Semantic Data Europe 2025

First Sessions Announced

We are busy developing an exciting agenda for Semantic Data: Taxonomy, Ontology, and Knowledge Graphs so keep your eyes peeled for details. 

Sign up for our mailing list to be the first to hear about updates on the agenda and speakers. 

If you are interested in speaking at this conference, please contact Feyisayo Borisade at FeyisayoB@henrystewart.co.uk

In the meantime, check out the first 5 sessions confirmed:

Delivering Trustworthy Answers 

Integrating Conversational Agents and Knowledge Graphs Within the Scholarly Domain

The Problem:

Large language models (LLMs) have revolutionised question answering, including within the scientific domain. However, scientific question-answering remains significantly challenging for the current generation of LLMs due to their reliance on highly specialised concepts.  

A promising solution: 

Integrate LLMs with knowledge graphs of research concepts (also known as Scientific Knowledge Graphs), ensuring that responses are grounded in structured and verifiable information. 

• One approach consists of retrieving a relevant portion of the knowledge graph and providing it as input for the LLM. In this way, the LLM will use the structure knowledge of the graph to provide more accurate responses. 

• An alternative, and more effective, strategy uses LLMs to translate questions posed in natural language into SPARQL queries, enabling the retrieval of relevant data. 

Reporting the state-of-the-art:

• Effective LLM-KG integration strategies

• Approaches to scientific question-answering

• Insights from Knowledge Media Institute’s research 

• Achievable robustness in scientific (and other highly technical) question-answering. 

• Current challenges and future directions for the field.

Angelo Salatino, Research Fellow, Knowledge Media Institute, The Open University

Building Knowledge Graphs to Support Answering Customer Questions

The objective: to build a robust knowledge graph that supports a large-scale customer facing question-answering system. 

Context: Empirical experience from the presenter’s work supporting question-answering functionality in virtual assistants (notably Amazon Alexa) and, more recent, work at Compare the Market. 

Covering: The talk will cover common pitfalls and proven solutions, including insights into what works - and what doesn’t - when deploying and scaling knowledge graphs in real-world environments.

• Use Case Design: What to consider when tailoring the graph to different domains, user profiles, machine-learning applications, and answer formats ranging from plain text and tables to graphical responses.

• Operationalising Development: Establishing high-level design principles, managing iterative development, addressing distributed ownership, handling change management and testing, and ensuring the system adapts to evolving data sources.

• Technical Challenges: We will explore trade-offs in accuracy, coverage, latency, expressivity and abstraction, and consider other potential requirements such as context-handling, semi-structured data and permissions levels. 

• Internationalisation: Strategies for scaling globally including language support, units of measurement, and localised content.

Attendees will come away with a clear framework for designing knowledge graphs that are both scalable and adaptable, grounded in practical industry experience.

Beth Holmes, Senior Manager Knowledge Graph, Compare the Market

Recipes for Factories

How Developing Pharmaceutical Manufacturing Processes is like Writing a Cookbook

Building a vendor agnostic, production-ready ontology

The requirement

The recipes used in pharmaceutical manufacturing (CMC) need to be developed and documented in a way that allows an identical product to be manufactured in different locations and potentially at different scales without sacrificing quality. 

For years this has been supported by the ISA-88 standard, but a lot has changed in the thirty years since it was first published. A more flexible, comprehensive representation of the recipes is needed. 

Our response to the need for change

The Pistoia Alliance along with our member companies (MSD, Amgen, GSK, Eli Lilly, AstraZeneca and J&J) and implementation partners (CrownPoint & ZS) is well on the way to building a vendor agnostic, production-ready ontology for Pharmaceutical CMC (Chemistry, Manufacturing and Control) Processes supporting today’s data science and exchange needs. 

Our completed Phase 1 has shown that a contemporary ontology is capable of capturing a more precise view of the processes used in pharmaceutical development and is able to facilitate data integration, exchange (tech transfer) and data insights. Specifically, we have shown that the ontology is capable of describing recipes and capturing data associated with runs allowing comparative analysis of different sites, equipment and scales (from a consumer stove baking a single cake to a commercial oven producing thousands of cakes for distribution). 

Phase 2 is now well underway and is extending the ontology beyond simple small and large molecules, ensuring both extensibility and interoperability with related ontologies, and defining a governance structure for future sustainability.

This briefing

Taking ontology to mean a structured framework that defines the concepts, entities, and relationships within a specific domain, enabling machines to understand and process information in a human-like manner the briefing covers:

• What we have learned so far.

• Challenges that lie ahead.

• Lessons of general application for developers of complex ontologies.  

Christian Baber, Chief Portfolio Officer, The Pistoia Alliance

Data Standards Interoperability and Governance - From putting the data into a graph to putting the graph into the data

Like many organizations AstraZeneca R&D faces the challenge of siloed data. We see adopting Findable, Accessible, Interoperable, Re-usable (FAIR) Data principles as a route to releasing value from our existing data as well as setting us up to be able to do so much more with new data we generate from here on. Semantic knowledge graphs are a proven approach to achieving this and we started by building Scientific Intelligence, a knowledge graph to support exploration and analysis of clinical data. 

Traditionally building knowledge graphs has focused on combining inconsistent, siloed data from multiple sources into a common data model, the graph. This requires developing a data model consisting of entities, the relationships between them and the attributes that describe them. Having created the model data from sources is then mapped into it and transformations created to align the data to a common set of standards. This is a non-trivial exercise often made harder by the lack of metadata to describe either the source or the individual fields within. As a result a huge amount of effort is spent understanding and fixing other people's data. As a consequence, capacity to scale the breadth and depth of the knowledge graph rapidly becomes limited. The Scientific Intelligence knowledge graph was adding significant business value but continuing to grow it would require more semantic engineering resources. Given the shortage of this skill set we had to think differently. The challenge was clear, how do we get everyone else to fix their data so we could focus on the semantics and the knowledge graph. The answer was to turn the problem on its head and move from putting the data into the graph to putting the graph into the data. This pivot now forms the core of data standards, interoperability and governance strategy for AstraZeneca R&D. In this presentation I will discuss this journey, the lessons learnt and describe the simple pragmatic services we have put in place to enable easy adoption and compliance.

Ben Gardner, Data Standards, Interoperability and Governance, Data Office, Data Science & Artificial Intelligence, R&D, AstraZeneca, Cambridge, United Kingdom