MLKN.lab

Research Platform

A living laboratory for testing hypotheses about the evolution of science.

Scientific Instrument

A high-precision tool for graph-theoretic analysis of knowledge networks.

Cognitive Interface

A generative system to augment human reasoning through interactive knowledge maps.

For Researchers

Uncover hidden connections between disciplines, identify emerging fields, and test metascience hypotheses.

Example: How does interdisciplinarity impact citation networks?

For Educators

Design interdisciplinary curricula and adaptive learning paths using MLKN.lab’s knowledge maps.

Example: Bridging Computer Science and Cognitive Psychology.

For Policymakers

Inform science policy by visualizing gaps, silos, and opportunities in research.

Example: Identifying underfunded interdisciplinary areas.

For AI Developers

Enhance AI reasoning with structured knowledge networks from MLKN.hypergraph.

Example: Grounding LLMs in scientific knowledge.

Theoretical Foundations

Built on unified theories of cognition, network science, systems theory, and computational epistemology. Models knowledge as a computational system where structure can be analyzed and optimized.

Mathematical Formalization

Formalizes knowledge as a multilayer hypergraph \( \mathcal{H} = (V, \mathcal{S}, \mathcal{P}, \mathcal{T}, \omega, \tau) \), with 5 ontological layers (Core Domains → Concepts) and hyperedges for cross-disciplinary connections.

Architecture

Organized into 3 core layers: Knowledge Representation (polyhierarchy), Network Analysis (graph theory, TDA), Dynamic Simulation (random walks, GNNs).

Formal Definition

A multilayer hypergraph where:

• \( V \): Nodes (Core Domains → Concepts).
• \( \mathcal{S} \): Hyperedges (cross-disciplinary connections).
• \( \mathcal{P} \): 5 ontological layers.
• \( \mathcal{T} \): Hierarchical edges (adjacent layers).
• \( \omega \): Weights (e.g., citations, semantic similarity).
• \( \tau \): Temporal function (evolution over time).

Key Properties

The framework is:

• Multilayer: 5 ontological layers (Core Domains → Concepts).
• Polyhierarchical: Nodes can have multiple parents.
• Dynamic: Tracks evolution via \( \tau \) and \( \mathbf{T}_{\mathcal{H}} \).

Mathematical Tools

Supports:

• Graph theory (adjacency, Laplacian matrices).
• Topological Data Analysis (persistent homology).
• Optimal Transport (Wasserstein distance).

Data Sources

We use OpenAlex as our primary data source, supplemented by Scopus, MeSH, and IEEE Thesaurus for thematic and conceptual mappings.

Classification Logic

Disciplines and subdisciplines are classified using OECD Frascati Manual and UNESCO Fields of Science, ensuring alignment with global standards.

Network Construction

We map connections between disciplines using co-occurrence analysis and citation networks to reveal interdisciplinary links.

Validation

Our hierarchy is validated through network metrics (e.g., centrality, modularity) and expert review to ensure accuracy.

Core Discipline Domains

Layer 1 embraces 6 Core Domains: Formal Sciences, Natural Sciences, Health Sciences, Social Sciences, Humanities, Applied Sciences.

Academic Disciplines

Layer 2 comprises 25 Academic Disciplines as distinct clusters, organized into their respective Core Domains.

Academic Subdisciplines

Layer 3 represents 235 specialized subfields within each discipline (e.g., AI within Computer Science).

Core Thematic Domains

Layer 4 contains thematic clusters within subdisciplines (e.g., Machine Learning within AI).

Main Thematics

Layer 5 includes fundamental thematics and concepts (e.g., Supervised Learning within Machine Learning).

François Papin

Researcher and creator of MLKN.lab. Background in cognitive psychology and education, with a passion for interdisciplinary knowledge mapping.

Mission

To bridge the gaps between scientific disciplines through a unified, hierarchical framework that reveals hidden connections and fosters collaboration.

Vision

A world where knowledge is seamlessly interconnected, enabling researchers, educators, and policymakers to address complex challenges collaboratively.

Research Platform

MLKN-lab is a living laboratory for testing hypotheses about the nature of science. Just as AI is both a subject of study and a tool for studying other phenomena, MLKN-lab allows researchers to:

• Study the evolution of disciplines over time.
• Identify fragmentation and convergence in scientific fields.
• Explore the structural patterns of knowledge networks.
• Investigate the emergence of interdisciplinarity.

This dimension positions MLKN-lab as a testbed for metascience, where researchers can ask and answer fundamental questions about how knowledge is organized and evolves.

Scientific Instrument

As a scientific instrument, MLKN-lab provides a rigorous framework for analyzing the structure and dynamics of knowledge. Like a microscope or telescope, it reveals patterns that are invisible to the naked eye:

• Graph Topology: The geometric and topological properties of knowledge networks.
• Centrality: Identifying the most influential disciplines and concepts.
• Disciplinary Bridges: Mapping the connections between fields.
• Emergent Clusters: Detecting new and growing research areas.
• Conceptual Diffusion: Tracking how ideas spread across disciplines.
• Epistemic Isolation: Identifying silos and gaps in knowledge.
• Historical Shifts: Analyzing how disciplines evolve over time.

This dimension makes MLKN-lab a powerful tool for bibliometrics, network science, and systems analysis.

Cognitive Interface

Finally, MLKN-lab serves as a cognitive interface—a way for humans to interact with and make sense of knowledge systems that have grown too complex for traditional academic structures. It enables users to:

• Navigate: Explore large-scale knowledge networks intuitively.
• Discover: Uncover hidden connections and patterns.
• Visualize: See the structure of knowledge in interactive maps.
• Learn: Understand complex fields through layered, interconnected representations.

This dimension transforms MLKN-lab into a tool for education, decision-making, and innovation, much like how AI augments human cognition.