This workshop is a detailed introduction to what taxonomies are (and are not), why they are useful, and covers everything needed to develop a successful one, from standards, planning and design, populating with concepts, to business implementation, governance, and maintenance.

Following a thorough introduction to taxonomies, their types, and how they relate to other controlled vocabularies, Hedden focuses on methods and best practices for designing and building taxonomies, whether for website/marketing content, technical content, or internal enterprise content. These methods include stakeholder interviews, content analysis, text extraction, and selective use of generative AI in concept research. Best practices include following standards for hierarchical relationships, using suitable labels for concepts, and using the most suitable taxonomy structures. The workshop includes interactive exercises of hierarchical relationship and alternative label creation.

The workshop is suitable for a wide audience, including those interested in getting into taxonomy work, as well as those currently managing taxonomy projects or collaborating with a taxonomist or taxonomy consultant.

Enterprise taxonomies are, by their nature, cross-organisational, but they still need a home base within an organisation. This workshop provides a solid grounding in different types of ownership models, including centralised and federated structures, governance committees, and hybrid approaches that balance authority with flexibility.

Drawing on real-world implementations across large public and nonprofit institutions, the workshop deep dives into how different ownership models influence priorities, funding, and long-term sustainability. It examines how the most common places for a taxonomy function to end up—for example IT, knowledge management, or communications—can enable or constrain successful governance efforts, along with the pros and cons of different approaches.

Participants will leave with frameworks and guiding questions to help them clarify and fit responsibility in their own organisations, enabling them to collaborate with confidence across different teams and with other digital and data governance structures.

The workshop also gives practical advice on establishing cross-functional governance structures that ensure continuity beyond individuals, teams, or tools and also covers ways to mitigate common pitfalls, including department reorganisation, staff turnover, and budget prioritisation.

The rise of AI has sparked intense speculation, but the evidence shows there has never been a better time to invest in knowledge management as a career or a company. Current generative AI frameworks suffer from "hallucinatory limitations" that can only be fixed through well-structured, human-guided knowledge and high-quality information. Join Ben Clinch, CDO & Partner at Ortecha, for this engaging and thought-provoking keynote as he demonstrates why KM and metadata are the cornerstone of the next wave of advanced systems, specifically Neuro-Symbolic AI—the hottest topic in the field. Drawing on extensive research, Clinch will share the opportunities and risks facing organisations and provide clear navigation strategies to bring increased value to your company and the industry. He will show why mastering knowledge management is the only way for AI to live up to its hype.

Check back soon for details.

AI hallucinations pose serious risks to organisations, including compliance issues, reputational damage, and poor decision making. In this session, Gower explains why hallucinations occur and how strong knowledge management can reduce them. Attendees will learn how AI behaves in real-world settings, why accuracy breaks down, and how governed, high-quality knowledge transforms AI into reliable, enterprise-grade intelligence.

Dave Clarke explores new use cases that emerge when LLM-based classification and information extraction are combined with taxonomies, ontologies, and AI. Taxonomies and ontologies define decision points, criteria, triggers, dependencies, and outcomes for business processes. AI agents can reference this structured knowledge to execute processes consistently. Agentic ontologies are process-oriented semantic graphs that enable AI agents to perceive, reason, decide, and act using enterprise domain knowledge, managing complex interactions with humans and other systems.

An essential session for any taxonomist who is working inside an organisation and facing one—or both—of the challenges of seeking to ensure their expertise is valued, and working with legacy systems, governance, and data. While these challenges can block progress, in this session. Our presenters bring their wealth of experience to offer tips and inspiration to help knowledge workers thrive.

The core of the taxonomy discipline lies in organising a vast number of things (such as documents, products, data, or people), and in recent years, the concept of the semantic layer or layers as a way of managing the connections between things has gained traction. Kasenchak cuts through the buzzwords to give a highly accessible introduction to the idea and how it can be used to help organisations find, manipulate, and understand complex data.

This session offers two perspectives on the human-powered value of good design in information systems. Jenkins presents a nontechnical overview of the decision points that might arise in designing a taxonomy, including structural principles, types of properties, and interoperability considerations. Stephenson tracks common information architecture threads across many recent trends in tech and AI and makes the clarion call for the value of IA in delivering flexible systems.

Language is at the heart of designing semantic systems of any kind, and this session offers two perspectives on ensuring that what goes into taxonomies is correct, useful, and manageable. Lehnert talks about ways to identify the “best” concepts out of those being suggested by machine learning or large language model (LLMs), ensuring the most enduring and relevant concepts are favoured instead of ephemeral ones. Chapman’s wealth of career expertise informs her view of the natural affinity between taxonomy and term bases and the commonalities and potential of both to drive consistency and efficiency in using terminology.

Metadata and taxonomy have always been the building blocks of successful information management and have enjoyed renewed prominence with the rise of AI and retrieval-augmented generation for search. Seasoned experts Levenson and Talisman show how the idea of “Minimum Viable Metadata” can be used to bring meaning and context to semantic data, while Hill describes how existing taxonomies can be enriched to support advanced AI implementations.

One key use case for taxonomies is in enabling data interoperability, where consistent language and identifiers are essential. Hollidge presents an in-depth case study of how taxonomies were deployed to join up previously separate content and data repositories across the United Nations. Reiz describes the standards, including SKOS, RDF that enables taxonomies to connect concepts, and data points across disparate systems.

This session is all about pushing the boundaries of what we can do with structured knowledge. Petkova presents a vital but sometimes-overlooked topic: how to embed meaning and consistency in content to support rich knowledge graphs and shared understanding. McComb uses the example of the open source “gist” ontology to show how ontologies and taxonomies are the bedrock of enterprise data and knowledge.

In recent years, graph-based Retrieval Augmented Generation (RAG) has emerged as a strong and practical use case for AI and semantic technologies. Complementing this is the ever-present need to ensure that search works for user needs and language. Rodriguez presents a deep dive into Elsevier’s work to use RAG to drive higher-quality retrieval of content in the health domain, while Wright presents an overview of creating taxonomies to drive great search experiences. For anyone interested in managing a search application, this session brings together these approaches.

AI depends on data patterns, but knowledge management depends on shared understanding and meaning. In most organisations, meaning is buried in unstructured formats, legacy silos, and inconsistent tagging or in traditional databases, where true data understanding or flexibility is limited by rigid, incomplete structures. In this keynote, Urbina shows how to design scalable knowledge architectures, based on proven technologies and methodologies, that prepare knowledge for the future of combined human consumption and AI-driven discovery and leverage. When the “shape” of information matches the way people think and work, everything flows better: engineers resolve issues faster, support teams get clearer insights, and AI systems behave in ways that feel explainable rather than opaque.

A taxonomy is only part of the toolbox for managing data. Combining the worlds of ontologies and thesauri to show the value of both, Appleby and Singh give an overview of the SKOS and SKOS-XL standards and how they can be used to enrich taxonomies. Hedden explains the ISO 25964-1 international standard for thesauri in relation to its alignment with SKOS and how it can be used to drive high-quality taxonomies.

Amidst the promise of AI technologies, there are many other social dimensions to consider. As the head of CILIP, Coiffait-Gunn is well-placed to support the role of the information management profession in pushing for AI applications that are safe, legal, secure, transparent, accurate, useful, and ultimately ethical, and his presentation will help that ongoing narrative. In a similar vein, Xemma is developing new taxonomy management software and has learned much about the challenges, trends and features that matter to taxonomists. The company is now keen to share its findings with the community.

This session unites two of the most prominent voices in the field of structured content management to provide essential insights for ensuring content is fit for purpose for both humans and AI. Bailie covers the groundwork needed to remediate content and improve its usability and quality, and thus realise the investments that organisations are making in AI and automation. Busch gives an overview of the two types of vocabulary—schema and value—that can be used to develop effective content architectures for content delivery and search.

Case studies from two very different organisations that both wanted to maximise the quality and usefulness of their data and content. Horan and Lapin’s case study from online marketplace Vinted describes their project to integrate external data standards to enrich incomplete and unstructured data submitted by sellers in order to maximise findability and usefulness for buyers. Alexander gives a comprehensive overview of how he managed Cancer Research UK’s content taxonomy through a content migration project involving 10,000 pages.

Far from negating the need for taxonomies and controlled language, AI has in fact led to even more use cases for the work we have always done, especially in the area of validation and fact-checking. Maddison’s candid case study from Adobe shows how the raw material of taxonomies was fed into large language model prompts to support AI tools for a range of business workflow applications. Reamy describes a case study from the US Department of Transportation, in which a taxonomy that was not optimised for the department’s content was turned around. A combination of human review, automatic tagging, and human-powered review of that tagging was used to ensure the taxonomy delivered on its potential.

Two case studies from the recruitment and legal sectors respectively; are presented: two domains with large volumes of data, metadata, and language to manage and exploit. Bryant showcases how taxonomy, user experience, engineering, and data science teams worked together to build a truly semantic infrastructure for process automation, data discoverability, and innovation. Oleynik talks about the taxonomy underpinning the legal research tool rangefindr, which helps lawyers and judges assess criminal penalties. Both talks are relevant to anyone who works with complex data and diverse user needs.

Reference data and linked data are not new ideas but, with the increased adoption of ontologies and AI, they are once again in the spotlight. The EU Publications Office exemplifies best practices for how to manage and disseminate reference data in complex systems, and Dechandon and Gerencsér outline the work done to build an ontology to support multilingual reference data. Kent describes how the British Library combined open source mapping data and an ontology to improve the quality of AI-powered search results.

It is important to choose the right tool for whatever kind of semantic project you might undertake. The Arches Lingo platform was developed for the cultural heritage field by the Getty Conservation Institute and World Monuments Fund, and Carlisle and Wuthrich give an overview of its features and some of the use cases it handles. Graphwise was born from the merger of Ontotext and Semantic Web Company, two established names in the semantic space. Nagy describes how they are building their own knowledge graph from the ground up and offers tips on identifying potential applications and measuring quality.

This session presents two perspectives on large language models (LLMs). Hallucinations from AI-generated text have been a source of media interest and even ridicule but, for anyone using LLMs inside their organisation, Alexander argues that there is an opportunity for knowledge workers to reposition themselves as custodians of good judgement and factual context. LLMs do of course have great potential to enrich taxonomies too, and Mitzias shows how AI-driven workflows can achieve this by surfacing latent knowledge, uncovering conceptual patterns, and iteratively refining taxonomies, all the while enabling high-quality human oversight.

This session presents detailed overviews on building out semantic layers to structure knowledge across systems. Bainey shows how using the right mix of ontologies, AI, and graphs can unlock siloed knowledge and drive adaptive systems. Kopytko introduces the concept of the semantic hypergraph, which is a memory and reasoning layer for AI. The hypergraph draws on taxonomies and the like and creates an adaptable layer which acts as a “cognitive substrate” that is capable of extending, querying, and explaining information by itself.