columns provide context for better matching.
🔹 High performance – Alligator achieves top-tier scores on benchmark datasets, even outperforming LLM-based models like TableLlama & GPToin certain cases.
🔹 Stand-alone usage – Available as an independent tool for diverse applications.
Recent Updates & Advancements
The latest developments in s-elBat focus on enhancing entity linking accuracy and improving interoperability with large-scale knowledge bases:
✅ Extensive benchmarking against LM- & LLM-based models (TURL, TableLlama, GPTo).
✅ Integration of new, mostly open-source entity linking algorithms to improve accuracy and flexibility.
✅ Optimization for Wikidata, ensuring compatibility with the most widely used encyclopedic knowledge base.
Why Use s-elBat?
End-to-end STI solution– Covers the full spectrum of table annotation and enrichment.
High precision & automation– Reduces manual effort while improving accuracy.
Seamless transformation to Knowledge Graphs– Converts structured data into graph-based formats for advanced analytics.
Fully open-source – Users can download, host, and run the tool independently.
Learn More & Try It Out!
s-elBat is designed for researchers, data scientists, and organizations looking to bridge the gap between tabular data and structured knowledge.
Want to explore Semantic Table Interpretation in depth? Check out our tutorial at ISWC 2024 for expert insights and hands-on learning!
https://unimib-datai.github.io/sti-website/tutorial/
Download s-elBat today and enhance your data enrichment workflows!