semanticClimate at the UK Knowledge Graphs Symposium 2026
About the event
-
Date: Friday, November 13, 2026
-
Location: City St George's University of London
-
Format: In person
The #semanticClimate team will participate in the UK Knowledge Graphs Symposium 2026, bringing together researchers, practitioners, and innovators working with knowledge graphs and their applications across science, society, and industry.
The symposium provides an opportunity to share how semantic technologies and knowledge graphs can help make scientific knowledge more accessible, connected, and reusable.
The #semanticClimate team will showcase its work on knowledge extraction from climate literature and the development of knowledge graphs for major climate assessment reports.
|
Poster
Rapid Scoping Review of Climate Literature using Encyclopedia and KGs
-
Presenting Author: Shweata Hegde
-
Co-Authors: Renu Kumari, Larisa Beni, Neel Kumar, Aleena Harold Peter, Anusha, Peter Murray-Rust, Simon Worthington
-
Abstract:
semanticClimate is a global online community of interns and volunteers creating Trustable semantic knowledge on climate Change. How do they explore a new subject - "El Nino in Latin America", "climate justice"...? They face complex language and acronyms ( ENSO, SST, Southern Oscillation, climate justice...) and many different tools.
We have created a powerful toolkit (AUKGET - Automatic Universal KnowledgeGraph Encyclopedia Toolkit - Python Apache2). A single automatic pipeline command (for any subject) will:
- query the recent scientific literature for a subject from OR load a corpus (e.g. UN-IPCC)
- create and clean a corpus (BAGIT)
- extract keyphrases (KBIR)
- build an encyclopedia based on Wikipedia entries for phrases
- find inter-entry relations through bibliography and Wikimedia proximity
- create and display a knowledgeGraph (corpus and encyclopedia tightly linked)
- remove false positives (visualising and analysing outlying clusters in the KG is particularly useful)
AUKGET is available as a colab script that can be used by highschool students or citizen groups or researchers. It is deliberately AI-free but the corpus is also usable as input to AI, e.g. an LLM RAG chatbot.
A Knowledge Graph for the IPCC’s Sixth Assessment Report (AR6)
-
Presenting Author: Simon Worthington
-
Abstract:
Problem: A text mountain
The IPCC's Sixth Assessment Report (AR6) is the world's most authoritative climate science summary, aiming to inform global policy through history's most comprehensive literature review. However, the 10,000 page report is not machine-readable.
Breaking down the problem
Climate Knowledge Graph (CKG) has created an AR6 knowledge graph that breaks down the corpus structure into an entity-relationship model (ER model). This allows data scientists to add semantics data to describe climate science in AR6 making it machine-readable.
AR6 ER model (KISS)
AR6 ER model has 5 datasets: Structure, bibliographic, authors, glossary, and acronyms. Quantifying AR6, it has: 7,524,958 words, 2,153 image files, 932 authors, 1,274 glossary terms, 1,910 acronyms, and cites 48,400 bibliographic references and 66,834 datasets.
KG construction
A knowledge graph (KG) has been built to hold AR6’s data and full text. It is available online as Open-Access and FAIR Data. The KG is made with Wikibase and uses MediaWiki for report browsing and has SPARQL and API interfaces.
Future work: More-is-more!
Extending the KG for figures, references, datasets, and granular edge-metadata.
Program Schedule : Click Here
← Back