Scientists in the past collected data in notebooks. In the digital age, we need scientific data and models to be Findable, Accessible, Interoperable, and Reusable, helping individuals, businesses, and governments make better informed decisions.

A fully connected information landscape using open, safe, accurate, “Wikipedia-like” sharing and linking of models can enable data-intensive science for decision making on a scale yet unimagined.

We want to share the methods and technologies we have built to achieve this vision. Join us to reach it faster.

k.LAB Stac
ARtificial Intelligence for Environment &Sustainability

The Integrated Modelling Partnership, begun in 2017, brings together institutions contributing to designing and building a fully integrated information landscape for the science of the future.

The partnership develops and maintains the IM worldview, the k.IM language and the k.LAB software stack. It provides training in semantic modelling and supports partners and users in creating unprecedented model-data integration in projects such as ARIES.

Become a partner to participate in building the vision, knowledge, and tools to support a more efficient, integrated, and democratic scientific process.

Learn moreBecome a partner

Four building blocks for one new approach

1. Semantics

The language used to describe scientific observations must be flexible and shareable, without ambiguity. It must efficiently address all the “W’s of information – what, where, when, why, and how – without becoming too large or complex to learn and use.

We have developed a base to define scientific worldviews – conceptualizations within which integration is practical, creating languages and tools to make the semantics of scientific observations easier to specify, understand, and use. We build and maintain one worldview, the IM worldview, geared toward describing scientific phenomena and problems related to Earth and its natural and human inhabitants.

2. Open, linkable data

Making data and models FAIR is complex and requires understanding of – and agreement on – the nature of all scientific information, including, but not limited to, adopting URIs and open standards.

We provide an immediately actionable framework, with working infrastructure, to enable publishing of semantically annotated data as first-class research objects, so that they can be found online, read and understood by computers and humans alike.

An important part of our work is to provide a semantic skeleton for domain ontologies where existing vocabularies and resources can be reused and flourish, while ensuring fully consistent semantics throughout the most diverse information landscapes.

3. Open, linkable models

While today’s dialogue on interoperability focuses on data, it can help to see models and model components as other ways to make scientific observations. This enables a single, consistent discussion on how to semantically connect data to models and how to build and validate complex models from simple ones.

Powered by efficient and correct semantics, artificial intelligence can transparently match the right data and models to the chosen time, place, problem, and scale. This way, much of the complexity of building and running models can be handled by machines, with substantial advantages for science and decision making.

4. Software infrastructure

We believe in contributions that can start making positive change immediately. Our theoretical semantic integration work is complemented by an open-source software stack built and maintained by the Partnership. The software provides tools and interfaces for end users, modelers, and network administrators, aimed at simplifying the tasks of semantically describing, coding, and distributing data and models as much as possible.

We provide and maintain documentation, community resources for discussion, user support and bug reporting, and are creating tools for participatory, graphical model building that can be directly translated into templates for working models.

Technological partners

install4j

install4j

JProfiler

JProfiler

Atlassian

Atlassian