Integrative Social-Ecological Knowledge Systems: Conceptual and Computational Foundations for Ecological Synthesis

28 Sept 2026, 12:00 → 1 Oct 2026, 15:30 Europe/Berlin

E010 (CIP)

Birgitta König-Ries (Heinz Nixdorf Chair for Distributed Information Systems), Robert Frühstückl, Tim Alamenciak (Carleton University & University of Waterloo), Tina Heger (TUM)

Ecological research increasingly relies on the integration of heterogeneous knowledge sources, including field observations, experimental results, theoretical models, and socio-economic data. However, relevant knowledge remains highly fragmented across disciplines, conceptual frameworks, and data infrastructures, which limits the capacity for ecological synthesis. Recent advances in large language models, semantic knowledge graphs, and agent-based AI systems make it technically feasible to formalize and computationally integrate ecological and social-ecological knowledge. Yet the conceptual, methodological, and epistemological foundations required for such integration remain largely unresolved. Without addressing these issues, AI-based synthesis risks producing scientifically unreliable or non-interpretable results. Initial progress on these issues has emerged from recent interdisciplinary collaborations, including a  research group at the Center for Interdisciplinary Research (ZiF) in Bielefeld , which brought together scholars from ecology, computer science, and philosophy of science, among them the organizers of this workshop. This work has identified key open questions:

·       What methodological limits and validation requirements arise when using AI-based approaches for ecological synthesis?

·       Which conceptual and semantic frameworks are suitable for representing ecological and social-ecological knowledge in  computational systems?

·       How can ecological and social science knowledge be integrated in interoperable data and knowledge models?

·       What epistemic standards are required to ensure transparency, interpretability, and scientific reliability in interdisciplinary knowledge integration?

Addressing these questions requires sustained dialogue between fields that rarely interact directly, including ecology, artificial intelligence, philosophy of science, and the social sciences. Existing disciplinary conferences rarely provide formats that allow such focused conceptual exchange. The workshop will therefore bring together an international group of researchers and structure an emerging research field on integrative social-ecological knowledge systems.

The primary scientific objectives are:

·       clarifying conceptual and epistemological foundations for social-ecological knowledge integration;

·       evaluating emerging computational approaches, including LLM-based and agentic AI systems;

·       identifying methodological and epistemological challenges in interdisciplinary ecological research;

·       developing priorities for future international research in integrative social-ecological knowledge systems.

 

We plan to publish the abstracts as well as documentation about the event in zenodo.

 

Abstract submission deadline was on  July 28, 2026.

 

Registration is now possible.

 

The registration fee will be €125 per in person participant and $ 50 for remote participation. It will be waived for student participants. 

 

We will be able to offer limited travel support for active participants (speakers and presenters of posters). Details on how to apply are  provided on the registration page. If you plan to apply for travel support, please do so by August 09 at the latest.

 

       supported by:

 

 

 

 

 

 

 

 

 

 

 

 

 

Monday 28 September

12:00 Registration and Lunch

1 Welcome to the Jena Workshop on Integrative Social-Ecological Knowledge Systems

The workshop organizing committee of Birgitta König-Ries, Robert Frühstückl, Tim Alamenciak, and Tina Heger will deliver introductory remarks on the proceedings of the week to come.

Speakers: Birgitta König-Ries (Heinz Nixdorf Chair for Distributed Information Systems), Robert Frühstückl, Tim Alamenciak (Carleton University & University of Waterloo), Tina Heger

Keynotes: About the relevance of transdisciplinary social-ecological biodiversity research: concepts and opportunities for knowledge integration

Keynote talks

2 Keynote: About the relevance of transdisciplinary social-ecological biodiversity research: concepts and opportunities for knowledge integration

Speaker: Prof. Marion Mehring (Das Institut für sozial-ökologische Forschung (ISOE) Frankfurt)

15:30 Coffee break

Paper talks: Session I: Place-based knowledge systems

Paper talks given as part of the conference proceedings, including four sessions: Place-based knowledge systems, What works where, Representing causality, and Machines in the loop

3 social–ecological networks for biodiversity and ecosystem services

The demand the human population is placing on the environment has triggered accelerated rates of biodiversity change and created trade-offs among the ecosystem services we depend upon. Decisions designed to reverse these trends require the best possible information obtained by monitoring ecological and social dimensions of change. Here, I present a social–ecological network framework and discuss the challenges of monitoring biodiversity and ecosystem services across spatial and temporal scales. I propose that spatially explicit multilayer and multiscale monitoring can help estimate the range of variability seen in social–ecological systems with varying levels of human modification across the landscape. To illustrate the framework, a conceptual case study on the ecosystem service of maple syrup production is used. I conclude that the use of analytical tools capable of integrating qualitative and quantitative knowledge of social–ecological systems (combining Essential Ecosystem Service Variables) provides a causal understanding of change across a network. Altogether, the SEN framework and EESV provide a foundation for establishing monitoring of socio-ecological systems.

4 Bridging Knowledge Systems for Adaptive Social–Ecological Governance: Synthesizing Scientific, Indigenous, and Local Knowledge in India’s Biodiversity Hotspots

This paper develops an integrative framework that combines scientific, Indigenous, and local knowledge to strengthen biodiversity conservation and adaptive governance in India’s Sundarbans and Western Ghats, two globally significant social–ecological systems facing climate vulnerability, biodiversity loss, and livelihood pressures. Although community-based conservation increasingly recognizes knowledge pluralism, practical approaches for synthesizing diverse knowledge systems remain limited. The framework builds on major contributions from social–ecological systems and sustainability science, including Berkes and Folke’s SES perspective, Cash et al.’s knowledge systems and boundary organizations, Folke et al.’s adaptive governance, Reed’s participatory approaches, Tengö et al.’s Multiple Evidence Base framework, Norström et al.’s principles of knowledge co-production, and Armitage et al.’s adaptive co-management scholarship. These perspectives emphasize resilience, participation, collaboration, social learning, and complementary knowledge systems. The framework integrates four components: participatory knowledge elicitation through interviews, mapping, workshops, and ethnography; ecological assessments of biodiversity and environmental change; boundary processes that facilitate dialogue among communities, researchers, governments, and civil society while respecting epistemological diversity; and adaptive governance mechanisms that translate synthesized knowledge into collective action through monitoring, feedback, experimentation, and learning. The framework is illustrated through the Sundarbans and Western Ghats, where Indigenous and local observations complement scientific assessments of climate adaptation, ecosystem resilience, biodiversity conservation, and traditional stewardship. Across both regions, knowledge co-production enhances legitimacy, trust, stakeholder participation, and adaptive governance.

5 A biocultural framework for categorising human-freshwater fauna interactions for social-ecological integration

Integrating social-ecological knowledge requires conceptual frameworks able to hold culturally-embedded, often intangible relationships between people and species alongside conventional ecological data, without flattening meaning or losing interpretability. Such knowledge remains scattered across disciplines that rarely share a vocabulary, even where the phenomena described overlap substantially. These relationships are multidirectional, multi-scale, and can be understood through the concept of biocultural diversity: that biological and cultural diversity are interrelated and co-evolved. This is particularly true in freshwater ecosystems, where intersecting human demands on water and biodiversity create an urgent need for social-ecological approaches to counter threatened biocultural diversity.
Drawing on existing relational frameworks and a review of 612 interdisciplinary sources, we developed a framework categorising biocultural interactions between people and freshwater fauna across seven interconnected domains from Consumptive Use, to Worldviews, Beliefs and Identities. We move away from binaries such as tangible/intangible, which segregate meaning from practice, and centre relationships rather than services, recognising these as multidirectional, reciprocal and not quantifiable.
The framework functions as a translating lens connecting species, the unit of natural science, to ethnospecies, the units by which communities conceptualise fauna. the units by which communities conceptualise fauna. This raises questions central to this session: what categorisation granularity is possible across epistemic contexts, how category overlap that reflects real-world plurality should be treated, and how relational, qualitative meaning can interface with more structured, quantitative data. The framework is a worked example of addressing relationships that resist quantification, a challenge future knowledge infrastructures must confront.

6 Contextualizing Ecological Synthesis: Indigenous Knowledge, Stewardship, and Environmental Archives

Ecological synthesis increasingly brings together diverse knowledge sources, yet synthesis can lose meaning when knowledge is separated from the contexts in which it is held, governed, and applied. This is especially important when Indigenous Knowledge is involved, where context also includes relationships to place, community priorities, consent, stewardship responsibilities, Indigenous Data Sovereignty, and decisions about what knowledge should be shared, represented, protected, or withheld.
This presentation asks how ecological synthesis can better account for these contexts. Drawing from examples of long-term environmental change across knowledge systems, I first consider how environmental observations become meaningful when linked to lived experience, reference points for environmental conditions over time, governance, and consequences for harvesting, access, safety, and stewardship. I then turn to early-stage PhD research being co-developed with Miawpukek First Nation at the Little River Estuary in Ktaqmkuk (Newfoundland, Canada).
Little River has been the focus of community-led Traditional Knowledge initiatives, including Traditional Use Study work, TEK mapping, marine inventory work, stewardship, and environmental monitoring. Building on this foundation, this presentation highlights the research design phase prior to data collection as an important part of how ecological synthesis is shaped. It considers how sedimentologic and environmental archives can be brought into conversation with Indigenous Knowledge and stewardship practices without treating community-held knowledge as extractable data.
By highlighting context before ecological synthesis, this presentation contributes to discussions in the broader literature on interdisciplinary environmental research and knowledge co-production about how multiple knowledge systems can be brought into conversation without collapsing distinct epistemologies into one another or treating one as validation for another.

Posters: Poster session and welcome reception

Posters being displayed as part of the workshop.

7 Adding context to network graphs to improve understanding of ecological theory and data

Network approaches are powerful tools to synthesize, connect, and visualise both ecological theory and data to improve understanding of real-world ecological systems and support management decisions. However, network approaches often lack context-dependence which limits their generalizability. I will present two case studies demonstrating how I have approached adding context to network graphs to increase the utility and interpretability of my work as a research ecologist. The first case study concerns a synthesis that categorizes and connects biological invasion hypotheses along a timeline of invasion to clarify conceptual links and contradictions between hypotheses. Separating hypotheses along the invasion timeline revealed that seemingly contradictory hypotheses can be complementary when considered in the appropriate invasion stage context. The second case study concerns the addition of non-trophic interactions and season to a boreal food web using interaction data from diverse sources. In this work, the additions of variation in interaction type and season created a more connected and holistic representation of boreal communities that cannot be captured by traditional food web methods alone. As complex and disparate types ecological data become increasingly available, context-rich network approaches provide a promising avenue for integrating diverse information and generating more ecologically realistic representations of ecological systems and processes.

8 Building Ecological Knowledge Graphs: Lessons from CAMO Annotation

Ecological knowledge is distributed across many formats, making findings difficult to compare and synthesize. The Causal Mosaic Schema (CAMO) is one approach to synthesize findings as structured data grounded in philosophical understanding. CAMO is being explored as part of EcoWeaver, an initiative to build a machine-readable knowledge base of ecological restoration findings that will allow users to examine causal relationships across hundreds of papers instantly.CAMO represents ecological findings as connected nodes and edges. A node represents an entity in a causal relationship through three components: an ecological entity, an attribute, and a change qualifier. Edges connect causal nodes by detailing the nature of the causal relationship described between them. They record evidential strength, supporting source passages, annotation confidence, and the geographic, temporal, and ecological context. Concepts are grounded in ontologies, including the Ecological Land Management Ontology (ELMO), to support standardized, machine-readable representation. As an EcoWeaver annotator, I will present my experience applying CAMO to three studies. These studies show challenges in selecting level of detail for nodes, representing context-sensitive findings, and determining what the evidence supports without overstating. They also highlight epistemic challenges, such as distinguishing species presence from an increase in abundance and determining how each finding should be represented. Preliminary annotation shows that these decisions affect what remains visible in the graph. For example, a treatment may have little detectable effect on species richness or diversity while still altering species composition. Representing individual species as separate nodes preserves information but increases annotation complexity and labour. Ecological knowledge-graph construction is not fact extraction; it requires interpretation, explicit standards, and documentation of context and uncertainty.

9 EU AI Act and implications for Global South Conservation governance: The regulatory classification of AI-based ecological synthesis tools

The use of AI tools for knowledge synthesis, such as the large language model and knowledge graph, has been proposed as a solution to ecological knowledge fragmentation, which combines field data, models, and socio-economic information to enable causal inferences to be made about social-ecological systems. The EU AI Act was not established to come into effect within the regulatory landscape as these tools transition from research prototypes to operational use, with consent considerations for environmental impact assessment, conservation funding initiatives, and biodiversity policy. The Act's risk-tiering was created with an emphasis on domains like, employment or law enforcement, where the nature of ecological synthesis tools could be deemed to have a direct impact on funding decisions, employment or crime, respectively. Additionally, there is no explicit clarity on whether ecological synthesis tools can be deemed to impact or not impact funding decisions, employment or crime, etc., and should be included in the risk tiering system or excluded. This paper explores whether AI systems based on ecological knowledge synthesis would fall within the definition of "high risk" under the EU AI Act, given the nature of their epistemic profile and the downstream socio-ecological implications. It then asks about the implications of this classification for the application of EU-developed or EU-classified tools in jurisdictions around the globe, including those in the Global South, whose data governance systems, conservation values, and capacities differ. If not well structured, the EU AI Act has the potential to become a European blueprint on ecological AI with Europe-centric perspectives exported to ecosystems and communities that were not consulted during the development process. The paper suggests a framework for assessing the risks posed by these downstream impacts and identifies specific questions that could be explored by various partnership stakeholders.

10 Integrating Heterogeneous Ecological Knowledge to Improve Automated Moth Species Classification

Over the past three decades, flying insect biomass has declined by more than 75% in protected areas, highlighting the urgency of monitoring insect populations. Moths represent approximately 9% of all known species on Earth and are among the most affected insect groups, despite their important roles as pollinators and food sources. Automated nocturnal light recorders are increasingly used to monitor moth populations, but extracting meaningful species information from the resulting images remains challenging due to the fine-grained differences between many species. Addressing this challenge is essential to provide entomologists and conservation researchers with more reliable tools for large-scale moth monitoring. However, automated moth classification pipelines have not kept pace with recent advances in large-scale training data and vision architectures. To address this gap, we build on the pipeline of Korsch et al. (2022), which provides a solid baseline for automated species recognition, and extend it in two ways. First, we treat web-sourced image collections, museum archives, citizen science platforms, and light trap cameras as heterogeneous external knowledge sources and integrate them by removing cross-dataset duplicates and resolving label ambiguity. Second, we investigate the impact of vision-language pretraining, exemplified by BioCLIP, an image-text model pretrained on ecological data, on fine-grained species classification. By aligning visual and taxonomic textual knowledge, this approach represents a step toward AI-based ecological synthesis, and we compare it against purely image-based convolutional and transformer architectures pretrained on generic versus ecological data. Our initial results show that BioCLIP achieves significant gains over the reproduced baseline from Korsch et al. (2022), though evaluation of the benefits of combining datasets is still ongoing.

11 Re-imagining rivers' social-ecological future by centering nonhuman perspectives: A multispecies role-playing game

Riverscapes are socio-ecological systems where human and nonhuman entities co-create and co-transform. However, anthropocentric water governance frameworks have long treated water as a “resource” or “service,” leaving nonhuman entities with no agency and dependent on human decisions for protection. Innovative participatory approaches like role-playing games (RPGs) offer experimental spaces for alternative perspectives and extending justice to ignored nonhumans. We developed the RPG “The Nonhuman Rhine Assembly,” composed exclusively of 9 nonhuman roles living in the Rhine River basin, one of Europe's most regulated ecosystems. The 9 roles include the Mother River, floodplains, aquatic and riparian vegetation, migratory fish, birds, amphibians, endangered native species, and invasive species, all framed as agentive subjects endowed with intrinsic, relational values. Between 2025 and 2026, we conducted 7 workshops in France, Poland, and Germany that brought together +110 participants from diverse backgrounds, who co-designed eco-centric restoration strategies for the Rhine's future rooted in social-ecological justice. Through analysis of the proposed scenarios, the teamwork dynamics, and participants' feedback, we show that RPGs can help operationalize nonhuman agency and stimulate ethical and political encounters, even when nonhuman representations remain mediated by human knowledge systems. The RPG also facilitated context-specific river commoning practices, e.g., recognition of the river as a living being and inclusive scenario-building grounded in interspecies negotiation, echoing the global Rights of Nature movements. Younger participants particularly demonstrated social learning outcomes that may promote environmental stewardship. We argue that integrating RPGs into real-world water governance, e.g., as pedagogical experiments for stakeholders/through the institutionalization of interspecies councils, can help move multispecies justice beyond a “utopian desire”.

12 Traditional Ecological Knowledge System and Resource Degradation Nexus: Insights from Ngoketunjia, North West Region, Cameroon

TRADITIONAL ECOLOGICAL KNOWLEDGE SYSTEM AND RESOURCE DEGRADATION NEXUS: INSIGHTS FROM NGOKETUNJIA, NORTH WEST REGION, CAMEROON

NUGAH Alvina NYALUM (PhD Student) Department of Geography and Planning, the University of Bamenda.
The traditional ecological Knowledge (TEK) of the Mbororo pastoralists is the cornerstone guiding resource management, adapting to environmental aspects especially rainfall variability in Ngoketunjia Division. This study seeks to explore the nexus between traditional ecological knowledge system and resource degradation, agro-pastoral conflicts. A mixed-methods approach was adopted to collect primary and secondary data. The observable impacts of the use of traditional ecological knowledge to manage resources have been pasture and landscape degradation, water pollution and agro-pastoral conflicts. The agro-pastoral code has not been able to remedy the situation given that. The Mbororo pastoralists in Ngoketunjia Division maintain an inextricable link to culture and traditional ecological knowledge system that shape how they manage resources. Pastoralists continue to engage in un-sustainable practices such as bush burning (90%) to regenerate pasture for cattle in the dry season, the cattle carrying capacity of the area is often exceeded especially in the dry season. State institutions have organized a series of workshops geared towards resources sustainability. Despite all, pastoralists (95%) do not turn up in meetings aimed at integrating traditional ecological knowledge systems into adaptation plans. These have led to resource degradation in Ngoketunjia Division. We argue that understanding traditional ecological knowledge systems in tandem with resource degradation is crucial for the sustainable management of resources in a mixed-farming zone as Ngoketunjia Division.Keywords: Traditional ecological knowledge systems, resource degradation, Mbororo pastoralists, Ngoketunjia Division, North West Region, Cameroon.

ALVINA NYALUM NUGAH (REVISED).docx

ALVINA NYALUM NUGAH (REVISED).pdf

Tuesday 29 September

Unconference planning: Tuesday unconference planning

The morning planning sessions will determine what happens in the open working group rooms.

Keynotes: Scales of context dependence in ecology — challenges for identifying which context actually matters

Keynote talks

13 Keynote: Scales of context dependence in ecology — challenges for identifying which context actually matters

“Context-dependence” is a ubiquitous challenge in ecological and social-ecological synthesis. Context-dependence occurs when a relationship varies depending on the conditions under which that relationship is observed. Unless we know what caused that variation (e.g. why is X correlated with Y in some circumstances but not others), we cannot effectively integrate the diverse evidence that exists for many relationships of interest. Much recent effort has gone into identifying and classifying different types of context dependence, which lay important groundwork for gaining a more mechanistic understanding of social-ecological systems. However, a less recognised point is the various scales at which context-dependence can arise, and the challenges associated with identifying which scale matters when looking for mechanistic explanations for context-dependence. Using plant-consumer interactions as an example, I will explore the scope of this issue. I will show how context-dependence can arise at the level of individual plants (e.g. intraspecific variation), species (e.g. different species respond in different ways to the same stimulus), local conditions (e.g. different soil resources in adjacent fields cause divergent trends) or regional trends (e.g. precipitation regimes and evolutionary history determine the possible range of ecological responses). This can cause problems when trying to integrate different studies into a knowledge graph, because without explicit measurement of relevant contexts at all these scales, it is difficult to know why the results of two studies may not be consistent, and thus what we can learn from their contrast. Strategies to meet this challenge will be a key part of knowledge synthesis.

Speaker: Dr Joshua Brian (King's College London)

10:15 Coffee break

Paper talks: What works where

Paper talks given as part of the conference proceedings, including four sessions: Place-based knowledge systems, What works where, Representing causality, and Machines in the loop

14 The TripleA Principle: Actionability, Applicability, and Auditability for Actionable Knowledge in Post-FAIR Infrastructures

The FAIR Principles have made biodiversity data findable, accessible, and interoperable, but not operationally usable. A critical layer is missing between semantic representation and context-sensitive action, one that distinguishes the structural capacity of a knowledge representation to support operations (actionability) from the epistemic reliability of using it in a specific context (applicability) based on an empirical evidence track of successful and failed applications (auditability). Conflating these has contributed to well-documented failures in conservation practice.

We propose the TripleA Principle (Actionability, Applicability, and Auditability) as a post-FAIR guiding framework, and argue that the concept of Action Units, implemented within the Semantic Units Framework, provides the representational foundation next-generation biodiversity knowledge infrastructures require.
We introduce Action Units as formal knowledge objects that extend plan specifications within the Semantic Units Framework by making applicability conditions explicit as first-class typed components. Three types correspond to three operation classes: epistemic action units ask ‘What is the case here, and does this knowledge apply?’; transformational action units ask ‘How can available representations be processed into usable form?’; and intervention action units ask ‘What can be done in this situation to achieve a desired outcome?’ Conditional action units, i.e., executable IF–THEN rules linking SPARQL context queries to workflow triggers, enable knowledge graphs to serve as AI-ready decision-support systems.
Applied to biodiversity science, the framework addresses conservation failures and enables practitioners to identify where actionable knowledge is missing.

This is a conceptual proposal; implementation is deferred to subsequent work.

15 Challenges and options for synthesis and extrapolation for ecosystem restoration outcomes at landscape level

To conserve biodiversity, ecosystem functions and nature’s contributions to people (NCPs) in agricultural landscapes, ecosystem restoration measures are implemented mainly targeting processes at different spatial and temporal levels. Analytically, the focus is often to assess past impacts of restoration measures for different taxa and different NCPs and underlying ecosystem functions. These mainly include the local and/or landscape level and the impact of governance instruments. For example, spatial cross- and multi-level analyses of social-ecological systems are still rare. These challenges raise the questions: (i) how can we synthesize these results across the multitude of dimensions including scales, and (ii) how can historic and current as well as local- and landscape-level results be extrapolated in space and time.
For (i), we will give an overview on the investigated dimensions, in particular the synthesis of indicators on different temporal and spatial levels, as well as our portfolio of synthesis approaches (e.g., meta-analyses, multi-scale synthesis) as a baseline. Hereon, we will elaborate on challenges and solutions for within-project synthesis as well as the integration with existing syntheses in the domain. A major challenge to be addressed is the spatial multi-level suitability of indicators. (ii) For spatial extrapolation beyond the landscape level, we will present ecological restoration potential archetypes for Europe to explore the potential for transferability of case study results. For temporal exploration, we will present participatory back- and forecasting scenarios highlighting pathways for addressing the restoration potential of agricultural landscapes in the future. The scenario development process will reveal suitable approaches to integrate social, economic and environmental drivers and to highlight the role of societal demands and governance.

16 TIB Knowledge Loom: Publishing Analysis-ready Scientific Knowledge to Accelerate Synthesis Research

Ecological research increasingly relies on integrating heterogeneous knowledge sources that are fragmented across disciplines, conceptual frameworks, and data infrastructures. This fragmentation is compounded by the fact that, despite significant advances in digital technologies, scientific results are still primarily communicated in text-based formats that machines cannot interpret without human assistance. This prevents us from taking full advantage of computational tools for knowledge organization and reuse.

These limitations are especially acute in synthesis research. Systematic reviews and meta-analyses require researchers to manually extract and organize knowledge from numerous publications, a time-consuming, error-prone process that can consume months or years of full-time work. While AI-based approaches offer new possibilities for knowledge integration, their effectiveness ultimately depends on access to structured, high-quality inputs—a foundational challenge that remains largely unaddressed.

The TIB Knowledge Loom is a novel Open Science digital library designed to address this challenge by supporting the production of machine-reusable scientific knowledge. In the Loom, scientific knowledge is curated at the level of statements that are linked to the underlying data, code, and analyses—capturing details typically absent from published articles, such as model parameters, test statistics, and uncertainty estimates. Using an ecological case study, we illustrate Loom record features and their potential to support synthesis research by strengthening reproducibility, improving data granularity, and enabling automated harvesting by downstream knowledge bases and AI systems. As the Loom is still in its early stages, we see this presentation as a valuable opportunity to gather perspectives from the synthesis research community on how the Loom could best support interdisciplinary knowledge integration and identify priorities for future development.

TIB Knowledge Loom_Snyder.pptx

17 Context-dependent coexistence of social parasites: fungus-growing ant communities as a case study

Parasite virulence (i.e. the harm caused by parasites to host survival and fitness) shapes host-parasite dynamics in both pathogenic and social parasites. While high virulence can paradoxically limit parasite persistence, outcomes are context dependent, sometimes even turning parasites into net mutualists. The fungus-growing ants (Myrmicinae: Attini) are part of a well-studied symbiotic relationship that is vulnerable to exploitation by social parasites. Although experimental work has explored the possible mechanisms behind the network of species interactions, a theoretical understanding of coexistence patterns among multiple social parasites competing for the same host has been missing. In this study, we use a patch occupancy model inspired by fungus-growing attine ants to investigate the dynamics of a community in which multiple socially parasitic species exploit a shared host. We use the model to identify the range of conditions that lead to coexistence of both generalist and specialist parasites with their host, with relative frequencies that correspond to field data. We show that the combined effects of context-dependency, differences in relative virulence, and the degree of specialization are key to explaining the coexistence and parasitism patterns of the community. We also identify the conditions under which the less virulent social parasite emerges as a net mutualist versus a net parasite to the hosts, which allow us to characterize the context- dependent nature of coexistence among competing natural enemies and their hosts.

18 Theoretical Syntheses based on model use: How to go from "snapshots" of current theories to "films" of how theories develop.

A pragmatic scientific theory is the conceptual basis that a group of scientists uses to conduct their scientific activities. This notion implies a definition of scientific theory based on what, from all available knowledge, scientists use to support their research. In other words, the theory is what scientists are using.

The recently published Pragmatic Approach to Theories is a method to produce theoretical syntheses by tracking the knowledge exchange in citations among publications. The method involves collecting scientific papers produced by a defined group of scientists and investigating which publications are consistently referred to. Afterwards, there is a content analysis to understand how these publications are being cited. This content analysis results in textual syntheses describing the most relevant models used by the group of scientists to conduct research activities. The theoretical synthesis produced assumes that such a pragmatic theory is an ever-changing entity, and the method takes snapshots of it.

The use of AI could advance several aspects of this process, from screening of papers produced by the group, through helping track scientific activities, to content analysis and model syntheses. However, there are several challenges relating to automating this process into a pipeline. The main issue is still the publishers' rights that prevent an automated content analysis. However, there is room for discussion about whether a synthesis produced by LLMs is equivalent to synthesis produced by humans (the tested way the method was implemented).

Being able to automate parts of this process through AI would allow for a fast implementation of each "snapshot" of the theory, leading to a more continuous analysis of theory development. In other words, it would allow us to create an "animation" of how a theory has been developing through time, among different geographical regions and even demographic profiles.

11:45 Lunch Break

Social Event: Visit to the Botanical Garden

19 Botanical Garden tour

Together we will take a tour of the Botanischer Garten Jena - the second-oldest botanical garden in Germany and one of the oldest in the world. Established in 1586, the garden contains more than 12,000 species of plants, including 900 species of trees in its outdoor arboretum.

Participation is limited to 25 people.

Social Event: Visit to the Haeckel House

Working groups: Bridging the Evidence Gap: A Collaborative Question-ranking Workshop for EcoWeaver

Working group sessions designed to be interactive and develop ideas, papers and projects.

20 Bridging the Evidence Gap: A Collaborative Question-Ranking Workshop for EcoWeaver

Restoration ecology often suffers from a disconnect between the availability of scientific evidence and the specific, site-level questions faced by practitioners. While high-level syntheses provide general guidance, practitioners frequently require precise answers to localized problems—for example, "What is the most effective method for managing European Buckthorn in a Carolinian grassland?" To address this, we propose a collaborative workshop designed to identify and prioritize the high-value questions that should drive the development of the EcoWeaver knowledge graph.

Drawing on Bill Sutherland’s environmental scan approach, this session will assemble a diverse group of restoration researchers and practitioners to iteratively identify, refine, and rank evidence needs specific to their ecosystem types. By systematically surfacing the "unanswered" questions of the field, the workshop aims to create a prioritized research and synthesis agenda that informs the architecture of EcoWeaver. Rather than relying on top-down priorities, this process ensures that the resulting knowledge graph is grounded in the actual needs of those implementing restoration on the ground.

The output of the session will be a ranked list of evidence gaps that will be directly integrated into the EcoWeaver framework, transforming the graph from a theoretical exercise into a tool for actionable local intelligence. Participants will leave with a shared understanding of the current evidence landscape and a direct role in shaping a digital tool designed to reduce the research-practice gap in restoration ecology.

Working groups: Creating Narratives for Ecoweaver/TReK

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Uncertainty and challenges working group

Working group sessions designed to be interactive and develop ideas, papers and projects.

21 Reference Tracking of “Uncertainty” in Ecological Research

Following the breakout session in November 2025, this proposal asks one question: when "Uncertainty" is grounded in ecologists' first-hand research cases rather than imposed by theory, what does it actually refer to and can a formal ontology recover those referents across different cases, or does it break down?
The group runs in two sessions. In the first session (ideally situated at the start of the workshop), participants will describe their research process, specifically where uncertainty arose. Their verbatim will be the basis for what the word “Uncertainty” refers to. After the recording of their accounts, we will demonstrate the realist referent-analysis pipeline (Progressive Constraint Referent Analysis) on one use case. The authors will then process each account outside the breakout session setting with the demonstrated pipeline extracting only positive statements about what was the case and assembles them into a theory-neutral graph model of the situation, structured with the Basic Formal Ontology (BFO 2020), and where useful the Environment Ontology (ENVO). We then project competing definitions of uncertainty onto that model to see which referents each one picks out, and whether the same referent types recur across cases. In the second session (ideally situated at the end of the workshop), the authors present the resulting graphs and projections. Participants then test the result against their own experience: does the model capture the uncertainty they described, and where does it miss?
We expect this break-out sessions to be interesting for a variety of researchers. For example, it offers an ontology-based qualitative method for analyzing subject verbatim for ecologists and the application of metaphysics and argument theory for the philosophers. For computer scientists, it generates empirical data for evaluating whether formal ontology-based methods can represent uncertainty across interdisciplinary cases.

15:55 Coffee Break

Unconference planning: Planning Tuesday Late Afternoon

The morning planning sessions will determine what happens in the open working group rooms.

Group game: The Nonhuman Rhine Assembly

Nine nonhuman roles in the Rhine basin; participants co-design eco-centric restoration strategies in character.

Working groups: Open working group C

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Open working group D

Working group sessions designed to be interactive and develop ideas, papers and projects.

Wednesday 30 September

Unconference planning: Wednesday unconference planning

The morning planning sessions will determine what happens in the open working group rooms.

Keynotes: Dr. Pier Luigi Buttigieg - A new niche for community ontologies in contemporary digital ecosystems: insights from the triumphs and trials of the Environment Ontology (ENVO)

Keynote talks

10:15 Coffee break

Paper talks: Session III: Representing causality

Paper talks given as part of the conference proceedings, including four sessions: Place-based knowledge systems, What works where, Representing causality, and Machines in the loop

22 Navigating research diversity in ecological synthesis: the role of openness and machine actionability

Multiple research fields from a range of disciplines contribute - each in their own ways - to the study of ecological systems and their internal and external interactions. Attempts at synthesizing the various contributions within and across fields into a coherent body of knowledge thus have to take into account the diversity of ways in which these fields perform, organize and share research. In this presentation, we will explore how openness and machine actionability can assist in the process. Specifically, we will consider four contexts in which openness and machine actionability can and do affect evidence synthesis in ecology: (i) the Diamond Open Access literature, (ii) nanopublications, (iii) research software, (iv) Wikimedia projects. Within each of these contexts, we will consider examples of ecological research questions and highlight challenges and opportunities that these contexts present in terms of evidence synthesis within and across fields.

23 FAIR, CLEAR, and Causal: A Semantic Framework for Ecological Knowledge Synthesis

Ecological synthesis produces knowledge that is human-interpretable but rarely machine-actionable. Attempts to address this through RDF/OWL-based semantic representations introduce a new problem: structures become opaque even to domain experts. We present the Semantic Units Framework, a modular approach to structuring ecological data and knowledge as logic-aware, independently addressable units aligned with the FAIR (Findable, Accessible, Interoperable, Reusable) and CLEAR (Cognitively interoperable, semantically Linked, contextually Explorable, easily Accessible, human-Readable and -interpretable) Principles.
The central contribution is embedding causal knowledge within this architecture. Universal causal statement units encode causal hypotheses as addressable graph entities forming a navigable semantic grid, like a spatial map, to which evidence, publications, and contextual qualifiers can be anchored. Individual observations can be linked to causal hypotheses they support or contradict, enabling queries such as “Which data support this causal claim?” This supports semi-automated causal reasoning, confounder detection, and AI-ready evidence-gap maps.
The framework decomposes knowledge graphs into (a) statement units as single propositions with a persistent identifier, provenance, and method links and (b) compound units grouping related statement units into reusable knowledge objects, both serializable as nanopublication FAIR Digital Objects. Machine-interpretable structures are decoupled from human-facing representations via dynamic labels and graphs, keeping OWL/RDF accessible without specialist knowledge. Currently conceptual with partial prototypes, we invite discussion toward its community-driven realisation.

24 EcoGraphRAG: Causal Evidence Graphs for Grounded Ecological Synthesis

Ecological synthesis is increasingly difficult not only due to the growing volume of ecological literature, but also because evidence is deeply contextual. A claim about range shifts or niche change depends on the taxon, habitat, method, driver, or response under consideration. Standard retrieval-augmented generation can retrieve relevant text but often leaves this structure implicit. We present EcoGraphRAG, a pilot workflow that represents ecological claims as causal evidence records and queries them through a provenance-preserving graph. We applied EcoGraphRAG to a 58-paper corpus of ecological studies on niche breadth, range shifts, specialization, and climate-related responses. Full-text sections were parsed, and we extracted evidence records with a task-specific causal evidence schema and validated records before graph construction. The corpus contained 296 validated records and a graph with 1,883 nodes and 3,126 edges linking papers, claims, passages, taxa, drivers, responses, methods, and concept mappings. We compared graph retrieval with a semantic retrieval baseline across synthesis questions using the same evidence corpus. Retrieved evidence was evaluated for relevance, passage support, ecological context, and concept mapping. Both approaches returned well-supported evidence, but the graph representation made driver-response structure, provenance, and concept coverage easier to inspect. A provenance-matching check recovered 289 of 296 extracted source texts from parsed papers, while flagging a small number of page-attribution and PDF parsing issues for manual review. Our work suggests that, rather than automating synthesis, EcoGraphRAG makes the evidence behind synthesis easier to inspect. By keeping claims connected to ecological context and source text, causal evidence graphs can support more transparent ecological synthesis when provenance, taxa, methods, and drivers matter as much as textual similarity.

25 Evaluating LLMs for Causal-Evidence Auditing in Ecology Literature: A Case Study on Biotic Resistance to Plant Invasions

Often scientific papers use conceptual causal diagrams to represent complex relationships. In ecology papers, these causal graphs can be used to depict species interactions, environmental conditions, ecological processes, etc. However, determining whether each of the causal relationships in such diagrams is strongly supported by the cited literature is difficult because the evidence is distributed across many reference papers and may vary in strength, direction and certainty levels. This project develops an LLM-assisted workflow for auditing such causal claims in ecology literature using the paper “Dynamics of biotic resistance to plant invasions” as a case study. The selected paper contains an explicit conceptual causal network describing hypothetical temporal dynamics of biotic resistance, including interactions among native communities, alien invaders, abiotic conditions, invasiveness, invasibility and biotic resistance or facilitation.

The workflow extracts text from the chosen paper and its cited references, processes the reference papers into structured sections, divides the extracted content into paragraph and sentence-level chunks, applies a custom causal-ecology term mapping and uses large language models to generate summaries of the referenced papers. These summaries are then used to support an author-feedback survey, where authors of the referenced papers are asked to evaluate whether the generated summaries accurately represent their work and whether the interpreted causal evidence is appropriate. The goal is not to directly challenge the original ecology paper or the referenced literature, but rather to evaluate whether LLMs can help trace and interpret causal evidence from cited literature. This study contributes a semi-automated workflow for causal-evidence auditing and highlights the need for expert validation when using LLMs for scientific literature interpretation.

26 Restoration ecology as a poietic practice

Thinking about causality has almost always conceptualised humans as knowers distant from the things known. This has informed our causal methodologies, including the experimental methodologies which are still considered our best methods. Even where separating ourselves from the target system is difficult, methodologies advise as much separation as possible. However, this conception is no good for practices like restoration ecology, where restoration practitioners exist in constant dynamic contact with their object, which is simultaneously their object of study and of intervention. We trace the history of theorising causal methods that positions us first as observers and then as interveners, and argue that such practices show that we need a better-suited concept of causality: poiesis. From the poietic point of view, humans are not merely observers or interveners; we are makers, creators of whole causal systems. Sometimes we create causal systems while ourselves living within them. We argue that this requires a shift in understanding our causal methodologies. This shift to causality as poiesis is needed in all the disciplines struggling with complexity, uncertainty, and other pragmatic constraints, the wild sciences. These include at least health complexity, sustainability science, geology, climate science, and ecology. Understanding our causal methodologies for ecology as observational, experimental and poetic gives us a much more complete and integrated view of those methodologies.

11:50 Lunch break

Head to the Mensa for lunch!

Working groups: EcoWeaver Workbench — hands-on schema prototyping

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Interweaving Western and Indigenous water knowledges on rivers

Working group sessions designed to be interactive and develop ideas, papers and projects.

27 Interweaving Western and Indigenous Water Knowledges on Rivers

The Blue Humanities have recently emerged as a subfield of Ecocriticism replacing land-centered knowledge systems and semantic frameworks through water-centered approaches of knowing and being in the world. Various countries are now acknowledging rivers as persons with inalienable rights. Such a process is often set in motion by local indigenous groups that use mythologies, situated knowledges, and activism to take freshwater bodies out of colonial contexts in which they have been straight-jacketed, dammed, polluted, and violated in various other ways. Material and symbolic Indigenous conceptualizations of water, for instance by Michi Saagiig Nishinaabeg scholar Leanne Betasamosake Simpson’s recent book theory of water: Nishnaabee Maps to the Times Ahead, create a water consciousness that values rivers in their own rights. Western new nature writers as, for example Robert MacFarlane in his book Is a River Alive? have made a reencounter and reevaluation of rivers possible.
The breakout group would start out by discussing a presentation of the key elements of the above-mentioned indigenous and changed ecological Western understandings of river: How do they shift epistemological, ontological, and representational ways of thinking and being with water? How do they allow us to re-encounter and reassess rivers? In a second step, we would sound out ways to interweave Western and indigenous knowledges as well as data and cultural sciences to arrive at alternative ways of assessing rivers and human-water relationships.

Working groups: Open working group E

Working group sessions designed to be interactive and develop ideas, papers and projects.

13:50 Short break

Working groups: Indigenous and local knowledge-weaving: territorial perspectives

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Open working group G

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Open working group G2

Working group sessions designed to be interactive and develop ideas, papers and projects.

15:15 Coffee break

Working groups: How to bring Ontologies to Ecologists (Josh, Ella)

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Indigenous data sovereignty in computational ecological knowledge integration

Working group sessions designed to be interactive and develop ideas, papers and projects.

28 Indigenous data sovereignty in computational ecological knowledge integration

Evidence synthesis in ecology increasingly integrates diverse knowledge sources, yet often overlooks Indigenous data sovereignty and community governance rights. This session presents three interconnected elements.
First, we present recent findings from a Nature Communications paper applying the CARE principles (Collective Benefit, Authority to Control, Responsibility, Ethics) to evidence synthesis workflows. We outline how CARE commitments can be operationalized through computational and methodological choices in systematic reviews and meta-analyses.
Second, we examine Can-Peat as a live case study. Can-Peat demonstrates practical implementation of Indigenous data sovereignty tools — specifically the Local Contexts Labels and Hub — for Canadian peatland research. We share lessons from supporting researchers in applying protocol and permission labels to existing datasets, documenting both successes and ongoing challenges.
Third, we facilitate group discussion on integrating Local Contexts into EcoWeaver, eliciting reflections and ideas on technical and ethical issues. We hope to advance some agenda points on how semantic knowledge structures and AI-based synthesis systems can, especially in the EcoWeaver context, accommodate Indigenous protocols for data access, use, and governance.
The session brings together ecology, Indigenous data governance, philosophy and computer science to ask: What does it look like to build ecological knowledge systems that genuinely center Indigenous authority? We welcome researchers, technologists, and community collaborators interested in responsible, ethical knowledge integration.

Working groups: Precovery (Daniel)

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: EcoWeaver planning session (OPTIONAL)

Working group sessions designed to be interactive and develop ideas, papers and projects.

Conveners: Birgitta König-Ries (Heinz Nixdorf Chair for Distributed Information Systems), Robert Frühstückl, Tim Alamenciak (Carleton University & University of Waterloo), Tina Heger

19:00 Dinner reception

Finger foods will be provided and posters on display from all of the working groups held so far.

Thursday 1 October

Unconference planning: Final day unconference planning

The morning planning sessions will determine what happens in the open working group rooms.

Keynotes: Dr. Christina B. Claß, Ernst-Abbe-Hochschule Jena - AI and Ethics: challenges for society and science

Keynote talks

10:15 Coffee break

Paper talks: Session IV: Machines in the Loop

Paper talks given as part of the conference proceedings, including four sessions: Place-based knowledge systems, What works where, Representing causality, and Machines in the loop

29 A global synthesis of herbivorous crop pests and their natural enemies using large language models

Large language models (LLMs) have recently emerged as a novel tool for knowledge extraction and evidence synthesis in a wide range of scientific disciplines, including ecology and conservation science. Whilst many taxa have been classified based on traits or the provision of ecosystem functions, taxonomic and geographic coverage is sparse, and knowledge of interactions between species, particularly invertebrates, is limited. In this study, we use a pre-trained LLM to analyse abstracts of a corpus of published material on biological pest control to obtain a global database of pest herbivores and their natural enemies (DAPHNE). Our methodology resulted in the extraction of 179,001 species interactions from 112,830 publications, containing 17,219 unique animal taxa resolved at the species level. Interactions include herbivory (granivory, frugivory and gall-formation), predation, parasitism and hyperparasitism, and comprise a range of additional information, such as species' taxonomy, pest status, pest importance, natural enemy importance, provision of biocontrol, associated plants and industries, invasiveness and the vectoring of pathogens and diseases. Our study thoroughly evaluated the extracted information, and shows that most data columns were extracted with high recall and precision (>90%) as evaluated on a held-out test set. Comparison with other sources shows a large degree of overlap with DAPHNE, but also highlights that our dataset was able to capture important information that has been missed by other sources. DAPHNE is openly available and is hosted online as an interactive shiny app (https://ddx5w5-daan-scheepens.shinyapps.io/daphne-interaction-network/).

30 LLM-based generation of spatiotemporal knowledge graphs for ecological synthesis

In ecology, systematic reviews and meta-analyses are important methods to synthesize scientific evidence. They allow to qualitatively and quantitatively integrate large bodies of literature to investigate cross-study research questions. Ecological evidence is described by numerous interrelated and complementary entities, many of which are associated with spatial and temporal information. Those entities and their relationships can be represented as spatiotemporal knowledge graphs (STKGs) enabling the analysis of complex ecological interdependencies, e.g., based on the identification of patterns and connections across studies.
The construction of knowledge graphs is increasingly being supplemented by large language models (LLMs). However, the spatiotemporal and complex nature of some domains, such as ecology, has received limited attention so far. Automated STKG construction methods could support publication analyses and synthesis by enabling efficient extraction, representation and integration of ecological knowledge. Despite this potential, LLM-generated knowledge graphs can be unreliable and insufficiently evaluated. Therefore, close interdisciplinary collaboration with domain experts is essential to integrate their knowledge and develop new benchmarks for assessing the quality and reliability of automatically constructed STKGs.
We work with researchers in the “AgriRestore” project, that aims to investigate key indicators for more resilient ecosystems in the field of ecological restoration by evaluating measures for ecosystem and landscape restoration in agricultural landscapes. The project’s meta-analyses, ecological studies and the domain-specific knowledge of the collaborators provide a valuable framework for developing, evaluating and interpreting generated STKGs. With our presentation, we aim to share our current progress on how to use LLMs to derive STKGs from publications and effectively support literature synthesis methods in the field of ecology.

31 Knowledge-Guided Species Classification with Multimodal Large Language Models

Biodiversity monitoring provides the systematic measurements required to assess ecosystem health, predict environmental collapse, and inform global conservation efforts. A key component is the automatic classification of species from image data, where recent deep learning methods have achieved remarkable performance. However, distinguishing visually similar species based on fine-grained characteristics remains challenging when relying solely on visual information. We propose a training-free framework that integrates human knowledge in the form of textual species descriptions using multimodal large language models (MLLMs). We leverage human-curated species descriptions from Wikipedia as an external source of taxonomic knowledge, providing them alongside a query image to an MLLM, which relates visual evidence to textual descriptions to support the classification decision. To accommodate the limited context length of MLLMs, we first narrow the search space by selecting a small set of candidate classes using image features alone. Because the framework requires no additional training and is compatible with different MLLMs, it can be readily applied to new ecological domains while naturally benefiting from ongoing advances in foundation models. We evaluate the proposed knowledge integration approach on three ecological classification tasks: birds, moths, and flowers. In addition to predicting the species label, the MLLM provides a natural-language explanation that relates visual evidence to the retrieved species descriptions, making the classification process inherently interpretable. Our analysis shows that while incorporating textual knowledge often improves classification performance, current limitations of MLLMs, such as hallucinations and self-contradictions, remain evident.

32 The LIB Open Knowledge Space: A semantic architecture for FAIR, CLEAR, and AI-ready biodiversity knowledge

Natural history collections risk remaining isolated archives unless their data and knowledge are transformed into connected, semantically structured, and machine-actionable resources. The LIB Open Knowledge Space is a planned FAIR-, CARE-, and CLEAR-compliant knowledge infrastructure. It is central to the data and knowledge science strategy of the Leibniz Institute for the Analysis of Biodiversity Change (LIB) to integrate collection data, research outputs, and curated knowledge into a unified environment serving as LIB's living memory and an anchor of distributed semantic interoperability.
The architecture rests on two frameworks: (1) The Semantic Units Framework organises biodiversity data and knowledge into modular, identifiable units of meaning, each with a globally unique persistent identifier and explicit epistemic status, supporting assertional, existential, prototypical, and universal statements. (2) The Semantic Ladder defines a progressive formalisation pathway from natural language text snippets through Rosetta Statement anchors to Description Logics-based representations, enabling incremental FAIRness without full formalisation at data entry. Across all levels, vector embeddings support similarity-based search and AI-assisted exploration, enabling hybrid symbolic-vector access that we plan to complement with LIBTellMe, a RAG-based chatbot.
Together, these frameworks support semantic digital twins of specimens capturing not only associated data but also object and object-data journeys, going beyond the extended specimen concept.

We present the conceptual architecture and research programme, and invite discussion on alignment with the goals of the EcoWeaver initiative.

33 You ask iAnswer: How Knowledge Graphs and LLMs combine in an accessible and reliable Question-Answering tool for Ecological Research Data

Recent advances in AI promise new possibilities for knowledge integration, synthesis, and dissemination. However, novel AI systems must produce reliable answers if researchers and stakeholders want to make trustworthy claims, predictions, and decisions. Carefully curated Knowledge Graphs can be the key for this reliability, as they represent research data in a machine-readable, -interpretable, and -actionable fashion. In this talk, we present an LLM (Large Language Model)-based Question-Answering Interface that leverages facts from a Knowledge Graph to produce grounded answers in plain language. This both bridges the gap between users and Knowledge Graphs – as the technical query language SPARQL is notoriously difficult to use – and allows users to trace how the answer to their question was formulated. The application allows users to converse with PhenObs, a global network of botanical gardens that observes plant phenology – the timing of biological events, such as flowering, fruiting, and senescence. Users can ask questions, generate plots and visualizations, and download the data relevant to their question for further analysis. The chatbot’s flexibility makes research data available to users, no matter their prior experience, and thus supports other researchers in their synthesis efforts while giving the general public and enthusiasts a tangible entry point into the research. Additionally, as the system follows a modular design, the Question-Answering Interface can easily be configured to answer questions over different Knowledge Graphs. We invite attendees to try the provided demo and give their suggestions on further functionalities.

34 Talking to Ontologies: Making Scientific Knowledge Accessible through MCP

Ontologies are widely used across scientific domains, e.g. life sciences and materials science to organize complex knowledge and describe relationships between concepts, data, and processes. However, accessing this knowledge often requires familiarity with ontology engineering, specialized terminology, or query languages such as SPARQL. This can limit the use of ontologies by researchers and practitioners who could otherwise benefit from them.
This work presents how the Model Context Protocol (MCP) can support more natural ways of interacting with ontologies. Through an MCP-based conversational interface, users could ask questions in ordinary language—for example, about biological entities, material properties, or relationships between concepts—and receive answers grounded in the underlying ontologies. The system can also point users to relevant terms, relationships, and data sources, helping them understand where an answer comes from.
An important aspect of this approach is that it can support access to knowledge from several ontologies at once. These may be hosted within the same repository or distributed across different ontology repositories. This makes it possible to connect information from different domains, compare related concepts, and answer questions that cannot be addressed by a single ontology alone. For example, a user may combine information about a biological process with material properties relevant to biomedical applications.
By lowering the technical barrier to ontology use, this approach could help non-ontologists explore and reuse structured scientific knowledge more easily. It may also support collaboration across disciplines by making life science and materials science ontologies easier to discover, query, connect, and interpret.

12:05 Lunch break

Off to the Mensa!

Working groups: Best practices for LLM

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Reference tracking of "uncertainty" — Part 2: graphs, projections and participant testing

Working group sessions designed to be interactive and develop ideas, papers and projects.

Working groups: Semantic modeling of causal mosaic nodes

Working group sessions designed to be interactive and develop ideas, papers and projects.

35 Closing debrief and discussion