Skip to content

T-Episteme: Theory Graph Language Pipeline

Welcome to the documentation for grund.Episteme, a cutting-edge pipeline for constructing theory graphs from scientific literature.

License Python 3.13 Status


What is T-Episteme?

grund.Episteme is a research library designed to construct theory graphs—structured representations of theoretical frameworks—from academic texts in German and English. While designed for computational epistemologists, NLP researchers, and philosophers of science, the pipeline itself focuses on an agnostic and approach for knowledge graph construction and may serve different use cases as well.

Our pipeline abstracts unstructured scholarly documents into queryable semantic structures. Unlike traditional knowledge graphs that focus on factual relationships, theory graphs capture the conceptual and argumentative structures that underpin scientific theories. The pipeline transforms unstructured scholarly documents into a property graph, that can be stored in any graph database of your choosing. The episteme-pipeline produced a three-layer architecture:

  1. Layer 1 (Provenance Embedding): Original documents embedded and queryable in the graph.
  2. Layer 2 (Knowledge Graph): Abstraction from original sources as a knowledge graph.
  3. Layer 3 (Theory Framework): Higher-level theoretical constructs and argumentative structures.

Documentation Pillars

The documentation is organized into six foundational pillars designed to serve researchers, system architects, and software engineers:


System Architecture & Processing Flow

Episteme System Architecture Overview