Skip to content

v0.3.0 — Pre-release: Movement-Preserving Semantic Phase Space

Pre-release
Pre-release

Choose a tag to compare

@m3data m3data released this 18 Dec 05:28
· 31 commits to main since this release

v0.3.0 — Pre-release: Movement-Preserving Semantic Phase Space

Status: Private research pre-release
Intended audience: collaborators, reviewers, future maintainers
Relation to publications: Companion implementation to forthcoming Springer chapter
and preprints on Entangled Cognition, Semantic Climate, and Trajectory Integrity.

This release marks the first internally coherent instantiation of the Semantic Climate
Phase Space
as a movement-first semantic analysis framework. The system treats meaning
not as a sequence of static states, but as trajectory through a constrained phase space,
with explicit attention to path-dependence, hysteresis, and temporal integrity.

This is not a public API release. Interfaces, naming, and internal structures remain
subject to change as theoretical and empirical work continues.


Conceptual Contributions (Primary)

The following elements constitute the core intellectual and architectural contributions
of this project:

  • Movement-preserving semantic framework

    • Meaning modeled as trajectory rather than pointwise state
    • Classification preserves how a configuration was reached, not only where it is
  • Trajectory Integrity

    • Coherence defined as the relationship between a path and its own history
    • Path-dependence, memory without lock-in, and gestalt structure treated as first-class
      analytic properties
  • Hysteresis-aware semantic dynamics

    • Explicit separation of entry and exit thresholds
    • Resistance to spurious state-flipping under noise or short-term perturbation
  • Phase-space architecture

    • Semantic dynamics represented via velocity, acceleration, curvature, and entropy shift
    • Soft membership across attractor basins rather than hard classification
  • Interpretive stance

    • Instrumentation designed to constrain interpretation without reifying metrics
    • Intended for use alongside first-person accounts and contextual analysis

These contributions are authored and integrated by Mathew Mytka (2025) and form the conceptual
basis for related academic publications.


Core Metrics (With Repairs and Reinterpretation)

This implementation incorporates semantic metrics originally proposed by
Daria Morgoulis (2025), with mathematical corrections, reinterpretation, and embedding
within a trajectory-based system:

  • Semantic Curvature (Δκ)

    • Frenet–Serret local curvature applied to semantic trajectories
  • Fractal Similarity (α)

    • Detrended Fluctuation Analysis (DFA) over semantic velocity
    • Corrected scaling behaviour and validity constraints
  • Entropy Shift (ΔH)

    • Shared clustering with Jensen–Shannon divergence
    • Interpreted as change in semantic dispersion, not static uncertainty

Metric repair, validation, and integration into the phase-space model were performed as
part of this project.

Architecture Overview

  • TrajectoryBuffer

    • Maintains temporal continuity for semantic velocity, acceleration, and curvature
  • Movement-preserving classification stack

    • TrajectoryStateVector
    • SoftStateInference
    • HysteresisConfig
    • Classification outputs retain temporal context and path structure
  • Attractor basin system

    • 10 canonical semantic configurations
    • Soft assignment via centroid proximity
    • Hysteresis-aware transitions

Web Application (Experimental)

  • Multi-provider LLM support
    • Together AI, Ollama, Anthropic, OpenAI
  • Real-time semantic coupling visualisation
  • Session export and replay
  • Optional integration with EarthianBioSense for biosignal coupling experiments

UI components are exploratory and not yet stabilised.

Test Coverage

  • 244 tests covering core metrics, trajectory logic, and classification behavior
  • Emphasis on regression detection for movement and hysteresis properties

License

Earthian Stewardship License (ESL-A)

This license reflects the project’s orientation toward responsible use, research
stewardship, and regenerative application. Licensing terms may evolve alongside
governance work.

Attribution

  • Original semantic metrics:
    Daria Morgoulis (2025), MIT License

  • Phase-space framing, trajectory integrity model, hysteresis architecture,
    movement-preserving classification, system integration, and interpretive framework:

    Mathew Mytka (2025)

Attribution reflects conceptual authorship, not only code provenance.