Phase 6: TheoryNet Projection¶

Overview¶

Phase 6 executes the TheoryNet Projection pass (Phase6Runner). Operating on the mined theory atoms and argument relations produced through Phase 4b and clustered in Phase 5, Phase 6 formalizes the graph into a structuralist theory network $T = \langle \mathcal{A}, \mathcal{R}, \rho, \alpha \rangle$.

It assigns plausibility functions ($\rho$) to nodes, weights ($\alpha$) to dialectical relations, and performs structural verification of empirical content via $Z_1$ path reachability.

Goals¶

Project raw argument components into formal theoretical atoms categorized by epistemic partition ($A$ vs. $B$).
Assign normalized plausibility values to all theory atoms.
Assign normalized weights to all dialectical relations.
Evaluate $Z_1$ empirical connectivity: verifying which theoretical hypotheses connect directly or transitively to empirical observations.
Commit formalized attributes back to the Neo4j graph store.

Steps¶

Projection from Artifact View:
Ingests Phase4ArtifactsView containing all cumulative TheoryAtom and TheoryRelation domain objects.
Invokes TheoryNetProjector.project(input) to generate a strongly-typed TheoryNet structure.
Plausibility Scoring ($\rho$):
For each atom in theory_net.atoms: $$\rho(a) = \begin{cases} \text{confidence}(a) & \text{if confidence is defined} \\ 1.0 & \text{otherwise} \end{cases}$$
Prepares atoms for batch upsert.
Relation Weighting ($\alpha$):
For each relation in theory_net.relations: $$\alpha(r) = \begin{cases} \text{confidence}(r) & \text{if confidence is defined} \\ 1.0 & \text{otherwise} \end{cases}$$
Preserves relation properties: scope ("local" vs. "global"), relation_type, and weight.
Graph Persistence:
Upserts updated TheoryAtom nodes in batches of 500 via graph_store.upsert_argument_components().
Upserts updated TheoryRelation edges in batches of 500 via graph_store.upsert_relations().
Empirical Content ($Z_1$) Verification:
Partitions atoms using SchemaConfig.component_partitions:
- Partition A: Theoretical Hypotheses (component_partitions[type] == "A")
- Partition B: Empirical Observations (component_partitions[type] == "B")
Constructs the theoretical adjacency graph $\mathcal{G}_A = (V_A, E_{A \times A})$.
Identates the direct empirical boundary $Z_1^{\text{direct}} \subseteq V_A$: theoretical atoms that share an edge with at least one $B$-atom.
Executes breadth-first search (BFS) starting from $Z_1^{\text{direct}}$ across $\mathcal{G}_A$ to identify all reachable $A$-atoms.
Computes empirical content metrics and logs ungrounded theoretical components.

Phase Data Flow¶

Input: Phase4ArtifactsView (cumulative theory atoms and relations).
Output: Phase 6 ArtifactCollection containing formal TheoryNet envelopes.
Graph updates:
ArgumentComponent / TheoryAtom nodes: plausibility: float
TheoryRelation edges: weight: float

Pluggability¶

Projector: Custom projector implementations can be injected via Phase6Runner(projector=MyProjector()).
Schema: Partition definitions and component types are governed by SchemaConfig.

Implementation¶

pipeline/phases/phase6_theorynet/__init__.py — Phase6Runner
pipeline/projection/theorynet_projector.py — TheoryNetProjector
pipeline/contracts/domain.py — TheoryNet, TheoryAtom, TheoryRelation
pipeline/config.py — Phase6Config